New Perspectives in Language, Discourse, and Translation Studies ...

Second Language Learning and TeachingSeries EditorMirosław PawlakFor further volumes:http://www.springer.com/series/10129

About the SeriesThe series brings together volumes dealing with different aspects of learning andteaching second and foreign languages. The titles included are both monographsand edited collections focusing on a variety of topics ranging from the processesunderlying second language acquisition, through various aspects of languagelearning in instructed and non-instructed settings, to different facets of the teachingprocess, including syllabus choice, materials design, classroom practices and evaluation.The publications reflect state-of-the-art developments in those areas, theyadopt a wide range of theoretical perspectives and follow diverse research paradigms.The intended audience are all those who are interested in naturalistic andclassroom second language acquisition, including researchers, methodologists,curriculum and materials designers, teachers and undergraduate and graduatestudents undertaking empirical investigations of how second languages are learntand taught..

Mirosław Pawlak lEditorsJakub BielakNew Perspectivesin Language, Discourseand Translation Studies

EditorsMirosław PawlakAdam Mickiewicz UniversityFaculty of Pedagogy and Fine ArtsDepartment of English StudiesNowy Świat 28-3062-800 KaliszPolandpawlakmi@amu.edu.plJakub BielakAdam Mickiewicz UniversityFaculty of Pedagogy and Fine ArtsDepartment of English StudiesNowy Świat 28-3062-800 KaliszPolandkubabogu@amu.edu.plISBN 978-3-642-20082-3 e-ISBN 978-3-642-20083-0DOI 10.1007/978-3-642-20083-0Springer Heidelberg Dordrecht London New YorkLibrary of Congress Control Number: 2011930818# Springer-Verlag Berlin Heidelberg 2011This work is subject to copyright. All rights are reserved, whether the whole or part of the material isconcerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting,reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publicationor parts thereof is permitted only under the provisions of the German Copyright Law of September 9,1965, in its current version, and permission for use must always be obtained from Springer. Violationsare liable to prosecution under the German Copyright Law.The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply,even in the absence of a specific statement, that such names are exempt from the relevant protectivelaws and regulations and therefore free for general use.Cover design: SPi Publisher ServicesPrinted on acid-free paperSpringer is part of Springer Science+Business Media (www.springer.com)

AcknowledgementThe editors would like to express their gratitude to Professor Piotr Stalmaszczyk(University of Łódź, Poland), who has kindly agreed to review the papers includedin the present volume. His invaluable comments and suggestions have withoutdoubt greatly enhanced the quality of this work.v

ContentsPart IPhonetics and Phonology1 Aspiration in Polish: A Sound Change in Progress? ................... 3Ewa Waniek-Klimczak2 Noise as a Phonological Element: On the Representationof Plosives and Affricates ................................................ 13Anna Bloch-Rozmej3 Word and Foot Minimality in English: A MetricalGovernment Analysis .................................................... 25Tomasz CiszewskiPart IIGrammar: Morphology and Syntax4 The Psychological Reality of Grammar. A Cognitive LinguisticsPerspective ............................................................... 43Henryk Kardela5 A Morphologist’s Perspective on “Event Structure Theory”of Nominalizations ....................................................... 61Maria Bloch-Trojnar6 A Recalcitrant Nature of Object Experiencers ........................ 77Sylwiusz Żychliński7 On the Representations of Motion Events: Perspectivesfrom L2 Research ........................................................ 91Jolanta Latkowskavii

viiiContents8 On the Interplay Between Prepositional Categories. The Caseof the Polish od–do Construction ...................................... 103Daria Bębeniec9 Governed Prepositions in English: A Corpus-Based Study ......... 117Jerzy GaszewskiPart IIIHistorical Linguistics10 Reflections on Structural Variation in Old English Verbs .......... 137Magdalena Charzyńska-Wójcik11 The Semantic Analysis of Old English unnan ........................ 153Agnieszka WawrzyniakPart IVPragmatics, Discourse Analysis and Sociolinguistics12 “When We Talk, It Never Materializes”: Functions of Off-RecordCommunication in Conflict Talk ...................................... 165Joanna Bobin13 Territorialization in Political Discourse: A Pragma-Linguistic Studyof Jerzy Buzek’s Inaugural Speeches ................................. 177Katarzyna Molek-Kozakowska14 From a Complaint through Therapy to Recovery: PatientIndexicality in Medical Case Reports ................................. 189Magdalena Murawska15 Beyond and within Standard English: Categories, CategoryBoundaries and Fuzziness .............................................. 201Maciej RatajPart VTranslation16 Construction Grammar as a Framework for DescribingTranslation: A Prolegomenon ......................................... 215Izabela Szymańska17 Crossing the Frontiers of Linguistic Typology: Lexical Differencesand Translation Patterns in English and Russian Lolitaby Vladimir Nabokov ................................................... 227Łukasz Grabowski

IntroductionRecent years have seen important developments in the field of linguistics, or thescientific study of language, which can be ascribed to the accumulation of empiricaldata, the emergence of new theories of language as well as the rapid growth of newtechnologies. As a result, apart from traditional areas such as, for example, historicallinguistics, phonetics, phonology, morphology, syntax or semantics, currentoverviews of the discipline recognize the fact that formal aspects of language haveto be considered within the context of their use, with the effect that pragmatic,discoursal and sociolinguistic dimensions also need to be taken into account. Inaddition to this, considerable emphasis is now laid on the study of languagedisorders and the relationship between language and the brain, both of which fallwithin the scope of neurolinguistics, the way in which language is represented in thebrain, the processes responsible for its acquisition, use and attrition, the phenomenonof multilingualism, the description of real-life language use made possible byadvances in computational linguistics, or the different purposes for which ourknowledge in these areas can be employed, which comes within the purview ofapplied linguistics. Worth noting as well is the emergence of new theories oflanguage, especially usage-based models, such as Langacker’s Cognitive Grammar,which have quite successfully begun to challenge, and in some cases to complement,the well-established formal and functional accounts and are potentially majorcontenders when it comes to explaining how languages work, how they are acquiredand how they are used.The present volume recognizes all of these developments and it is a collection ofpapers representing most recent developments in Polish linguistics, specifically inthe fields of language, discourse and translation studies. The book is divided intofive parts, each including papers representative of traditionally distinguished linguisticsubdisciplines such as phonetics and phonology, morphology and syntax,historical linguistics, pragmatics, discourse analysis and sociolinguistics, and translation.The perspectives from which these areas are considered in specific contributions,however, are far from traditional since in many cases they strive to reconcilemore conventional and innovative perspectives by drawing, among others, uponix

xIntroductionlatest advances in corpus linguistics, cognitive linguistics, construction grammar orpragmatics. Since the contributions included in the publication touch upon such avariety of subdisciplines of linguistics and do so from such diverse theoreticalpositions, it is likely to provide an impulse for further theorizing and research forscholars, as well as constituting an important point of reference for graduatestudents and lecturers teaching courses in linguistics.Mirosław Pawlak

Biographical NotesDaria Bębeniec, PhD, is a lecturer at the Department of English, Maria Curie-Skłodowska University, Lublin. Her research interests include polysemy, vagueness,meaning representation, case semantics, language-specific encoding of spatialrelations, constructional meaning and methodological issues in prepositionalanalysis.Dr hab. Anna Bloch-Rozmej is affiliated with the Celtic Department and theHistory of English Department in the Institute of English at John Paul II CatholicUniversity of Lublin as well as the Institute of English at the Holy Cross University,Kielce. She specializes in linguistics, phonology in particular. She is author ofnumerous articles devoted to the phonology of the Celtic languages, English,German and Polish, published both in Poland and abroad. Her latest monographis Melody in Government Phonology exploring the phonological phenomena pertainingto the subsegmental structure of sounds. Other areas of interest involvemethodology of English teaching and phonetics.Maria Bloch-Trojnar, PhD in Linguistics, is a lecturer in the Department ofCeltic, John Paul II Catholic University of Lublin. Her major research interestsinclude morphology and its interfaces with other grammatical components, inparticular de-verbal nominalizations, lexicology, English, Celtic and Slavic languages.She is author of Polyfunctionality in morphology – A study of verbal nounsin Modern Irish (2006, Lublin, Wydawnictwo KUL) and editor of Perspectives onCeltic languages (2009, Lublin, Wydawnictwo KUL). She has published amongothers with E´igse: A Journal of Irish Studies and Journal of Celtic Linguistics.Joanna Bobin, MA, is a teacher at the State School of Higher ProfessionalEducation in Gorzów Wlkp. and a doctoral student of linguistic pragmatics at theSchool of English, Adam Mickiewicz University, Poznań. Her academic interestsinclude pragmatic theories of politeness and impoliteness, conflict typology andthe importance of context for conflict analysis, as well as literary theory andmodern drama.xi

xiiBiographical NotesMagdalena Charzyńska-Wójcik, PhD, is Assistant Professor in the Department ofEnglish Historical Linguistics at John Paul II Catholic University of Lublin.Research interests: Old English syntax with special focus on impersonals, Experiencerverbs, passive constructions, (clausal) ditransitive verbs, subjectless constructions,the status of clausal constituents and the syntax of the Old English verb.Tomasz Ciszewski, PhD, is a lecturer at the Institute of English, University ofGdańsk. He received his MA and PhD degrees from the University of Łódź in 1994and 2000. His fields of expertise include linguistics, phonetics and phonology. Hismost important publications are The English stress system: Conditions and parameters.Studia Językoznawcze Wszechnicy Mazurskiej w Olecku (2005), “Thestem-suffix conflict? The sources of variability of stress patterns”. In Issues inaccents of English, 195–211, Cambridge Scholars Publishing (2008), and “Theoreticalmodelling of word-level isochrony in English”, In Issues in accents ofEnglish 2: Variability and norm, 261–275, Cambridge Scholars Publishing (2010).Jerzy Gaszewski is a doctoral student at the University of Łódź. His main researchinterests are comparative linguistics and language typology. His research so far hascombined syntactic and semantic issues and has focused on minor lexical classeslike particles and prepositions. The latter are investigated in his doctoral dissertationas well as in the present paper.Łukasz Grabowski, PhD, is Assistant Professor at the Institute of English Studies,Opole University, Opole. He specializes in corpus linguistics, translation studies(theoretical and applied), bilingual lexicography, language typology and sociolinguistics.He has authored over 15 articles on, among others, application of Englishand Russian language corpora in didactics of translation and lexicography. He isalso interested in computer-assisted methods of text analysis.Henryk Kardela is Professor of Linguistics at the Department of English, MariaCurie-Skłodowska University, Lublin. His fields of interest include syntax, semanticsand philosophy of science. He is author of 3 books (A grammar of Polish andEnglish reflexives, Lublin: UMCS, 1985, WH-movement in English and Polish.Theoretical implications, Lublin: UMCS – both written in the spirit of NoamChomsky’s Government and Binding Theory; and a book written in the frameworkof Ronald Langacker’s Cognitive Grammar, Dimensions and parameters ingrammar. Studies on A/D asymmetries and subjectivity relations in Polish, Lublin:UMCS, 2000); editor and co-editor of 12 linguistic volumes, and author of manyarticles on syntax, semantics and lexicography.Jolanta Latkowska (PhD in Applied Linguistics, University of Silesia, 1998) isAssistant Professor in the Department of English Studies at the University ofSilesia, Poland. Her research addresses cognitive and linguistic aspects of bilingualism,including L2-induced restructuring of the L1, semantic and conceptualtransfer, as well as bidirectional interaction of languages within the construct ofmulticompetence.

Biographical NotesxiiiKatarzyna Molek-Kozakowska, PhD, is Assistant Professor at the Institute ofEnglish Studies, Opole University, Opole. She specializes in pragma-linguistics,Critical Discourse Analysis and cultural/media studies. She has published overtwenty critical studies of mass-mediated political discourse, focusing on persuasion,identity polarization, personalization, stereotyping or metaphors. She is alsointerested in the methodology of qualitative research and critical media literacy.Magdalena Murawska is Assistant Professor in the Department of Sociolinguisticsand Discourse Studies in the School of English at Adam MickiewiczUniversity, Poznan, Poland. Her main research interests lie in the field of sociolinguistics.Her PhD dissertation concerned patient imaging in case reports fromBritish and American medical journals, with particular attention to the presentationof the patient’s experience of illness.Maciej Rataj obtained his Master’s degree in English at the University of Gdańskin 2006. He is now an Assistant Reader at the Institute of English and a translator ofPolish. He has written several papers on sociolinguistics and normative linguisticsand has presented his research at several conferences in Poland and abroad.Izabela Szymańska, PhD, is Assistant Professor at the Institute of English Studies,Warsaw University. Her interests range from theoretical and contrastive linguistics,especially the Construction Grammar framework, through the teaching of Englishfor academic purposes, to translation studies. In this last field, her research focuseson text-linguistic and pragmatic issues, on the interface between linguistic andcultural aspects of translation, and on the translation of children’s literature.Ewa Waniek-Klimczak is Professor and Head of the Department of EnglishGrammar and Phonetics at the University of Łódź, Poland. She teaches phoneticsand phonology of English, sociolinguistics and accents of English courses. Herresearch interests include bilingual Polish-English speech and the pronunciation ofEnglish in the EFL/EIL contexts. She organizes conferences on native and nonnativeaccents of English and publishes books, articles and collections of editedpapers on various issues in applied linguistic phonetics.Agnieszka Wawrzyniak, PhD, graduated from Adam Mickiewicz University,Poznań, and received a doctoral degree from this university in 2006 on the basisof a dissertation entitled The emergence of epistemic senses in preterite-presentverbs in Old English. She is currently employed in the Department of EnglishStudies, Faculty of Pedagogy and Fine Arts, Adam Mickiewicz University, Kalisz.Sylwiusz Żychliński is a PhD student of syntax in the School of English at AdamMickiewicz University, Poznań, working in the Department of Polish-EnglishComparative Linguistics, interested mainly in the syntax and semantics of psychologicalpredicates and reflexives, also a member of a grant-holding team working onthe selected issues of control in Polish and English.

Part IPhonetics and Phonology

Chapter 1Aspiration in Polish: A Sound Changein Progress?Ewa Waniek-KlimczakAbstract Aspiration in the production of /p/t/k/ before a stressed vowel has beentraditionally believed to be typical for English, but not Polish. In the speech ofPolish-English bilinguals, the length of the Voice Onset Time in English and Polishhas been found to be conditioned by phonetic universal effects as well as thestylistic and attitudinal factors, with aspiration functioning as one of the markers.Recently, the use of aspiration in Polish seems to be growing not as a sign ofEnglish-accented speech but rather an element of emphatic style. The studyreported here explores the use of the longer VOT values in Polish from theperspective of stylistic conditioning, adopting the attention-to-speech paradigmand discussing the relationship between the observed variability and thepredictions formed on the basis of universal phonetic tendencies. The observedtendency to lengthen the VOT values in the universally lengthening contexts isclaimed to be further affected by the attention paid to speech, which suggests apossible prestigious function of aspiration in Polish. The intriguing questionsemerging from the study are whether the change is the effect of language experienceand whether it spreads from bilingual/advanced learners of English to monolingualspeakers.1.1 IntroductionAs a phonetic phenomenon, aspiration can be defined perceptually as a strong burstof air that may accompany the release of a plosive, or acoustically – as a relativelylong time-lag between the release of the plosive and the onset of voicing for thefollowing vowel. It is the acoustic approach that has been most commonly used eversince Lisker and Abramson (1964) demonstrated that the length of the VoiceOnset Time corresponds to distinct phonetic categories, such as (pre-) voiced,E. Waniek-Klimczak (*)University of Łódź, Łódź, Polande-mail: ewaklim@uni.lodz.plM. Pawlak and J. Bielak (eds.), New Perspectives in Language,Discourse and Translation Studies, Second Language Learning and Teaching,DOI 10.1007/978-3-642-20083-0_1, # Springer-Verlag Berlin Heidelberg 20113

4 E. Waniek-Klimczakvoiceless unaspirated (corresponding to short lag VOT values) and voicelessaspirated (long lag values). Although more categories have been recognized onthe perceptual basis (including voiced aspirated), the VOT approach showed thatlanguages tend to contrast between the two categories, with one or three-way oppositionlimited to specific contexts (Cho and Ladefoged 1999). In Polish and Englishvoicing is signalled by two-way oppositions specific to particular contexts, with theexception of a word-final position in Polish where voicing tends to be neutralized (seeJassem and Richter 1989 for a detailed discussion). The choice of the phoneticcategories contrasting voicing differs in a pre-stressed vowel position, where Polishcontrasts pre-voiced (lead) against unaspirated voiceless plosives, while Englishdifferentiates between voiceless unaspirated and voiceless aspirated plosives in mostdialects. The choice of the implementation strategy for the voicing contrast has beenfound not to be accidental: out of the three categories, voiceless unaspirated has beenclaimed to be the only “unmarked” realization (Nathan 1997), corresponding to thecategory boundary found in new-born babies (Jusczyk 1997). Following universalpreferences, language systems can be expected to contrast voiceless unaspirated stopswith one of the marked phonetic categories: aspirated or voiced ones; if the twomarked options occur, especially in language contact situations they are used invariation with the preferred, marked-unmarked choice (Nathan 1997).When treated as an element of the physical reality of speech, the three categoriescorrespond to a range of the VOT values measured, with the individual instancesinterpreted by means of the “goodness-of-fit” criterion (see e.g. Cho and Ladefoged1999). Most typically, the voiced category corresponds to the voice lead (negativeVOT), voiceless unaspirated to the VOT between 0 and 30 milliseconds (ms), andvoiceless aspirated to values above 30 ms. However, the actual duration of the VOT isstrongly dependent on phonetic conditioning, including such factors as the place ofstricture, the height of the following vowel, the degree of stress and the length of theword (Maddieson 1997). Universally, the further back the place of closure, the longerthe VOT; the VOT is also longer before a high than non-high vowel, in one-syllablewords and under stress. The articulatory contextual factors have been referred to as“mechanical” universals as they reflect aerodynamic properties of the vocal tracts. Theeffect of stress, on the other hand, is more variable, and has been described as atendency rather than a regularity (“ecological” universals in Maddieson 1997).The above considerations form an interesting background for the study ofaspiration in Polish. On the one hand, an early study of the VOT values in Polishconducted by Keating et al. (1981) showed that the phonological system of Polishdoes not use the phonetic category voiceless aspirated. The mean VOT values forthe three voiceless plosive categories /p/, /t/, /k/ cited in the study (see Table 1.1)Table 1.1 The VOT measurements in milliseconds for Polish (afterKeating et al. 1981) and English (Lisker and Abramson 1964)/p/ /t/ /k/Polish 22 28 52English 59 67 84

1 Aspiration in Polish: A Sound Change in Progress? 5range from 22 ms for bilabials to 52 ms in velars. On the other hand, however, themean value for velar plosives reported in the study exceeds 30, which can suggestthat the mechanical effect of a velar lengthening results in the tendency for weakaspiration in /k/, even if there is no reason to claim that Polish uses aspiration at thephonetic category level. In fact, when compared to the mean values for Englishvoiceless aspirated stops, the VOT values for Polish follow the lengthening pattern,with a stronger effect of velar lengthening when compared to the bilabial plosive(2.3 for Polish vs. 1.4 in English).The evidence for the tendency to lengthen the VOT in a velar plosive in Polish inthe data from early 1980s provides motivation for further studies into the use ofaspiration in Polish. Although mentioned as an element of the system of Englishdifficult for Polish learners (e.g. Sobkowiak 1996) and often treated as typical forEnglish-accented Polish, aspiration can be increasingly heard in Polish, especiallyin emphatic contexts. While the variable use of aspiration can be possibly explainedon the basis of phonetic universals, there seems to be another intriguing possibilityof the potential source of what might be considered a potential change in progress inPolish: the effect of English.1.2 Acoustic StudyThe study reported here assumes that aspirated plosives are possible variants inpre-stressed vowel word-initial position in Polish. The assumption is based on thefindings by Keating et al. (1981) with respect to the length of the VOT valuefor the velar plosive /k/, phonetic universal tendencies for the lengthening ofthe VOT in specific phonetic contexts, including emphasis, and on personalobservation pointing to an increased use of this variant by Polish speakers intheir Polish mode. The design of the study reflects the expected effect of phoneticuniversals and a possible effect of the phonetic transfer from English, as one ofthe possible sources of change explored here is the effect of language experience.The data presented here come from a larger study aiming to explore theeffect of style, language experience and phonetic context on the use ofthe VOT values (Waniek-Klimczak 2011); additionally, the data reportedfor bilingual Polish-English speakers (Waniek-Klimczak 2005) are used forcomparison.1.2.1 MethodTwo groups of Polish speakers with different degrees of English experience recordedexperimental words in carrier sentences and dialogues. The two groups consisted of10 speakers each, with an equal proportion of male and female speakers. Bothgroups represented young, educated Poles (age range 20–30, university education).

6 E. Waniek-KlimczakGroup 1, monolingual Polish speakers (“Pol”), claimed minimal experience inEnglish; Group 2, on the other hand, included proficient Polish speakers of English,who use English in their every-day work in Poland (“Staff”). The speakers recordedthe following test words:One syllable wordspił ‘he drank’, pan ‘mister’, kin ‘cinemas’, kat ‘executioner’Two syllable wordspiwo ‘beer’, parking ‘parking’, kipi ‘boils over’, kanał ‘channel’The test words contained two voiceless plosives: bilabial /p/ and velar /k/ in thefollowing context of a high and non-high vowel in one or two syllable words.Following the predictions made on the basis of mechanistic phonetic universals, theVOT is expected to be longer in velars than bilabials, before a high than a non-highvowel and in shorter, monosyllabic words. In order to check the stylistic effect ofemphasis, the words were recorded in two styles: word-reading in identical carriersentences (I am saying .... .) and in humorous dialogues where an emphatic contextwas created for the test words. The predicted lengthening effect of the emphasis wasbased on implicational, “ecological” universals. Moreover, longer VOT values inthe text rather than word reading were expected to reflect the variable nature ofaspiration in Polish. Following the attention-to-speech paradigm (Labov 1972),more standard-like articulation corresponds to a higher level of attention andlanguage awareness; consequently, the more careful word reading style can bepredicted to be less conducive to aspiration than less-controlled, more emphaticdialogue reading.1.2.2 ResultsThe results presented in this section refer to the mean values for the VOT measuredfrom the waveform and spectrographic analysis of the recordings performed withthe use of the Wavesurfer software. The general results obtained from the meanmeasurements across the styles and word length for the two groups of speakerspoint to the effect of language experience in the higher range of values consistentlyused by the ‘Staff’ group, with the parallel effect of the place of articulation and theheight of the following vowel (Fig. 1.1).The relationship between the two groups proves consistent within each style, i.e.the more experienced group regularly produces longer VOT values in all the testwords in both styles (Figs. 1.2 and 1.3).The comparison of the results between the styles for the two groups shows thatalthough there are differences in the degree of lengthening in relation to individualphonetic context characteristics across the styles, both groups maintain asimilar pattern of universally conditioned VOT length. Interestingly, however, thepattern proves to be more regular in the case of more English- (and aspitation-)experienced group in the more careful, word reading style. Moreover, the samegroup seems to be more sensitive to emphatic, dialogue reading style. The above

1 Aspiration in Polish: A Sound Change in Progress? 7Fig. 1.1 Mean VOT (ms) values for /p/ and /k/ in the context of high and non-high vowels acrossthe styles and word-length by the two groups of speakers, N ¼ 10 4 in each groupFig. 1.2 Mean VOT values (ms) for words read in carrier sentences by the two groups of speakers(N ¼ 10 for each group)tentative observations may suggest that the “Staff” group not only use longerVOT values across the context, but are also more likely to lengthen the VOT inemphatic speech. In fact, the comparison of the mean values across the style forboth groups illustrates this tendency for the “Staff” group more than “Pol”(Figs. 1.4 and 1.5).Apart from the distribution of values, let us notice that both groups use the VOTvalues in the range higher than expected for unaspirated plosives in all the casesexcept for /p/ in the following context of a non-high vowel, where the values drop to20 ms for the “Pol” group and barely exceed 30 ms for the “Staff”. It is in thiscontext, however, that we can observe an interesting effect of style in the case of theword pan in the more English-experienced, “Staff” group.

8 E. Waniek-Klimczak1009080706050403020100927468675950 50 505445343137362018piłT piwoT panT parkingT kinT kipiT katT kanałTStaffPolFig. 1.3 Mean VOT values (ms) for words read in dialogues by the two groups of speakers(N ¼ 10 for each group)Fig. 1.4 Mean VOT (ms) in two styles for the group “Staff”, N ¼ 10Fig. 1.5 Mean VOT (ms) in two styles for the group “Pol”, N ¼ 10

1 Aspiration in Polish: A Sound Change in Progress? 91.3 DiscussionDue to a small number of observations, the data presented in the previous section donot make it possible to draw any firm conclusions as to the effect of individualvariables. However, the overall results seem to be clearly suggestive of the generaltendency for the speakers to use longer VOT values than expected in the production ofthe phonetic category “voiceless unaspirated”, with the exception of /p/ in the Englishinexperienced,“Pol” group. Although the tendency to lengthen the VOT as the effectof English experience seems much stronger than the effect of style, both these factorscontribute to the VOT lengthening. Contrary to our expectations though, it is the morecareful, slower speech style that elicits longer VOT values, suggesting that the effect ofemphasis and a possible tendency not to produce aspiration in a more controlled styleof speech in Polish may not be relevant. These tentative results might be interpreted interms of the acceptance of the aspirated plosive as a possible variant, with both the“Staff” and the “Pol” group aspirating the initial plosives in words such as pił, piwo,kin, kipi. Moreover, the “Staff” group aspirates plosives in non-high vowel contexts –the tendency which may spread as the result of their English experience, overriding thephonetic universal factors.The tendency to use aspiration in Polish in the “Staff” group may suggest thepresence of the elements of an English accent in their Polish, the tendency oftenmentioned in connection with Polish-English bilinguals, as a highly marked featureof “accented Polish”. The every-day use of English in this group would certainlymake them possible candidates for bilinguals, in spite of the Polish surroundings.The fact that the “Staff” members tend to have a predominantly Polish-accentedEnglish experience in their every-day work may in fact correspond to the bilingualexperience of foreign-accented target language in their experience in the Englishspeaking countries. Let us remember, however, that the continued amount of PolishFig. 1.6 Mean VOT values for /p/ and /k/ in Polish: monolingual vs. bilingual Polish-Englishspeakers; (data for Pol 1981: Keating et al. 1981; bilinguals: Waniek-Klimczak 2005)

10 E. Waniek-Klimczakexperience is undoubtedly much richer in the Polish rather than immigrant setting.In other words, if Polish-English bilinguals use aspirated plosives in their Polish,their production will be interpreted as the result of a cross-linguistic effect; however,when a similar process takes place internally in the system, we need toconsider the possibility of a different, language-internal change. Interestingly,however, the comparison between the data discussed above and the resultsreported in the course of the study into the so-called Early and Late immigrants,with the age of immigration below 12 and over 18, shows a parallel tendency(Fig. 1.6). When compared to the data from 1981, the mean values for the /p/ andfor /k/ increased for all speakers. The difference is small in the case of the Englishless-experienced“Pol” group; in the case of the “Staff”, however, the mean valuesused exceed the ones measured for the Early bilinguals. While these results need tobe treated with caution as they come from different studies, investigating the VOTin different contexts, the magnitude of the difference and the tendency for changesuggest an interesting tendency.1.4 ConclusionThe results presented in the study offer tentative support for the main claim: thetendency to use the VOT values typical for aspirated voiceless plosives has beenfound in Polish spoken by young educated speakers. The degree of aspirationvaries, with velar plosives aspirated more than bilabials. The effect of phoneticuniversals is attested for all speakers; however, the use of aspiration cannot beexplained by the emphatic style of speech. Although emphasis does createfavourable conditions for aspiration, the degree of lengthening and the extent towhich aspiration is used depends on the speakers’ experience of English. Whilemuch further work is needed before it might be possible to determine the source ofthis change, the use of aspiration in Polish does seem to be related to the experienceof aspiration in English. The fact that proficient speakers of English aspirate inPolish may not be surprising as an element of language interaction; however, thelack of the tendency to avoid aspiration in the most careful speech style, elicitingmost target-like forms, suggests that the use of aspiration may well be a change inprogress that might lead to the need to include voiceless aspirated as one of thephonetic categories in Polish.ReferencesCho, T. and P. Ladefoged. 1999. Variation and universals in VOT: Evidence from 18 languages.Journal of Phonetics 27: 207–229.Jassem, W. and W. Richter. 1989. Neutralization of voicing in Polish obstruents. Journal ofPhonetics 17: 205–212.

1 Aspiration in Polish: A Sound Change in Progress? 11Jusczyk, P. 1997. The discovery of language. Cambridge, MA: The MIT Press.Keating, P., M. Mikoś and W. Ganong III. 1981. A cross-language study of range of voice onsettime in the perception of stop consonant voicing. Journal of Acoustical Society of America70: 1260–1271.Labov, W. 1972. Sociolinguistic patterns. Philadelphia, PA: University of Pennsylvania Press.Lisker, L. and Abramson, A. S. 1964. A cross-linguistic study of voicing in initial stops: Acousticmeasurements. Word 20: 384–422.Maddieson, I. 1997. Phonetic universals. In The handbook of phonetic science, eds.W. J. Hardcastle and J. Laver, 619–639. Oxford: Blackwell.Nathan, G. S. 1997. On the non-acquisition of an English sound pattern. In Second languagespeech, eds. A. James and J. Leather, 181–185. Berlin: Mouton de Gruyter.Sobkowiak, W. 1996. English phonetics for Poles. Poznań: Bene Nati.Waniek-Klimczak, E. 2005. Temporal parameters in second language speech: applied phoneticsapproach. Łódź: Wydawnictwo Uniwersytetu Łódzkiego.Waniek-Klimczak, E. 2011. Aspiration and style: A sociophonetic study of the VOT in Polishlearners of English. In Achievements and perspectives in SLA of speech: New Sounds 2010, eds.M. Wrembel, M. Kul and K. Dziubalska-Kołaczyk, 303–317. Frankfurt am Main: Peter Lang.

Chapter 2Noise as a Phonological Element: On theRepresentation of Plosives and AffricatesAnna Bloch-RozmejAbstract This paper is devoted to the problem of the noise component in thephonological representation of consonantal segments, plosives and affricates inparticular. In this brief discussion, arguments will be supplied in favour of awardingthis property the status of a phonological primitive capable of manifesting itself asan independent segment as well as combining with other such primary elements toform larger, more complex melodic structures.After the phonetic characteristics of noise have been introduced, we shall explorea number of melody-related phenomena which clearly indicate that noise should berecognised as an independent phonological element. In particular, special attentionwill be devoted to lenition processes. The data will be taken from two differentlanguages: English and Basque. It will be demonstrated that noise participates inphonological processes independently of other primes. One kind of phenomenawe shall focus on will be the so-called edge and anti-edge effects exhibitedby affricates. Further, certain cross-boundary events will be elaborated on. Theproblem of the phonological status of noise will be regarded through the opticof Government Phonology. We shall adhere to the major assumptions of ElementTheory which is part of this framework. Applying this model of melodic representation,internal structure of plosives and affricates will be examined.2.1 Theoretical IntroductionThe aim of the present paper is to submit evidence testifying to the existence of thephonological primitive of noise as a legitimate member in the structure of releasedobstruents, plosives and affricates in particular. In this brief discussion, argumentswill be supplied in favour of awarding this property the status of a phonologicalelement capable of manifesting itself as an independent segment as well asA. Bloch-Rozmej (*)John Paul II Catholic University of Lublin, Lublin, Polande-mail: abloch@kul.lublin.plM. Pawlak and J. Bielak (eds.), New Perspectives in Language,Discourse and Translation Studies, Second Language Learning and Teaching,DOI 10.1007/978-3-642-20083-0_2, # Springer-Verlag Berlin Heidelberg 201113

14 A. Bloch-Rozmejcombining with other such primary elements to form larger, more complex melodicstructures.After the phonetic characteristics of noise have been introduced, we shall explorea number of melody-related phenomena which clearly indicate that noise should berecognised as an independent phonological element. In particular, special attentionwill be devoted to lenition processes. The data will be taken from two differentlanguages: English and Basque. It will be demonstrated that noise participates inphonological processes independently of other primes. One kind of phenomenawe shall focus on will be the so-called edge and anti-edge effects exhibited byaffricates. Further, certain cross-boundary events will be elaborated on.Importantly, our attention will be focused on the internal structure of theaforementioned segment types with a view to depicting the function of the noiseprime in their phonological representations. The model of intra-segmental structureadopted in this presentation is non-linear, autosegmental and molecular, inwhich autonomous elements reside on separate tiers, each being independentlysynchronized with a skeletal slot and having a unique phonetic interpretation. Themajor assumptions underlying such a mode of representation are summarized in(1) below: 11. Elements as minimal phonological units• Elements are autonomous;• They are directly co-indexed with skeletal slots;• Each prime resides on its autosegmental tier;• Elements are gathered under class nodes which dominate them;• Each element has a unique phonetic interpretation;• Elements can combine to form complex melodic structures;• Their attachment to slots requires an autosegmental licence;• Only elements associated to skeletal positions can be manifested phonetically;• Within segmental structures, primes can enjoy different status (head, operator/dependent), the head prime defines the salient property of the segment.In this molecular approach to segmental structure, segments consist of one ormore elements which amalgamate in accordance with the licensing constraints, i.e.co-occurrence conditions, built in the system of a given language. 2 Since, as indicatedabove, primes can enjoy different status, varying theoretical configurations of elementscan be distinguished. These are schematized in (2) below.1 The major characterisctics of phonological primes are defined in Harris (1994), Harris andLindsey (1995) as well as Bloch-Rozmej (2008).2 A particular language, for instance, can impose a bar on the combination of given two primes as aresult of which they will never be found to co-occur within a single segment. Such a licensingconstraint can be identified in English where the elements I (frontness, palatality) and U (roundedness)do not combine within vocalic segments. Consequently, English does not possess frontrounded vowels.

2 Noise as a Phonological Element: On the Representation of Plosives and Affricates 152. (a) x (b) x (c) x (d) x (e) x| | | |A A A A| |BBAs depicted in (2a), a prosodic position can dominate no segmental material,thus remaining melodically mute, it can be mono-elemental with this prime beingeither head or dependent (in (2b) and (2c) respectively), or more elements can builda segment with one of them being its head or with none of the primes in the headstatus (the structures in (2d) and (2e) respectively). Significantly, the head element(underlined) determines the salient property of the melody. In other words, itscontribution to the phonetic shape of the segment is the largest.This brief characterization of melodic structure proposed by non-linear frameworks,Government Phonology in particular, 3 reveals two important features of elements asphonological entities. Namely, thanks to their autonomous interpretation, elementsdirectly determine the phonetic outcome of a given phonological structure. Secondly,their residence on separate autosegmental tiers allows them to participate in phonologicalprocesses independently of the remaining elements belonging to particular segmentalrepresentations. Consequently, the recognition of the existence of a givenphonological element rests upon two kinds of evidence. Firstly, a phonological primehas to have significant phonetic exponent, which is most obviously supported by theoccurrence of melodies defined exclusively by this very element. Secondly, we shouldbe able to detect phonological processes which directly target the posited prime, oralternatively, all the other elements building a segment except the prime in question.Finally, phonological significance of an element derives from its ability to define someimportant aspect of the system, for example its susceptibility to licensing constraints orother language-specific principles (e.g. being able to occupy the head position of thesegment or, alternatively, being a persistent non-head).2.2 Noise as a Phonological PrimeLet us now turn to the phonetic effect of noise and discuss evidence implying itsphonological status. Government Phonology (henceforth GP) argues in favorof recognizing three manner-defining phonological primitives: occlusion, noise andnasality. 4 To specify the distinctive characteristics of h, let us quote Harris (1994:123): “the elemental pattern of h may be defined as noise, which maps onto the speech3 The framework first proposed in Kaye et al. (1985, 1990) and considerably refined in e.g. Harris(1994), Gussmann (2002), Cyran (2003), or Scheer (2004).4 The element of occlusion defines the effect of stoppness, or an abrupt decrease in overallamplitude. Nasality will naturally reside in the structure of nasal vowels and consonants (Harrisand Lindsey 1995). Further, there are the so-called resonance elements specifying the place ofarticulation of consonants: I (palatality), U (labiality) and A (coronality). The laryngeal primes Hand L, in turn, are responsible for defining voicelessness and voicing respectively.

16 A. Bloch-Rozmejsignal as aperiodic energy. The articulatory targeting of this effect involves a narrowedstricture which produces turbulent airflow. This element contributes a noise componentto the class of obstruent consonants”. As further argued in Harris and Lindsey(1995c: 70), “the absence of any supralaryngeal gesture in the independent articulationof [h] is entirely the function of the fact that it lacks its own resonance property.In compounds with other elements, however, the location of the noise-producinggesture will be determined by whatever resonance element may also be present”.Defined in this way, noise will naturally reside in fricatives and affricates,sounds which at least in some phase of their production involve turbulent airflow.Yet, in orthodox feature-based models of segmental representation, this componenthas not been recognized as phonologically significant in plosives. It has to beobserved, however, that in order to comply with the phonetic facts, phonologyshould be able to account for the aperiodic energy characterizing the release phaseof genuine plosives which manifests itself in the form of a noise burst. However,before bringing up evidence supporting the presence of noise in the lexical structureof plosives, let us focus on noise itself. As indicated in the introduction, eachphonological prime must have a unique phonetic manifestation. This requirementis satisfied in the case of noise since its independent interpretation is that of theglottal fricative [h]. Also, as already stated, in complex segments representedas elemental compounds, noise defines the aperiodic energy identifiable in thephonetic signal. To illustrate this situation with some examples, the structures ofselected English fricatives have been depicted in (3).3. [h] [s] [T] [f] [v] [S]x x x x x x| | | | | |h h h h h h| | | | |A A U U A| | | |H H H I|H(3) discloses one more important dimension of segmental structure thatpertains in an obvious way to noise-containing sounds, namely headedness relations.As indicated in (3), head elements are underlined. The head-dependent relation is aGP-specific means of representing elemental salience. In this way, we possess aninstrument for representing the difference between strident and non-strident fricativesand affricates. Thus, h-headed segments will display greater stridency, or noisiness,than ones in which h is just a dependent.Moving on to the structure of obstruent stops, our aim now is to uphold theirrepresentations involving the element of noise, despite its being less significant thanocclusion. The latter defines the acoustic effect of an abrupt decrease in overallamplitude, articulatorily effected as constriction. More specifically, we shalladdress the data illustrating the participation of noise in phonological processing

2 Noise as a Phonological Element: On the Representation of Plosives and Affricates 17as an independent prime. In particular, the following processes indicate the presenceof h:• Spirantisation,• Debuccalisation,• Tapping,• Glottalling.2.3 Noise as a Process Target and Result: The Structureof PlosivesThe acoustics of plosives indicates the necessity of representing them by means oftwo independent elements: occlusion and noise. Even though the narrowing in thevocal tract typical of plosives is practically equal to zero, their release isaccompanied by a short noise burst. However, more compelling evidence in favorof h in plosives comes from lenition processes.One of them is the process of vocalization, which results in the weakening of thesegment. In Element Theory, 5 such an effect is represented as decomplexificationof a melody carried out through the loss of elements which become delinked fromtheir skeletal slot. From the phonological perspective, the advantage of such effectsderives from their ability to indicate independent elements defining the place andmanner of articulation of sounds. Thus, the process of debuccalisation targets theplace-defining prime and discloses the existence of manner primes. With respect toobstruents, two processes seem to be of utmost importance: spirantisation and lossof audible release. Since each process targets one element at a time (Clements1985), weakening can be characterized as the suppression of some aspect of theelementary content of a segment. In other words, lenition involves progressivedecrease in the melodic complexity of a segment. It seems reasonable to expectthat the weakest segments, which are pre-deletion targets, are the least complex,i.e. monoelemental. In this way, lenition testifies to the existence of individualprimes.With that in mind, let us examine the stages of one of the opening processes,quoted in Harris (1994):4. Opening trajectory (Harris 1994: 120)Spirantization > Aspiration > DeletionPlosive > Fricative > h > ØDefined in terms of progressive decomplexification, the process of lenition revealsthe existence of the noise prime which shows up as the least complex segment betweenthe two extremes of a released obstruent and zero. In this type of lenition process, the5 Element Theory is part of the Government Phonology framework. It pertains to the nature andphonological behaviour of elements within melodic structures.

18 A. Bloch-Rozmejelement targeted is that of occlusion, whereas the one that remains is noise. Such adevelopment can be found in Ireland and Merseyside area of England. A couple ofillustrative examples come from Harris (1994:121):5. get [gɛs] letter [lɛsə]at [æh] not [nɒh]that [dæh] but [bʊh]As the [h]-final forms indicate, spirantisation has given way to debuccalizationto [h] in function words. Thus, the element of noise turns up as the most primitiveunit of the lenited segments (see (9) below).Further, we can identify a process which directly targets the noise prime, oftenleaving the remaining elements intact. The development that is meant here is that oftapping, or flapping. The process is firmly established in most of North America,Australia, Ireland and parts of England. Examples depicting this phenomenon arelisted in (6):6. pity [pɪɾɪ] pretty [prɪɾɪ]fit us fi[ɾ] us get off ge[ɾ] offcity [sɪɾɪ] set up se[ɾ] upAs can be seen, the intervocalic context yields a coronal tap, a segment consistingof the occlusion prime and one specifying the place of articulation. It loses, however,the element of noise (see (9) below). A more radical form of weakening resulting in theloss of both noise and resonance primes takes place in the case of glottalling:7. Glottalling >> DeletionPlosive > ? > ØIn this lenition type, prevalent in Scotland and England, the loss of the coronalgesture is accompanied by the delinking of noise, the residual reflex being realizedwith glottal stricture, as in8. pit [pɪ?] bit [bɪ?]pity [[pɪ?ɪ] peter [pi:?ə]The three weakening developments occurring in the varieties of English can bestructurally depicted as follows:9.

2 Noise as a Phonological Element: On the Representation of Plosives and Affricates 19It can be clearly noticed that the processes of lenition operating in English disclosethe existence of an independent element of noise which can be both the target of theprocess and the sole resident of the segment after the delinking of the remainingprimes has been effected.In what follows, we shall look into the structure of affricates, whose behaviourimplies the presence of both manner-defining primes in their structure: that ofocclusion and that of noise.2.4 Affricates: Noise and Occlusion Together and ApartAffricates are often considered as consonants having bi-segmental structure. Such aconception stems from the articulatory evidence which reveals the presence of twoconsecutive phases: constriction and narrowing producing friction. The approachesto affricates vary. On the whole, we can distinguish the following theoreticalpossibilities of representing affricates:10. (a) One segment attached to a single skeletal slot(b) Two segments attached to two adjacent slots(c) Two segments linked to one slot (contour structure)i. Whole segments are attached to the slotii. Elements from different tiers are linked to a single slot• With an ordered sequence of stop + fricative parts• With an unordered sequence of stop + fricative partsThe first option was proposed in Rubach (1994), where affricates were treated asstrident stops. Yet, the evidence on the so-called edge effects concerning differentbehavior of affricates with respect to the segments on their right and to their lefthandneighbors implies a bi-partite analysis. The data on stress placement inEnglish, on the other hand, exclude the analysis of affricates as a sequence of twoindependent positions attached to a stop and to a fricative. More specifically, inEnglish verbs, stress falls on the last heavy syllable and should the last one be light,on the penult. Consider the words in (11): 611. (a) Torment (b) Maintain (c) EditCollapse Arouse CancelIf affricates consisted of two positions, syllables containing them should behavejust like ones containing long vowels or a final VCC sequence. This, however, doesnot take place, as depicted below:6 Letters indicating the stressed vowels are in bold type.

20 A. Bloch-Rozmej12. Manage, damage, encourageHence, we infer that the affricate-containing word-final syllable is light. This, inturn, suggests that affricates are linked to a single skeletal position. Such a representationis known as a contour structure. It has to be acknowledged at this stage thateven among phonologists who opt for the contour analysis, there are ones who arguein favor of the Ordered Component Analysis, with a strictly determined sequence ofa stop plus fricative parts (e.g. Hoard 1967; Campbell 1974; Hockett 1975; Ewen1982; Clements and Keyser 1983; Sagey 1986; Steriade 1994; van de Weijer 1996).The Unordered Component Hypothesis represented by Lombardi (1990) and Hualde(1988a, b, 1991), assumes that the order of affricate components is irrelevant. Thesetwo opposing viewpoints are possible and justifiable since affricates exhibit onthe one hand edge effects, and on the other hand, anti-edge effects. Put differently,they sometimes behave with respect to the left-hand segments as stops and to theright-hand ones as fricatives, but on other occasions this order does not matter.Consider the past tense allomorphy or plural forms in English:13. wanted [tɪd] but messed [st] watched [tʃt]cats [ts] buses [sɪz] watches [tʃɪz]Clearly in the above examples, with respect to the right-hand segments,affricates behave like fricatives. However, Lombardi (1991) brings up data fromBasque which undermine the Ordered Component Analysis. In Basque, stops arevoiced after a nasal or lateral segment, as in14. afal-tu [afaldu] ‘have dinner, perf.’ken-tu [kendu] ‘take away, perf.’Yet, affricates resist the voicing operation:15. afal-tsen [afaltsen] ‘have dinner, imperf.’ken-tsen [kentsen] ‘take away, imperf.’The facts addressed so far bring to light the following properties of affricatesegments:16. Affricates• Are prosodically simplex.• Are melodically complex.• Exhibit inconsistent behavior with respect to the processes targeting themfrom the right and from the left, which depends not on the sequentialordering of their parts but the presence of a given feature or the characterof the process.To clarify the last point, it has to be added that for the outcome of the process itseems to matter how much of the internal structure of the affricate is visible to theprocess. This point will be illustrated presently.

2 Noise as a Phonological Element: On the Representation of Plosives and Affricates 21GP argues in favour of the geometric organization of elements in the subsegmentalplane. This hierarchical arrangement is depicted in (17).17. Geometric configuration of elements: Elements grouped under class nodesxRoot • hLaryngeal • N• PlaceHLA I U (R)Individual elements are organized under class nodes: the Root, Place/Resonanceand Laryngeal. A phonological process can involve either a single prime or targetthe whole class node, thus affecting all the elements gathered under it. 7GP has its own answer to the affricate dilemma which is based on the recognitionof class nodes. In detail, Harris (1994) proposes a contour structure with an internalgeometry of elements which are arrayed under class nodes. This structure isdepicted in (18)18.xRoot nodes• ••hPlace nodeA[t ]IThe representation rests upon the assumption that elements can co-exist under asingle skeletal position without actually undergoing fusion, unlike in plosives forinstance. 8 The structure assumes that within a single melody, two independentmanner elements, occlusion and noise, are dominated by two different Rootnodes, which prevents fusion. 9 Both Roots are linked to a single Resonance orPlace node, which expresses the homorganicity of the affricate parts. This structureaccounts for the fact that with respect to the processes depending on syllableweight, affricates behave as prosodically simplex segments. Importantly, the twoRoots reside on a single autosegmental tier, which secures a sequence of the7 For more information on element geometry, see Harris (1994) and Bloch-Rozmej (2008).8 In plosives noise and occlusion are fused under a single slot.9 The idea of two Root nodes goes back to Clements (1987). McCarthy (1988) proposed thatmanner elements should be dominated directly by the Root.

22 A. Bloch-Rozmejstopness and noise phases. Harris’s structure of an affricate is very useful in theaccount of affrication processes attested in English. In English, apart from the twolexical affricates, the coronal [t] and [d] can undergo affrication to [ts] and [dz],especially in the intervocalic position, as in [pitsi], [tsen] [dzip]. This effect ischaracteristic of the London vernacular speech (Wells 1990). In some cases, thisprocess leads to spirantisation, whereby [t]>[ts]>[s]. In terms of GP, the affricationof plosives can be analysed in terms of the Root node split. Namely, a single Rootundergoes cloning, whereby each of the new Root nodes comes to dominate its ownmanner element, both Roots being still synchronized with a single Place node.Interestingly, and luckily for our contour analysis, affrication takes place before avowel. It must be stressed that between a plosive, which is a constricted sound, andthe vowel, which is characterized by considerable opening in the vocal tract, thearticulatory distance is huge. Hence, affrication can be seen as a mechanismreducing this distance by the progressive opening of the full constriction of thestop part of the affricate, through the narrowing of the noise phase of its latter partbefore eventually a vowel succeeds.constriction > narrowing > opening2.5 ConclusionIn conclusion, we hope to have demonstrated that the element of noise is alegitimate member in the structure of all released obstruents, including fricativesand plosives. However, the phonetic contribution of the prime to the manifestationof these segment types is different. More specifically, in the structure of plosives,noise has to fuse with occlusion. In fricatives, h is the sole manner-definingelement. Further, it has been shown that noise possesses a unique phonetic manifestationand can be a single resident of a consonantal segment. In isolation, h isinterpreted as a glottal fricative.Furthermore, noise can be directly targeted by phonological processes and itremains in the structure of a consonant after spirantisation/aspiration processeshave lenited this segment. Both manner-defining primes – occlusion and noise – areneeded in the structure of affricates as they enable us to express both theirbi-segmental phonetic nature and the puzzling behaviour of affricates with respectto phonological processes. We have demonstrated that a special contour structureshould be employed in the phonological representation of affricates. In suchconfigurations, noise and occlusion have to be independently dominated by separateRoot nodes which are both linked to a single Place/Resonance node. Thisgeometric arrangement of primes enables us to provide an account of the phonologicalphenomena involving affricates much more efficiently than has been done inthe traditional, non-autosegmental frameworks.

2 Noise as a Phonological Element: On the Representation of Plosives and Affricates 23ReferencesBloch-Rozmej, A. 2008. Melody in government phonology. Lublin: Wydawnictwo KUL.Campbell, L. 1974. Phonological features: problems and proposals. Language 50: 52–65.Clements, G. N. 1985. The geometry of phonological features. Phonology Yearbook 2: 223–50.Clements, G. N. 1987. Phonological feature representation and the description of intrusive stops.CLS 23, Part 2: Parasession on autosegmental and metrical phonology, 29–50.Clements, G. N. and S. J. Keyser. 1983. CV phonology: A generative theory of the syllable.Cambridge, MA: MIT Press.Cyran, E. 2003. Complexity scales and licensing strength in phonology. Lublin: WydawnictwoKUL.Ewen, C. 1982. The internal structure of complex segments. In The structure of phonologicalrepresentations. Vol. 2, eds. H. van der Hulst and N. Smith, 27–67. Dordrecht: Foris.Gussmann, E. 2002. Phonology: Analysis and theory. Cambridge: Cambridge University Press.Harris, J. 1994. English sound structure. Oxford: Blackwell.Harris, J. and G. Lindsey. 1995. The elements of phonological representation. In Frontiers ofphonology. Atoms, structures, derivations, eds. J. Durand and F. Katamba, 34–79. London:Longman.Hoard, J. E. 1967. Complex phonological segments in distinctive feature theory. Paper presentedat the annual meeting of the LSA, Chicago, Ill. Ms.Hockett, C. F. 1975. If you slice it thin enough it’s not baloney. American Speech 47: 233–55.Hualde, J. E. 1988a. On Basque affricates. Anuario del Seminario de Filologia Vasca Julio deUrquijo XXII-2: 379–389.Hualde, J. E. 1988b. Affricates are not contour segments. Proceedings of the West Coast Conferenceon Formal Linguistics VII: 143–157.Hualde, J. E. 1991. Basque phonology. London: Routledge.Kaye, J., J. Lowenstamm and J.-R Vergnaud. 1985. The internal structure of phonologicalelements: A theory of charm and government. Phonology Yearbook 2: 305–328.Kaye, J., J. Lowenstamm and J.-R Vergnaud. 1990. Constituent structure and government inphonology. Phonology 7: 193–231.Lombardi, L. 1990. The nonlinear organisation of the affricate. Natural Language and LinguisticTheory 8: 375–425.Lombardi, L. 1991. Laryngeal features and laryngeal neutralization. Unpublished PhD dissertation,University of Massachusetts, Amherst, MA.McCarthy, J. 1988. Feature geometry and dependency: A review. Phonetica 45: 84–108.Rubach, J. 1994. Affricates as strident stops in Polish. Linguistic Inquiry 25: 119–143.Scheer, T. 2004. A lateral theory of phonology: What is CVCV, and why should it be? Vol 1.Berlin: Mouton de Gruyter.Steriade, D. 1994. Complex onsets as single segments: The Mazateco pattern. In Perspectives inphonology, eds. J. Cole and C. Kisseberth, 203–291. Stanford, CA: CSLI Publications.Sagey, E. 1986. The representation of features and relations in nonlinear phonology. UnpublishedPhD dissertation, Massachusetts Institute of Technology, Cambridge, MA.van de Weijer, J. M. 1996. Segmental structure and complex segments. HIL dissertations no. g.Holland Academic Graphics.Wells, J. C. 1990. Longman pronunciation dictionary. London: Longman.

Chapter 3Word and Foot Minimality in English:A Metrical Government AnalysisTomasz CiszewskiAbstract Possible foot inventories have been widely discussed in the literature(Hayes 1995; Halle and Vergnaud 1987; Burzio 1994, among others). While binaryfeet are generally accepted in all typologies, other foot types (unary, ternary,unbounded) have been subject to debate. In this paper we argue that unary feetare ill-formed due to their inability to support a contour tone. This prohibition isfurther reflected not only in word minimality requirements in English, i.e. the banon light open monosyllables, but also in the absence of short vowels word-finallyand the avoidance of final stress on light syllables. The acoustic analysis ofmonosyllabic CVC words shows that in isolated pronunciations the vowel issystematically lengthened in order to meet the tonally-driven foot minimalityrequirements. Thus, given the abundance of light monosyllabic words in theEnglish lexicon, the word and foot minimality requirements must be treated asnon-identical. These empirical observations will be formally captured by governmentrelations that hold between nuclei. In particular, we will argue that minimalfoot binarity follows from the fact that a full nucleus must always govern anothernucleus to its right. A phonetic result of such internuclear government is thereduction of the latter. Thus, CV monosyllables are both lexically and metricallyexcluded; CVC monosyllables, on the other hand, are lexically possible since thefull nucleus governs the final empty one. However, in order to meet the footminimality requirement the nucleus must increase its duration to accommodate acontour tone (cf. Gordon 2000). The minimal word and foot requirements, therefore,differ in one aspect only, namely that the head nucleus in the minimal wordneeds a nuclear position to govern, whereas the head of the foot needs to govern aphonetically full nucleus.T. Ciszewski (*)University of Gdańsk, Gdańsk, Polande-mail: angtc@univ.gda.plM. Pawlak and J. Bielak (eds.), New Perspectives in Language,Discourse and Translation Studies, Second Language Learning and Teaching,DOI 10.1007/978-3-642-20083-0_3, # Springer-Verlag Berlin Heidelberg 201125

26 T. Ciszewski3.1 IntroductionIn this paper we discuss the English word and foot minimality requirements. Thediscussion is set in the framework of Metrical Government Phonology, which ismeant to be a prosodic extension of Standard Government Phonology. Thus, wewill first refer to Csides’ (2007) original formulation of Metrical Governmentwhich, however, follows the assumptions of “Strict CV” offspring of Standard GP.The advantage of our proposal lies in the fact that the theoretical assumptions wemake are empirically testable (as opposed to the empirically doubtful status ofempty nuclei within the Strict CV version of Government Phonology) and ampleacoustic evidence directly supports the analysis. In particular we argue that theword and foot minimality both follow from obligatory metrical governmentrelations holding between nuclei. Thus, the minimal word and foot are bothdefined as internuclear governing domains but they differ in the stringency ofmelodic requirements on the nuclear governee, i.e. in the minimal word thegovernee may be the word-final empty nuclear position (CVCØ), whereas in theminimal foot the position must be filled with a melody (CVCV/CVV). Thisexplains the absence of CV # words in English as well as the phonotactic ban onshort unreduced vowels word-finally (in which the internuclear governing relationdoes not obtain) and the systematic vowel lengthening in isolated and phrase-finalCVC forms and final stressed CVC “syllables” which is imposed by the requirementof internuclear metrical government. The acoustic analysis of CVC wordsimplies that the vowel lengthening is tonally driven since it is invariablyaccompanied by a significantly greater pitch slope on the vowel in question. Assuch, it is fully supportive of Gordon’s (2000) cross-linguistic study of the tonalbasis of final weight criteria.3.2 Metrical GovernmentTo the best of our knowledge, the term “metrical government” has been first usedby Csides (2004, 2007), 1 but it seems to have been given little attention, evenfrom government phonology (GP) circles. The proposals that stem from his conceptof “metrical government” – hopefully inseminating for mainstream GP in the longrun – are also of certain relevance to our analysis.Thus, we propose that metrical government relations are bidirectional and maybe non-local. Ultimately, all government within a morphologically simple domainemanates from the head nucleus of the domain final maximal foot. The heads of1 We wish to thank the anonymous reviewer for drawing our attention to Charette (1991) andespecially to Manfredi’s article (1993), who had apparently formulated the principles of metricalgovernment prior to Csides.

3 Word and Foot Minimality in English: A Metrical Government Analysis 27the foot can only govern from left to right. Recessive nuclei govern from right toleft. The internuclear government relation is obligatory for all well-formed phonologicaldomains. Here, we agree with the original proposal of metrical governmentrelations (Csides 2004: 29–30) which claims that: “[g]overnment directlyaffecting melodic complexity takes place at the foot level. In a trochaic system thestressed vowel governs the farthest contentful vocalic position to its right withinthe foot silencing it relatively in the form of vowel reduction and therebyproviding it with governing licence. The government licensed contentful vocalicposition then governs the farthest contentful vocalic position to the oppositedirection within the foot”. We think, however, that Csides (2004) is not radicalenough, i.e. he seems to reject strict directionality and simultaneously adheres tofoot binarity (cf. “a trochaic system” he mentions). Our proposal differs in that itprovides a formal framework that accounts both for the governing relations withintrochaic/dactylic feet with the same set of principles and explicitly excludesunaries.3.3 Structural and Acoustic Ill-Formedness of “Unary Feet”Virtually all past non-government accounts of word stress and foot structure havebeen based on a pre-theoretically unwarranted assumption that the syllable is anindispensable mediating constituent between the segmental and metrical structure.Our model is grounded on the tenet that foot structure is a derivative of theinternuclear relations within a phonological domain. Since intrapedal governmentis contracted between skeletal positions, no intermediate constituent is necessary.Accordingly, the traditional structural n-ary labels do not refer to the number ofsyllables within a foot, but rather to the number of skeletal positions that participatein metrical government. Thus, unary, binary and ternary feet contain one, two orthree skeletal positions, respectively. This is shown in (1) below (Intermediateconstituent levels Rhyme/Nucleus have been omitted).1. (a) unary (b) binary (c) ternaryF F F..... .............. ..........................x x x x x xIn this way the model is able to directly relate the segmental strength andmetrical structure via the rejection of the syllable constituent. In the followingsection we will provide theoretical as well as phonologically relevant acousticarguments against skeletally unary F(x) metrical feet.

28 T. CiszewskiA pivotal theory-internal argument is that a solitary non-branching nucleus maynot form a foot itself since feet are defined as left-headed internuclear governingdomains. In the absence of another non-empty nucleus to its right, the potentialhead of the putative unary foot may not metrically govern.2. *FNØxA theoretically conceivable GP mechanism, which could “restore” the minimalstructural binarity of light monosyllables, i.e. the incorporation of the final emptynucleus into the structure in (2), 2 must be ruled out for a few reasons. First of all, asit has been emphasised before, empty nuclei are claimed to be inactive in footconstruction. Thus, they would have to be made visible to stress in this particularinstance only, i.e. in cot Foot {cotø} but not in cotton Foot {cotton}ø. Secondly, directmetrical government imposed on foot recessive rhyme is able to destroy its vocalicmelodies only partially, i.e. through the reduction to schwa. Note that in structurallysimilar forms, e.g. better [betə], the same government relation in unable to silencethe foot recessive nucleus completely, unless it is followed by a word-final highsonority onset [m, n,1], as in bottom [bɒtm], button [bʌtn] or bottle [bɒt1], which isacoustically capable of supporting some (albeit reduced) pitch and intensitycharacteristics of the lost vowel. 3 In such cases, however, the available vowel slotis linked directly to the melody of the high sonority consonant. By contrast, the finalempty nucleus in (3a) cannot bear a tone as it is inherently devoid of phoneticcontents. Thus, the representations of cot (3a) and cotton (3b), for instance, may notbe treated as metrically identical.3.The empty word final nuclei in (3a) and (3b) follow from theory-internallicensing requirements, according to which an onset must always be licensed2 As proposed by Burzio (1994), for example. For arguments against, see Ciszewski (2005).3 This suggests that a radical a priori exclusion of onsets from foot structure should be approachedwith caution. While onsets are indeed unable to contribute directly to foot weight or carrydistinctive pitch, they may indirectly reinforce the acoustic robustness of adjacent vowels(cf. Gordon 2005).

3 Word and Foot Minimality in English: A Metrical Government Analysis 29by the following nucleus (for arguments in favour of such analysis see, e.g. Harrisand Gussmann 1998 or Gussmann 2002: 96–101). While we agree with theobservation that word final consonants do not behave like codas and, consequently,we do not treat word final syllables that end in a single consonant asheavy 4 (which has important consequences for the theory of metrical governmentproposed here), we reject on acoustic/perceptual grounds the possibility thatempty nuclei are directly accessible to metrical structure. However, an unstressedgoverned nucleus enclosed within two consonants between which a right-to-leftinteronset government obtains, as in (3b), despite being phonetically absent, doesindirectly participate in intrapedal government since its skeletal slot is ultimatelylinked with a melody of the high sonority consonant. It is the interonset governmentbetween [t] and [n] that licenses the vowel syncope in (3b) on the one handand makes the vowel position available for intrapedal government on the other. In(3a) the melodically empty nucleus N 2 may not participate in intrapedal governmentas it is not licensed by the interonset government. Thus, (3a) does notrepresent a well-formed binary foot. Structurally it is identical to the prohibitedunary foot in (2).A collateral problem is the lexical absence of CV monosyllables and wordfinalshort unreduced vowels. The distributional asymmetry between CV andCVC/CV: syllables domain-finally follows naturally from the assumption thatboth minimal words and minimal feet must constitute left-headed internucleargoverning domains. Therefore, their well-formedness is based upon the ability ofthe potential head nucleus to govern the following nuclear position. Unreducedshort vowels are banned both word and foot finally for two reasons: (1) theycannot be metrical governors and (2) they cannot be governees, since they areunreduced. Thus, due to the absence of a nuclear position to their right they maynot realise their potential to govern. The minimal word and foot requirements,however, differ in one aspect only, namely the head nucleus in the minimal wordneeds a nuclear position to govern, whereas the head of the foot needs aphonetically full nucleus. Metaphorically speaking, one cannot be the Head ofthe Department, if no staff is employed. The staff, however, may either fake work(empty nuclei) or be fully involved in it (full nuclei). In the former case, thedepartment merely exists, in the latter it properly functions. By inference, theminimal word and the minimal foot requirements are non-identical. 5 They areinterrelated by the fundamental condition of internuclear government, but theydiffer in the stringency of requirements on the government target. This isillustrated in (4) below. Note that the requirements on monosyllables are identicalto those on domain-final syllables in polysyllabic forms. (The broken line4 Note that the double association of [n] with onset and the preceding vowel position does not makethe second syllable of cotton phonologically heavy.5 Cf. for instance Lahiri (2001: 1357), who claims that “[t]he minimal word must be at least a foot,or two syllables, or bimoraic, or some other constraint”.

30 T. Ciszewskiseparating the minimal and subminimal forms represents the word/foot wellformednessthreshold.)4.Due the unrealised governing potential of the nucleus in (4a), domain-finalunreduced vowels and CV monosyllables are banned both lexically and metrically. 6Final and monosyllabic CVC forms in (4b) are lexically well-formed, althoughmetrically subminimal, i.e. the internuclear government obtains but its target isempty. This is where the word and foot minimality depart. In terms of their phoneticrealisation, we may expect that for the head nucleus to be properly pedified andstressed it either has to lengthen or N 2 must be filled with a melody, e.g. post-lexically.The forms in (4c) pass both the word and the foot minimality tests, since two adjacentfull nuclear positions may freely contract government. This accounts for the lack ofdistributional restrictions for domain-final reduced vowels (since they are always infoot-recessive position) and long vowels/diphthongs (since they constitute wellformed– albeit minimal – binary feet). Thus, the lexical presence of monosyllabiccontent words with non-branching nuclei, e.g. pit, pet, pat, putt, pot, put does notprovide a direct falsifier for our ban on unary feet once a sharp distinction is madebetween “stressability (¼no reduction)” and “being stressed (¼pedification)”. In ourmodel, these forms are lexically unstressed since they do not form feet, but potentiallystressable once the minimal foot requirement is fulfilled.Bearing in mind the relational nature of stress, the theoretical exclusion of unaryfeet is also acoustically and perceptually justified, since linguistic stress relies onfluctuations of acoustic energy and pitch differences both within and between consecutivenuclear and/or rhymal positions. Monosyllables in which the rhyme dominatesonly one nuclear slot are thus unable to support a distinctive pitch contour. 7 Phoneticallythey suffer from what has been traditionally referred to in the metrical literature6 Note that citation forms of function words which end in a vowel (a, the, her, she, your, to, there)always contain a branching nucleus.7 Although Terken and Hermes (2000: 89) treat all monosyllables as prosodically prominent, sincethey ‘stand out from silence’. Under this assumption, even an isolated fricative, for example, couldconstitute a peak of acoustic prominence.

3 Word and Foot Minimality in English: A Metrical Government Analysis 31as the “unstressable word syndrome” (Hayes 1995: 110–113; cf. (4a) and (4b) above).This is shown in (5) below, where T and I stand for tone and intensity, respectively.5. *FNxT XI X> T Y> I YA possible counterargument against treating the putative foot structures in (2),(4a) and (4ab) as metrically ill-formed, as we do, may be that light monosyllablesare capable of carrying nuclear stress in a phrase, e.g. It’s a nice [cot, cat, kit, pet,book, etc.] and therefore they cannot be “footless”, since we take the presence ofstress to be a primary indication of the presence of a foot. (Otherwise, the footbecomes a theoretically superfluous constituent.) However, as it has been shownexperimentally (Duanmu 1996), light monosyllables in English in order to carrynuclear stress require a substantial increase in duration, especially in a pre-junctureposition, where their vowels were observed to have lengthened by as much as 75%.Other analyses, e.g. Umeda (1975), suggest that the increase in duration may bewithin 70–100% of the canonical duration of a vowel. As reported by Gordon(2000, 2001), tonal languages impose severe restrictions on the syllable types whichare allowed to carry contour tones. It is only on the V: or VC SON syllables thatcontour tones may occur without any restrictions. This has a direct bearing on ourtypology of well-formed English feet. Since foot internal head-dependent relationsin English are phonetically realised as pitch and energy fluctuations, sufficientduration is required in order to execute the two different tones, 8 or to avoid whathas been referred to by Gordon (2000) as “tonal crowding”. A light monosyllable isinherently unable to carry a contour tone on its single nuclear position. Thus, it mustlengthen its vowel by more than a half in order to be eligible for nuclear (prejunctural)stress. Formally, the pre-juncture lengthening may be expressed as a(post-lexical) addition of a prosodically recessive nuclear position (6).6. (a) *F (b) FN O N Ox x# x x x#T 1 T 2T 1 T 2An isolated pronunciation of a light monosyllabic (C)VC word is contextuallyidentical to a pre-junctural or pre-pausal one. Therefore, there are reasons to expect8 The increased duration of stressed syllable rhymes is, therefore, a derivative of tonal restrictions.

32 T. Ciszewskithat the acoustic characteristics of the nucleus in an isolated (C)VC monosyllableshould be systematically similar in terms of pitch, intensity and duration to the samenucleus in a word uttered pre-juncturally, i.e. [#bet#] vs. [....bet#], and systematicallydifferent from the occurrence of the same stressed nucleus non-finally, i.e.[#bet#] vs. [(#)betty#]. The acoustic analysis of relevant examples (Ciszewski2010) fully supports our predictions concerning the phonological non-identity ofstressed nuclei in isolated light monosyllables and those in well-formed trochees.The mean durations of nuclei in isolated CVC forms are consistently greater thanthose of the corresponding CVC + s items in an identical context (by 59% whenfollowed by a voiceless consonant and by 77% when followed by a voiced one).A viable formal alternative to lengthening, which may seem extravagantly“derivational” within GP, is to postulate that the nucleus in a CVC monosyllableis bipositional i.e. C[xø]C, which would in effect satisfy both lexical and metricalminimality requirements. Since the final empty nucleus is unable to properly governthe other empty nucleus within the C[xø]C domain, the latter would have tophonetically surface (7a), which amounts to lengthening (by analogy, the CV:Cmonosyllables would also retain their long nucleus). However, once the final emptynucleus is filled with a melody post-lexically, e.g. as in the inflected form betting(7b), it “regains” its proper government potentials and is able to silence the extranuclear position in the CVC stem and prevent the lengthening.7.This, however, would be grossly inconsistent with our assumption that phoneticallyempty nuclei are visible to metrical structure or available for metrification.Secondly, the lexically absent CV monosyllables would have to be represented inthe same way, i.e. C[vø]. Then, one would expect them to surface as long vowelsbecause the empty position – in the absence of a potential proper governor to itsright – would not be properly governed either. Thirdly, and most importantly, indisyllabic (C)V:CV(C) forms, like paper, the second position of the branching headnucleus, by virtue of being properly governed by the final full nucleus, would haveto be silenced *[pe(ø)pə]. Thus, we find little empirical support for this theoreticallycostly solution.The phonological ill-formedness of unary feet F(x) is also confirmed by theirdefective distribution within a phonological domain. English disyllables, forinstance, are never parsed into two consecutive F(x) feet, e.g. *[ Foot1 (CV 1 )Foot2(CV 2 )C]. Sequences of two (or more) unary F(x) feet are not attested domaininitially and internally in polysyllabic forms, either. Such footing would obviously

3 Word and Foot Minimality in English: A Metrical Government Analysis 33violate the rhythmic distribution of stresses and result in word-internal stress clash. 9Its ill-formedness is also independently predicted and accounted for by metricalgovernment. Since intrapedal structure is a reflection of governing relationscontracted between nuclei/rhymes and the government emanating from the headnucleus is head-initial, N 2 has no target to its left, unless it is compensatorilylengthened, e.g. in a pre-juncture position. Otherwise, it may not form an internucleargoverning domain, i.e. a foot. If N 1 , on the other hand, governed N 2 , it shouldresult in the reduction of N 2 to schwa. If this did happen, N 2 could be no longertreated as the head of the second foot, since foot heads cannot be reduced. But then,the resulting structure would be that of an even trochee, i.e. F(xx), rather than twoconsecutive unaries F(x)F(x). 10 This government “deadlock” is illustrated in (8)below (V stands for any full short vowel).8.Further distributional curtailments are observed domain initially and internally.While heavy (V: or VC) rhymes, whichformbinaryfeet,may–undercertain conditions – appear domain internally, e.g. sentimentality, Halicarnassus,or initially, e.g. ambition, unary feet are not attested in this position. The reasonsfor this ban are also related to the inability of two consecutive unreduced nucleito contract metrical government. In a hypothetical context whereby a unary footis followed by another well-formed foot within the same domain, e.g. F 1 [(x)]F 2 [(xx)(x)], interpedal government between the putative F 1 and F 2 does notobtain since it must be simultaneously licensed by intrapedal government withinboth participant feet. F 1 , however, is not an internuclear governing domain, hence9 We use the term “stress clash” without recourse to the syllable. Thus, the term is applicable onlywhen two consecutive nuclear positions, rather than syllables, are stressed. For instance, in thefrequently quoted example thirteen vs. thirteen men, both “syllables” (thir- and -teen) are metricalgoverning domains F(xx), and consequently, their head positions are non-adjacent, hence no stressclash occurs. The stress shift in thirteen men, thus, is a result of a much more complex interplaybetween interpedal government and syntactic relations.10 Direct government between heads of consecutive feet is not allowed in our model. The primary/secondary stress distinctions result from interpedal government which is contracted between twoconsecutive feet, and not between foot heads.

34 T. Ciszewskiits potential head nucleus is not projected to the level of interpedal government.The inability of the initial nucleus to metrically govern – resulting in its inabilityof projection – is phonetically manifested by vowel reduction (cf. agenda,veranda, America, etc.). Nuclei trapped in this particular position are in factunpedified. Consequently, a non-initial nucleus may never evade pedification asit is invariably preceded by another full nucleus by which it is obligatorilygoverned, hence it must be reduced and pedified under the foot-recessive branch.Therefore, unary feet are never found in polysyllabic domains in initial orinternal position. In the domain final context, on the other hand, the nucleus ofsuch foot is expected to lengthen to form a legal binary foot. This is confirmed bythe analysis of polysyllabic words in which the final stressed rhyme contains alexically short vowel.Although the EPD lists about 1,100 non-monosyllabic items which are markedas finally stressed, the group constitutes a fairly exotic collection of borrowings,mainly of French, Chinese, Russian, Spanish, Arabic and African origin, or Englishplace names. The only homogenous subset of English native forms is that ofdisyllabic verbs (a’ttack), prefixed verbs (re’set), isolated compounds in whichthe stressed member is a final light monosyllable (-cook, -with, etc.) or truncatedforms (le’git~le’gitimate or ce’leb~ce’lebrity) where the stress of the non-truncatedform is preserved.Out of approximately 840 finally stressed borrowings, more than 25% have annon-finally stressed variant, which is often the dominant one, and 54 items retaintheir final stress but lexically lengthen the final nucleus. A selective analysis of theirlexical frequency in the British National Corpus (~100 million words) confirms thatthe number of occurrences of particular items in usually negligible. More than halfof them do not occur in the corpus at all. This raises the question of how representativesuch items may be and whether any generalisations concerning the Englishmetrical structure may follow from their analysis. The answer is apparently negative.Native speakers of English, however, must in principle be able to process suchforms, both articulatorily and perceptually. The argument that the English pronunciationof such non-native forms is a copy of their original pronunciation in thesource language is easily refutable, since rarely do English native speakers have adirect auditory access to the original form. It is equally unlikely that they possess arudimentary knowledge of such diverse sound systems as, for instance, Chinese andArabic. Instead, we suggest that the exceptional stress on a lexically non-branchingfinal rhyme functions as a strong boundary marker. Given the minimal lexicalfrequencies of the finally stressed borrowings, their phonetic information loadmust be maximal, in the sense of Harris (2005). The hyperarticulation of theword-final nucleus in lexically rare borrowings, i.e. through enhancing the vowelquality and tonally-driven pre-junctural lengthening, draws the perceptual attentionof the listener to the word boundary. It seems plausible to assume that suchanomalous acoustic characteristics of the word final stressed nucleus is aimed todelay the word recognition point until all phonetic information has been presented.This account seems fully consistent with the “cohort theory” of word recognition

3 Word and Foot Minimality in English: A Metrical Government Analysis 35(Marslen-Wilson and Tyler 1980). A finally stressed item, e.g. Hamas, in a languagewith a generally non-final stress pattern must in all probability be the lastcandidate in the potential cohort of disyllabic items with initial [hæm-], if weassume that the structure of the cohort reflects the lexical frequencies of the itemsit contains. 11 Thus, the phonetic execution of stress on a lexically non-branchingfinal rhyme corresponds to the word recognition point, since no more candidates inthe cohort remain.The special status of the final stressed nucleus is acoustically marked by an extraH + tone on the onset of the final nucleus with a sharp drop towards its end. Theactual frequency of such final H tone is remarkably higher than of a high toneassigned to a binary foot head in an identical context, e.g. Hamas (287 Hz) vs.massive (175 Hz). As far as the duration of the stressed nucleus [æ] is concerned,the one in the finally stressed Hamas is more than twice the length of that in massive(204 ms and 88 ms, respectively). The acoustic and perceptual energies of thenuclei in question show similar divergences. Moreover, the pitch maximum onthe final nucleus of Hamas tends to be even higher than the average focal pitch onthe accented nucleus in massive (220 Hz) (e.g. in a carrier phrase: Massive, notpassive).This entails a further, and perhaps a more radical, inference that stress clash doesnot exist, either. In English simple domains consecutive short stressed vowels neveroccur. In suffixed forms the suffix either has a reduced vowel (¼foot recessiveposition) or it is autostressed (¼bipositional, i.e. a binary foot). Within a phrase,non-finally stressed items will never cause stress clash (¼they end in foot recessivenucleus). Finally stressed items and light monosyllabic content words, on theother hand, may be also be pedified (hence stressed) post-lexically when theyare followed by an unstressed syllable within the same phrase. The followingunstressed syllable may be either a monosyllabic unstressed function word or theinitial unstressed (lexically unpedified) syllable of the following word, e.g. hitand... or hit another. In such cases, the following unstressed nucleus forms therecessive branch the foot whose head is the stressed light monosyllable. This isconfirmed by the fact that the word-final [t]s in hit will undergo lenition in bothexamples, similarly to the foot medial post-head [d] in hidden [hɪɾ/ʔən]. Then, theT 1 > T 2 contour tone is distributed over two nuclear slots belonging to two separatesyllables, hence no compensatory lengthening of the foot head is required. Formally,it is captured by post-lexical intrapedal government (N 1 ! N 2 ) whereby theN 1 becomes the head of the foot and governs the following nucleus N 2 in the phrase.11 This of course refers to an isolated pronunciation whereby no syntactic or semantic/pragmaticcontext supports the word recognition by delimiting the number of words in a given cohort.Interestingly, while the word Hamas is more likely to be finally stressed in isolation, the initiallystressed variant seems more natural in connected speech, especially in a well-established (e.g.political) context.

36 T. Ciszewski9.Finally, light monosyllabic content words and finally stressed items may befollowed by an initial stressed nucleus of the next word, in which case they alsoneed to lengthen to form a binary foot as well. Interestingly, English phrasestructure rules successfully prevent the construction of longer sequences of stressedmonosyllabic content words, 12 e.g. Bob hit Rod, by separating them with unstressedfunction words, which restores minimal foot binarity through the post-lexicaladjunction of the following unstressed nucleus, e.g. Foot (Bob and) Rod. Thus, twoconsecutive nuclear positions will never be both stressed. Apparently, stress clash isa representational illusion experienced only by those who have their feet firmly onthe level of the syllable.The formal exclusion of unary feet, however, needs to be independentlysupported by acoustic data. For this, we must first establish which aspects of theacoustic characteristics of nuclei and rhymes are a true reflection of phonologicallyrelevant regularities. One may suspect that the actual pitch and intensity values ofnuclei and rhymes, which are natural exponents of pragmatic speaker- and contextdependentcommunicative functions, may be to a large extent non-phonological.Similarly, the duration of nuclei may also be influenced by stylistic functions. Thus,we assume that it is not the absolute acoustic values that encode phonologicalregularities but rather the systematic networks of relations between the acousticcorrelates of stress that are both phonetically and phonologically relevant. Theserelations may then be expressed in terms of metrical governing relations. It isprecisely why we take the systematic lengthening of the nucleus in isolated CVCmonosyllables and finally stressed polysyllables to be phonologically relevant.As far as unary feet are concerned, neither metrical government obtains norsystematic acoustic relations can be empirically established. For instance, if oneassumes that light monosyllables are inherently stressed, the actual acoustic valuesof stress correlates in such forms prove completely irrelevant, since they cannot beset against corresponding pitch and intensity values of adjacent rhymes/nuclei.Thus, the pitch and intensity differences between two or more isolatedpronunciations of a given (C)VC item must be treated as non-phonological, andperhaps even extralinguistic. In other words, whether the nucleus in a monosyllabic12 It may be accidental, but nonetheless intriguing, that function words following a content wordare syntactically limited in number. This observation requires an extensive corpus-based analysis,but, if confirmed, these facts must be rhythmically related.

3 Word and Foot Minimality in English: A Metrical Government Analysis 37word is pronounced with the pitch maximum of 500 Hz or 150 Hz and the meanintensity of, let us say, 84 dB or 70 dB, seems irrelevant for the perception of stress.However, if, hypothetically, the same values are mapped on a disyllabic wordrespectively, i.e. 500 Hz/84 dB on the initial nucleus and 150 Hz/70 dB on thefinal one, stress is unarguably perceived as initial. Note that in a monosyllabic wordpitch and intensity values of approximately 150 Hz/70 dB 13 would also have to beaccepted as perceptually sufficient for the perception of stress. Interestingly, thesevalues are quite similar to those of the final [ə] in a disyllabic initially stressedword, e.g. header (160 Hz/76 dB). This empirical paradox is resolved only if lightmonosyllables are assumed to be lexically unstressed and monosyllabic F(x) feetexcluded from the typology, since no absolute acoustic properties of the nucleuscan be indentified which correspond to the minimum threshold of stress. Bothphonologically and acoustically/perceptually stress remains a relative property ofnuclei and rhymes. As far as light monosyllabic words are concerned, it is onlythrough the increase in vowel duration or left-to right post-lexical government thatpitch and intensity differences bound to two adjacent non-empty nuclear positionsmay become phonologically relevant and perceived as stress. In our model, this isformally related to the requirement of minimal foot binarity F(xx).It needs to be emphasised that the strict ban on unary feet F(x) is first of all aconsequence of their inability to support a contour tone T 1 T 2 (e.g. HL). 14 Sincestatements and questions in English are respectively marked by L and H terminaltones, isolated unmarked (statement-like) pronunciations of light monosyllablesrequire a substantial increase in their duration to accommodate the extra L tonerequired in statements. The same holds true for the marked question-like isolatedpronunciations, whereby the extra H + tone needs to be supplied. In either case, theincrease in duration is required for articulatory reasons. 15 In disyllabic (and longerwords) the temporal space necessary for the execution and perceptual robustness ofa complex contour tone (HL or LH) is stretched over two consecutive nuclei, henceno lengthening of the head nucleus occurs. In government theoretical terms, acontour tone requires government between two adjacent nuclear positions.Drawing on Gordon’s idea of “tonal crowding” (2000), we attribute thesedifferences in the realisation of the stressed nucleus in final vs. non-final positionto the cross-linguistic absence of contour tones on light syllables. Since contourtones are combinations of two distinct level tones (HL or LH), their execution isimpossible over one timing slot. Thus, a unary foot F(x), both in light monosyllablesand in polysyllabic forms (regardless of its position within a word), would be13 Compare with the isolated (EPD) pronunciations of: (1) Bob (133Hz/77dB) and (2) kit(493.5 Hz/73 dB).14 Since this is a purely articulatory requirement, we infer that the ill-formedness of unary feet iscross-linguistic.15 Rather than perceptual ones. As shown experimentally by Janse (2003), the perception of speechsignal in artificially accelerated normal speech does not affect intelligibility. By contrast, when thesubjects were asked to increase the speed of delivery they were unable to double the normal rateand their intelligibility plummeted.

38 T. Ciszewskiineligible to carry two distinct tones. Hence, the avoidance of final stress on a lightsyllable and the absence of F(x) in polysyllables. The exceptional phoneticbehaviour of both word-final stressed light syllables and light monosyllabic itemsis thus a derivative of the phonological ill-formedness of unary feet.3.4 ConclusionIn our model unary feet are not “degenerate” but non-existent. Thus, the CVCwords and final stressed CVC syllables in order to become stressable must eitheracquire an extra nuclear slot (lengthening) or post-lexically govern the followingnucleus/nuclei in order to satisfy the requirements of metrical government. Sincethe notion of a governing domain necessarily entails the co-existence of the head(governor) and the complement (governee), the unary foot F(x) does not represent awell-formed metrical governing domain.ReferencesBurzio, L. 1994. Principles of English stress. Cambridge: Cambridge University Press.Charette, M. 1991. Conditions on phonological government. Cambridge: Cambridge UniversityPress.Ciszewski, T. 2005. The English stress system: Conditions and parameters. Studia JęzykoznawczeWszechnicy Mazurskiej w Olecku.Ciszewski, T. 2010. Is metrical foot a phonetic object? Research in Language 8: 115–134.Csides, C. 2004. Farewell to strict directionality: Evidence from English. The Even Yearbook6: 29–48.Csides, C. 2007. A strict CV approach to consonant lenition: Bidirectional government in Englishphonology. Language Sciences 29:177–202.Duanmu, S. 1996. Pre-juncture lengthening and foot binarity. Studies in the Linguistic Sciences 26:95–115.Gordon, M. 2000. The tonal basis of final weight criteria. Chicago Linguistics Society 3: 141–56.Gordon, M. 2001. A typology of contour tone restrictions. Studies in Language 25: 405–444.Gordon, M. 2005. A perceptually-driven account of onset-sensitive stress. Natural Language andLinguistic Theory 23: 595–653.Gussmann, E. 2002. Phonology: Analysis and theory. Cambridge University Press.Halle, M. and J.-R. Vergnaud. 1987. An essay on stress. Cambridge, MA: MIT Press.Harris, J. 2005. Vowel reduction as information loss. In Headhood, elements, specification andcontrastivity: Phonological papers in honour of John Anderson, eds. P. Carr, J. Durand andC. J. Ewen, 119–132. Amsterdam: John Benjamins Publishing.Harris, J. and E. Gussmann. 1998. Final codas: Why the west was wrong. In Structure andinterpretation in phonology: Studies in phonology, ed. E. Cyran, 139–162. Lublin: Folia.Hayes, B. 1995. Metrical stress theory: Principles and case studies. Chicago: University ofChicago Press.Janse, E. 2003. Word perception in fast speech: Artificially time-compressed vs. naturally producedfast speech. Speech Communication 42: 155–173.

3 Word and Foot Minimality in English: A Metrical Government Analysis 39Lahiri, A. 2001. Metrical patterns. In Language typology and language universals, eds.K. Ekkehard and M. Haspelmath, 1347–1367. Berlin: Mouton de Gruyter.Manfredi, V. 1993. Spreading and downstep: Prosodic government in tone languages. In Thephonology of tone: The representation of tonal register, eds. H. van der Hulst and K. Snider,133–84. Berlin, De Gruyter.Marslen-Wilson, W. D. and L. K. Tyler. 1980. The temporal structure of spoken languageunderstanding. Cognition 8: 1–71.Terken, J. and D. Hermes 2000. The perception of prosodic prominence. In Prosody: Theory andexperiment. Studies presented to G€osta Bruce, eds. M. Horne, 89–127. Dordrecht: KulwerAcademic Press.Umeda, N. 1975. Vowel duration in American English. Journal of the Acoustical Society ofAmerica 58: 434–445.

Part IIGrammar: Morphology and Syntax

Chapter 4The Psychological Reality of Grammar.A Cognitive Linguistics PerspectiveHenryk KardelaAbstract The paper deals with the issue of “psychological reality” of grammar.It is claimed that a grammar is “psychologically real” to the extent it meets twocriteria of psychological reality: (1) the criterion of explanatory adequacy in thesense of Chomsky (1965) and (2) the criterion of “realizability” by a psychologicalmodel of language use. The first criterion is connected with the language acquisitionprocess – a grammar is psychologically real if it can be learned by a child,while the second, with the “explicit realization mappings” to the PDP neural modelof language use – a grammar is “real” if its elements can be associated withanalogous constructs in a processing model of language use.4.1 IntroductionAs defined by Joan Bresnan (1978: 3), a linguistic concept is psychologically realto the extent that it contributes to the explanation of behavior relative to linguisticjudgments [...] [The issue relates] to the question as to whether constructions which aresuited for description of one form of verbal behavior (intuitive judgments) are equallysuited to the description of other verbal processes (the comprehension and retention ofsentences) [...].A realistic grammar, Bresnan continues (ibid.),must not only be psychologically real in this broad sense, but also realizeable. That is, weshould be able to define for it explicit realization mappings to psychological models oflanguage use. These realizations should map distinct grammatical rules and units in such away that different rule types of the grammar are associated with different processingfunctions.H. Kardela (*)Maria Curie-Skłodowska University, Lublin, Polande-mail: henkar@klio.umcs.lublin.plM. Pawlak and J. Bielak (eds.), New Perspectives in Language,Discourse and Translation Studies, Second Language Learning and Teaching,DOI 10.1007/978-3-642-20083-0_4, # Springer-Verlag Berlin Heidelberg 201143

44 H. KardelaThe paper examines the issue of psychological reality of grammar from a cognitivelinguistics perspective. It is claimed that a grammar is “psychologically real” if itmeets:1. The criterion of explanatory adequacy in the sense of Chomsky (1965);2. The criterion of the “explicit realization mappings” to the PDP neural psychologicalmodel of language use.We discuss these criteria in detail below.4.2 The Explanatory Adequacy of GrammarAccording to Chomsky (1965: 25), for a grammar to be “psychologically real” and,eo-ipso, be “explanatorily adequate”, it must be able to select, as Chomsky puts it,“a descriptively adequate grammar on the basis of primary linguistic data” and thusaccount “for the intuition of the native speaker on the basis of an empiricalhypothesis concerning the innate predisposition of the child to develop a certainkind of theory to deal with the evidence presented to him”.A somewhat different formulation of explanatory adequacy of grammar can befound in Radford (1981, 1988), where it is stated that a grammar meets explanatoryadequacy ifit correctly predicts which sentences are and are not well-formed in the language, correctlydescribes their structure, and also does so in terms of a highly restricted set of optimallysimple, universal, maximally general principles which represent psychologically plausiblenatural principles of mental computation, and are “learnable” by the child in a limitedperiod of time, and given access to limited data. [...] to attain explanatory adequacy, agrammar must in effect be “psychologically real”; in addition, it must be maximallyconstrained (Radford 1981: 26; emphasis added).Because the universal, “maximally general principles which represent psychologicallyplausible natural principles of mental computation”, are, according toChomsky, “radically underdetermined by evidence available to the languagelearner”, they must – so the argument goes – be supplied, in the process of languageacquisition, by Universal Grammar - UG (Chomsky 1981: 3). UG, let us add, isdefined by Chomsky as an “innate component of the human mind”, geneticallydetermined, which, to use Chomsky’s (1986: 3) parlance, “yields a particular languagethrough interaction with presented experience”.It should come as no surprise, then, that, seen in this light, the idea of psychologicalreality of grammar is of considerable theoretical importance to both cognitivistsand the adherents of generative grammar, although the basic lines of argumentationdeveloped by the two linguistic camps differ markedly. Says Langacker (2000: 2):Regarding the issue of innate specification I make no apriori claims. I do however subscribeto the general strategy in cognitive and functional linguistics of deriving language structureinsofar as possible from more general psychological capacities (e.g. perception, memory,categorization), positing inborn language-specific structures only as a last resort.

4 The Psychological Reality of Grammar. A Cognitive Linguistics Perspective 45And he adds (ibid.):If one aims for psychological reality, it cannot be maintained on purely methodologicalgrounds that the most parsimonious grammar is the best one. Should it prove that thecognitive representation of language is in fact massive and highly redundant, the mostaccurate description of it (as a psychological entity) will reflect that size and redundancy.The “massive redundancy” of the cognitive representation of language, alludedto by Langacker, is reflected in the way cognitive models of grammar approachlinguistic structure. According to Langacker (2000: 1–2), for instance, generativegrammarhas always tried to minimize what a speaker has to learn and mentally represent inacquiring a language. Its minimalism was originally based on economy: the best grammarwas the one that did the job with the fewest symbols. In recent years, the emphasis hasshifted to positing a richly specified universal grammar, so that the role of experience inlearning a language involves little more than the setting of parameters.Cognitive linguistics rejects the minimalist approach to language and grammar.For Langacker, the grammar of a language is an “inventory of conventionalizedlinguistic units” (1987: 57), among which one finds, apart from lexical items, formssuch as stock phrases, collocations and other fixed expressions which cannot bederived by means of syntactic or morphological rules, but which, as Langacker (1987:35) notes, “thoroughly permeate [language] use”. To them, Langacker explains,belong expressions such as “take it for granted, hold...responsible for, express aninterest in, great idea, tough competitor, [...], good to see you, mow the lawn, turn thepage, let the cat out, have great respect for, ready to do, play fair, I’ll do the best I can,answer the phone, and never want to see...again, etc.” (ibid.). It is expressions of thiskind – “prefabs”, as they are sometimes called – that have recently attracted attentionof theorists of language acquisition. The reason for this is twofold.First, the latest research on L1 acquisition appears to indicate that childrenacquire a language piecemeal: making use of available expressions and varioussimple constructions, they abstract away schemas from them, which are subsequentlyused to sanction more complex new expressions they encounter(cf. Tomasello 2000; Dąbrowska 2004; Wray 2002).There is yet another reason as to why L1 acquisition theorists are interested informulaic language. The argument comes from language processing research,where it is claimed that (1) formulaic language “offers shortcuts” for efficientconveying of message thereby lessening the burden of rule-governed computationof meaning and (2) it helps fill in gaps in messages uttered in “poor communicativeconditions”. Thus commenting on Segalowitz’s (1997: 105) observation about theimportance of formulaic language in “understanding a spoken message in a noisyroom”, that is, in a situation in which the person “will not be able to engage in all thecomplex processing that the situation requires”, Alison Wray (2005: 15–16)observes thatit seems reasonable that “the main reason for the prevalence of formulaicity in the adultlanguage system appears to be the simple processing principle of economy of effort”(Perkins 1999: 56). This economy occurs because it gives us access to “ready-made

46 H. Kardelaframeworks on which to hang the expression of our ideas, so that we do not have to gothrough the labor of generating an utterance all the way out from ‘S’ every time we want tosay something” (Becker 1975:17).Seen in this light, it should be clear that in order to meet the criteria of explanatoryadequacy and thus be “psychologically real”, a grammar must be able to offer aviable account of “formulaic language” as well. Good candidates in this respectappear to be cognitive, “bottom-up” models of language, i.e. models in which moregeneral constructions and schemas are abstracted from lower level units, rather than“top-down” rule-governed models of language of the generative grammar type.In what follows we take a look now at the cognitive, bottom-up approach tolanguage acquisition.4.3 The Cognitive Approach to Language AcquisitionAs already remarked, the adherents of cognitive linguistics reject the “minimalist”as well as “universalist” approach to language acquisition as advocated by thefollowers of generative grammar. Tomasello (1995: 138; quoted in Dąbrowska2004: 70), for example, observes, somewhat sarcastically, thatMany of the Generative Grammar structures that are found in English can be found in otherlanguages—if it is generative grammarians who are doing the looking. But these structuresmay not be found by linguists of other theoretical persuasions because the structures aredefined differently, or not recognized at all, in other linguistic theories.Challenging generative grammar on the issue of the rule-governed linguisticcreativity in the context of L1 acquisition, Tomasello (2000) notes that linguisticcreativity in the case of language acquisition by children has been “overstated”.Although it is true, he argues, that, children do develop generalizations, transformingforms from, say, Allgone paper to Allgone sticky “glue”, on the other hand, the child,upon hearing a sentence such as The window broke is unable – unless he has alreadymastered other usages with the verb break – to form such sentences as He broke it or Itgot broken. This, according to Tomasello, should be taken to mean that at this stagesmall children have not yet developed the predicate-argument structures and are usingcontext-determined “concrete” verbs, mastering their syntactic behaviour in the processof language acquisition. Dąbrowska (2004: 219) argues in the same vein. She says:Because, according to CG, language is a structured inventory of conventional linguisticunits, during the learning process children acquire a repertory of linguistic units—formulaicexpressions like Do you want milk, together with appropriate connections betweenthem. This is so because utterances of this sort, addressed to children by adults are used in arich and predictable context, which enables the child to gradually develop some idea abouttheir meaning.By way of example Dąbrowska describes a situation in which one day duringbreakfast a child looks at a milk carton while his mother pours some milk fromit and says: Do you want milk? During the weeks that follow, Dąbrowska(2004: 219) notes,

4 The Psychological Reality of Grammar. A Cognitive Linguistics Perspective 47the child hears: You want milk, D’you want milk?or Do you want milk? while the mother isholding the milk carton and looking at him, walking towards the fridge (which the childknows contains the milk), and in other mealtime contexts.The outcome of this is that[e]ventually [the child] learns that you want milk is used when he wants milk and is offeredit — in other words, he acquires a fixed formula with a communicative function(Dąbrowska 2004: 219–220).Below we give, based on Dąbrowska (2004: 182), some of the children’s earliestquestion formulas appearing in order of emergence in the acquisition process ofthree children with age range 1.8–2.4, as found in MacWhinney’s (1995) database:What’s this/that?; Where NP?; What’s NP?; Who VP (VP ¼ did that)?; What (is) NPdoing; Is it AP?; What happened to NP; Why S?; Is NP going to work?Formulas of this kind, the results of carefully structured studies on questionswith long distance dependences of the sort What did you say they thought Mariabelieved Chris needed? (p. 196), studies on contraction and inverted/uninvertedquestions involving auxiliary placement (pp. 191–196) – all this, in Dąbrowska’sview, argues against a rule governed mechanism of question formation à laChomsky, in favor of the “piecemeal learning” thesis as advocated by cognitivists.Says Dąbrowska (p. 200):On the standard generative account, the acquisition of questions involves learning to applytwo rules, WH fronting and subject-auxiliary inversion. Both of these rules are verygeneral, but subject to certain constraints. Such constraints are thought to be unlearnable,and hence the fact that humans apparently obey them is sometimes regarded as providingthe clinching evidence for the existence of innate Universal Grammar [...].For Dąbrowska, the evidence available now thanks to the advances in modernacquisition research suggests an entirely different scenario of language development.Thus she says (ibid.):[...] early usage is highly stereotypical and [...] development proceeds from invariantformulas through increasingly general formulaic frames to abstract templates. [...] developmentis piecemeal, in the sense that children learn the correct word order independentlyfor each construction, and the same changes occur at different times in differentconstructions. We also found no evidence of abrupt changes in behaviour which mightindicate a shift to a different productive mechanism: instead we see slow, steady progress.This suggests that the endpoint of development – that is to say, adult grammar – might berather similar to the kinds of representations that children use; in other words, that adultknowledge comprises not constraints on movement and such like, but rather fairly generaltemplates like Can NP VP?, Who VP? And What BE NP TRANS.V-ing? These would alsoallow speakers to produce and understand a wide variety of different question types, thusaccounting for the observed flexibility of adult behaviour. However, they do not involvemovement, and hence do not require innate constraints or an abstract level of syntacticrepresentation distinct from surface structure. 11 Some evidence that speakers make use of ready-made templates and insert new material in themwhen producing long distance wh-question formations rather than follow rules and constraintscomes, in Dąbrowska’s view, from the grammaticality judgments of speakers of English involving

48 H. KardelaIf Dąbrowska is right – if speakers of a language do indeed “fill in the slots” in theready made templates acquired piecemeal in the process of language acquisition –then a more plausible grammar and thus a more “realistic” one should be a grammarthat makes principled use of formulaic structures of this sort rather than a grammarwhich proposes rules and constraints on them. Needless to say, seen from this angle,all usage-based, bottom-up models of grammar which are currently being developedin the general framework of cognitive linguistics, Ronald Langacker’s CognitiveGrammar included, can be said to be “psychologically real” grammars.As we shall see presently, Cognitive Grammar appears to meet another criterionof psychological reality, namely the criterion of satisfying “explicit realizationmappings” to a theory of language processing. A good candidate for a languageprocessing theory appears to be a PDP neural model of language use, to which weturn now.4.4 The PDP System and the Realizability of a GrammarThe Paralel Distributed Processing system – PDP – is a system of highly complexcomputer simulation programs of information processing (also referred to as Connectionism).A “typical” PDP model is presented in Fig. 4.1 below. The PDP modelconsists of units which are treated as being equivalent to biological neurons. Unitsare linked with other units by connections whereby each unit receives input fromanother unit and provides input to other units in the system. Neural network modelsof this kind involve units, activations, connections and connection weights.prototypical and non-prototypical structures of this sort. If, as Dąbrowska notes, people werefollowing general rules, there would be “no [felt] difference between prototypical andunprototypical questions, provided that the corresponding declaratives are equally acceptable. If,however, they use lexically specific templates, they should rate the prototypical questions as moreacceptable than the unprototypical ones”. (p.199) The data presented by Dąbrowska, including theacceptability ratings, appear to support this claim (p. 199):Prototypical interrogative: Where do you think they sent the documents? Mean: 4.5Unprototypical interrogative:Prototypical declarative:Unprototypical declarative:Where will the customers remember they sentthe documents?You think they sent the documents to theHead Office.The customers will remember they sent thedocuments to the Head Office.Mean: 3.3Mean: 4.0Mean: 3.7Ungrammatical: *Who do you think that left? Mean: 1.9*He left in the office and had to go back.

4 The Psychological Reality of Grammar. A Cognitive Linguistics Perspective 49Fig. 4.1 A typical PDPmodel (Stufflebeam 2006; seealso a discussion in Bechteland Abrahamsen 1991 andCattel 2006)Output unitsoutput informationHidden unitsInput unitsinput informationLeastactivelessactiveaveragemoreactivemostactive0 .1, .2, .3, .4, .5, .6, .7, .8, .9 1discrete(digital)Non discrete (analog)discrete(digital)Fig. 4.2 The activation value system (Stufflebeam 2006)A connectionist model contains three layers of units, including input units, outputunits and hidden units. These three types of units, which are represented diagrammaticallyby circles, are linked by connections, symbolized by lines/arrows. In a socalledfeed/forward network, information flows forward from the network’s INPUTunits, via its HIDDEN units, to its OUTPUT units. A hidden unit receives signalsfrom input units, processes these signals and sends them – in the form of activations(or activation value) represented by numbers – as OUTPUT to other units.Neural networks are dynamic systems: when information is processed, someunits send stronger, while others weaker activations. The activation values arerepresented by numbers and they oscillate between 0 and 1 for each unit. Theactivation value system is shown in Fig. 4.2.The connections between the input and output units are assigned “weights”,which represent the function of synapses in the real brain. In order to calculate thetotal connection weight, (cj), between the input units and output units one has tomultiply the given weight and the activation value of the unit. The weight calculationin a neural network can be presented diagrammatically, as depicted in Fig. 4.3.Now, although Cognitive Grammar and the generative theory both aim atproviding a systematic, explicit description of the internal representation of

50 H. KardelaCOMBINED INPUT (Cj)wj1wj2wj3...weightsxa1x a2 x a3...activations... =cjFig. 4.3 The weight calculations in a neural network (Stufflebeam 2006). cj ¼ the sum of eachINPUT activation multiplied by its connection weightlanguage structure, of the two it is the cognitive account of such internal representationthat can be more readily, it seems, “realized” within a connectionist model oflanguage processing. According to Langacker (1991: 533), Cognitive Grammar,has, as he puts it, (1) a “natural affinity to connectionism” as neither model makesuse of explicit rules, claiming that they “are unnecessary for a viable account ofmental processing” and (2) Cognitive Grammar conceives of the speaker’s knowledgeof “grammatical patterns and restrictions” in such a way “that a connectionistaccount of its internal representation is quite imaginable”. One can “imagine”a connectionist account of “the internal representation” because the “organizationalfeatures within the cognitive processing” are, as Langacker notes (p. 534),emergent rather than fundamental; they reside at higher levels of organization that emergein processing activity whose basic character is roughly as envisaged in neural networkmodeling. [...] linguistic knowledge is stored in connection weights, and no direct analogof linguistic rules is discernible at the level.Yet, although “no direct analog of linguistic rules is discernible” at the level ofprocessing activity, this should not be taken to mean, Langacker observes, that “noanalog of linguistic rules can be discerned at any level of cognitive organization”. Inhis view, linguistic constructs postulated for linguistic description – although they“are unlikely to be directly and individually represented at the level of weightedneural connections” – have for him “some kind of cognitive validity” as they (p. 534)might [...] correspond to certain aspects of the processing supported by these connections.Thus given a strongly motivated and psychologically plausible linguistic description, aswell as a successful PDP model of the same phenomena, it would be both natural and [...]quite profitable to search for points of contact between the two.But where exactly should one look, “quite profitably”, for “points of contact”between the two theoretical systems? The shortest answer is: one should look forthem in the network of schematizations. For, according to Langacker, rules inCognitive Grammar have the form of schematizations of expressions and are, ashe puts it, “inherent in [...] the patterns of cognitive activity in whose occurrence

4 The Psychological Reality of Grammar. A Cognitive Linguistics Perspective 51their instantiations reside”. This being the case, linguistic rules, Langacker explains(p. 535),have a straightforward connectionist interpretation that invokes the standard notion of a statespace. A PDP system with n units defines an n-dimensional state space, in which eachdimension comprises a range of values that correspond to a particular unit’s possible levelsof activation. [...] As the system operates, it changes state from one instant to the next [...]and thus traverses a path through state space. [...] [i]f mental experience reduces to patterns ofneural activation, then [...] entertaining a concept, or invoking a linguistic structure—can beidentified either with a location in state space or a path (a series of locations).The parallels between a schema and a “diffuse location (or series of locations)” inthe PDP system established, the extraction of a schema from its instantiations can nowbe described in terms of a neural account. Thus, according to Langacker (1991:528),when patterns of activation corresponding to specific structures appear within a certainregion of state space, they trigger adjusting connection weights which induce theoccurrence of activation patterns falling in that region. PDP models can be said to berealistic from the standpoint of categorization, because they can, for instance, give riseto prototype effects.What sort of linguistic phenomena then should a PDP system be able to accountfor under the “realization mappings” condition between Cognitive Grammar anda neural network? In Langacker’s view, a linguistically viable PDP model mustaccount for complex structured conceptualizations such as the conceptual embeddingof complex sentences, quantifier scope. The model must be able to implementthe co-activation of distinct structures and establish correspondences between themwhile retaining their separate identity. Examples include analyzability, metaphoricalstructuring and the correspondences between elements of different mental spaces inthe sense of Fauconnier (1985) and Fauconnier and Turner (2002). An adequatePDP model must be able to handle conventional imagery, including profiling,trajector/landmark asymmetry, vantage point, and subjectivity.In what follows we concentrate on one aspect of linguistic structure, namely thereflexive – antecedent relation, a relation that crucially involves profiling, salience,trajector/landmark asymmetry and vantage point.4.5 Reflexivisation in Cognitive GrammarReflexivisation in Cognitive Grammar, together with possessivity, metonymy andso-called “raising” phenomena, is analyzed in terms of the so-called reference pointrelationship (Langacker 1991). In a typical reference point configuration,exemplified by the possessive construction John’s picture, for instance, the conceptualizer(the speaker/hearer) gains, through the reference point, access to therange of targets, i.e. to concepts and meanings associated with the concept JOHN(such as, for example, picture, hat, book, girl-friend, etc.), which are located in theso-called dominion for the concept JOHN.

52 H. KardelaExactly the same mechanism is responsible, according to Langacker, for theestablishing of the antecedent - reflexive relation in sentences such as John washedhimself (cf. Langacker 1999). In this sentence, the antecedent John sets upa reference point and establishes a dominion in which the pronoun himself (target)gets its interpretation. The antecedent is construed as a viewer, who “looks at” thereflexive from his vantage point. Whereas, as Langacker notes, pronouns likereflexives, are “immediately accessible” to the conceptualizer in the current discoursecontext, a full nominal (here John), which is potentially treated as areference point, is not. What actually determines whether a particular nominal iseligible or not for a reference point and how large the context over which it rangesis expected to be is determined by a number of factors including the salience ofthe prospective reference point and the nature of the conceptual connections withother elements. On the Cognitive Grammar account, the decreasing accessibility ofreference points to the targets should be treated as an extent to which a givenreference point configuration should be judged to depart from the prototypicalconfiguration. In what follows we explore in some detail instances of departurefrom the prototype in question.Consider the following examples, which involve the reference point relationshipproposed for reflexives (cf. Kardela 2009: 262):1. Peter saw himself in the mirror.2. John talked to Greg about himself.3. (a) John talked to Mary about herself.(b) *John talked about Mary with herself.4. A picture of himself hanging in John’s office appears to frighten Tom’s brother.5. Jan obiecał Marysi [– kupić sobie płaszcz].‘John promised Mary to buy himself a coat.’6. Jan kazał Marysi [– kupić sobie płaszcz].‘John asked Mary to buy himself/herself a coat.’(1) is a well formed sentence: the first NP that is encountered by the conceptualizeris John, which, in the reference point relationship, fits the specification of personand number required for the reflexive. In (2), there are two potential referencepoints for the target, namely John and Greg. The sentences in (3) show an interestingcontrast: whereas, as expected, of the two possible candidates for antecedents in(3a), it is Mary that is eligible for the reference point (on account of satisfying theperson and number feature specifications); this is not so in the (3b) sentence, whereno mental path can be established between Mary and herself. In (4) the referencepoint relationship between the target and the reference point is not easy to establishas the reflexive is not only situated to the left of the potential antecedent, but alsothe potential antecedents, John and Tom, enter the reference point relationshipswith office and brother, respectively. Still, himself is associated with Tom’s brother,not with John or Tom. Certainly, this sentence is not easy to process; it definitelydeparts from the prototypical reference point relationship for the reflexives. (5) is

4 The Psychological Reality of Grammar. A Cognitive Linguistics Perspective 53unproblematic: the reflexive sobie in this sentence can only refer to the subject Janas the verb obiecać ‘promise’ is a “subject-oriented” verb; (6), however, does posea problem as the reflexive sobie here can either refer to the subject or to the objectdespite the fact that kazać ‘order’ is an object-oriented verb (hence sobie shouldrefer to the object of the main clause, that is to Marysia, not to Jan).4.6 Parameters on Reflexivisation as Weights-RelatedRealization Mappings to a PDP SystemThe sentences involving reflexives discussed above and, indeed, any sentenceinvolving the reflexive-antecedent relationship, can, as already observed, be arrangedaccording to the decreasing accessibility of reference points (antecedents) to thetargets (reflexive pronouns). That is, the argument goes, the more difficult it is fora reflexive to enter a reference point configuration with its antecedent, the lessprototypical this configuration should be judged to be. The question now is how tomeasure the degree of this prototypicality.We would like to suggest, and this solution was adopted in Kardela (2009), thatthe departure in question should be formulated in terms of a set of parameters onreflexivisation. The parameters on reference point configuration for reflexives, inturn, we would like to claim now, should have – under the “realizability condition”– some analogs in the PDP theory. It seems to us that the best candidates for suchanalogs in a PDP model are weights.First, consider the parameters on the reference point relationship for reflexivisation.We shall discuss two such parameters: the Accessibility of Reference Points Parameter –ARPP and the Deagentivisation Parameter – DP (cf. Kardela 2009). The role of suchparameters is, to reiterate, to signal the varying degrees to which a given referencepoint relationship should be judged to depart from the prototypical reflexive antecedentrelationship.Let us look at the ARPP parameter first. In Kardela (2009) we claim that thisparameter crucially involves the notion of chain of command principle in the senseof Kuno (1987: 96; see also Deane 1992: 206). This being the case, theprototypicality scale involving the ARPP for the Polish reflexives, for instance,might look as follows (the most prototypical antecedent-reflexive configurationsare instances in (7a)(i) and (7b)(i), while the least prototypical are those in (7a)(iii)and (7b)(iii); cf. Kardela 2009: 263):7. The ARPP and the prototypicality scale for the Polish reflexives(a) The Syntactic Scale: The accessibility of the subject of a verb for a reflexivevaries in strength according to the syntactic role of the anaphor and is(i) Strongest in the case of the verb’s direct object anaphor: Jan umył się‘John washed himself’,(ii) Middle in the case of the object of a complement clause: Jan kazał Marysi[SUB kupić sobie płaszcz] ‘J. Asked Mary to buy him/her a coat’,

54 H. Kardela(iii) Weakest (unpredictable) in the case of się in prefixed inchoative verbs: Janza-siedział się-refl ‘Jan stayed-Perf. too long’ vs.*Jan nie u-siedział-Perf.się-refl’ vs. Jan nie usiedział ‘J. did not hold out staying in one place’.(b) The Semantic Scale: The accessibility of the subject of a verb for a reflexivevaries in strength according to the semantic/discourse nature of the reflexiveand is(i) Strongest in the case of definite animate NP: Jan umył się ‘John washedhimself’,(ii) Middle in the case of definite inanimate NP: Gałąź się złamała ‘Thebranch broke refl-się’,(iii) Weakest in the case of impersonal constructions -no, -to pragmaticallygoverned: Nie odrobiło się lekcji, co? ‘One/you did not do you classes,did you’ (reproach).We will take a look now at the parameter-based prototypical reference pointrelationship for reflexives from the point of view of the deagentivisation process,which in Polish is marked by, inter alia, the presence of the reflexive particle się.In order to do so, we have to introduce first the notion of energy chain.In Ronald Langacker’s model of Cognitive Grammar an event can be looked atfrom two perspectives: from the point of view of the so-called force dynamicsconstrual and from the point of view of the so-called absolute construal. 2 Thesetwo perspectives involve the so-called energy chain, in which an event is seen asinvolving the flow of energy which is transmitted from one participant to another(Langacker 1991: 283). The energy chain, in turn, gives rise to the so-calledcanonical event model, which underlies the prototypical transitive constructionand which, as Langacker (1991: 285–286) puts it, represents “the normal observationof a prototypical action”.The canonical event model includes elements which provide specifications forsemantic roles such as agent, patient, instrument, experiencer, etc. In contradistinctionto traditional approaches, however, these roles, called by Langacker “rolearchetypes”, are treated by Cognitive Grammar not as linguistic roles but ratheras pre-linguistic conceptualisations. The archetypal agent is thus a person whovolitionally “initiates physical activity”. As a result, the transfer of energy takesplace to an external object. The archetypal patient “absorbs” the energy andundergoes a change of state. The instrument is a physical object which ismanipulated by the agent and serves as the “intermediary” in the transmission ofenergy. The experiencer is a person who is engaged in mental activity, and themover (or theme) is an entity which changes its location.As already observed, the canonical event model represents the prototypicaltransitive construction: the participant who is typically an agent initiates the energy2 For a detailed discussion of the absolute construal of events, see Langacker (1991); for adiscussion of the absolute construal of the się-constructions in Polish, see Kardela (2000, 2009).

4 The Psychological Reality of Grammar. A Cognitive Linguistics Perspective 55which is transmitted onto the patient. However, there exist constructions whichdepart from their transitive prototype in that they form a hierarchical organizationin which each instance of such departure represents a “less transitive” type. Thedepartures from the prototype form a cline, represented by the transitivity hierarchy(Maldonado 1992: 63).8. The transitivity hierarchyTransitive > reflexive > oblique intransitive > middle > intransitive > absoluteThe hierarchy in (8) can be exemplified by the following examples from Polish:9. Marysia zapamiętała słowa Piotra. (transitive)‘Mary remembered Peter’s words.’10. (Ja) zapamiętałem siebie jako małego chłopca skorego do bójki. (reflexive)‘I remembered myself as a boy ready to fight.’11. Piotr długo zastanawiał się nad problemem przeludnienia. (obliqueintransitive)‘Peter reflected on the problem of overpopulation for a long time.’12. (Ja) pamiętałem o jego słowach. (middle)‘I remembered (of) his words.’13. Długo myślałem zanim napisałem pierwsze zdanie. (absolute)‘I reflected a lot before writing the first sentence.’According to (8), the most prototypical transitive construction is (9), since theenergy flows from the agent (Marysia) to the patient (słowa Piotra ‘Peter’s words’).A less prototypical construction is (10), in which one participant is coded by boththe subject, ja ‘I’, and the reflexive siebie ‘myself’. The least prototypical is (13), inwhich only one participant appears.We can now introduce the Dagentivisation Parameter.14. The Deagentivisation Parameter:The use of się signals the departure from a prototypical transitive construction(with a (human) agent and a patient).In accordance with (14), out of the following examples, the sentence in (18)represents a fully deagentivized construction.

56 H. Kardela15. Jan uderzył Piotra.‘John hit Peter.’16. Mężczyźni pozdrawiali się/siebie nawzajem.‘The men greeted each other.’17. Piotr widział siebie/się w lustrze.‘Peter saw himself in the mirror.’18. Lód stopił się.‘The ice melted-refl.’The greatest degree of deagentivization of (18) is due to the fact that this sentencecodes an event in which the last element of the energy chain is profiled, withthe agent and instrument being present in the base, but remaining unprofiled(cf. Langacker 1991).This brings us to our initial problem, i.e. how to establish the “realizationmappings” between the parameters on the reference point relationship for reflexivesand a PDP system. Suppose that, as already mentioned, the parameters for thereference point configuration for reflexives, formulated in (7), have, in a PDP system,their analogs in the form of weights. If so, the prototypicality effects in the case of thereference point relationship for reflexives should be seen as corresponding to theactivation values-as-reflected-in-the-weights of the PDP system. This being the case,a diagrammatic presentation of a computer-simulated program for the “realizationmappings” of the parameters on reflexives to the PDP system of weights might lookas depicted in Fig. 4.4 (arrows symbolize realization mappings). 34.7 Conclusion and Prospects for Further ResearchIt should be clear that the psychological reality of grammar (and its explanatoryadequacy), defined either in terms of language acquisition theories or in terms ofJoan Bresnan’s “realization mappings” to a language processing model (like PDP)has no direct link with the actual behaviour of the brain when engaged in informationprocessing. This is so because all connectionist theories are computersimulations of what may be going on in the actual biological brain milieu and are,3 Note that we say nothing here of what exactly a computer-simulated neural network programshould look like. We have nothing to say about the realization mappings of the sort presented inMichael Fortescue’s (2009) study A neural network model of lexical organization; nor do we claimto have aspired and developed in this paper an actual program of the sort proposed by Rummelhartand McClelland (1986) to account for the acquisition of linguistic structure by the child. Rather,we have attempted to show what kind of theorizing might lie behind an attempt to associate thelinguistic parameters on the reference point phenomena for reflexives as presented in Sect. 4.6 withtheir analogs in a PDP system.

4 The Psychological Reality of Grammar. A Cognitive Linguistics Perspective 57weightsweightsweightsweigthsweightsweightsweightsweights weights weightsAccessibility ofReference PointsParameterDeagentivisationParameterParameterXParameterY(i) The SyntacticScale: The accessibilityof thesubject of a verbfor a reflexivevaries accordingto the syntacticrole of the anaphor.transitive > reflexive> obliqueintransitive> middle >intransitive >absolute………………(ii) The SemanticScale: The accessibilityof thesubject of a verbfor a reflexivevaries accordingto the semantic/discoursenatureof the reflexive.Fig. 4.4 A diagrammatic presentation of a computer-simulated program for the “realizationmappings” of the parameters on reflexives to the PDP system of weightsfrom this point of view, treated as so-called neural bridging theories. This is not tosay, of course, that the theoretical import of such theories is insignificant. SaysTomasz Krzeszowski (2010: 48–49; trans. mine):

58 H. KardelaModels based on [the neural theory of language] simulate the ways in which neuralconnections perform simple cognitive tasks such as the acquisition of concepts and spatialterms (cf. Regier 1996), the acquisition of verbs expressing hand movement (Bailey 1997)or, finally, some issues pertaining to the meaning of conceptual metaphors. Neural bridgingtheories are criticized by their critics on the grounds that they describe the behaviour ofvirtual connectionist models and not the behaviour of the brains of concrete individuals. Bythe same token, the bridging theories – while attempting to define the mechanism of neuralconnections – completely ignore the issue of their localization in the brain. It is hard tooverestimate the theoretical value of such models even though they have to be verified onthe grounds of research on real brains.It is a verification of bridging theories by neurolinguistic brain research thatcould bring a real breakthrough in answering the fundamental questions about therelation between the real brain processing, language and reasoning, includingthe following (Krzeszowski 2010: 48; see also Wróbel 2003 for related discussion):(1) In what way are the neurons in the brain, linked to each other and with the wholeof the human body, related to reasoning or, just to plain thoughts? (2) What do thecomputational mechanisms underlying neurons which are responsible for thinkinglook like? (3) What is the import of the fact that the neural connections in the brainare also connected with other parts of the body? (4) How does the neural structurelearn, reflect and relate to conceptual structure and to language corresponding toit? Until these questions are satisfactorily answered, the psychological reality ofgrammar is bound to be defined in terms of the formulaicity-based theories oflanguage acquisition or the “realization mappings” à la Bresnan involving neuralbridging theories of the PDP type.ReferencesBailey, D. 1997. When push comes to shove. A computational model of the role of motor controlin the acquisition of active verbs. Unpublished PhD dissertation, University of California atBerkeley.Bechtel, W. and A. Abrahamsen. 1991. Connectionism and the mind: An introduction to parallelprocessing in networks. Oxford: Blackwell.Becker, J. 1975. The phrasal lexicon. In Bolt Beranek & Newman Report no. 3081, AI Report no.28. Reprinted in Theoretical issues in natural language processing, eds. R. Shank and B. L.Nash-Webber, 60-63. Cambridge, MA: Bolt Beranek and Newman.Bresnan, J. 1978. A realistic transformational grammar. In Linguistic theory and psychologicalreality, eds. M. Halle, J. Bresnan and G. Miller, 1-59. Cambridge, MA: The MIT Press.Cattel, R. 2006. An introduction to mind, consciousness and language. London: Continuum.Chomsky, N. 1965. Aspects of the theory of syntax. Cambridge, MA: The MIT Press.Chomsky, N. 1981. Lectures on government and binding. Dordrecht: Foris Publications.Chomsky, N. 1986. Knowledge of language: Its nature, origin and use. New York: Praeger.Dąbrowska E. 2004. Language, mind and brain. Some psychological and neurological constraintson theories of grammar. Edinburgh: Edinburgh University Press.Deane, P. 1992. Grammar in mind and brain. Berlin: Mouton de Gruyter.Fauconnier, G. 1985. Mental spaces: Aspects of meaning construction in natural language.Cambridge, MA: The MIT Press.Fauconnier, G. and M. Turner. 2002. The way we think. Conceptual blending and the mind’shidden complexities. New York: Basic Books.

4 The Psychological Reality of Grammar. A Cognitive Linguistics Perspective 59Fortescue, M. 2009. A neural network model of lexical organisation. London: Continuum.Kardela, H. 2000. Dimensions and parameters in grammar. A study in A/D asymmetries andsubjectivity relations in Polish. Lublin: Maria Skłodowska Curie University Press.Kardela, H. 2009. Cognitive grammar as a parametrized continuum of linguistic units. InNew pathways in linguistics, eds. S. Puppel and M. Bogusławska-Tafelska. Olsztyn:Uniwersytet Warmińsko-Mazurski.Krzeszowski, T. 2010. Perspektywy semantyki. In Studia językoznawcze: od językoznawstwateoretycznego do stosowanego, ed. J. Fisiak. Kraków: Tertium.Kuno, S. 1987. Functional syntax. Harvard University Press.Langacker, R. 1987. Foundations of cognitive grammar. Vol. 1. Theoretical prerequisites.Stanford: Stanford University Press.Langacker, R. 1991. Foundations of cognitive grammar. Vol. 2. Descriptive application. Stanford:Stanford University Press.Langacker, R. 1999. Grammar and conceptualization. Berlin: Mouton de Gruyter.Langacker, R. 2000. A dynamic usage-based model. In Usage-based models of language, eds. M.Barlow and S. Kemmer, 1-63. Stanford: CSLI Publications.MacWhinney, R. 1995. The CHILDES project: Tools for analyzing talk. Hillsdale, NJ: LawrenceErlbaum.Maldonado, R. 1992. Middle voice. The case of Spanish ‘SE’. Unpublished PhD dissertation,University of California, San Diego.Perkins, M. R. 1999. Productivity and formulaicity in language development. In Issues in normal& disordered child language: From phonology to narrative. Special Issue of The NewBulmershe Papers, eds. M. Garman, C. Letts, B. Richards, C. Schelletter and S. Edwards,51-67. Reading: University of Reading.Radford, A. 1981. Transformational syntax. Cambridge: Cambridge University Press.Radford, A. 1988. Transformational grammar. A first course. Cambridge: Cambridge UniversityPress.Regier, T. 1996. The human semantic potential: Spatial language and constrained connectionism.Cambridge, MA: The MIT Press.Rummelhart, D. and L. McClelland. 1986. Parallel distributed processing: Explorations in themicrostructure of cognition. Cambridge, MA: The MIT Press.Segalowitz, N. 1997. Individual differences in second language acquisition. In Tutorials inbilingualism: Psycholinguistic perspectives, eds. A. M. B. Groot and J. F. Kroll, 85-112.Mahwah, NJ: Lawrence Erlbaum.Stufflebeam, R. 2006. Connectionism: An introduction. http://www.mind.ilstu.edu/curriculum.Accessed 15 August 2010.Tomasello, M. 1995. Language is not an instinct. Cognitive Development 10: 131-156.Tomasello, M. 2000. Do young children have adult syntactic competence? Cognition 74: 209-253.Wray, A. 2002. Formulaic language and the lexicon. Cambridge: Cambridge University Press.Wróbel, S. 2003. Spór inżynierów: koneksjonizm a klasyczna architektura umysłu. PrzeglądFilozoficzno Literacki 4.

Chapter 5A Morphologist’s Perspective on “EventStructure Theory” of NominalizationsMaria Bloch-TrojnarAbstract The paper pinpoints problems of a morpheme-based approach to deverbalnominalizations as advocated by Grimshaw (1990) and Alexiadou andGrimshaw (2008). It is argued that nominalizations are capable of exhibiting theprocess/result dichotomy regardless of their formal exponent and there are twotypes of regular nominals which differ in terms of their aspectual characteristics.Nominalizations in -ing, which in Grimshaw’s model are regarded as ComplexEvent Nominals analyzable in terms of aspectual distinctions and having anassociated argument structure, may display the process/result dichotomy andsome concrete -ing formations can pluralize (building). Zero nominalizations, inturn, which are regarded as Simple Event Nominals with no a-structure, aredemonstrated to be capable of appearing with satellite phrases corresponding toverbal arguments. They are telic, i.e. denote events and therefore have an aspectualanalysis. Regardless of their actional interpretation, Grimshaw treats zeroderivatives on a par with result nominalizations because they are countable andcan pluralize, whereas, in fact, countability points to their telic character. Beingtelic they do not accept aspectual modifiers (for an hour, in an hour) or modifierslike frequent or constant, unless they are in the plural. Also their limited ability toaccept NP satellite phrases may be explained in terms of their aspectualcharacteristics. Consequently, the properties of nominalizations are a reflection ofthe properties of two productive lexical rules which generate countable anduncountable nominalizations and lexicalization phenomena. Formal exponentsshould be regarded merely as clues rather than determinants of meaning/function.M. Bloch-Trojnar (*)John Paul II Catholic University of Lublin, Lublin, Polande-mail: bloch@kul.plM. Pawlak and J. Bielak (eds.), New Perspectives in Language,Discourse and Translation Studies, Second Language Learning and Teaching,DOI 10.1007/978-3-642-20083-0_5, # Springer-Verlag Berlin Heidelberg 201161

62 M. Bloch-Trojnar5.1 IntroductionThis paper is an appraisal of a morpheme-based approach to deverbalnominalizations as advocated by Grimshaw (1990) and Alexiadou and Grimshaw(2008). 1Grimshaw’s model (1990) undoubtedly pinpoints the differences in the syntacticpatterning of regular and lexicalized deverbal nominalizations and there is merit tothe idea that the analysis of nominalizations should take into account aspectualdistinctions. However, she attributes the properties of nominalizations to the affixesinvolved, which results in an inaccurate description of linguistic facts and hasinsufficient explanatory potential. By introducing the level of Argument Structure(henceforth AS), Grimshaw (1990) acknowledges the lack of isomorphism betweensemantic and syntactic structures. 2 However, she strictly adheres to isomorphismbetween lexical and grammatical categories and their phonological manifestation.In her approach bound morphemes have lexical entries and morphologicaloperations involve concatenation of morphemes in accordance with their subcategorizationrestrictions.In the model of Lexeme Morpheme Base Morphology (Beard 1995) advocatedhere “semantics, derivation, and affixation represent three distinct levels ofmorphological operations, which require two distinct mapping systems” (Beard1998: 55). The rules determining the phonological representation of bound1 By choosing the “event structure theory” of Grimshaw we also refer indirectly to the “structuremodel”, in which nouns that have Argument Structure in their representation contain VPs and/orverbal functional layers (e.g. Fu et al. 2001; Alexiadou 2001, 2009; Roeper 2005; Harley 2009).Alexiadou and Grimshaw (2008: 10) compare both models and conclude that “the principaldifficulties arise from the non-uniformity of deverbal nominalization patterns: different affixationtypes exhibit different behaviour. The successes and failures of the two models occur on exactlythe same questions. What one describes, the other describes. What one fails to explain, the otherfails to explain, and for fundamentally the same reasons.”2 The lexical representation of verbs has two facets, i.e. syntactic representation (ArgumentStructure) and semantic representation, which following Hale and Keyser (1986) has come to beknown as the “lexical conceptual structure” (LCS). LCS serves as a representation of the verb’smeaning and among others specifies semantic relations obtaining between participants and spatiotemporalorganization of the event. Despite close correlation between the two structures it isimportant to keep them distinct since there is no simple one-to-one mapping of theta roles andvarious syntactic roles or surface cases. Only true arguments (Pustejovsky 1995: 63–64) arenecessarily mapped onto/expressed as syntactic constituents (John arrived late), whereas defaultarguments (John built a house out of bricks), and shadow arguments (Mary buttered her toast withan expensive butter) are not. The opposite situation is just as conceivable, i.e. there may be verbswith supernumerary syntactic arguments which contribute nothing to the semantics, e.g. reflexivearguments of perjure oneself, avail oneself of X. Furthermore, synonymous verbs may feature indifferent syntactic configurations, e.g. replace X with Y, substitute Y for X (Jackendoff 2010:16–17). The level of Argument Structure introduced by Grimshaw (1990) is that part of the lexicalentry which interfaces LCS representation and deep structure. It is derived from Lexical-ConceptualStructure via thematic and aspectual hierarchy.

5 A Morphologist’s Perspective on “Event Structure Theory” of Nominalizations 63morphemes are independent of the rules targeting the lexical or morphosyntacticrepresentation, which is known as the Separation Hypothesis. 3In what follows we shall demonstrate that nominalizations are capable ofexhibiting the process/result dichotomy regardless of their formal exponent andthat there are two types of regular nominals which differ in terms of their aspectualcharacteristics. Consequently, the properties of nominalizations are a reflection ofthe properties of two productive lexical rules which generate countable anduncountable nominalizations and lexicalization phenomena. Formal markersshould be regarded merely as clues rather than determinants of meaning/function.5.2 Event Structure Model5.2.1 Grimshaw (1990)Grimshaw (1990) distinguishes two types of deverbal nominalizations, which differin their Argument Structure licensing potential. There are complex event nominals(CE nominals) which are analyzable in terms of aspectual distinctions and whichhave an associated AS like verbs, and simple event or result nominals which lackboth. In contradistinction to verbs the external argument in CE nominalizations isoptional. If the external argument is present (either as a NP in the genitive case or aby-PP), the internal argument is obligatory.1. Complex Event Nominalsshooting, destruction, assignment(The enemy’s) destruction of the cityThe destruction of the city (by the enemy)The shooting of rabbits by BillSimple Event/Result Nominalsdestruction, assignment, buy, offer, race, war, storm, tripThe destruction was complete.A good buy/race*A good buy of clothesGerundive nominals possess all properties typical of CE nominals, i.e. theyallow only the definite determiner thus precluding the indefinite article, the3 Apart from Beard (1976, 1985, 1995) it is also argued for by Laskowski (1981), Szymanek(1985), Malicka-Kleparska (1985, 1988), Aronoff (1994), and Bloch-Trojnar (2006). Lexical andsyntactic rules are abstract operations, which apply to the grammatical representation of a lexeme,whereas affixation and other morphological processes (prosodic variation, internal modification,∅, etc.) are effected in an autonomous postsyntactic Morphological Spelling Component.

64 M. Bloch-Trojnardemonstrative or pronominal one, never pluralize and never occur predicatively.The opposite is true of simple event/result nominals.2. Complex Event NominalsThe shooting of rabbits is illegal.*A/one/that shooting of rabbits is illegal.*The shootings of rabbits are illegal.*That was the shooting of rabbits.Simple Event/Result NominalsThey studied the/an/one/that assignment.The assignments were long.That was the/an assignment.The varying patterns and differences in terms of determiners and modifiers that thetwo types of nominals allow are attributed to the lexical specification of the affixesinvolved and the type of their external argument, i.e. Ev or R (Williams 1981; diSciullo and Williams 1987; Higginbotham 1985). The nominal suffix -ing isassociated with AS with an external non-thematic argument Ev, Latinate suffixeslike -ation or -ment are specified for either Ev or R, whereas zero derivationintroduces R and zero-derived forms are never CE nominals, as depictedschematically below.3. Complex Event Nominals Simple Event/Result NominalsLatinate suffixes;Latinate suffixes;-ing – N, (Ev)Zero derivatives – N, (R)shooting N, (Ev (x(y)))observation N, (Ev (x(y)))observation N, (R)offer N, (R)5.2.2 RefinementsIn Alexiadou and Grimshaw (2008: 2) and Alexiadou (2009) we find a slightlymodified version of this approach, where nouns like examination are three wayambiguous, i.e. they have a complex event reading, simple event reading and aresult reading.4. (a) The examination of the patients took a long time. (Complex Event nominal)(b) The examination took a long time.(Simple Event nominal)(c) The examination was on the table.(Result nominal)Notably, the basis for this classification is their ability to take obligatoryarguments, license event-related PPs and the ability to pluralize. CE nominalsbehave like verbs since they license event-related PPs (in an hour, for an hour),they “have arguments which are obligatorily present” (Alexiadou and Grimshaw2008: 2) and they cannot be made plural. On a “simple event” reading, like CE

5 A Morphologist’s Perspective on “Event Structure Theory” of Nominalizations 65nominals they denote an event, but are not associated with an event structure andhence lack argument structure because they do not license event-related PPs andadmit plural formation. Result nominals do not denote events and behave like nonderivednouns (dog, event).They maintain that “nominals derived from verbs with no (overt) affix behave assimple event nouns and/or individual nouns”. Hence the ungrammaticality of thefollowing:5. *The constant offer of credit cards to students...*(The) frequent report of looting ...Furthermore, bare nominals such as stop, end, change, unlike their Latinatecounterparts, do not nominalize transitively (Smith 1972):6. The climate’s change*Global warming’s change of the climateThis is to be taken as evidence that “zero derived nominalization never preservesevent structure” (Alexiadou and Grimshaw 2008).These and other differences between the two types are concisely summarized anddiscussed in Alexiadou and Grimshaw (2008: 3) and some of them are listed below:7. Complex Event Nominals Simple Event/Result NominalsObligatory argumentsNo obligatory argumentsEvent readingNo event readingAspectual modifiersNo aspectual modifiersModifiers like frequent, constant appear with Modifiers like frequent, constant only withsingularpluralMust be singularMay be pluralZero affixation is opaque to argument transfer, the suffix -ing is argumentpreserving, whereas -(a)tion and -ment are ambiguous/unspecified. 48.“Zero” affixation[N [V] ∅] (Aspect: ...)Affixation with -ing[N [V] -ing] (Aspect: telic/atelic/ ...)Affixation with -(a)tion; -ment[N [V] -(a)tion] (Aspect: ...)[N [V] -(a)tion] (Aspect: telic/atelic/ ...)4 In the “structure model” argument taking properties of nominalizations depend on where the affixis attached: -ing is specified for “high/outer cycle attachment” and contains verbal projections, ∅is specified for “low/root attachment”, contains no verbal projections and, therefore, has noargument structure, whereas Latinate suffixes can be doubly specified or underspecified.

66 M. Bloch-Trojnar5.2.3 ProblemsAlexiadou and Grimshaw (2008) admit that there is no formal apparatus to predictwhy specific morphemes have the properties they have. In their approach identicalN–V pairs are regarded as zero affixation to maximize parallels with other affixationprocesses, despite the fact that an affix which is null is cognitively lessplausible than no affixation at all. They need a separate lexical entry which isphonologically null since if there was no affix the derivative would lack the head.They are also puzzled by the fact that zero-nominals which look most like verbs donot preserve AS.The first two of these problems are theory internal, and can be circumvented if weopt for the Separation Hypothesis. Then, it becomes a matter of lexical accidencethat a particular meaning is mapped onto a particular affix or group of affixes. Thereis also no need for zero affixes. On this view, morphology is not merely concatenationof morphemes. What matters is the system of relations or contrasts thatmorphemes create. If there is a number of co-functional affixes, no marking at allalso performs a contrastive function. However, there are far more serious problemswith Alexiadou and Grimshaw’s proposal. A closer look at linguistic facts makestheir division of affixes difficult to uphold. It turns out that by rights allformal exponents should be regarded as ambiguous between an actional and resultinterpretation since zero-derived nominals display an actional reading (albeit differentfrom that of -ing nominalizations) and -ing derivatives display lexicalizedsenses. Furthermore, zero-derived nominals can be accompanied by NPs whichcorrespond to NP arguments of the verbal base whereas the arguments of -ingnominals aren’t as obligatory as they are professed to be. In a nutshell, the propertiesof nominalizations are not a direct consequence of the type of phonological markerthat they bear.5.3 The Data5.3.1 Configuration PropertiesIn their classification, Grimshaw (1990) and Alexiadou and Grimshaw (2008) insiston the obligatoriness of arguments in CE nominals and the treatment of zeroderivatives as lacking an aspectual interpretation and AS. Alexiadou (2009)weakens the former statement and speaks in favour of optionality of licensing ofAS in the nominal system but the latter “is a strong tendency, which still awaits anexplanation” (Alexiadou 2009: 257). In his critique of Alexiadou (2009), Bierwisch(2009) notes that “the distinction between simple and complex nominals rests onthe appearance of complements realizing argument positions. If however, argumentpositions of nouns are generally optional, as must be assumed for independent

5 A Morphologist’s Perspective on “Event Structure Theory” of Nominalizations 67reasons, the distinction between the absence of a position in AS and an unrealizedoptional complement becomes spurious, and the distinction between simple andcomplex event nouns collapses” (Bierwisch 2009: 22). Thus, the differencebetween simple and complex nominalizations can be attributed to the optionalityof nominal complement positions and not the presence or absence of AS.If nominalization is a lexical process, in which a verb is turned into a noun, thestructures with derived nominals should resemble structures with morphologicallysimple nouns, i.e. there is no reason why their complements should be obligatoryas in their verbal bases. Deverbal nominalizations depending on their numberspecification are preceded by appropriate determiners, and are modified byadjectives and prepositional phrase complements, which are optional. A fairamount of evidence has accumulated, which can be used in support of thisapproach.Cetnarowska (1993: 71–84) investigates the inheritance of Predicate ArgumentStructures in bare nominalizations outside complex predicates and concludes that itis effected in actional readings as in, e.g.9. The purchase by India of howitzers from the Swedish firm of BoforsThis would imply that contrary to Grimshaw zero derivatives are ambiguousbetween argument-taking and non-argument-taking readings just as Latinatenominalizations. Consider the example below and contrast it with the one in (6)above:10. Climate change – generally refers to the change of the climate by man’s emissionof Greenhouse gasses 5Alexiadou and Grimshaw (2008: 3) do not attempt to conceal cracks in theiranalysis by citing David Embick’s examples, in which zero nominals show a fullrange of arguments and co-occur with modifiers like frequent and constant, afeature of CE nominals (Grimshaw 1990: 67). Even more examples can be foundin Alexiadou (2009: 257) from Newmeyer (2009) and Harley (2009):11. My constant change of mentorsThe frequent release of the prisoners by the governorAn officer’s too frequent discharge of a firearmThe ancient Greek’s practice of infanticideAlexiadou (2009: 257) following Newmeyer and Harley suggests that only zeroderived nominals from French/Latinate roots allow CE interpretation. However, inWierzbicka (1982) we find examples of native bare nominals where NPs in thegenitive correspond to the direct object of the verb.5 http://www.astralweb.co.uk/smart-meters-glossary.html.

68 M. Bloch-Trojnar12. John had a lick of Mary’s ice cream.Johnny let Jimmy have a throw of his new boomerang.He went outside and had a kick of his football.May I have a borrow of your pencil?Have a sip of my wine.Newmeyer (2005) in his critique of Roeper’s (2005) article demonstrateshow arguments adduced to argue for CE status of suffixed nominals can be appliedto zero-derivatives. Examples where they occur with an anaphor do so are to pointto the presence of a VP node in nominalizations:13. John’s destruction of the city and Bill’s doing so tooAmerica’s attack on Iraq was even less justified than the latter’s doing so to Kuwait.Randall’s (1988: 137) proposal that a deverbal noun derived from a transitiveverb without an accompanying prepositional thematic phrase must receive a nonactionalinterpretation is not corroborated by the facts, e.g.14. “But it’s not the sex I miss so much”, she continues, “it’s the kissing...”.In hospital she’ll be under observation all the time.Cetnarowska (1993: 71, 43–44) gives many examples of actional barenominalizations without thematic prepositional phrases:15. She drenched my face with a kiss.A good throw was answered by a good catch.In the literature we find other examples, which show the erratic behaviour of CEnominalizations, where an -ing nominal is preceded by the indefinite article and theplural is accompanied by arguments:16. There was a capsizing of the boat by Mary (Mourelatos 1978: 425).The climber’s two ascents of the mountain (Filip 1993: 89).In sum, in the process of transposition nouns inherit LCS rather than AS of baseverbs. Event participants may appear as optional nominal complements. That iswhy the actional reading is available if there are no satellite phrases and it ispossible to find examples where zero derived nominals are accompanied byNPs corresponding to arguments of the base verb. It is true that non-countnominalizations are far more likely to occur with satellite phrases and aspectualmodifiers than their zero-derived count counterparts. Bare nominals occur mostly incomplex predicates/light verb constructions (LVCs), in informal registers withnative nominalized items (Adams 2001: 29). Stylistic considerations aside, it willbe argued below that their reluctance to occur with satellite NPs has to do withtheir aspectual characteristics and the presence of suffixed nominals based on thesame root.

5 A Morphologist’s Perspective on “Event Structure Theory” of Nominalizations 695.3.2 Aspectual Characteristics and the Ability to Pluralizeand License Satellite PhrasesThe analogy between count entities and events, on the one hand, and mass entitiesand activities, on the other has been of keen interest not only to linguists but also tophilosophers and cognitive psychologists. This analogy is established on ontological,semantic and syntactic grounds (Mourelatos 1978; Langacker 1987; Krifka1992; Brinton 1998; Willim 2006 and the references therein).Both events (i.e. telic situations) and things or objects (the referents of countNPs) are intrinsically countable and not subdivisible. Both have precise limits anddelimit time or space (Krifka 1992: 30). Note the count modifiers with eventnominalizations.17. Vesuvius erupted three times. > There were three eruptions of Vesuvius.Mary capsized the boat (once). > There was a capsizing of the boat by Mary(Mourelatos 1978: 425).Continuities (i.e. atelic situations) and “stuffs” (the referents of mass NPs) arenon-countable. They are homogenous, they lack precise limits, and they “fill” timeor space (Krifka 1992: 30).18. Jones pushed the cart for hours. > For hours there was (*a) pushing of the cartby Jones (Mourelatos 1978: 427).Brinton (1998) investigates the mapping of aktionsart properties of a verb to acorresponding nominalization and concludes that unlike Latinate suffixes, -ingnominals and ∅-derivatives fail to preserve the aktionsart of the verb. Whereasthe -ing suffix “has the effect of converting the situation into an activity, of makingthe situation durative, atelic and dynamic” (Brinton 1998: 48), conversion “is ameans of converting the situation into an event (an accomplishment, achievement,or semelfactive) by adding the feature of telicity” (Brinton 1998: 50).In line with Wierzbicka (1982), Brinton (1998: 50) regards “complexpredicates” an important means of telicizing activities, “that is, of convertingactivities into accomplishments or achievements, without the necessity of statingan explicit goal”. Willim (2006: 119) also argues for their singulative or partitiveeffect and observes that in English when the nominal in the LVC is derived from apunctual verb a single event is denoted, whereas those based on homogenousactivity verbs refer to an extended activity with an endpoint in its semanticrepresentation (give a jump vs. have a walk).The telicizing nature of LVCs in English is achieved by the use of countablezero-derived nominalizations, which carry the semantic load of the predicate andimpose clear spatiotemporal limits on the situation denoted by the verb. Telicactivities can be multiplied whereas count nouns pluralized. Brinton (1998: 50)explains: “the result of multiplying situations (no matter what their type) a specific

70 M. Bloch-Trojnarnumber of times is a situation of the accomplishment type”. Zero derivatives canco-occur with cardinal numbers, enumerative determiners and take plural morphology(Cetnarowska 1993: 43).19. Give it a few kicks.Have another guess.Take a second look.I took several rides in his car.Can I have two guesses?Adverbials of “span” or “duration” denote a specific quantity of an atelicsituation (20a). With in-adverbials atelic situations also receive a telic interpretation(20b). LVCs (20c) are incompatible with both types (Brinton 1998: 48, 51;Willim 2006: 119; Filip 1993: 113).20. (a) He walked for ten hours.(b) He swam in an hour.(c) *He had a walk for ten hours./*He had a swim in an hour.Willim (2006) explains that if have a walk is telic, i.e. its semantic representationcontains an endpoint, it is not possible to combine it with a delimiting expressionwith an independent endpoint, as this would violate the “one delimitation per event”constraint put forward by Tenny (1994: 79).Being telic, zero derivatives accept neither aspectual modifiers (for an hour, inan hour) nor modifiers like frequent or constant, unless they are in the plural. In hercharacterization of the progressive, Brinton (1988: 9) notes that this form “portraysthe situation as progressing, that is continuing, ongoing or developing, at a certaintime. Normally, a situation so portrayed must be durative in nature, though punctualsituations, if repeated, may be viewed as ongoing”. By parity of reasoning, atelicnominalizations can occur with constant and frequent, whereas telicnominalizations are incompatible, unless they are pluralized to imply repetition.Also their limited ability to accept NP satellite phrases may be connected withtheir aspectual characteristics. Brinton (1998: 38–9) observes that “the entire VPenters into the expression of aktionsart”, e.g. run is an activity verb (atelic), but thepredicate run (home, to the corner) contains an endpoint/goal and is thus anaccomplishment (telic). For this reason atelic nominalizations may be more likelyto accept satellite phrases that correspond to the internal argument which measuresout or delimits the event denoted by the base verb. Nominalizations which aredelimited (telic) show no need for further individuation.5.3.3 Semantic Considerations and Formal MarkersNominals with the suffix -ing, which are considered CENs, may display theprocess/result dichotomy, in which case some concrete -ing formations can pluralize(building, warning, painting, filling (in a tooth), recording) and as such they do

5 A Morphologist’s Perspective on “Event Structure Theory” of Nominalizations 71not require obligatory satellite phrases corresponding to the arguments of the baseverb. The nominalization in (21a) can only denote a concrete result (Quirk et al.1985: 1290), whereas that in (21b) is ambiguous.21. (a) Brown’s paintings of his daughter(b) Brown’s painting of his daughterIt can be interpreted as a CE nominal (22a) or a result nominal (22b).22. (a) Brown’s painting of his daughter is a pleasure to watch.(b) Brown’s painting of his daughter is a pleasure to look at.This would imply that the interpretation is context dependent and is not aninherent property of the affix involved. The meaning of action nouns, regardlessof the derivational type they belong to, is subject to the inexorable process oflexicalization (cf. Marchand 1969: 303; Malicka-Kleparska 1988). Consider someexamples of lexicalized senses of each type:23. -ing Latinate suffixes Øbuilding(s) possession(s) cut(s)filling(s) in a tooth solution(s) find(s)warning(s) government(s) exhibit(s)stuffing condensation deposit(s)roofingdisposalattendanceIt should be borne in mind that lexicalized senses are not equally frequent withall markers. Latinate nominals are regarded as lexically listed and described interms of redundancy statements (Malicka-Kleparska 1988), whereas -ing and zeroderivatives are generally regarded as productive, hence generated by rule (Malicka-Kleparska 1988; Cetnarowska 1993; Bauer 2001).With regard to semantics, traditionally we talk about the process/result dichotomy,whereas in fact nominalizations display two kinds of actional reading, i.e. -ingnominalizations are interpreted as “action or process of V-ing”, whereas the NomenActi reading, i.e. “a single instance of V-ing” is prevalent in zero derivatives(Cetnarowska 1993: 112–113; Adams 2001: 28–29). Latinate nominalizations areregarded as ambiguous, but this ambiguity is more complex than generallyassumed. The difference between CE and SE nominals does not lie in theobligatoriness of arguments but in different aspectual characteristics.Quirk et al. (1985: 1551) suggest that there is an aspectual contrast between thenominalisations in -ation, -ment, etc. and the -ing verbal noun, with the formerreferring to actions in their entirety, including their completion.24. His exploration of the mountain took/will take three weeks.His exploring of the mountain is taking a long time.Malicka-Kleparska (1988: 30) refutes this argument by demonstrating that thedifference is due exclusively to the context, and provides examples of non-ing

72 M. Bloch-Trojnarnominalizations which do not imply completion. This, in turn, shows that suffixednominalizations can denote a process or an action in its entirety without losing theability to license arguments.25. The punishment of the boy is taking a long time.The organization of the party is taking a long time.The placement of the stone is taking a long time.The position that Latinate nominalizations are in fact three-way ambiguous is alsoadvocated by Pustejovsky (1995: 170). In his analysis, Latinate nominalizations arelogically polysemous items, which form a lexical conceptual paradigm (lcp)constructed from two base types with three senses available (cf. Pustejovsky andAnick 1988):26. Process.result_lcp ¼ {process.result, process, result}This means that only nominals which have the process and result reading canhave the additional event interpretation (the dot object reading).27. (a) The house’s construction was finished in two months. (event)(b) The construction was interrupted during the rains. (process)(c) The construction is standing on the next street. (result)That is why -ing nominalizations as in (28) below cannot be polysemous; this isbecause they are not interpreted as the result of an event. 628. The launching of the space shuttle occurred at 10.30.The launching of the space shuttle was aborted.To this we could add that there is a zero-derived noun which has the semantics tofill the remaining slots in the lcp of the same verbal stem. It has the event reading(the dot object) and the object reading. Consider the examples from CCAD (note theof-NP following the nominal):29. This morning’s launch of the space shuttle Colombia has been delayed.The launch of a ship was a big occasion.The captain was on the deck of a launch, steadying the boat for the pilot.Pustejovsky (1995) does not discuss zero derived nominals in any detail. However,he notes that purchase is logically polysemous between the event and theobject involved in the transaction. Thus, we would use the nominal purchasing torefer to the process and purchase to an event.6 Malicka-Kleparska (1988: 165) observes that civilizing and civilization are equivalent andinterchangeable on the actional/process interpretation but not in the lexicalized sense....to attempt the civilization of the Australian aboriginesthe civilizing of the Highlands of Scotland ...the ancient civilizations / *civilizings

5 A Morphologist’s Perspective on “Event Structure Theory” of Nominalizations 7330. (a) The purchasing of office supplies by State Agencies in Mississippi 7The purchase by India of howitzers from the Swedish firm of Bofors(b) For hours there was pushing of the cart by Jones.Information is called up at the push of a button.However, -ing nominals such as launching, purchasing, pushing should becontrasted with nominals such as painting discussed at the beginning of this section.Since the latter display the result reading, the event interpretation should also beavailable, which immediately prevents an event interpretation of the noun paint.31. Brown’s painting of his daughter is taking a long time.Brown’s painting/*paint of his daughter took one evening.Brown’s paintings of his daughterThe derivatives terminating in -ing and bare nominals can be part of one lcpwhere they show an aspectual contrast as demonstrated in (30) above. However,they may be derived from two homophonous lexemes, which differ in transitivity. 8This would explain why some bare nominals do not nominalize transitively (Smith1972). They can be accompanied by an of-NP which corresponds to the subject ofthe verb since it is the only participant in their LCS representation. According toMalicka-Kleparska (1988: 103), there is a strong tendency for -ing to occur withprototypical transitive verbs with the theta grid (Agent, Theme), whereas Marchand(1969: 374) points out that the bulk of bare nominals is derived from intransitiveverbs.32. (a) The campaign for the end of slavery gained momentum in Great Britain ... 9Economic factors were paramount in dictating Britain’s ending of slave carrying fromAfrica in 1807. 10(b) This will usher in the start of World War Three. 11If he supports America’s starting of WW3 with Iran. 12In sum: lexicalization affects lexical items (not classes), which explains why alltypes of nominals regardless of the formal marker may show the result reading(building, civilization, launch). Actional readings are of two types, process andevent. Since only nominals which display the result reading in addition to theprocess reading can be three-way ambiguous, this ambiguity characterizes listedLatinate nominalizations and a fraction of -ing nominals. It follows that nominals7 http://www.procurement.msstate.edu/officesupplies.pdf.8 In Levin (1993: 2–3) many verbal diathesis alternations (e.g. causative-inchotative) are regardedas rule governed conversions, i.e. a particular word is used in two lexical categories, because theyshow clear changes in argument structure, e.g. The window broke. vs. The little boy broke thewindow.9 www.guyana.org/features/guyanastory/chapter44.html.10 http://eh.net/bookreviews/library/0676.11 http://www.threeworldwars.com/world-war-3/ww3.htm.12 http://www.bbc.co.uk/blogs/theeditors/2007/05/prime_ministers_questions.html.

74 M. Bloch-Trojnarwhich denote events and results (zero derivatives) cannot refer to processes. Barenominals will not show an event reading if there is a Latinate nominalization basedon the same verbal root. With native roots a bare nominal and an -ing nominal showan aspectual contrast (launching “atelic” – launch “telic”) or correspond to verbsdiffering in transitivity (ending – end).5.4 ConclusionThe fact that Grimshaw specifies the differences in the syntactic patterning ofactional and lexicalized nominalizations and takes into account aspectualdistinctions shows her model to good advantage. However, it has two flaws.It has been demonstrated that the classification of nominalizations based on theformal markers involved cannot be maintained since nominalizations are capable ofexhibiting the process/result dichotomy regardless of their formal exponent. This isdue to lexicalization which targets items not types. Configurational properties candifferentiate only between regular and lexicalized senses to the effect that LCSparticipantsof the verbal base may accompany regular (actional) nominalizationsbut may not accompany lexicalized ones.Zero nominalizations, in turn, which are regarded as simple event nominals withno a-structure, can appear with satellite phrases corresponding to verbal arguments.They are telic, i.e. denote events (an entire process of verbing) and therefore havean aspectual analysis. Regardless of their actional interpretation, zero derivativesare treated on a par with result nominalizations because they are countable and canpluralize whereas, in fact, countability points to their telic character. Thus, there aretwo types of actional (Complex Event) nominals. There are uncountable/atelicnominals (-ing, Latinate suffixes) and countable/telic nominals (zero derived,Latinate suffixes). The former denote activities and are more likely to beindividuated contextually by means of NPs and PPs, whereas the latter denoteaccomplishments and achievements and due to their aspectual characteristics areless likely to be further individuated in the syntax.ReferencesAdams, V. 2001. Complex words in English. London: Longman.Alexiadou, A. 2001. Functional structure in nominals. Amsterdam: John Benjamins.Alexiadou, A. 2009. On the role of syntactic locality in morphological processes: The case of(Greek) nominals. In Quantification, definiteness and nominalization, eds. A. Giannakidou andM. Rathert, 253–280. Oxford: Oxford University Press.Alexiadou, A. and J. Grimshaw. 2008. Verbs, nouns and affixation. In Working papers of the SFB732 Incremental specification in context 01, ed. F. Sch€afer, 1–16.Aronoff, M. 1994. Morphology by itself. Stems and inflectional classes. Cambridge, MA: MITPress.Bauer, L. 2001. Morphological productivity. Cambridge: Cambridge University Press.

5 A Morphologist’s Perspective on “Event Structure Theory” of Nominalizations 75Beard, R. 1976. A semantically based model of a generative lexical word-formation rule forRussian adjectives. Language 52: 108–120.Beard, R. 1985. Is separation natural? Studia Gramatyczne 7: 119–134.Beard, R. 1995. Lexeme morpheme base morphology. Albany, NY: SUNY Press.Beard, R. 1998. Derivation. In The handbook of morphology, eds. A. Spencer and A. M. Zwicky,44–65. Oxford: Blackwell Publishers Ltd.Bierwisch, M. 2009. Nominalization – lexical and syntactic aspects. In Quantification, definitenessand nominalization, eds. A. Giannakidou and M. Rathert, 281–320. Oxford: Oxford UniversityPress.Bloch-Trojnar, M. 2006. Polyfunctionality in morphology. A study of verbal nouns in ModernIrish. Lublin: Wydawnictwo KUL.Brinton, L. 1988. The development of English aspectual systems. Cambridge: Cambridge UniversityPress.Brinton, L. 1998. Aspectuality and countability: A cross-categorial analogy. English Languageand Linguistics 2: 37–63.Cetnarowska, B. 1993. The syntax, semantics and derivation of bare nominalisations in English.Katowice: Uniwersytet Śląski.Collins cobuild advanced dictionary of English. 2009. Glasgow: Harper Collins Publishers.Di Sciullo, A. M. and E. Williams. 1987. On the definition of word. Cambridge, MA: MIT Press.Filip, H. 1993. Aspect, situation types and nominal reference. Unpublished PhD dissertation,University of California, Berkeley.Fu, J., T. Roeper, and H. Borer. 2001. The VP within process nominals: evidence from adverbs andthe VP anaphor do so. Natural Language and Linguistic Theory 19: 549–582.Grimshaw, J. 1990. Argument structure. Cambridge, MA: MIT Press.Hale, K. and S. J. Keyser. 1986. Some transitivity alternations in English. Lexicon ProjectWorking Papers 7. Cambridge, MA: Center for Cognitive Science, MIT.Harley, H. 2009. The morphology of nominalizations and the syntax of vP. In Quantification,definiteness and nominalization, eds. A. Giannakidou and M. Rathert, 321–343. Oxford:Oxford University Press.Higginbotham, J. 1985. On semantics. Linguistic Inquiry 16: 547–593.Jackendoff, R. 2010. Meaning and the lexicon. The parallel architecture. 1975 – 2010. Oxford:Oxford University Press.Krifka, M. 1992. Thematic relations as links between nominal reference and temporal constitution.In Lexical matters, eds. I. A. Sag and A. Szabolcsi, 29–53. Stanford, CA: Center for the Studyof Language Information.Langacker, R. W. 1987. Nouns and verbs. Language 63: 53–94.Laskowski, R. 1981. Derywacja słowotwórcza. In Pojęcie derywacji w lingwistyce, ed.J. Bartmiński, 107–126. Lublin: Uniwersytet Marii Curie-Skłodowskiej, Zakład JęzykaPolskiego.Levin, B. 1993. English verb classes and alternations. A preliminary investigation. Chicago andLondon: The University of Chicago Press.Malicka-Kleparska, A. 1985. The conditional lexicon in derivational morphology. A study ofdouble motivation in Polish and English. Lublin: Redakcja Wydawnictw KUL.Malicka-Kleparska, A. 1988. Rules and lexicalisations. Selected English nominals. Lublin:Redakcja Wydawnictw KUL.Marchand, H. 1969. The categories and types of present-day English word-formation. A synchronicdiachronicapproach. M€unchen: C. H. Beck.Mourelatos, A. 1978. Events, processes, and states. Linguistics and Philosophy 2: 415–34.Newmeyer, F. 2005. Some remarks on Roeper’s remarks on Chomsky’s ’Remarks’. A commenton Tom Roeper: Chomsky’s remarks and the transformationalist hypothesis. Special issueof SKASE Journal of Theoretical Linguistics: 26–39. http://www.pulib.sk/skase/Volumes/JTL03/04.pdf.

76 M. Bloch-TrojnarNewmeyer, F. J. 2009. Current challenges to the lexicalist hypothesis: An overview and a critique.In Time and again: Papers in honor of D. Terence Langendoen, eds. W. D. Lewis, S. Karimiand H. Harley, 91–117. Amsterdam: John Benjamins.Pustejovsky, J. 1995. The generative lexicon. Cambridge, MA: MIT Press.Pustejovsky, J. and P. Anick. 1988. On the semantic interpretation of nominals. In Proceedingsof the 12 th Conference on Computational Linguistics. Vol. 2, ed. D. Vargha, 518–523.Budapest: Association for Computational Linguistics.Quirk, R., S. Greenbaum, G. Leech and J. Svartvik. 1985. A comprehensive grammar of theEnglish language. London and New York: Longman.Randall, J. 1988. Inheritance. In Thematic relations: Syntax and semantics, ed. W. Wilkins,129–146. San Diego: Academic Press Inc.Roeper, T. 2005. Chomsky’s remarks and the transformationalist hypothesis. In Handbookof word-formation, eds. P. Štekauer and R. Lieber, 125–146. Dordrecht: Springer.Smith, C. 1972. On causative verbs and derived nominals in English. Linguistic Inquiry 3: 36–38.Szymanek, B. 1985. English and Polish adjectives. A study in lexicalist word-formation. Lublin:Redakcja Wydawnictw KUL.Tenny, C. 1994. Aspectual roles and the syntax-semantics interface. Dordrecht: Kluwer.Wierzbicka, A. 1982. Why can you have a drink when you can’t *have an eat? Language58: 753–799.Williams, E. 1981. Argument structure and morphology. Linguistic Review 1: 81–114.Willim, E. 2006. Event, individuation, and countability: A study with special reference to Englishand Polish. Kraków: Jagiellonian University Press.

Chapter 6A Recalcitrant Nature of Object ExperiencersSylwiusz ŻychlińskiAbstract Object experiencers have been the object of study for many years now.The extensive analysis by Beletti and Rizzi (1988) proposed to treat those verbs asunaccusative. This approach was later revisited by Pesetsky (1995), who providedarguments against the wholesale unaccusative treatment of object experiencerverbs, introducing a finer-grained semantic division among them. Recently,Landau (2010) has come up with an analysis which at first blush reconciles thetwo earlier treatments. The crucial part of the analysis concerns the status of verbalpassives. Following a thorough cross-linguistic study, Landau concludes thatlanguages fall into two types with regard to the presence of verbal passives, thefirst type exemplified by languages such as English, Finnish or Dutch, whichpossess verbal passives for eventive verbs only; the second type represented byItalian, French or Hebrew, which have no verbal passives at all. However, this neattypology seems to run into problems in Polish. It appears that Polish, which sharessome properties of each of the two groups, falls out of the proposed classification.This paper briefly reviews the previous treatments of experiencer verbs andhighlights the areas where Polish stands out, putting forth possible ways to refinethe existent analyses.6.1 IntroductionEver since the inception of the generative enterprise, there has been an ongoingconflict between the proponents of syntax as the autonomous module of grammarand the adversaries of such a contention, holding that a continuous recourse to otherS. Żychliński (*)Adam Mickiewicz University, Poznań, Polande-mail: sylwiusz@ifa.amu.edu.plM. Pawlak and J. Bielak (eds.), New Perspectives in Language,Discourse and Translation Studies, Second Language Learning and Teaching,DOI 10.1007/978-3-642-20083-0_6, # Springer-Verlag Berlin Heidelberg 201177

78 S. Żychlińskiareas of cognition, e.g. semantics or pragmatics, is inevitable. After more than halfa century, the disagreement has not been settled. The class of psychologicalpredicates, which is at the core of this article, exemplifies the gist of the disagreementvery well. This recalcitrant group of verbs, which are characterized by thepresence of experiencer and theme thematic roles, and involve the experiencerundergoing a psychological change of state, have long posed a problem forsubsequent analyses because of the mapping/linking problem they display. Overthe course of the last decades, various attempts have been made to disambiguatethis issue, some of which involve purely syntactic machinery, whereas others callfor fine-grained semantics to help explain the superficially surprising arrangementand properties of psych verbs. This paper focuses on a very narrow selection ofproperties of psychological predicates, the status of the experiencer argument beingthe most important among them. As more attention has been devoted to the questionof unaccusativity, which is central to the proper treatment and representation ofargument structure of psychological predicates, it is the class of object experiencer(or ObjExp) verbs that is the primary focus of investigation. The aim of this articleis twofold. On the one hand, it aspires to be a fairly comprehensive overview andcritique of different treatments of ObjExp verbs so far, mainly Belletti and Rizzi’s(1988), Pesetsky’s (1995) and Landau’s (2010). On the other hand, it considershow ObjExp verbs in Polish fit into the unaccusative scenario 1 and, typologically,where they belong (or at least where they certainly do not).6.2 The Beginnings: Psych Predicates and TheirThematic Structure6.2.1 Belletti and Rizzi’s Seminal AnalysisOver the years, psychological predicates have been the object of numerousanalyses. Of the myriads of accounts, Belletti and Rizzi (1988), Grimshaw(1990), Pesetsky (1995) and Landau (2010) are all among recent authors ofcomprehensive theories striving to cast light on and explain the diversity ofdata.1 Polish psych verbs were also studied by, inter alia, Klimek and Rozwadowska (2004) and Biały(2004). Their findings are not included here as the present paper’s main focus is on the difficultiesone meets trying to implement the locative hypothesis into Polish object experiencers.

6 A Recalcitrant Nature of Object Experiencers 79The standard problem with psychological verbs is how to reconcile their thematicstructure with UTAH. 2 The canonical patterns manifested by psych verbs arepresented in (1–3) 3 (Belletti and Rizzi 1988: 291):1. Gianni teme questo.Gianni.Nom fears this.Acc2. Questo preoccupa Gianni.this.Nom worries Gianni.Acc3. (a) A Gianni piace questo.to Gianni.Dat piace questo.Nom(b) Questo piace a Gianni.this.Nom pleases to Gianni.DatAt first blush the same y-roles are distributed arbitrarily, with the Nominativeexperiencer in the subject position and the Accusative theme in the object positionin (1) (frequently referred to as class I of psychological verbs), the Nominativetheme in the subject position and the Accusative experiencer in the object positionin (2) (class II), and the dative experiencer and the Nominative theme in bothorderings in (3) (class III). As such freedom of syntactic positioning is not otherwisewidely attested, a convincing argument to explain this phenomenon was much2 It has been one of the challenges to prove UTAH right or wrong. In its original formulation, it is asfollows:1. Uniformity of Theta Assignment Hypothesis (UTAH)Identical thematic relationships between items are represented by identical structuralrelationships between those items at the level of D-Structure (Baker 1988: 45).A weaker version of UTAH was proposed by Perlmutter and Postal (1984), which reads asUAH.2. Universal Alignment Hypothesis (UAH)There exist principles of UG which predict the initial relation borne by each [argument] in agiven clause from the meaning of the clause.In Zero syntax (1995), Pesetsky argues that it is actually possible to keep UTAH intact on the novelassumption that the repository of thematic roles is governed by finer-grained semantics, making itpossible to distinguish among different types of the theme role.3 Only the Italian examples are cited, but English, Russian or Polish, to mention just a fewlanguages, exemplify a similar pattern, e.g. in Polish:3. Janek boi się tego.John fears this4. To przeraża Janka.This frightens John5. Jankowi podoba się to.To John (Dat) pleases REFL itTo podoba się Jankowi.It pleases REFL to John

80 S. Żychlińskineeded. On the basis of a thorough investigation of Italian facts, Belletti and Rizzi(1988: 293) proposed one uniform underlying representation for sentences (2–3)(they treat (1) in an uncontroversial way, with the surface arguments marking theirDP positions):4.SNPecVpreoccupapiaceVPV’ NPGianniNP a GiannquestoThe diagram in (4) shows that psych verb sentences with different surfacestructures project the same configuration at the underlying level, with theexperiencer asymmetrically c-commanding the theme, formalized in (5) (Bellettiand Rizzi (1988: 344).5. Given a y-grid [Experiencer, Theme], the Experiencer is projected to a higherposition than the Theme.The motivation for movement out of the VP to the subject position comes fromBelletti and Rizzi’s interpretation of Burzio’s Generalization (1988: 332):6. V is a structural Case assigner if it has an external argument.The practical implementation of (6) leads to the conclusion that verbs of thepreoccupare and piacare type (2–3) do not have external arguments, with thesubject position considered as athematic. To derive (2), then, we assume thatthe experiencer gets the inherent Accusative, 4 and the theme by virtue of (6),moves to the subject position to get Case. In (3), since Dative is assigned by apreposition, the experiencer is free to stay in the VP or move to the subjectposition.6.2.2 Landau’s ReanalysisIn The Locative Syntax of Experiencers 5 (Landau 2010), Belletti and Rizzi’sconclusions regarding the case of the experiencer are taken one step further. It isthe author’s original idea that all object experiencers are not only inherently casemarked, but also governed by prepositions, thus Oblique. This applies equally wellto Dative arguments and, albeit in a less obvious way, to Accusative ones. Such an4 This is possible if unaccusativity is restricted to the lack of structural case.5 If not indicated otherwise, all references regarding Landau will consider this monograph.

6 A Recalcitrant Nature of Object Experiencers 81assumption, to treat the source of Dative and (inherent) Accusative on a par, is notnecessarily novel as it basically follows the argumentation of Emonds (1985), whonoticed that inherent case is assigned by either overt or null prepositions, dependingon language-specific factors. But this is not where Landau’s original contributionstops. The author goes on to present evidence that experiencers are in fact mentallocations 6 and, as such, locatives. To make it more plausible, periphrastic psychconstructions are cited (Landau 2010: 11):7. (a) Nina is in love (with Paul).(b) There is in me a great admiration for painters (Arad 1998: 228).It is easy to reproduce relevant examples in Polish:8. Wzbiera we mnie gniew na myśl o spotkaniu.wells up in me anger at the thought of meeting‘I get angry at the thought of the meeting.’However, apart from conceptual reasoning Landau amasses more empirical evidenceto corroborate the locative nature of experiencers, drawing heavily on thecross-linguistic adjunct control and Super-Equi facts. The gist of this argumentationwill be presented in Sect. 6.6, where Landau’s findings are confronted with thePolish data. For the time being, two crucial conclusions follow. One is the logicalconsequence of the claim that object experiencers are locatives, namely they mustundergo (covert) locative inversion to a higher subject position. The other conclusionstates that the nature of case on class II and class III object experiencers isbasically the same, 7 i.e. both groups of verbs are governed and marked for case byprepositions.6.2.3 Psych Unaccusatives RevisitedOne of the key diagnostics for unaccusativity comes from the domain ofpassivization. As observed by Marantz (1984: 144–149), unaccusative verbs areincompatible with passive morphology:9. (a) Passive morphology absorbs the external (underscored) Y-role.(b) Vacuous dethematization is impossible.6 Landau follows here the tradition of Jackendoff’s decompositional analysis of mental states(1990), refined in the works of Bouchard and Arad, among others.7 The distinction is finer-grained, though its details are not of primary significance in the context ofthis presentation. As a matter of fact, there is a structural difference between stative/eventiveobject experiencers as well as agentive/non-agentive ones. For further reference, see Landau(2005, 2010).

82 S. ŻychlińskiHowever, English ObjExp passives are relatively frequent (Pesetsky 1995: 22):10. Bill was angered by Mary’s conduct.11. The paleontologist was pleased by the discovery of the fossil.12. Bill was irritated by the loud noises coming from the next door.13. Bill would not be satisfied by half measures.Belletti and Rizzi go to great lengths showing that these passives are adjectival, notverbal, therefore not subject to facts noted by Marantz. Still, Pesetsky manages toshow that some passives in English are unambiguously verbal: 814. Sue was continually being scared by odd noises.6.3 Partial Reconciliation of FactsLandau (2010) pieces together all the facts compiled by his predecessors and takeshis conclusions one step further, offering a cross-linguistic generalization onunaccusativity among class II psych predicates. A painstaking analysis of verbal/adjectival passives in a cross-linguistic perspective allows him to divide languagesinto two groups with respect to the status of their class II psych predicates:15. Psych PassivesType A Languages: Only eventive (non-stative) Class II verbs have verbalpassive (English, Dutch, Finnish).Type B Languages: Class II verbs have no verbal passive (Italian, French,Hebrew) (Landau 2010: 47).His predictions depart from the reasoning offered by Pesetsky, as Pesetsky assumedthat languages are basically similar with respect to verbal/adjectival passives, andpointed out flaws in the selection of examples in Belletti and Rizzi whereas Landaupartly agrees with both sides.With the case on the object being determined as inherent, Landau starts out byexplaining what two strategies can be employed to passivize a quirky object.16. Strategies for Passivization of Quirky Objects:(a) P-stranding: The preposition that governs the object is stranded andreanalyzed with the verb.Pseudopassive: [TP [DP Exp]1 [T’ Aux [VP [V VPASS + Ø_ ][DP t1 ] ]]](b) Pied-Piping: The preposition that governs the object is carried along to thesubject position.Quirky passive: [TP [PP Ø_ [DP Exp]]1 [T’ Aux [VP VPASS [PP t1 ] ]]](Landau 2010: 48)8 For Pesetsky, the presence of a progressive aspect and the choice of prepositions are some of thediagnostics for the verbal status of passives.

6 A Recalcitrant Nature of Object Experiencers 83English and Dutch are among languages showing the former strategy, whereasFinnish uses the latter one (Landau 2010: 48):17. This bed was slept in.18. Daar werd over gepraat. (Dutch)There was talked about19. Sinu-sta pidet€a€an. (Finnish)you.ELA like.PASS‘You are liked.’This leads Landau to formulate a cross-linguistic generalization about theavailability of verbal passives in ObjExp verbs (Landau 2010: 49):20. Verbal passives of non-agentive ObjExp verbs will only be available inlanguages allowing either pseudopassives or (oblique) quirky passives.Neither of the strategies described above for the passivization of quirky objects isattested in Italian, French and Hebrew (these three were tested by Landau), which istaken as a clear indication that they do not have verbal passives and, thus, belong toType B Languages. Contrary to Pesetsky’s criticism of Belletti and Rizzi’s treatmentof Italian participles, Landau is able to show that even though their analysis isnot adequate for English, it still passes muster in Italian.6.4 Inconvenient Truth (About Polish ObjExp Verbs)The classification of participles put forth in the last section will only hold water onthe assumption that object experiencers are PPs. What if they are not? All evidenceleads to the inevitable conclusion that the nature of case on the object experiencer isdifferent in Polish. It has been commonly assumed that the object experiencer bearsinherent case. One of the universally accepted diagnostics for that is case suppression.Case suppression is clearly manifested in Russian, where the Genitive ofNegation rule is in full operation. This means that objects, which in positivesentences come in Accusative, switch to Genitive in negated sentences. The ruledoes not work with class II psych predicates, in which Genitive is not possible:21. *Ètot šum ne pobespokoil ni odnoj devocˇki.that noise.NOM not bothered not one girl.GEN‘That noise did not bother a single girl.’22. *Ego neudacca ne ogorcˇila materi.his failure.NOM not upset mother.GEN‘His failure did not upset mother.’(Legendre and Akimova 1993, ex. 40)This is to be expected given that “[a] standard account for this contrast exploitsthe fact that inherent case is fixed in the lexicon; GN, which is a syntactic rule,cannot override this case” (Landau 2010: 25). Polish, however, despite the fact that

84 S. Żychlińskiit is very similar to Russian in terms of the Genitive of Negation rule, surprisinglyproduces well-formed sentences:23. Ten hałas nie zaniepokoił ani jednej dziewczyny.this noise-NOM not worried any single girl-GEN‘This noise did not worry a single girl.’24. Jego porażka nie zdenerwowała matki.his failure-NOM not upset mother-GEN‘His failure did not upset his mother.’Apart from case suppression, the non-inherent nature of case is also furthersubstantiated by reflexivization facts. Neither in English-type languages nor inItalian-type languages can object experiencer verbs reflexivize:25. (a) Tedious talks irritate me.(b) I irritate myself.26. *Gianni si preoccupa.Gianni si worries‘Gianni worries himself.’A standard account for such a state of affairs involves the restriction onreflexivization (Landau 2010: 35):27. Reflexive si/se may absorb accusative or dative case, but not oblique case.Given that Accusative in Polish has all the hallmarks of structural case, it seemsa correct prediction to assume that Polish reflexives should be well-formed. 9 Thisis, indeed, the case:28. Złe wieści martwią Jana.bad news.NOM worry Jan.ACC29. Jan martwi się (złymi wieściami).John worries Cl.się (bad news.INST)‘John worries himself.’As already stated, it is Landau’s underlying assumption that all object experiencersare introduced by prepositions and marked as Oblique. As prepositional phrases areislands, extraction out of them is predicted to yield ungrammatical results. Theexamples from Polish do not give straightforward results, the judgments being atleast divided. 109 An anonymous abstract reviewer pointed out that in Russian similar reflexivization facts obtain:6. Ivan volnuetsja.Ivan worries-himselfThis, in fact, may be a further fact obscuring Landau’s analysis. Alternatively, it may suggest thatthe clitic reflexivization is not a true instance of reflexivization.10 Similarly, different facts corroborating the locative nature of object experiencers are notunambiguous in Polish, e.g. object control into adjunct clauses and Super-Equi control facts.

6 A Recalcitrant Nature of Object Experiencers 8530. ?Czyją irytowało to siostrę?whose irritated it sister31. ?Czyją straszyłeś siostrę?whose frightened you sister32. *Czyją irytują nocne telefony siostrę?whose irritate night calls sister?Interestingly enough, the (at least partial) acceptability of (30) and (31) seems tobe related to the (phonologically) light status of the subject more than to theiragentive versus non-agentive subjects. In (32), whose subject is phonologicallyheavier, the judgment is straightforwardly ungrammatical.6.5 More Trouble with Polish ParticiplesIn the light of the discussion on participles in the preceding sections, Polishparticiples should be considered now in greater detail. Similarly as in the priordiscussion, only class II ObjExp verbs are under observation, as they belong to theonly problematic group as regards the question of unaccusativity. Verbs of this classinclude, among others, przygnębiać ‘depress’, irytować ‘irritate’, straszyć‘frighten’, niepokoić ‘worry’, kłopotać ‘embarras’:33. Krzyki za oknem irytują Piotra.shouts outside irritate Peter34. Burza z piorunami przestraszyła dzieci.thunderstorm scared children35. Wieści o kryzysie niepokoją obywateli.news about crisis worry citizens36. Brak biletu zakłopotał pasażera.lack of ticket embarrassed passengerFollowing Landau’s line of reasoning, it should be possible to determine whether ornot a language has verbal passive participles on the basis of (20), repeated below:37. Verbal passives of non-agentive ObjExp verbs will only be available inlanguages allowing either pseudopassives or (oblique) quirky passives.Crucially, Polish does not exhibit either of these two constructions. This followsif object experiencers are marked for structural Accusative. Therefore, a straightforwardconclusion would require the classification of Polish as a Type B language(see (15)), which could be further supported by examples below:38. ?Pogoda wciąż przygnębia wczasowiczów.weather.NOM still depresses holidaymakers.ACC39. *Wczasowicze są bez przerwy przygnębiani (przez pogodę).holidaymakers.NOM are continuously being depressed (by the weather.OBL)

86 S. ŻychlińskiThe following two sentences, however, show considerable improvement:40. Krzyki na zewnątrz wciąż irytują Piotra.screams.NOM outside still irritate Peter.ACC41. ?Piotr jest wciąż irytowany przez krzyki z zewnątrz.Peter.NOM is still being irritated by screams.OBL from outsideWith more examples, the picture seems to be blurred even further. As hasalready been argued before, progressive aspect is said to be incompatible withpassive morphology on ObjExp verbs. However, (42) shows that the progressive ispossible with a class II psych predicate in the active voice, and (43) proves thatpassive is also possible with the same verb. In the light of the discussion so far, thiswould indicate that niepokoić ‘worry’ is an eventive predicate, thus notunaccusative, and it forms verbal passive participles.42. Wieści o kryzysie właśnie teraz wyjątkowo niepokoją obywateli.news.NOM about crisis just now exceptionally worry citizens.ACC43. Obywatele są teraz niepokojeni wieściami o nadciągającym kryzysie (choćledwie cocitizens.NOM are now being worried news about forthcoming crisis.(eventhoughbyli niepokojeni doniesieniami z rynku wschodniego).they have just recently been worried about news from the eastern marketIt becomes apparent, then, that the classification under (37) is insufficient as itdoes not predict the behavior of Polish ObjExp verbs.6.6 Further ImplicationsFinally, let us see whether Landau’s most innovative finding, i.e. the locative statusof experiencers, bears out the intuitions about Polish and its somewhat unrulybehavior.6.6.1 The Case of Super-Equi in EnglishThe fact that object experiencers, by virtue of their locative nature, always undergomovement to the subject position higher in the structure would be harder to acceptwere it not for the evidence that Landau supplies. The relevant argument 11 for thepresent purposes concerns Super-Equi facts. These facts were first observed byGrinder (1970) and relate to the observation that VP-internal infinitival clauseswhich undergo extraposition or intraposition create problematical accounts of11 Other arguments relate to Adjunt Control, Functional Readings and Forward Binding.

6 A Recalcitrant Nature of Object Experiencers 87control facts. The puzzle is presented by the paradigm below (Landau 2001:113–114):44. Mary thought that it pleased John [PRO to speak his/*her mind].45. Mary thought that it helped John [PRO to speak his/her mind].46. Mary thought that [PRO to speak his/her mind] would please John.47. Mary thought that [PRO to speak his/her mind] would help John.Although in each of these sentences two possible controllers exist for the PROsubject of the infinitival clause, in (44) the possibility of non-obligatory control isexcluded, John being the local and only available controller for PRO. No suchproblem arises for (45), where the only difference lies in the main clause predicate,which is not a psychological verb. The examples featuring intraposition (46–47)both allow for either the subject of the main clause or the object of the embeddedclause to act as controllers, instantiating, respectively, cases of non-obligatory(long-distance) and obligatory (local) control. In a more systematical way, thefollowing facts obtain in English (Landau 2001: 115):48. (a) In a structure [...X ...[it Aux Pred Y [ S PRO to VP]]], where Y and S arearguments of Pred:i. If Pred is psychological, Y must control PRO.ii. If Pred is non-psychological, either X or Y may control PRO.(b) In a structure [...X ...[ S [ S PRO to VP] Pred ...Y]], either X or Y maycontrol PRO.To provide the analysis of the Super-Equi facts that would explain all attestedpatterns of control Landau has recourse to a number of stipulations. The mostimportant is the OC Generalization 12 (Landau 2001: 118):49. The OC GeneralizationIn a configuration [...DP 1 ...Pred ...[s PRO 1 ...] ], where DP controls PRO:If at LF, S occupies a complement/specifier position in the VP-shell of Pred,then DP (or its trace) also occupies a complement/specifier position in thatVP-shell.Once again, such an assumption reduces OC to a certain syntactic configuration.In the case of Super-Equi constructions under discussion here, it follows that aslong as the infinitival clause stays within its host VP (either in the specifier or thecomplement position), then only the controller which is located in the same VP cancontrol the PRO. Since both extraposition and intraposition force the infinitival outof the VP, it is logical to expect the impossibility of OC control in these cases.The question which has not been tackled so far is why infinitival clauses undergoextraposition. The stipulation that Landau makes in order to answer this question isfound in (50) below (2001: 120):12 OC stands for Obligatory Control.

88 S. Żychliński50. ExtrapositionVP-internal clauses must be peripheral at PF.The operation of extraposition is taken to be a case of VP adjunction(an extraposed infinitival is not dominated by VP), and its motivation is derivedfrom cross-linguistic observations. 1351. Chain interpretationAny link in a chain may be the LF-visible link.The condition on chains follows naturally if, after Chomsky (1995), traces aretreated as full copies (Landau 2001: 120). It ensures the freedom of choice of thecontroller (the copy within the VP may be co-indexed locally with the VP-argumentor a long-distance relation may be established between the extraposed clause andthe subject of the matrix clause).Lastly, the thematic make-up of psychological predicates has to be systematizedas in (52) in order to reconcile the strange control facts of psych verbs (Landau2001: 118):52. Argument Projection(a) EXPERIENCER is generated above CAUSER.(b) CAUSER is generated above GOAL/PATIENT/THEME.Without going into too many details that are of no direct relevance here, it is (52)that makes it possible for Landau to explain the strange behavior of experiencerswith respect to Super-Equi constructions. From (52) it follows that in a structurewith the experiencer argument it is generated above the causer argument, thusmaking the causer VP-final, and annihilating the need for the infinitival clause tomove ((50) is satisfied in situ). Importantly, non-obligatory control can only beestablished once the infinitival clause has left the VP domain. So the correctprediction is that obligatory control is the only option available as the causernever leaves the VP. However, in his recent book (2010) Landau no longer assumes(52). The experiencer is now the higher argument only in stative psych verbs, whilein eventive psych verbs the causer is the external argument (Landau 2010: 104).The motivation for the infinitival clause to right-adjoin to the VP is still in place,then, as the external argument position is clearly not VP-final. However, thisoperation does not put the extraposed clause outside the scope of the experiencer.As must be remembered, experiencers (as locative PPs) raise to a higher position atLF. This means that in terms of scope their VP-internal placement does not marktheir final position in the derivation. When VP-internal subjects undergoextraposition to a VP-external position, they are situated outside the scope of theVP-internal arguments, as is the case in (45). But this picture changes in (44), where13 As Landau (2001: 120) says, (50) “corresponds to the cross-linguistic observation that embeddedclauses are typically peripheral to the VP and seldom intervene between a predicate and otherinternal arguments”.

6 A Recalcitrant Nature of Object Experiencers 89the experiencer argument, having moved up to Spec,TP, still scopes over theinfinitival subject right-adjoined to the VP.6.6.2 Polish Super-Equi FactsYet again Polish slips out of Landau’s neat picture as the facts about control intosubject infinitival clauses do not confirm the predictions from English. Thesentences below are due to Bondaruk (2004: 262), who first discussed these factsfor Polish. In terms of their illustrative power (49–52) are roughly equivalent to(44–47):53. Marek 1 uważa, że Ewę 2 może irytować [PRO 1/arb poprawianie popełnianychprzez nią 2 błędów].‘Marek thinks that it may irritate Ewa to correct the mistakes made by her.’54. Marek 1 uważa, że wymowę Ewy 2 może poprawić [PRO 1/arb poprawianiepopełnianych przez nią 2 błędów].‘Marek thinks that it may improve Ewa’s pronunciation to correct the mistakesmade by her.’55. Marek 1 uważa, że [PRO 1/arb poprawianie popełnianych przez nią 2 błędów]może irytować Ewę 2 .‘Marek thinks that correcting mistakes made by her may irritate Ewa.’56. Marek 1 uważa, że [PRO 1/arb poprawianie popełnianych przez nią 2 błędów]może poprawić wymowę Ewy 2 .‘Marek thinks that correcting the mistakes made by her may improve Ewa’spronunciation.’Unlike English, Polish does not allow for any relation of obligatory (local)control to be established. This runs counter to Landau’s expectation that at LF theextraposed infinitival remains within the scope domain of the raised experiencer.If this were so, obligatory control would be expected, which is not the case.6.7 ConclusionsLandau’s comprehensive treatment of object experiencers cannot accommodatePolish facts. The property of having verbal passives is not linked to thepassivization options present in languages studied by Landau. Also, nothingseems to support the inherent nature of case on the experiencer, which ruins thelocative hypothesis. Thus, the following two conclusions seem inescapable:• Polish object experiencers do not show many of the syntactic characteristicstypically associated with this group.• Polish object experiencers are not locatives.

90 S. ŻychlińskiWhat is more, Polish potentially causes concerns for Burzio’s Generalization,which relates the capacity to assign structural Accusative to the presence of theexternal argument, suggesting that more has to be said about this relation, especiallygiven that in Polish there is at least one more known situation where Burzio’sGeneralization seems to run into trouble, namely impersonal passive constructions:57. Napisano listy.were written letters.ACC‘The letters were written.’58. Nie napisano listów.not were written letters.GEN‘The letters were not written.’Not only is Accusative preserved on the object, but also the operative Genitive ofNegation confirms its structural status. It becomes necessary, then, to subject agreater number of languages to a painstaking analysis and find out if Polish standsalone as an exception to the locative hypothesis, or whether the locative hypothesishas to be reworked.ReferencesArad, M. 1998. VP-structure and the syntax-lexicon interface. Unpublished PhD dissertation,University College London.Baker, M.C. 1988. Incorporation: A theory of grammatical function changing. Chicago: Universityof Chicago Press.Belletti, A. and L. Rizzi. 1988. Psych-Verbs and Theta-Theory. Natural Language and LinguisticTheory 6: 291–352.Biały, A. 2004. Polish psychological verbs at the lexicon-syntax interface in cross-linguisticperspective. Ms. University of Wrocław.Bondaruk, A. 2004. PRO and Control in English, Irish and Polish: A minimalist analysis. Lublin:Catholic University of Lublin Press.Chomsky, N. 1995. The minimalist program. Camdridge: MIT Press.Emonds, J. 1985. A unified theory of syntactic categories. Dordrecht: Foris.Grimshaw, J. 1990. Argument structure. Cambridge, MA: MIT Press.Grinder, J. T. 1970. Super Equi-NP deletion. Chicago Linguistic Society 6: 297–317.Jackendoff, R. 1990. Semantic structures. Cambridge, MA: MIT Press.Klimek, D. and B. Rozwadowska. 2004. From Psych Adjectives to Psych Verbs. Poznań Studies inContemporary Linguistics 39: 59–72.Landau, I. 2001. Control and extraposition. The case of Super-Equi. Natural Language andLinguistic Theory 19: 109–152.Landau, I. 2005. The locative syntax of experiencers. Ms. Ben Gurion.Landau, I. 2010. The locative syntax of experiencers. Cambridge: MIT Press.Legendre, G. and T. Akimova. 1993. Inversion and antipassive in Russian. In The 2nd annualworkshop on formal approaches to Slavic linguistics, eds. S. Avrutin, S. Franks andL. Progovac, 286–318. Ann Arbor, MI: Michigan Slavic Publication.Marantz, A. 1984. On the nature of grammatical relations. Cambridge: MIT Press.Perlmutter, D. and P. Postal. 1984. The 1-advancement exclusiveness law. In Studies in relationalgrammar 2, eds. D. Perlmutter and C. G. Rosen, 81–125. Chicago: University of Chicago Press.Pesetsky, D. 1995. Zero syntax. Cambridge: MIT Press.

Chapter 7On the Representations of Motion Events:Perspectives from L2 ResearchJolanta LatkowskaAbstract Talmy’s (1985, 2000) influential typology of motion event constructionsdivides languages into categories based on typical lexicalization patterns of motionwithin the sentence. In the literature, priority is given to satellite-framed(S-languages) and verb-framed languages (V-languages), which encompass themajority of languages spoken in the industrialized world. This paper sets out todiscuss the rationale behind Talmy’s classification, and to present evidence thatcalls into question the legitimacy of his claims. This encompasses data showingbehavior that diverges from the category norm, as well as an analysis of thelinguistic salience of Manner and Path in two typologically related languages:Polish and English. Although both are classified as satellite-languages, they showmarked differences in the way they encode Path. Polish prioritizes Path by encodingit throughout the predicate, while English confines Path information to post-verbalsatellite constructions. The paper closes with an overview of research into theacquisition of motion patterns by speakers of typologically contrasting languages,and makes a number of predictions about how Polish-English bilinguals maylexicalize motion events in L2 English. Overall, the paper accentuates the need toreevaluate the typology, which is especially important in view of the fact that,despite its inaccuracies, it inspired a spate of studies in the area of linguisticrelativity, speech production and second language acquisition.7.1 Theoretical BackgroundIt is widely accepted that motion constitutes a change of location of an object withrespect to another object. Talmy (1985) posits that a motion event is made up ofseveral semantic components, which in addition to the Motion itself include Path,J. Latkowska (*)University of Silesia, Katowice, Polande-mail: jolanta.latkowska@us.edu.plM. Pawlak and J. Bielak (eds.), New Perspectives in Language,Discourse and Translation Studies, Second Language Learning and Teaching,DOI 10.1007/978-3-642-20083-0_7, # Springer-Verlag Berlin Heidelberg 201191

92 J. LatkowskaManner, Ground and Figure (Talmy 2003; cf. Lakoff 1987). Path refers to thetrajectory or directionality of a moving object (Figure) and specifies the Source, e.g.from the supermarket, and Goal of the activity, e.g. to the bus stop, as well as itsMedium, e.g. through the Red Square. Ground is the reference object, whileManner covers the motor patterns of movement, its rhythm and pace, and the effortput into it. Examples (1a) and (1b) present the main elements of a simple motionevent in English (Slobin 2005; Talmy 1985).1. (a) John ran into the room.Figure Motion + Manner Path Goal(b) The bottle rolled off the table.Figure Motion + Manner Path GroundLanguages differ with regard to how these components are encoded syntactically.Overall, Talmy (2003) speaks of verb-framed languages (V-languages),which encode Path and Motion in the main verb of a clause, e.g. enter, exit, andsatellite-framed languages (S-languages), where Path is encoded in a satellite,i.e. an attached verb particle (in, out), prepositional phrase (PP, e.g. into the cave)or a verb prefix, as can be observed in Slavic languages, e.g. w-biec in Polish (Eng.in-run). Both language types also differ as regards the Manner of motion, which insatellite languages, often referred to as high-Manner languages, is conveyed by theverb. Consequently, such languages have a larger stock of Manner verbs, the lessfrequent of which convey finer semantic distinctions. Verb-framed languages, bycontrast, code Manner of motion by means of an adjunct or simply do not mention it(Slobin 2003), and are hence labeled low-Manner languages. In Talmy’s typology,Manner and Cause of motion are treated as semantic equivalents, which arelexicalized in a similar way.At the discourse level, S-languages accumulate Path particles and prepositionalphrases next to the main verb and encode both Path and Manner in a single clause(Allen et al. 2007), while V-languages typically use two or more separate clauses,as shown in example (2), or omit Manner information altogether.2. English original: I ran out the kitchen door, past the animal pens, towardsJasón’s house.Spanish translation: Sali por la puerta de la cocina (I exited the kitchen door),pase por los corrales (passed by the animal pens), y me dirigi a casa de Jasón(and directed myself to Jasón’s house) (Slobin 2005).This general formula was subsequently modified by Aske (1989), who drawingon an analysis of how Spanish verbs lexicalize motion, discovered a generaltendency of V-languages to use Manner verbs only in combination with atelicPath phrases, i.e. those which do not denote event completion or boundary-crossing,as shown in example (3).3. We walked along the beach (French: Nous avons marché le long de la plage;Pourcel 2003: 55).

7 On the Representations of Motion Events: Perspectives from L2 Research 93Since S-languages make use of both atelic and telic, i.e. directional andresultative Path satellites, of which the latter clearly indicate an activity’s endpoint and/or boundary crossing (Pourcel 2003; Papafragou et al. 2008; Cadiernoand Lund 2004), the differences between the two language types are limited to howthey encode bounded (telic) situations, where V-languages allow Path verbs only.Satellite-framed languages include Germanic, Celtic, Slavic and Finno-Ugriclanguages, while verb-framed languages comprise Greek, Basque, Japanese,Korean, Romance, Turkic and Semitic languages (Riemer 2010). An additionalcategory proposed by Talmy includes languages such as Atsugewi and Navajo,which conflate Figure and Motion in the verb, as exemplified by the Englishsentence It rained in through the bedroom window (Aske 1989).Talmy’s three-way typology was subsequently extended by Slobin (2004, 2006),who introduced yet another category of equipollently-framed languages(E-languages). It encompasses serial-verb languages, which express Path andMotion by using two or three verbs in a clause, none of which is marked forfiniteness, as in the Thai sentence chán dəən (paj) (Eng. I walk go ‘I am walking’)(Slobin 2006: 4).7.2 Conflicting Research ResultsAlthough Talmy and his associates continue to refine their framework, evidence ofmajor inconsistencies has steadily been accumulating. To begin with, it hasbeen shown that the categorical contrasts suggested by Talmy are by no meansabsolute. This is certainly the case with the Manner/Path distinction as speakers ofboth V- and S-languages tend to use patterns that diverge from the category norm.In fact, Beavers et al. (2010) note that most languages straddle more than onecategory. For instance, even though Turkish is a V-language, where Manner andPath are specified in separate clauses, some of its motion verbs convey bothelements in a way typical of S-languages, e.g. the Turkish word for climb (Jarvisand Pavlenko 2008). Similar trends have been reported by Pourcel and Kopecka(2005) for French, which allows Manner to be encoded in the main verb and Path ina gerund, as in courir en montant ‘run ascending’. Moreover, French has a numberof frequent motion verbs, which contain a prefix expressing Path, e.g. a-ccourir‘run to’ and a-tterir ‘touch down’, ‘land’ (Kopecka 2006). By the same token, bothPolish and English, despite being satellite languages, are capable of encodingManner in an adjunct (adverbial of manner), as in opuścić salę biegiem ‘leave aroom at a run’. Likewise, Brown and Gullberg (2008) found that contrary totypological trends, Japanese (V-language) speakers use a wide range of Mannerverbs as well. The tendency has come to light thanks to the findings of Ohara (2004;cited in Brown and Gullberg 2008), who on analyzing the translation of Tolkien’sThe Hobbit into Japanese, found the verb to be the prevalent means of conveyingManner information. What is more, the verbs often conflated Manner and Path.Further still, Beavers et al. (2004) demonstrate that English allows motion verbs to

94 J. Latkowskaencode neither Path nor Manner since these may be encoded in the satellites, asportrayed in example (4):4. John moved stealthily out of the room.In some S-languages, e.g. English, the coding of Manner is optional. Talmy(1985) enumerates the following English verbs which carry Path but not Manner:enter, ascend, descend, cross, pass, circle, advance, proceed, approach, arrive,depart, return, join, separate, part, rise, leave, near and follow. At the opposite endare languages such as Russian, which encodes Manner obligatorily (Jarvis andPavlenko 2008) but with varying degrees of Manner salience. Finally, in Mandarin,three options, i.e. V-, S- and E-framing are possible (Beavers et al. 2010: 357).In the light of the above, the inescapable conclusion, and one that was drawn byTalmy himself (Talmy 1985: 62), is that the Manner/Path distinction only reflectsthe most characteristic, frequent and colloquial ways of talking about motion,which in turn might be determined by factors such as ease of processing (Slobin2004) and frequency of use (Papafragou and Selimis 2010). Also, it should notescape notice that individuals have at their disposal a variety of linguistic and nonlinguisticdevices that may compensate for the lack of category specific lexicalization.In this connection, Slobin (2004) lists adverbials of Manner, ideophones, i.e.words which imitate sounds and may function as adverbials, and gestures, whichsupply information about Manner that has not been encoded linguistically.Somewhat overlooked by both Talmy and Slobin is the fact that in somelanguages verbs of Manner implicate Path in addition to encoding Manner. Thatthis may be the case has been pointed out by quite a few researchers with regard to arepresentative number of languages. For instance, in English verbs like walk, run,swim and fly lexicalize both Manner and the direction of motion, which proceedstowards an unspecified goal (Tsujimura 2007: 403). Likewise, climb a treeexpresses both Manner and upward movement. Malt et al. (2010) used videoclips to elicit Manner verbs for the two basic biomechanical categories of runningand walking. They found that contrary to typological taxonomies Spanish, Englishand Japanese used conventional Manner verbs to encode the distinction. What ismore, the Japanese never used their main walking term to refer to walking in placeor backwards, which in the authors’ opinion indicates that for the Japanese forwardmovement is part of the meaning of walk. In English, by contrast, movement in adirection other than forwards has to be specifically qualified by an adverb or PP, asshown above. Other studies along these lines are Allen et al. (2007) and Férez(2007). The latter enumerates a number of Spanish and English verbs which, shecontends, encode Path as well as Manner information. Her list includes flee, slink,scurry, scuttle, charge, track, stalk, and rove together with their Spanish analogues.Allen et al. (2007), in turn, divide English Manner verbs into those that conveythe idea of Manner causing a change of location, and those where Mannerdoes not have such an effect (cf. Beavers et al. 2004; Willim 2006). For example,run denotes continuous movement forward, while rotate does not convey thisnotion. Obviously, one could speculate that Allen’s observations allude to theprototypical scenarios developed for specific Manner verbs rather than their

7 On the Representations of Motion Events: Perspectives from L2 Research 95semantic composition (c.f. Jackendoff 1990). Although clearly dependent on theadopted theoretical stance, this perspective is of relevance, particularly in view ofthe fact that pragmatic and contextual factors (Willim 2006; Beavers et al. 2004)often conspire to give the impression of motion along a trajectory, as shown inexample (5), where climb expresses Path but not Manner even though, overall, ittends to be classified as a Manner verb.5. The plane climbed to 9000 feet.Taken together, these considerations create the impression that in linguisticrenditions of directed motion, Path is more semantically salient than Manner (cf.Jackendoff 1990). Indeed, Talmy’s (2000) modified framework includes Path underthe core schema of motion events, together with Aspect, State Change, and others.Manner and Cause comprise supporting relations, i.e. those that may be but do notneed to be conveyed. It is the locus of the core schema either in the verb root or inthe satellite that determines classification as a V- or S-language. Relations withinthe core schema are implicational in the sense that if Path is encoded in the verb, theremaining components will also be encoded. Although this modified version offersa more comprehensive perspective on the lexicalization of motion, the currentdiscussion will concentrate on the Path/Manner contrast, since it became a pointof departure for numerous investigations in the field of SLA and bilingualism. Asregards the advantage of Path over Manner, the evidence in support of this claimwill be discussed on the basis of Polish, although it is likely to apply to other Slaviclanguages, too (cf. Pavlenko’s (2010) analysis of the motion lexicon in Russian).7.3 Path or Manner: Evidence from PolishOverall, Polish classifications of motion verbs are more inclusive than the originalTalmyan framework, and consequently incorporate many of the dimensionsthat Talmy (1985) was found to be lacking. Thus, Kubiszyn-Mędrala (http://celta.paris-sorbonne.fr/anasem/papers/Motion/CzasRuch.pdf; cf. Laskowski 1999: 176;Bojar 1979: 24–31) speaks of two semantically distinct verb categories, which refereither to change of location or to motion that does not result in locational change. Ofrelevance to the current discussion are change of location verbs, which can befurther subdivided into directed motion verbs and indeterminate motion verbs. Theformer denote one-way movement along a specific trajectory, e.g. iść, biec ‘walk’and ‘run’, respectively, while the latter encode habitual and/or iterative movementalong an unspecified path or in multiple directions, e.g. chodzić ‘walk repeatedly’.Laskowski (1999: 50) enumerates the following semantic components of motionevents in Polish (see example 6):6. Obiekt przemieszczał się przez pole od lasu ku rzece.Figure Motion Path Source Goal

96 J. Latkowska‘An object was moving across a field from the woods towards the river.’He also points out that not all of these components are lexicalized at the sentencelevel as sentences with a complete motion representation are relatively rare inPolish. This is because quite a few motion components are implicated by thecontext.Polish motion verbs convey information about Manner, which Laskowskibroadly defines in terms of intentionality, instrument (use of a vehicle or limbs),environment (water, air), speed, degree of contact with the surface, attitude and soon (Laskowski 1999; cf. Slobin 2006). Some verbs combine Manner and directionalityof motion, that is, Path, e.g. biec ‘run’ and iść ‘walk’ both imply continuousmovement forward. Polish has very few verbs that express Path without simultaneouslyencoding Manner, e.g. przybyć ‘arrive’, wyruszyć ‘depart’, przemieszczaćsię ‘move from one place to another’, zbliżać się ‘approach’, podążać ‘follow’,przedostać się ‘get through’, mijać ‘pass’, udać się, wybrać się ‘make one’s way to’and dotrzeć ‘reach’.In accordance with Talmy (1985), Polish, being a satellite-framed language,encodes Path by means of a prefix. The prefix specifies the beginning and endof the trajectory, as well as the trajectory itself. It also entails the position of thespeaker, expectancy of arrival, departure from the trajectory’s starting point andapproach to the end point (Krucka 2006). Janowska (1999) assigns spatial prefixesto three categories depending on the relation of the Figure to the Ground. Adlativeprefixes such as od-, wy-, z/s-, roz-, and u- specify the starting point of motion, be itthe inside of an object (Ground), as in w-yjść z domu ‘leave home’, the object’ssurface (z-ejść z drogi ‘get off the road’) or its boundary (od-płynąć od brzegu‘sail/swim away from the shore’). Ablative prefixes: do-, na-, nad-, pod-, przy-,w-, za-, z/s- , denote the Figure’s proximity to the Ground, specifying the numberof objects or individuals involved, as in z-jechały się tłumy‘crowds arrived’,boundary crossing (w-szedł do pokoju ‘he entered the room’) and the exact point ofcontact (na-jechał na rower *‘he drove onto a bike’), along with the degree ofproximity to the Ground (do-jechać vs. pod-jechać, ‘reach the destination’ vs. ‘stopshort of the destination’). Willim (2006) explains that in Polish, verbs referring tocaused/directed motion and change of location events require a spatial prefix wheneverthe event in question involves the crossing of a location boundary, as in w-biec nascenę ‘to run onto the stage’. Change of location verbs that do not have a prefixobligatorily require a directional PP, as in u-dać sie do Paryża ‘leave for Paris’(Willim 2006: 218). Generally, spatial prefixes telicise events by delimiting theirendpoints (see above). Consequently, the events have a perfective reading.The last of Janowska’s categories consists of perlative prefixes: prze- and o-.These signify the distance covered during a motion event, as in prze-płynąć rzekę‘swim across the river’. Quite uncharacteristically for satellites, in Polish, perlativerelations may be expressed by an instrumental noun phrase. This typologicalidiosyncrasy is additionally illustrated by the following examples: szli drogą,ścieżką, lasem ‘they walked the road, path, *the woods’ (Nagórko 1998). Onthe whole, Janowska’s (1999) analysis focuses on 18 spatial prefixes, which by

7 On the Representations of Motion Events: Perspectives from L2 Research 97virtue of their number and nuanced differences in meaning can yield greatprecision of description. The way prepositional phrases complement the wealth ofdetail provided by prefixation is discussed below.Generally, prefixed verbs of directed motion connote prepositional phraseswhere the preposition often echoes the prefix, e.g. dojechać do centrum miasta(Eng. to-come to the city centre) and wjechać w drzewo (Eng. in-drive in-to a tree).Sysak-Borońska (1974) stresses the functional interdependence of prepositions andcase endings in the complement noun phrase, which she sees as exponents of spatialrelations. Notably, case endings may determine the character of the entire phrase byintroducing a contrast between a locative and adlative reading. This is shown inexample (7)7. Locative: Leżeć pod tramwajem (instrumental) ‘to be lying under a tram’,adlative: wpaść pod tramwaj (accusative) ‘to fall under a tram’.In Polish, prepositional phrases can denote both bounded and unbounded events.There is disagreement over their syntactic status since Spencer and Zaretskaya(1998) see them as adjuncts while Willim (2006) argues that they function as verbcomplements, because they are obligatory in cases when a verb of motion does nothave a prefix (see above). Most crucially, a goal/directional PP further qualifies thePath of the motion event and can even impose a change of location reading on abasic activity verb, as in Pchnęli samochód na bok ‘They pushed the car to the side’(Willim 2006: 214).The above evidence seems robust enough to warrant the conclusion that Polishlexicalization patterns of directed motion prioritize Path, which may be encodedthroughout the predicate. A for Manner, it appears to be downgraded to the proverbialred herring, which despite being finely synchronized with Path and perceptuallysalient, semantically constitutes an addition to an otherwise complete accountof a motion event. In a similar vein, Pourcel (2003) found that English usersperceive the trajectory of motion rather than Manner as the defining feature. Itmay also be worth pointing out that in V-languages, too, Path information oftenexceeds that conveyed by the verb. This is illustrated by Riemer (2010: 402–403),who shows that Spanish (see examples (8a) and (8b)) does not restrict Pathinformation to the verb but further qualifies it in the satellite, which may beindicative of a general prominence of Path in linguistic expressions of motion.8. (a) La botella entró a la cueva (flotando)Eng. The bottle moved-in to the cave (floating)(b) El globo subió por la chimenea (flotando)Eng. The balloon moved-up through the chimney (floating).Finally, to put the current debate into perspective, let us consider the criticismsmade by Cadiot et al. (2006), who drawing on data from French, dismissed the ideaof using displacement as a principal semantic determinant of motion. The rationalebehind their approach is that many of the allegedly typical motion verbs tend to beused in ways that have nothing to do with change of location. A case in point isthe expression la route monte ‘the road goes up’, which does not imply any

98 J. Latkowskaform of movement, or indeed, la photo est bien sortie ‘the photo came out well’,which likewise does not convey a hint of displacement. Moreover, preoccupationwith physical components of motion overshadows the emotional, subjective andimpressionistic dimensions of meaning that are intricately interwoven into semanticnetworks, and as such, reflect the complexity of human experience. Cadiot et al.argue that by abstracting displacement from a variety of uses of motion verbs,Talmy narrowed down his investigations to just one aspect of their meaning, whichshould not be treated as privileged or determinative (Riemer 2010). The questionthat arises from this critique is whether and to what extent a fragmentary portrayalof the semantics of motion verbs can reliably be used as an explanatory basis forresearch and the ongoing debate on whether and to what extent Talmy’s typologyreflects the cognitive and linguistic operations of the (bilingual) mind. This isparticularly relevant in view of the fact that it became a springboard fordevelopments in linguistic relativity research, and consequently acquired considerabletheory-generating potential (cf. Slobin’s (1996) thinking for speakinghypothesis).7.4 The Bilingual PerspectiveA logical consequence of investigating lexicalization patterns across a broadspectrum of languages is that the obtained data can be used as a benchmark forexplaining aspects of SLA and bilingual speech. Seen from this perspective, L2sentence construction appears to be influenced by the L1, which limits the learners’choices primarily to those aspects of meaning that are encoded in L1 syntax. Inreferring to these processes, Slobin (1996: 89) aptly observes that the L1 trains itsusers to pay different kinds of attention to events and experiences when talkingabout them, to the extent that children as young as three consistently followlanguage-specific lexicalization patterns (Allen et al. 2007). What is more, sucheffects are exceptionally resistant to restructuring in adult second language acquisition(Slobin 1996: 89). The practical implication is that the L2 learner/user is setto habitually attend to and/or encode L1-based conceptualizations in the L2.Interestingly but not surprisingly, this does not preclude L2-inducedrestructuring of L1 patterns. As shown by Papafragou et al. (2008), who askedGreek and English subjects to verbally describe videoed motion events, the Greekspeakers were equally likely to use either a Path or a Manner verb to describe telicscenes. The authors explain that one of contributing factors might have beenadvanced proficiency in L2 English, which sanctions such usage. A similar conclusionhas been drawn by Brown and Gullberg (2008) who analyzed oral depictions ofmotion events by Japanese–English bilinguals. The evidence they collected bearsmarks of bidirectional L1–L2 transfer and convergence because the examinedbilinguals used more Manner verbs in their L1 Japanese than Japanesemonolinguals but fewer verbs of this type than English-only subjects. By thesame token, in their L2 English there was less encoding of Manner than in the

7 On the Representations of Motion Events: Perspectives from L2 Research 99monolingual English group. Not surprisingly, gesture use in the bilingual groupexhibited features of both L1 and L2. Hohenstein et al. (2006) report paralleltendencies with regard to the use of Path and Manner verbs in oral film retellsby Spanish–English bilinguals. Other research further validates these tendencies.For example, Polish-French bilinguals tend to use an adjunct to double-codemanner of movement, which violates logicality constraints in L1 Polish, as shownin example 9:9. Doszedł do nas biegając (Eng. He walked up to us running).An exception to this trend is oral film retellings collected from late Russian-English bilinguals by Pavlenko (2010), who demonstrates that the L1 Russianmotion lexicon is relatively resistant to cross-linguistic influence, and that of allthe syntactic markers of motion, Path prefixation is the most stable. The fewinstances of L2-induced changes that Pavlenko found in her corpus included apreference for imperfective verbs and loss of a semantic distinction betweenwalking, riding and driving, which manifested itself as the use of idti/walk, mostlikely by analogy to the English go.Another issue of interest, particularly in SLA contexts, is the semanticexponents of motion that L2 learners choose to lexicalize. Working along theselines, Cadierno and Lund (2004) hypothesize that, when learning an S-language,speakers of V-languages will avoid less frequent manner verbs, especially inreferring to boundary-crossing situations, and will express manner in a separateunit. By contrast, S-language users will add manner information to sentences in a V-language, and will fail to distinguish between boundary-crossing and non-boundarycrossingscenarios. They warn, however, that linguistic behavior will to some extentbe linked to proficiency in a particular language.Finally, an issue well worth looking into is how speakers of a Slavic S-language,e.g. Polish, go about expressing directed motion in L2 English, also an S-language.Keeping in mind the differences in the way both encode Path, it is possible to form anumber of hypotheses about the possible lexicalization patterns in L2 English.These run as follows:10. Given the tendency of Polish to lexicalize Path throughout the predicate, Polishusers of L2 English will prioritize Path at the expense of Manner verbs. This willmanifest itself as a preference for Path-only verbs, a tendency reinforced bytheir frequent use in English. In fact, The Oxford 3000 Wordlist (www.oup.com) features most of the English Path-only words, as listed by Talmy (1985). 111. Since Manner verbs, especially those conveying finer distinctions, are lesscommon, they are also less likely to be acquired (Cadierno and Lund 2004)and/or used productively by L2 learners. Consequently, the learners will rely onthe most frequent stock of Manner vocabulary and, if necessary, will attempt toexpress Manner by resorting to other parts of speech, such as adverbs.1 Exceptions include: ascend, descend, cross (v), circle (v), depart, part (v) and near (v).

100 J. Latkowska12. Since Polish users of L2 English expect the verb to express at least some Pathinformation, they will initially use deictic English Path verbs, such as come andgo. This is because their semantic composition is broad enough to accommodatethe nuanced meanings of Polish spatial prefixes. There will also be lesspressure to encode spatial information in the prepositional phrase, by analogyto L1 Polish. As a result, Polish–English users will produce less elaborate andshorter Path PPs in L2 English.Despite their intuitive appeal, these hypotheses only intimate possible linguisticbehavior and should not be interpreted as descriptions of definite trends. Indeed,with the benefit of hindsight, it seems more practical to adopt a data-drivenapproach to explicating these issues, rather than fall into the trap of overprediction,which for years blighted theory-oriented research into cross-linguistic influence.Finally, even though research in this area appears to be limited to a confinedsemantic field (but see Cadiot et al. 2006), its findings will find wide applicationin other domains, including syntax and stylistics. Given the challenges these presentto L2 users, insights into their modus operandi may turn out to be yet anothermissing piece of the puzzle as to why it is so difficult to achieve native-like/monolingual levels of L2 proficiency.ReferencesAllen, S., A. Özy€urek, S. Kita, A. Brown, R. Furman, T. Ishizuka and F. Mihoko. 2007. Languagespecificand universal influences in children’s packaging of Manner and Path: A comparison ofEnglish, Japanese, and Turkish. Cognition 102: 16–48.Aske, Jon. 1989. Path predicates in English and Spanish: A closer look. In Proceedings of theFifteenth Annual Meeting of the Berkeley Linguistics Society, eds. K. Hall, M. Meacham andR. Shapiro, 1–14. Berkeley, California: Berkeley Linguistics Society.Beavers, J., B. Levin and S.W. Tham. 2004. A morphosyntactic basis for variation in the encodingof motion events. Unpublished manuscript. http://comp.ling.utexas.edu/~jbeavers/publications.html. Accessed 10 July 2010.Beavers, J., B. Levin and S.W. Tham. 2010. The typology of motion expressions revisited.Linguistics 46: 331–377.Bojar, B. 1979. Opis semantyczny czasowników ruchu oraz pojęć związanych z ruchem.Warszawa: Wydawnictwo Uniwersytetu Warszawskiego.Brown, A. and M. Gullberg. 2008. Cross-linguistic influence in L1-L2 encoding of Manner inspeech and gesture. A study of Japanese speakers of English. Studies in Second LanguageAcquisition 30: 225–251.Cadiot, P., F. Lebas and Y-M. Visetti. 2006. The semantics of motion verbs: Action, space, andqualia. In Space in languages, eds. M. Hickmann and S. Robert, 175–206. Amsterdam:Benjamins.Cadierno, T. and K. Lund. 2004. Cognitive linguistics and second language acquisition: Motionevents in a typological framework. In Form-meaning connections in second language acquisition,eds. B. Van Patten, J. Williams, S. Rott and M. Overstreet, 139–153. Mahwah, NJ:Lawrence Erlbaum Associates.Férez, P. C. 2007. Human locomotion verbs in English and Spanish. IJES 7: 117–136.

7 On the Representations of Motion Events: Perspectives from L2 Research 101Hohenstein, J., A. Eisenberg and L. Naigles. 2006. Is he floating across or crossing afloat? Crossinfluenceof L1 and L2 in Spanish-English bilingual adults. Bilingualism: Language andCognition 9: 249–261.Jackendoff, R. 1990. Semantic structures. Cambridge, MA: MIT Press.Janowska, A. 1999. Funkcje przestrzenne przedrostków czasownikowych w polszczyźnie.Katowice: Wydawnictwo Uniwersytetu Śląskiego.Jarvis, S. and A. Pavlenko. 2008. Crosslinguistic influence in language and cognition. New Yorkand London: Routledge.Kopecka, A. 2006.The semantic structure of motion verbs in French: Typological perspectives. InSpace in languages, eds. M. Hickmann and S. Robert, 83–101. Amsterdam: Benjamins.Krucka, B. 2006. Kontrasty polsko-rosyjskie w zakresie morfologii a zjawiska interferencji. Łask-Warszawa: Oficyna Wydawnicza LEKSEM.Kubiszyn-Mędrala, Z. Polskie bezprzedrostkowe czasowniki ruchu w perspektywie kognitywnej.Unpublished manuscript. http://celta.paris-sorbonne.fr/anasem/papers/Motion/CzasRuch.pdf.Accessed 23 May 2008.Lakoff, G. 1987. Women, fire and dangerous things: What categories reveal about the mind.Chicago: Chicago University Press.Laskowski, R. 1999. Czasowniki ruchu liniowego (czasowniki przemieszczania) w pespektywiekontrastywnej. In Semantyka a konfrontacja językowa, eds. Z. Greń and V. Koseska-Toszewa,49–58. Warszawa: Slawistyczny Ośrodek Wydawniczy.Malt, B. C., S. P. Gennari and M. Imai. 2010. Lexicalization patterns and the world-to-wordsmapping. In Words and the mind: How words encode human experience. eds. B. C. Malt andP. Wolff, 29–57. Oxford: Oxford University Press.Nagórko, A. 1998. Zarys gramatyki polskiej. Warszawa: PWN.Ohara. K. M. Manner of motion in Japanese and English. Unpublished manuscript.Papafragou, A., J. Hulbert and J. Trueswell. 2008. Does language guide perception ? Evidencefrom eye movements. Cognition 108: 155–184.Papafragou, A. and S. Selimis. 2010. Event categorization and language: A cross-linguistic studyof motion. Language and cognitive processes 25: 224–260.Pavlenko, A. 2010. Verbs of motion in L1 Russian of Russian-English bilinguals. Bilingualism:Language and Cognition 13: 49–62.Pourcel, S. 2003. Linguistic relativity and motion events in English. Durham working papers inlinguistics 9: 53–67.Pourcel, S. and A. Kopecka. 2005. Motion expression in French: typological diversity. Durham &Newcastle working papers in linguistics 11: 139–153.Riemer, N. 2010. Introducing semantics. Cambridge: Cambridge University Press.Slobin, D. I. 1996. From “thought and language” to “thinking for speaking”. In Rethinkinglinguistic relativity, eds. J. J. Gumperz, and S. C. Levinson, 70–96. Cambridge: CambridgeUniversity Press.Slobin, D. I. 2003. Language and thought on line: Cognitive consequences of linguistic relativity.In Language and mind. Advances in the study of language and thought, eds. D. Gentner andS. Goldin-Meadow, 157–192. Cambridge, MA: The MIT Press.Slobin, D. I. 2004. The many ways to search for a frog: Linguistic typology and the expression ofmotion events. In Relating events in narrative: Vol. 2. Typological and contextualperspectives, eds. S. Str€omqvist and L. Verhoeven, 219–257. Mahwah, NJ: Lawrence Erlbaum.Slobin, D. I. 2005. Linguistic representations of motion events: What is signifier and what issignified? In Iconicity inside out: Iconicity in language and literature 4, eds. C. Maeder,O. Fischer and W. Herlofsky, 307–260. Amsterdam/Philadelphia: John Benjamins.Slobin, D. I. 2006. What makes manner of motion salient? Explorations in linguistic typology,discourse, and cognition. In Linguistic systems and cognitive categories, eds. M. Hickmannand S. Robert, 59–81. Amsterdam/Philadelphia: John Benjamins.Spencer, A. and M. D. Zaretskaya. 1998. Verb prefixation in Russian as lexical subordination.Linguistics 36: 1–39.

102 J. LatkowskaSysak-Borońska, M. G. 1974. Some remarks on the spatio-relative system in English and Polish.Poznan studies in contemporary linguistics 3: 185–208.Talmy, L. 1985. Lexicalization patterns: Semantic structure in lexical forms. In Languagetypology and syntactic description. Volume III: Grammatical categories and the lexicon, ed.T. Shopen, 57–149. Cambridge: Cambridge University Press.Talmy, L. 2000. Toward a cognitive semantics: Vol. I. Concept structuring systems. Cambridge,MA: MIT Press.Talmy, L. 2003. Toward a cognitive semantics: Vol. II. Typology and process in conceptstructuring. Cambridge, MA: MIT Press.The Oxford 3000 Wordlist. http://www.oup.com/elt/catalogue/teachersites/oald7/oxford_3000/oxford_3000_list?view¼Print&cc¼global. Accessed 10 June 2010.Tsujimura, N. 2007. An introduction to Japanese linguistics. Oxford: Blackwell.Willim, E. 2006. Event individuation and countability. A study with special reference to Englishand Polish. Kraków: Jagiellonian University Press.

Chapter 8On the Interplay Between PrepositionalCategories. The Case of the Polish od–doConstructionDaria BębeniecAbstract The Polish prepositions od and do, being opposite in terms of theirspatio-geometric and functional content, display a significant degree of compatibilityand thus tend to co-occur in a number of different contexts. As a result, the twoprepositions appear to have developed a number of joint meanings, and the aim ofthe present article is to account for their motivated character by means of one ofthe recent cognitive linguistic methodologies, the Principled Polysemy model(Tyler and Evans 2003). Specifically, it is argued that the od–do constructionconstitutes a polysemous category, which consists of several distinguishable yetinterrelated meanings. It is posited that the central spatial meanings of theprepositions od and do, taken together, form the basis for the remaining extendedsenses, the majority of which cannot be predicted from the individual prepositionalnetworks.8.1 IntroductionThe Polish prepositions od and do apparently constitute each other’s polaropposites. 1 In spatial terms, the former encodes the trajector’s orientation awayfrom the landmark and the latter – the trajector’s orientation towards the landmark.From a functional perspective, the landmark of od constitutes the trajector’s source,whereas in the case of do it corresponds to the trajector’s goal. 2 The symmetrybetween the two prepositions extends well beyond the spatial primary meanings, as1 The present paper is based on my dissertation devoted to the semantics of directional prepositions,including Polish od and do (Bębeniec 2010).2 For a discussion of the prefixal counterparts of the two prepositions, see for example Dąbrowska(1996), Suchostawska (2005) and Przybylska (2006).D. Bębeniec (*)Maria Curie-Skłodowska University, Lublin, Polande-mail: daria@hektor.umcs.lublin.plM. Pawlak and J. Bielak (eds.), New Perspectives in Language,Discourse and Translation Studies, Second Language Learning and Teaching,DOI 10.1007/978-3-642-20083-0_8, # Springer-Verlag Berlin Heidelberg 2011103

104 D. Bębeniecod and do have also developed several opposite senses. For example, while od isused to refer to distance, detachment, distinction and subtraction, do is frequentlyemployed in the contexts pertaining to proximity, attachment, similarity and addition.3 In consequence, the two prepositions exhibit a substantial degree of compatibility,which has in turn resulted in a number of joint instantiations, such as forinstance examples (1)–(10) below.1. Od Annasza do Kajfasza ‘from pillar to post’2. Od Sasa do Lasa ‘~as different as chalk and cheese’3. Od łyczka do rzemyczka ‘~one thing leads to another’4. Od ucha do ucha ‘from ear to ear’5. Od czasu do czasu ‘from time to time’6. Od góry do dołu ‘from top to bottom’7. Od deski do deski ‘from cover to cover’8. Od początku do końca ‘from start to finish’9. Od stóp do głów ‘from head to toe’10. Od szczegółu do ogółu ‘from the specific to the general’In view of this, the aim of the present paper is to account for the status of theafore-mentioned instances by means of one of the recent cognitive linguisticmethodologies, the Principled Polysemy model (e.g. Tyler and Evans 2001, 2003;Evans and Tyler 2004a, 2004b). 4 It will specifically be argued that that the Polishod–do construction constitutes a higher-order polysemous category, 5 whose emergencemight primarily be attributed to the frequent interaction between theprepositions od and do. 6 First, the central spatial senses of the two prepositionswill be taken as the point of departure for the category proposed, and thesemeanings will be treated as a kind of joint primary sense motivating the remaining3 The following are but a few examples: Mieszkamy daleko od centrum miasta ‘We live a long wayfrom the city centre’, Ma blisko do pracy, ale ciągle się spóźnia ‘He lives close to his workplace,but he is always late’, Ta sukienka mocno przylega do ciała ‘The dress is a tight fit’ and Nie mógłoderwać się od książki ‘He could not tear himself away from the book’.4 Among the hallmarks of the Principled Polysemy framework one could enumerate the constructof a proto-scene, which is a schematic representation of a spatial relation and consists of spatiogeometricand functional content, and principled criteria for isolating the central prepositionalmeaning and distinguishing conventionalized senses from contextual uses. For a detailed expositionof the assumptions of Principled Polysemy see Tyler and Evans (2003).5 In Construction Grammar e.g. Fillmore 1988; Kay and Fillmore 1999; Croft 2001; Goldberg1995, 2006), constructions are taken to be such pairings of form and meaning whose semantic andsyntactic properties are not predictable from the properties of the relevant constituent parts. For anaccount of constructional polysemy, as exemplified by the English ditransitive and caused-motionconstructions, see for instance Goldberg (1995). In Goldberg’s terms, polysemy, along withmetaphorical extension, subsumption and instance, constitutes one type of inheritance linksmotivating the systematic relationships among constructions.6 In other words, since the central spatial senses of the prepositions od and do respectively encodethe starting point and end point of the trajector’s potential relocation, which, in dynamic contexts,constitute crucial components of a motion event, it is only natural that there should exist caseswhere the two components are given explicit realization.

8 On the Interplay Between Prepositional Categories 105members of the od–do construction. Further, the isolated groups of contexts will beconsidered as akin to distinct senses, which are modelled into the semantic networkin its own right. Even though some of the meanings can still be predicted from theindividual prepositional categories, it will be argued that quite a number of othersmust have emerged as a result of the afore-mentioned interplay between the twoprepositions. 78.2 The Central SenseThe central meaning of the od–do construction is motivated by the central spatialmanifestations of each of the two prepositions. 8 The trajector’s (¼TR’s) location isspecified relative to two landmarks (¼LMs), which occupy two different positionsin space (see Fig. 8.1).The following examples instantiate the central meaning of od and do, depictingthe TR’s movement between two different LMs, which respectively correspond tothe starting point and end point of the TR’s relocation.11. Od Łodzi do Torunia jechali pociągiem. ‘They went by train from Łódź toToruń.’12. Dziewczynka pobiegła od mamy do taty. ‘The girl ran from her mother to herfather.’13. Od kuzynki pojechałam do sklepu. ‘From my cousin’s I went to the store.’14. Skakać od jednej latarni do drugiej ‘to jump from one lamp post to another’15. Pobiec od drzewa do drzewa ‘to run from one tree to another’It should be noted that the present meaning does not seem to be represented bypurely orientational examples associated with some of the uses of the primaryFig. 8.1 The proto-scene forthe od–do constructionLM1TRLM27 Put differently, sometimes we can speak of a less conventionalized association of senses than inother instances, yet for the sake of the completeness of the picture all these cases will be includedin the discussion.8 In addition, there is also the od–do phrase, which has begun to function as a unit, especiallyin informal language (e.g. pracować w godzinach od–do ‘to have fixed hours of work’, limitod–do ‘a fixed limit’, and Stowarzyszenie OD-DO ‘The TO-FROM Association’).

106 D. Bębeniecsenses of the individual prepositions, 9 which is not surprising inasmuch as thecompatibility between od and do follows from the fact that each of them encodesone essential component of a motion event. Furthermore, owing to the character ofthe TR-LM relation in the case of od, the central meaning of the od-do constructiondoes not cover cases in which the TR’s route starts within a particular LM, which inexamples such as (11) gives rise to the interpretation involving some spatialsegment of a larger path extending beyond both the starting point provided by theod-phrase and the end point supplied by the LM of do. Symmetrical as the twoprepositions are, it appears that do more readily than od participates in suchreadings where the TR is at some stage in contact with the inside of the LM.In this respect, od competes with the preposition z(e), and itself admits theTR’s relocation from the confines of the LM only in instantiations involving thePERSON FOR PLACE metonymy, such as (13) above. 108.3 Other Meanings8.3.1 The Spatial Repetitive SenseThe first of the extended meanings, labelled as the Spatial Repetitive Sense, seems tobe related to such instantiations of the central meaning in which the implicature ofrepetitiveness is detectable, as for example in (14) and (15) above. Recurrence isespecially manifested in the contexts in which the TR moves from one entity toanother (as well as from entity to entity, to capture the characteristics of form), withthe two entities corresponding to two or more distinct tokens of the same type (seeFig. 8.2 below). Besides, the contribution of imperfective verbs is also not insignificant,the more so as it seems to be the concomitant of the present meaning. Theinstantiations representing the Spatial Repetitive Sense, quoted below, display varieddegrees of spatiality, which is especially visible in examples (18), (19) and (20).16. chodzić od wsi do wsi/od sklepu do sklepu/od wystawy do wystawy ‘to go fromvillage to village/from shop to shop/from one exhibition to another’ 1117. Listonosz chodził od domu do domu/od drzwi do drzwi. ‘The postman walkedfrom house to house/from door to door.’9 A relation of orientation is exemplified for instance by the sentence Stał odwrócony tyłem od/dowidowni ‘He stood with his back towards the audience’.10 It thus appears that the contexts of the prepositions od and z(e) highlight different aspects in thesemantics of do, which could respectively be rendered as a boundary and a goal (cf. jechać od Łodzido Torunia and jechać z Łodzi do Torunia ‘to go from Łódź to Toruń’). Notice that in English, dueto the broader scope of the preposition from, such nuances are generally lost, or rather they areretrievable from the general context of the utterance rather than the relevant prepositional meaning.11 A similar meaning could be expressed by the preposition po followed by a plural noun,e.g. chodzić po domach/po sklepach/po wystawach ‘to go to different houses/shops/exhibitions’.As noted by Przybylska (2002: 465, 467), po is used to refer to situations where the TR moves fromone container-like entity to another, or from one person to another, and the LM constitutesa collection of containers or persons.

8 On the Interplay Between Prepositional Categories 107Fig. 8.2 The implicature ofrepetitivenessLMTR...LMTR18. Być odsyłanym od Annasza do Kajfasza ‘to be driven from pillar to post’ 1219. Tułał się od pracy do pracy. ‘He wandered from job to job.’20. Przez wiele lat poniewierała się od krewnych do krewnych. ‘She wanderedfrom relatives to relatives for many years.’8.3.2 The Spatial Limitative SenseIn another spatial meaning, instead of the TR’s relocation from one point to anotherwe can observe the notion of a path stretching between two positions in spaceor between two spatial limits. The Spatial Limitative Sense is thus compatible withthe information retrievable from the central spatial meaning of the current network,which supplies the starting point and end point of the TR’s relocation. 13 However,path, as used here, is no longer part of any motion event, but it corresponds toa certain demarcated region where a specific state or state of affairs prevails orwhere a particular activity is performed. As the following examples demonstrate,motion, if at all present, is of a fictive nature. 1421. Motorówki patrolowały Wisłę od Warszawy do Płocka. ‘Motor boats patrolledthe Vistula from Warsaw do Płock.’22. Ścieżka była zagrabiona od jednej ambony do drugiej. ‘The path was rakedfrom one hide to another.’23. Wiedzą o tym od USA do Sri Lanki. ‘They know about it from the US to SriLanka.’12 This example makes reference to some historical events, when Jesus was arrested and taken firstto the high priest Annas and then to his son-in-law Caiaphas to stand trial before the Sanhedrinprior to his Crucifixion. The English translation, in turn, most probably refers to the ancient gameof real tennis involving both pillars and posts, though its etymology is a matter of dispute.Interestingly, the German equivalent of the phrase is “von Pontius zu Pilatus” (from Pontius toPilate), which in yet another way conveys the futility of somebody’s actions.13 According to Tyler and Evans (2003: 217–218), the concept of path presupposes the connectionbetween the locational source and the locational goal.14 See for example Talmy (2000).

108 D. Bębeniec24. Łąka ciągnie się od jeziora aż do lasu. ‘The meadow stretches from the lake tothe forest.’25. Uśmiechać się/ziewać od ucha do ucha ‘to smile/yawn from ear to ear’26. Te spodnie są proste od góry do dołu. ‘These trousers are straight from top tobottom.’8.3.3 The Spatial Holistic SenseAs can be observed from, say, examples (25) and (26) above, the spatial extentprovided may pertain to the whole of a particular entity, especially when it refersto some salient outermost parts of the entity concerned. Such instantiations, inwhich a given spatial region could not be extended any further, appear to constitutea bridging context for the Spatial Holistic Sense, where the holistic interpretationpredominates over the notion of spatial extent. The main focus of the presentmeaning is on the accuracy or wide scope of a particular action or state, ratherthan on its spatial limits.27. Przeszukali go od góry do dołu. ‘They searched him from top to bottom.’28. Wczoraj umyłam kuchnię odgóry do dołu. ‘Yesterday I cleaned the kitchenfrom top to bottom.’ 1529. Zamierzamy wyremontować mieszkanie od góry do dołu. ‘We are planning toredecorate our home from top to bottom.’30. Przeczytaliśmy ten przewodnik od deski do deski. ‘We read this guide fromcover to cover.’31. Od stóp do głów była ubrana na niebiesko. ‘She was dressed from head to toe inblue.’32. Ten film jest przebojem od jednego krańca świata do drugiego. ‘The film isa smash hit all over the world.’8.3.4 The Limitative SenseThe Limitative Sense, for its part, is basically predictable from the limitative sensesof the prepositions od and do and as such, it appears to belong to lessconventionalized senses in the semantic network of the od–do construction. Here,the TR fluctuates between the two values provided by appropriate LMs, and thesevalues are typically ranked in ascending order, with the od-phrase and do-phraserespectively providing the lowest and highest possible amount or quantity. 16 Since15 Notice that the action of cleaning is not necessarily performed from one end to another or in thevertical direction suggested by the choice and order of the LMs.16 Since both LMs, taken together, determine the scope of the TR, which is to say that the TR isdescribed with reference to a certain closed set of possible values, it is also conceivable that the

8 On the Interplay Between Prepositional Categories 109the TR may also assume some intermediate values (i.e. the ones positioned anywherein the region demarcated by the LMs), the present meaning can also be usedto express ballpark estimates, as visible in example (37) below.33. Weź od 2 do 4 jabłek. ‘Take from 2 to 4 apples.’34. Mężczyźni w wieku od 15 do 40 lat ‘men aged from fifteen to forty’35. Ceny pralek wahają się od 1,000 do 5,000 złotych. ‘The prices of washingmachines range from 1,000 to 5,000 zloty.’36. Stężenia niebezpiecznych substancji były od 10 do 300 razy wyższe oddopuszczalnych. ‘The concentrations of dangerous substances were from 10to 300 times higher than permitted.’37. Mógł mieć od 20 do 25 lat. ‘He could be 20–25 years old.’8.3.5 The Range SenseThe Range Sense is apparently motivated by such contexts of the previous limitativemeaning in which the quantitative distance between two relevant LMs is so significantthat it produces an implicature of a multitude of alternatives that could beadopted by the TR. 17 The present meaning thus encompasses instantiations wherethe TR ranges widely between the two LMs, which are in turn not necessarily selectedfor the purpose of stating some limits, but rather with a view to highlighting diversity.reverse ordering will be used, as for example in Ceny pralek wahają się od 5,000 do 1,000 złotych‘The prices of washing machines range from 5,000 to 1,000 zloty’. Admittedly, such an arrangementappears somewhat marked in comparison with examples such as (35) above. Moreover, partof the present meaning might also be instantiations that indicate the internal organization orordering of the TR, such as for example the following: od najmniejszego do największego ‘fromthe smallest to the biggest’, idiomy od A do Z ‘idioms from A to Z’ and zestaw witamin i minerałówod A do Z ‘a combination of vitamins and minerals from A to Z’. The last two examples, onaccount of the character of the LMs, additionally give rise to a holistic, though non-spatial,interpretation, and as such, they could also be classified as instantiations of the subsequentmeaning (i.e. the Range Sense).17 As far as the present meaning is concerned, it should be observed that it competes with the od-poconstruction (cf. Można tu kupić wszystko – od ubrań po kosmetyki ‘You can buy anything here -from clothes to cosmetics’). Other possibilities for expressing diversity encompass the followinglexicalized expressions: zaczynając od– kończąc na(e.g. różne gatunki literackie, zaczynając odwierszy dla dzieci a na powieściach kończąc ‘various genres, ranging from children’s verse tonovels’) and począwszy od – poprzez – (aż) po (e.g. Sklepy pełne są upominków z HarrymPotterem, począwszy od breloczków, naszyjników, poprzez kubki, kielichy, aż po miotły ‘Shopsare full of Harry Potter souvenirs, ranging from key rings and necklaces through mugs and gobletsto besoms’). As noted by Przybylska (2002: 484–485), the prepositions od and po also co-occur inspatial contexts, in which case the TR of po is conceptualized as a surface extending in manydirections towards the LM which constitutes a linear boundary, as in one of Przybylska’sexamples, Nizina Polska ciągnie się od Odry po Bug ‘The Polish Lowlands stretch from theOder to the Bug’.

110 D. BębeniecBesides, the LMs are characterized by homogeneity in that they are comprised bya single domain or some other coherent subset of experience. 1838. Ich okazałe kwiaty mają różne barwy, od czystej bieli przez wszystkie odcienieróżu do ciemnej czerwieni. ‘Their splendid flowers vary in colour from purewhite through various pinks to dark red.’39. Nasz zespół zajmuje się wszystkim – od pisania do składu komputerowego.‘Our team does everything – from writing to computer typesetting.’40. Członkowie tej partii wywodzą się zróżnych środowisk, od Sasa do Lasa. ‘Themembers of this party are drawn from various backgrounds, as different aschalk and cheese.’ 1941. “Od buta Gortata do opaski Podolskiego” ‘“From Gortat’s shoe to Podolski’sarmband”’ 2042. “Od akordeonu do zasmażki” ‘“From an accordion to a roux”’ 218.3.6 The Temporal SenseThe Temporal Sense of the od–do construction is derivable from the respectivetemporal meanings of the two prepositions. It expresses duration of an action, stateor event, delimited by two points in time which indicate the beginning and end ofa particular action. The time period encoded by od in tandem with do may begrounded in the future, present or past. Moreover, the instantiations of the TemporalSense are less dependent on the moment of speaking than is the case with thetemporal meanings of the individual prepositions, where the moment of speakingoften indicates either the beginning or end of duration (that is, the other referencepoint not encoded by the preposition in question). 2243. Od marca do maja będę za granicą. ‘I will be abroad from March to May.’44. Zajęcia trwają od piątkowego wieczoru do pory obiadowej w niedzielę. ‘Theclasses last from Friday evening to Sunday lunchtime.’18 As can be observed in example (42), however, the LMs that do not typically belong to a commoncategory can be combined for humorous effect.19 It appears that the primary motivation for this example is no longer active, as it was derived fromthe political situation in Poland in the first half of the eighteenth century, when the Polish nobilityat one time supported Augustus II Strong in his claims to the throne, only to turn to StanisławLeszczyński soon afterwards. If so, the original meaning of the phrase probably expressedvacillation between two different options or states, and as such, it would be classified as aninstance of the Change-of-state Sense discussed below.20 This example is the title of an Internet article describing how scores of sportspeople paid tributeto the victims of the plane crash in Smoleńsk on 10 April 2010 (retrieved 12 April 2010, fromhttp://www.tvn24.pl/).21 The title of one of the albums recorded by the Polish cabaret band “Otto”.22 Cf. Czekam na ciebie od godziny ‘I’ve been waiting for you for an hour’ and Masz jeszcze dużoczasu do występu ‘You have plenty of time before you performance’.

8 On the Interplay Between Prepositional Categories 11145. Jej mąż był brytyjskim ambasadorem w Grecji od 1998 do 2001. ‘Her husbandwas the British ambassador to Greece from 1998 to 2001.’Moreover, some temporal phrases tend to highlight the holistic view of aparticular event, especially when they refer to the natural beginning and end ofthis event (or of one of its fundamental parts). By extension, some of these phrasesare used to express the entirety of a given state or situation, as in examples (48) and(49) below.46. Koncert był pasjonujący od pierwszej do ostatniej nuty. ‘The concert wasthrilling from the first to the last note.’47. Obejrzeć film od początku do końca ‘to see a film from beginning to end’48. Być uczciwym od początku do końca ‘to be honest from start to finish’49. Być drużyną od początku do końca ‘to be a team from start to finish’ 23Consider also the following example, where the duration of an event is delimitedby spatial rather than temporal LMs.50. Spała od Warszawy aż do Gdyni. ‘She slept from Warsaw to Gdynia.’ 24Furthermore, as indicated by (51) below, the Temporal Sense has a repetitivevariant, which may be motivated by the shift in focus from duration delimited onboth ends just to those two salient points.51. Od czasu do czasu/od przypadku do przypadku/od wypadku do wypadku/odokazji do okazji ‘from time to time’ 25Finally, in the contexts where the LMs are conceptualized as time periods ratherthan points in time, the temporal meaning of the od–do construction acquiresa limitative interpretation, and, just as the Limitative Sense discussed above,it can also be used to express approximate values, as in the example below.52. Czas oczekiwania wynosi od 2 dni do 2 tygodni. ‘The waiting time is from2 days to a fortnight.’23 Admittedly, it may also be the case that these examples omit to mention a background eventwhose duration determines the overall temporal extent of a state or situation evoked, e.g. Musimybyć drużyną od początku do końca mistrzostw ‘We have to be a team from the beginning of thechampionship to the end’. Nevertheless, the attested productivity of the phrase od początku dokońca indicates that it is approaching unit status.24 Langacker (2008: 74–75), discussing a similar spatio-temporal example, i.e. She’s been asleepfor thirty miles, asserts that it could be explained by reference to the viewer’s motion, which itselfconstitutes a modification of the canonical (stative) viewing arrangement.25 Consider also the sentences Z˙yli od wypłaty do wypłaty/od cudu do cudu ‘They lived from handto mouth/from one miracle to another’, which perhaps do not express frequency as such, but theystill refer to the repetitiveness of an event of a particular type, marking an important stage inanother event (in this case, one’s existence). So the focus here, just as in the phrase od czasu doczasu ‘from time to time’, is on the salient points and not on the duration of an event taking place inbetween these temporal points.

112 D. Bębeniec8.3.7 The Change-of-State SenseThe Change-of-State Sense appears to be motivated by such instantiations of thecentral meaning in which the TR moves from one entity to another, and thus, giventhe experiential correlation between states and locations, may easily be exposed to aparticular state or circumstances associated with a particular LM. The presence oftwo LMs in the region of the TR’s relocation, in turn, is likely to create animplicature of the TR’s transition from one state to another. For instance, thefollowing quasi-spatial sentences present conceptual proximity and distanceobtaining between two states rather than between representations of spatial entities.53. Od miłości blisko jest do nienawiści. ‘There is a thin line between love and hate.’54. Od pomysłu do realizacji jest długa droga. ‘It is a long way from an idea to itsimplementation.’Some other instances associated with the present meaning are presented below.Notice that in some cases the order of the LMs appears to be more motivated, as in(56) and (57). In other cases, such as for example in (55), the reversal of the LMswould not result in any significant meaning variation. 2655. Cierpiał na huśtawki nastrojów – od euforii do przygnębienia. ‘His emotionssee-sawed from euphoria to dejection.’56. Od szczegółu do ogółu/od ogółu do szczegółu. ‘from the specific to the general/ from the general to the specific’57. Od nowicjusza do eksperta ‘from novice to expert’58. Najwyższy czas przejść od słów do czynów. ‘It’s high time to put words intoactions.’8.3.8 The Progression SenseThe Progression Sense, apparently related to the previous meaning, expressesa transition from one state to another. There is a causal connection between thetwo relevant states, and together they amount to or precipitate yet another state orevent. Alternatively, it may be the case that the states at issue represent a certainqualitative progression or accumulative sequence, as a result of which (and as partof the ordinary course of events) a particular final state, not necessarily a positiveone, is reached.26 The present meaning also seems to be very productive, as it often occurs in article titles, booktitles, conference session titles and headlines. Among the recently encountered instantiations thereare for example the following: od cukru do octu ‘from sugar to vinegar’, od rośliny do lekunaturalnego ‘from a plant to a natural drug’, od biletu do cywila ‘from a ticket to civvy street’, odbezradności do bezpieczeństwa ‘from helplessness to safety’, od stresu do pewności siebie ‘fromstress to self-confidence’ and od chaosu do porządku ‘from chaos to order’.

8 On the Interplay Between Prepositional Categories 11359. Od słowa do słowa – i pokłócili się. ‘One thing led to another and they hada fight.’60. I tak, od łyczka do rzemyczka, powoli zaczął się ich romans. ‘And so, one thingled to another and their affair began.’ 278.4 The Semantic Network of the od–do ConstructionOverall, as presented in Fig. 8.3 below, the semantic network of the od–do constructionconsists of the central meaning and as many as eight distinguishable senses.Fig. 8.3 The semanticnetwork for the od–doconstruction8.5 Concluding RemarksTo conclude, it seems plausible to claim that the od–do construction constitutesa distinct polysemous category including several interrelated meanings. As such,the od–do construction displays both intra- and inter-categorial fuzziness. Firstly,the senses discussed are on the one hand distinguishable, and on the other, they form27 This example is part of a larger proverb, saying that starting from misdemeanours, one mayeasily get involved in serious offences and end up being hanged in consequence (cf. Od łyczka dorzemyczka, od rzemyczka do koniczka – będzie szubieniczka).

114 D. Bębenieca continuum of uses, which pertains not only to the pairs of meanings in which onesense is claimed to motivate the other, but can also be discerned in the similarimplicatures that arise even when the relevant senses do not seem to be directlyrelated (as is the case with the limitative, repetitive and holistic variants emerging ina number of different meanings). Secondly, the whole constructional networkcompetes at various points with other ablative and adlative prepositionalconstructions, such as the following:• The z-do construction (e.g. Inflacja spadła z 10 do 8 procent ‘Inflation wentdown from 10 to 8%’, Z dalekiej podróży do własnych kątów ‘From a longjourney to one’s own place’) 28• The z-w construction (e.g. wpadać z jednego dołka w drugi ‘to fall from one holeinto another’)• The z-na construction (e.g. zbiegać zgórki na pazurki ‘to hurtle helter-skelterdown the hill’, zgóry na dół ‘from top to bottom’, z dnia na dzień ‘from day today’, Dobry tancerz chodzi z wesela na wesele ‘He who dances well goes fromone wedding reception to another’) 29• The od-po construction (e.g. od morza po góry ‘from the sea to the mountains’)• The od-na construction (e.g. od wschodu na zachód ‘from the east to the west’)• The od-ku construction (e.g. od dołu ku górze ‘from the bottom to the top’)• The z-ku construction (e.g. z niemocy ku sprawności ‘from powerlessness toeffectiveness’)Furthermore, it should be noted that even though some of the members of theod–do construction do not appear to be fully conventionalized, since for examplethe central meaning can be predicted on the basis of the individual semanticnetworks of the prepositions od and do, their role for the category at issue canhardly be neglected. Without positing the primary spatial meaning, for instance,it would not be possible to account for the motivated character of a number ofextended senses. As far as the remaining meanings of this kind are concerned,recall that the Limitative Sense seems to have contributed to the emergence ofanother distinct meaning and the Temporal Sense includes several variants, noneof which could be explained without positing the two meanings in the semanticnetwork of under discussion.Finally, since the foregoing analysis draws on the study of the category ofdirectional prepositions, it hopefully testifies to the benefits of analyzing prepositionalcategories displaying some common patterns. Specifically, the juxtapositionof prepositions that encode either similar or opposite spatio-configurationaland functional content may contribute to the emergence of some distinctivecharacteristics that could not be brought to the fore unless embedded in a largercontext.28 The title of some rehabilitation programme for Children of Alcoholics, encountered 15 April2010 at http://www.od-do.org/.29 This example is a translation of the Spanish proverb “Quien bien bayla, de boda en boda seanda”.

8 On the Interplay Between Prepositional Categories 115ReferencesBębeniec, D. 2010. Directional prepositions in Polish and English. Towards a cognitive account.Unpublished PhD dissertation, Maria Curie-Skłodowska University, Lublin.Croft, W. 2001. Radical construction grammar. Oxford: Oxford University Press.Dąbrowska, E. 1996. The spatial structuring of events: A study of Polish perfectivizing prefixes. InThe construal of space in language and thought, eds. M. P€utz and R. Dirven, 467–490. Berlin:Mouton de Gruyter.Evans, V. and A. Tyler. 2004a. Rethinking English ‘prepositions of movement’: The case of to andthrough. InAdpositions of movement, eds. H. Cuyckens, W. de Mulder and T. Mortelmans,247–270. Amsterdam: John Benjamins.Evans, V. and A. Tyler. 2004b. Spatial experience, lexical structure and motivation: The case of in.In Studies in linguistic motivation, eds. G. Radden and K. Panther, 157–192. Berlin/New York:Mouton de Gruyter.Fillmore, Ch. 1988. The mechanisms of construction grammar. Proceedings of the BerkeleyLinguistic Society 14: 35–55.Goldberg, A. 1995. Constructions: A construction grammar approach to argument structure.Chicago: Chicago University Press.Goldberg, A. 2006. Constructions at work: The nature of generalization in language. Oxford:Oxford University Press.Kay, P. and Ch. Fillmore. 1999. Grammatical constructions and linguistic generalizations: TheWhat’s X doing Y construction. Language 75: 1–34.Langacker, R. 2008. Cognitive grammar: A basic introduction. New York: Oxford UniversityPress.Przybylska, R. 2002. Polisemia przyimków polskich w świetle gramatyki kognitywnej. Kraków:Universitas.Przybylska, R. 2006. Schematy wyobrażeniowe a semantyka polskich prefiksów czasownikowychdo-, od-, prze-, roz-, u-. Kraków: Universitas.Suchostawska, L. 2005. Space and metaphor: Polish verbal prefixes expressing the relationsINTO-OUT OF and TO-AWAY FROM. Wrocław: Oficyna Wydawnicza PolitechnikiWrocławskiej.Talmy, L. 2000. Toward a cognitive semantics. Vol. 1: Concept structuring systems. Cambridge,MA: MIT Press.Tyler, A. and V. Evans. 2003. The semantics of English prepositions: Spatial scenes, embodiedmeaning and cognition. Cambridge/New York: Cambridge University Press.Tyler, A. and V. Evans. 2001. Reconsidering prepositional polysemy networks: The case of over.Language 77: 724–765.

Chapter 9Governed Prepositions in English:A Corpus-Based StudyJerzy GaszewskiAbstract The paper is a corpus-based analysis of English governed prepositions,which are, due to their considerable frequency, an important part of Englishgrammar. It starts with a discussion of the concept of government, its differentapplications in linguistics and confusion it often creates. Government is a syntacticrelation in which one constituent determines the grammatical form of itsdependent. It also has a semantic side to it – the grammatical marking resultingfrom government denotes the relational meaning of the dependent in the contextof the given head word. The paper presents and discusses results of a series ofcorpus case studies. Each case study examined a grammatical pattern (a headword with a dependent preposition) against a sample of occurrences of thehead from the British National Corpus. Detailed analysis of the results aimed todecide whether the given pattern involves government. A general objective wasto identify quantitative and qualitative characteristics of government (and nongovernment).From the empirical perspective of language corpora, governmentseems to be best envisaged as a gradable phenomenon. While all case studiesshowed considerable variation in the observed patterns of the head word, much ofit is explicable by syntactic, pragmatic or stylistic considerations and does notcontradict government. However, some of the tested patterns were judged to benon-governed. These involved inter alia the tested pattern being part of anantonym set or contextual synonym set.The research has been financed from the budget funds for science in the years 2010–2012 as aresearch project.J. Gaszewski (*)University of Łódź, Łódź, Polande-mail: jgaszewski@op.plM. Pawlak and J. Bielak (eds.), New Perspectives in Language,Discourse and Translation Studies, Second Language Learning and Teaching,DOI 10.1007/978-3-642-20083-0_9, # Springer-Verlag Berlin Heidelberg 2011117

118 J. Gaszewski9.1 IntroductionThe present paper reports results of a series of case studies which form a corpusbasedanalysis of the syntactic phenomenon of government. The exact focus is ongoverned prepositions, i.e. those instances of English prepositions which are fixedformally by their syntactic heads. Governed prepositions are an important part ofEnglish in statistical terms. Just those that have verbal heads (see Sect. 9.2.2) occuralmost 5,000 times per million words. They are common across registers andoutnumber phrasal verbs 1 (Biber et al. 1999: 415).Each case study featured a syntactic pattern (a head word plus a dependentpreposition) which was tested against a sample of occurrences of the head from theBritish National Corpus. Altogether, the case studies involved more than 4,500excerpted examples. The analysis of each case study aimed to decide whether thegiven preposition is governed. The general goal was to pinpoint quantitative andqualitative differences between governed and non-governed patterns. Authenticlanguage data present a number of problems when theoretical constructs are appliedto them. Still, the paper offers a way of deciding which patterns involve governmentand which do not. It transpires, however, that (non-)government is actually gradable.A continuum might be a more appropriate model than a discrete binaryopposition.The paper deals with government in English but as it is a part of a larger Polish-English contrastive project, reference will also be made to ideas coming fromPolish linguistics.9.2 Theoretical Issues9.2.1 The Notion of GovernmentThe concept of government stems from classical grammar (Latin rectio). Let usbegin with an overview of definitions from various linguistics encyclopaedias:[T]he way in which the use of one word requires another word to take a particular form,especially in highly inflected languages. In Latin, prepositions govern nouns: ad is followedby an accusative of movement (ad villam ‘towards the villa’), in by either an accusative ofmovement (in villam ‘into the villa’) or an ablative of location (in villa ‘in the villa’)(Chalker 1992: 445).[A] type of grammatical relation between two or more elements in a sentence, in whichthe choice of one element causes the selection of a particular form of another element.(...) In German, for example, the preposition mit “with” governs, that is requires, the1 The decisive test for a sequence being a prepositional verb (and not phrasal) is that a pronominalobject is placed after the preposition (Biber et al. 1999: 404–5; cf. also Huddleston 2002: 281).Phrasal verbs are beyond the scope of this paper.

9 Governed Prepositions in English: A Corpus-Based Study 119dative case (...): Peter kam mit seiner Schwester. “Peter came with his sister”. Wheresein “his” has the dative feminine case marker –er (Richards et al. 1992: 158–9).[A] kind of syntactic linkage whereby one word (or word class) requires a specificmorphological form of another word (or class). For example, prepositions in Latin aresaid to ‘govern’ nouns (Crystal 1997: 171).[A syntactic relation in which] the governing constituent is linked with the governed onefreely while the governed constituent must be joined with a syntactic marker whose form isdetermined by the governing constituent (Karolak 1999: 503; trans. JG).The shared element of the definitions is the idea that one syntactic constituent(dependent) must take a particular grammatical form which is specified by someother constituent (head). A few clarifications are in order. By far, the best termto use is “constituent” and not “word”, which is an ambiguous term. Moreover,government may involve constituents that are larger than word by any standard.PPs, under scrutiny here, are such multi-word constituents. There is also asignificant difference between relations that link words and word classes (cf.Crystal’s definition above). Government as understood here applies only to theformer. The governing constituent is in each case a particular lexeme and itsrequirement of a given grammatical form of the dependent is its lexical property.To use an example from Palmer (1971), the Latin verb pareo, parere ‘to obey’requires one of its arguments to be in the dative case and “this has to be indicatedin the dictionary” (103). In English, we observe that e.g. the verb approve needsone of its arguments to be a PP with of. Similarly, famous requires an analogicaluse of for, etc.If government seems a neatly defined term now, the impression is misleading. Inactual practice it is used in different ways, which causes confusion. Traditionallythe term was underused – applied virtually only to verbs and prepositions governingcases (cf. Karolak 1999: 504). Today there is a tendency to overuse it. Dixon(2010), in his harsh criticism of the term (“linguistics would better off withoutit”), notes that it is typically said that for instance in the sentence She took coals tothe fireplace the verb take governs the object coals (231). Since there is no formalmarking of the object here, this has little to do with government applied rigorously.The notion of syntactic dependency suffices. In generative linguistics government ispractically adopted in its traditional meaning (see Cook and Newson 1996: 50–1),but the formal definition refers ultimately to the lexical class of the “governor” andthe relation of “c-command” (240). Thus, the relation of government holdingbetween two syntactic elements follows directly from the structure of the phrasemarker and it need not manifest itself formally at all. 2A crucial issue is the applicability of government to English. Crystal (1997) andChalker (1992) explicitly deny it. Since government has typically been associatedwith selecting a case form from a paradigm, such opinions are natural. However, ifwe do not restrict the scope of government to single-word dependents, we find anumber of analytic dependents in English that fit the notion perfectly. Governed2 The paper does not adopt the generative concept of government developed primarily by Chomsky(1981).

120 J. Gaszewskiprepositions emerge as one of the most significant groups here. 3 Thus, approvegoverns of, famous governs for, etc., much like pareo, parere governs the dativecase in Latin. 4Although government is a syntactic phenomenon, it has a semantic side to it.First of all, many lexemes license a number of dependents with different meaningsthat may all be governed. A classic example is the German verb geben ‘to give’which has two objects – one in the accusative, the other in the dative. The formerdenotes the theme and the latter the recipient, e.g.1. Er hat mir einen Kuli gegeben ‘He gave me a pen.’Mir is the dative of ich ‘I’, and einen Kuli bears the accusative masculinemarker –en. In real-life communication, which noun phrase is which participantwould be deducible from general knowledge and context. However, both objectshave fixed grammatical forms (i.e. they are governed). Each case renders aseparate semantic content – this is semantics behind government. 5The meaning of the head is important for government as well. A polysemousword may have different government patterns for different meanings – compareask (‘pose a question’) about and ask (‘request’) for (cf. Lesz-Duk 1998: 23; andalso Latin in in Chalker’s definition above). Thus, government holds between ahead lexeme with a specified meaning and a particular syntactic marker of itsdependent. 69.2.2 Governed Prepositions (?)Governed prepositions are the very theme of this paper, but their existence is notuncontroversial, whence the question mark in the title of the section. In fact, recentmonumental works on English grammar adopt opposite views. For example, theverb wait typically co-occurs with the preposition for as in the phrase wait for me,which can have two constituent analyses:3 Chalker (1992) mentions the possibility of such an extension. See also Karolak’s (1999) criticismof the inconsistencies in the traditional use of the term. Amendments much in line with hisarguments can be found in Saloni and Świdziński (1998). Also Gołąb et al. (1968) include PPsin the scope of government (519).4 This approach assumes that prepositions are like modifiers (or “analytic inflections”) that serve tolink the following NPs to superordinate structures rather than like true legitimate heads of phrases.This is in fact advocated by Biber et al. (1999: 74) and also Dixon (2010: 49, 232). NB some worksin Polish linguistics use the term “prepositional-nominal phrase” (trans. JG) for prepositionalphrases, cf. the very titles of Kosek (1999) and Grybosiowa (1979). The perspective adopted hereruns contrary to the one presented by Pullum and Huddleston (2002: 598–9).5 Exactly the same analysis applies to equivalents in other languages, e.g. Latin do, dare or Polishdać/dawać.6 Latin pareo, parere is also polysemous. The meaning ‘to obey’ has an object in the dative and themeaning ‘to appear’ does not.

9 Governed Prepositions in English: A Corpus-Based Study 1212. [wait for] [me]3. [wait] [for me]The first analysis assumes that the sequence contains a multi-word verb (called“prepositional verb”) wait for and its object (Biber et al. 1999: 74, 414). The secondoption prefers to envisage wait as a single-word verb and for me as its prepositionalobject. The label “prepositional verb” can be kept, now referring to wait only andimplying that it takes a PP as an object (Huddleston 2002: 274).Arguments can be raised for both analyses. The first one is supported by thesupposed semantic unity of the sequence of the verb and preposition. The allegedlymulti-word verb wait for can be replaced by the single-word synonym expect. 7“Semantic unity” is an unclear concept though - Herbst (1999) dismisses it altogether.The argument ignores the fact that the verbs in question are used without thepreposition:4. So you have been waiting. 8An elliptical sentence like (4) confirms that the verb is the locus of the meaning andthe preposition marks the dependent when it is present. Huddleston (2002:275–276) gives several structural arguments for the second analysis such as mobilityof the prepositional phrase as a whole and insertion of adjuncts, as in:5. Terry waited impatiently for me.Huddleston (2002: 277) explicitly rejects the first analysis, as does Herbst(1999). Another structural argument for the second analysis is that derivedwords and phrases typically involve only the verbal element, cf. waiting roomand not *waiting-for room. Finally, the second analysis is more encompassing. Itcan be easily applied to heads other than verbs as in:6. [allergic] [to dust]7. [attempt] [at denial]The extension of the first analysis would mean assuming, rather infelicitously,the existence of a multi-word adjective allergic to and a multi-word nounattempt at. 9 This paper accepts the second analysis. The verb (e.g. wait) takesan object whose form is a PP with a specific preposition (e.g. for). Thus, thepreposition is governed by the verb.7 Biber et al. (1999: 414) give more such pairs, e.g. look like and resemble, think about andconsider.8 All the examples of English sentences in the paper are excerpted from the BNC.9 The data below (Table 9.2) show that with attempt such an analysis is unacceptable for statisticalreasons.

122 J. Gaszewski9.3 Case Studies9.3.1 DesignEach case study involved a pattern of a given head and a putatively governeddependent preposition, e.g. attempt(n) at. Each head was homogeneous syntactically(occurrences of the verb attempt were excluded) and semantically (if attemptmeant ‘attack on a well-known individual with an intention of murder’, the examplewas dismissed). Both heads and dependents were defined in semantic terms (seeTable 9.2).Each head word was queried in the BNC and sampled results underwent manuallabelling that divided them into four groups. The first group contained the testedpattern, e.g.8. In the UN, the inevitable instrument of most attempts at mediation was theSecurity Council resolution (...).The second group included other syntactic patterns (as it were competing withthe tested one) used to render the same meaning of the dependent, e.g.9. Venables has been unable to keep his word because of chairman Alan Sugar’sattempt to oust him from the club.The data may (and usually did) reveal more than one competing pattern, e.g.10. No relief attempt was in evidence.The third group included results not containing any syntactically dependent elementwith the researched meaning, e.g.11. At my next attempt I passed into Eton a whole form from the bottom of theschool.Lastly, the queries also included some irrelevant examples, e.g.12. This series attempts to combine the best of traditional and modern mathematics(...).where the word attempt is a verb but has been misannotated in the corpus as anoun.To sum up this exemplification, (8) was labelled as having “at + NP”, thedependent form of the tested pattern. (9) was labelled as “to + infinitive” and(10) as “preposed noun”, both being competing patterns. (11) fails to realise thesought meaning of the dependent and was labelled “zero pattern”. (12) was classifiedas “irrelevant”. Labelled results were counted and put into percentages showingfractions of all the relevant examples from the given query that each patternaccounted for. The analysis of the tested pattern vs. competing patterns led toclassifying the pattern as either government or not.

9 Governed Prepositions in English: A Corpus-Based Study 123All tested patterns are presented in Table 9.1. The motivation for the choice ofthese particular heads was to represent different lexical classes and to include bothfrequent (e.g. speak) and less frequent words (e.g. conducive).Table 9.1 Overview of all case studiesCase studyAssumed to representRelevantexamplesIrrelevantexamplesAll analysedexamplesAll resultsin the BNCAttempt(n) at Government 424 30 454 13,189Afraid of Government 238 65 303 5,437Blame(v) for Government 488 17 505 3,762Conducive to Government 279 12 291 291Effective in Lack of government 134 271 405 9,763Hide(v) behind Lack of government 182 122 304 5,899Proud of Government 304 98 402 3,007Reason(n) for Disputable government 517 92 609 28,971Speak of Disputable government 803 106 909 24,278Speak to Disputable government 803 106 909 24,278Vote(v) for Lack of government 254 103 357 5,003The paper analyses just a dozen patterns while thousands of English words areavailable as heads. It was then necessary to make sure the few researched headswould show differing behaviour. Thus, some of the analysed patterns were chosenwith an assumption that they would represent government. Others were selected aslikely examples of lack of it. The classification as one or the other was based ongeneral command of English and linguistics practice. Another group was seen asproblematic for an aprioristic decision. These assumptions were only made toensure that the results would vary between the case studies. They did not constituteany research hypothesis per se.Since the analysis involved time-consuming manual labelling of all theexamples, it was necessary to sample query results. 10 Representativeness ofthe sample was obtained by quantifying the labelled examples for a smaller sample,then enlarging it and quantifying it again. The figures for the original and theenlarged sample were subjected to the chi-square test. When the test showed ahigh p-value (very low probability of the differences between the samples beingstatistically significant), it was assumed that the sample was representative andfurther enlarging would not alter the results.9.3.2 ResultsThe tables below provide semantic and statistical details for individual case studies.Tables 9.2–9.6 present the studies that were assumed to represent government.10 For conducive all examples from the BNC were analysed.

124 J. GaszewskiTable 9.2 Results of casestudy of attempt(n) atHead’s meaning: act of putting effort into achieving sthDependent’s meaning: thing one wants to achieveDependent’s structureOccurrencesat+NP/gerund 33 7.78%on+NP 1 0.24%of+NP 1 0.24%Preposed noun 15 3.54%to+infinitive 328 77.36%Zero 46 10.85%Total 424Irrelevant examples 30Table 9.3 Results of casestudy of afraid ofHead’s meaning: feeling fearDependent’s meaning: reason for the feelingDependent’s structureOccurrencesof+NP/gerund 98 41.18%(that)+clause 44 18.49%to+infinitive 43 18.07%Zero 53 22.27%Total 238Irrelevant examples 65Table 9.4 Results of casestudy of blame(v) forHead’s meaning: ascribe responsibility for an undesirablesituationDependent’s meaning: bad situation that is blamedDependent’s structureOccurrencesfor+NP/gerund 240 49.18%NP 62 12.70%at+NP 1 0.2%about+NP 1 0.2%Adverb 1 0.2%that+clause 1 0.2%Zero 182 37.3%Total 488Irrelevant examples 17As the examples of attributive adjectives never realised the tested or competingpattern, they were excluded from the count and labelled as irrelevant. The resultsrefer only to predicative use. This applies to all adjectival heads in the paper.

9 Governed Prepositions in English: A Corpus-Based Study 125Table 9.5 Results of casestudy of conducive toHead’s meaning: being a favourable conditionDependent’s meaning: thing/process favoured by the conditionDependent’s structureOccurrencesto+NP/gerund 268 96.06%for+NP/gerund 2 0.72%of+NP 1 0.36%Zero 8 2.99%Total 279Irrelevant examples 12Table 9.6 Results of casestudy of proud ofHead’s meaning: having the feeling of high self-valueDependent’s meaning: reason for the feelingDependent’s structureOccurrencesof+NP/gerund 172 a 56.58%to+infinitive 58 19.08%that+clause 9 2.96%Zero 65 21.38%Total 304Irrelevant examples 98a The figure for the tested pattern includes the following example,an obvious typo:13. One footnote the new clinic is proud off; surgeons fromelsewhere in Britain are asking for details of Hereford’s runawaysuccess.Table 9.7 Results of casestudy of reason(n) forHead’s meaning: thing that causes another thingDependent’s meaning: thing caused by the reasonDependent’s structureOccurrencesfor+NP/gerund 119 a 23.02%behind+NP 1 0.19%to+infinitive 34 6.58%(that)+clause 47 9.09%why+clause 67 12.96%Zero 249 48.16%Total 517Irrelevant examples 92a The figure for the tested pattern includes the example with anarchaic derivative of for:14. The Captain’s mirth over the Butcher’s Assistant’s love-lifeand the reasons therefore reached the ears of the Old Stager (...).

126 J. GaszewskiTable 9.8 Results of casestudy of speak ofHead’s meaning: utter words to convey meaningDependent’s meaning: topic of speakingDependent’s structureOccurrencesof+NP 86 10.71%about+NP 40 4.98%on+NP 13 1.62%zero 664 82.69%Total 803Irrelevant examples 106Table 9.9 Results of casestudy of speak toHead’s meaning: utter words to convey meaningDependent’s meaning: addressee of the messageDependent’s structureOccurrencesto+NP 183 22.79%with+NP 7 0.87%Zero 613 76.34%Total 803Irrelevant examples 106Tables 9.7–9.9 document the case studies of patterns of “disputable government”.The last three tables present the case studies that were assumed to represent lackof government (Tables 9.10–9.12).Table 9.10 Results of casestudy of effective inHead’s meaning: producing considerable effects for whoseproduction sth is usedDependent’s meaning: activity, domain or field in which theeffects are obtainedDependent’s structureOccurrencesin+NP/gerund 16 11.94%for+NP 4 2.99%at+NP 1 0.75%in terms of+NP 1 0.75%to+infinitive 1 0.75%Zero 111 82.84%Total 134Irrelevant examples 271

9 Governed Prepositions in English: A Corpus-Based Study 127Table 9.11 Results of casestudy of hide(v) behindHead’s meaning: position sth in a particular place to haveprotection from sth and deny access and/or visibilityDependent’s meaning: thing that provides protectionDependent’s structureOccurrencesbehind+NP 11 6.04%beneath+NP 5 2.75%underneath+NP 1 0.55%under+NP 4 2.2%in+NP 37 20.33%inside+NP 1 0.55%within+NP 1 0.55%among+NP 4 2.2%Adverb 18 9.89%under+NP or in+NP 1 0.55%under+NP or behind+NP 1 0.55%Zero 98 53.85%Total 182Irrelevant examples 122Table 9.12 Results of casestudy of vote(v) forHead’s meaning: take part in a formalised democratic decisionmakingprocess and support one/some of the possible optionsDependent’s meaning: option in the decision-making processDependent’s structureOccurrencesfor+NP 85 33.46%NP 27 10.63%in favour of+NP 5 1.97%against+NP 24 9.45%against+NP rather than for+NP 1 0.39%Zero 112 44.09%Total 254Irrelevant examples 1039.4 Discussion9.4.1 General RemarksSeveral general observations are in order. First of all, raw percentages of the testedpatterns vary considerably both with patterns assumed to represent government andlack of it. Thus, percentages taken on their own cannot be significant. All the casestudies involve examples of zero pattern and competing patterns.Examples of zero pattern do not contradict government. A governed dependentmay be unrealised because of ellipsis or underspecification (see (4)). Consider also

128 J. Gaszewski15. Carol with proud dad Johnwhere the noun dad implies the semantic content of the unrealised dependent ofproud.Attested examples of competing patterns prima facie undermine government.If there are several forms utilised for the same meaning of a dependent after thesame head, how can we talk of it requiring a particular form of the dependent?Competing patterns surfaced in all the case studies. Should we conclude that thereis no government in any of the studies? On the other hand, we know that blame“takes” for, proud “takes” of, etc., and the data prove this. Authentic languagematerial is never neat and includes statistical noise. Unanalysed corpus data donot prove anything on their own (cf. Rusiecki 2006). It is necessary to investigateof what kind the competing patterns are, where they occur and, lastly, hownumerous they are.Bearing in mind the unsure status of government in all the case studies, we willproceed from ones where it must clearly be rejected to those where it can be upheldunder certain provisions that apply to government in general.9.4.2 Antonym Sets: Strong Case Against GovernmentThe clearest example of non-government is hide behind. The tested patternaccounts for only 6% of examples, it has almost a dozen competing patterns, andsome of them are more frequent (hide in – c. 20%). We can also have co-ordinationof dependents with different patterns:16. Never hide spare keys under doormats or in flowerpots.17. And to hide the microphone under a table or behind a curtain, whileguaranteeing the naturalness of the content, does little for the acoustic quality(...).A sentence combining any of the patterns would be acceptable.Clearly, hide does not govern behind 11 – a PP with this preposition is not thenecessary form for rendering the thing that provides cover. This result was ofcourse easy to predict. Behind names the exact spatial configuration of the hiddenentity and the cover-provider. Other configurations are possible and these arerendered by the competing patterns. Crucially, behind and its competitors form acoherent contrastive set. Thus, they are not competing patterns, but differentrealisations of the same overarching category – localisation in space. Thus, hidesemantically requires a location argument which may be formally realised in aplethora of ways. 12 Since the actual form of this argument is not specified, there is11 However, hide would likely prove to govern from (with a different meaning).12 Some spatial prepositions did not surface in the results since the spatial configurations theydenote are unsuitable for hiding.

9 Governed Prepositions in English: A Corpus-Based Study 129no government. 13 NB the contrastive set evidenced in the researched dependentsof hide has general validity in the language system.The case study of vote for yields similar results. Vote does not govern for butopens a slot for a coherent contrastive set of prepositions (for, against), which againis generally valid in the language.9.4.3 Contextual Synonymy: Lack of Government (?)The case study of speak of also yielded a couple of prepositions that can all functionas the researched dependents. Apart from speak of we have speak about and speakon. Unlike in Sect. 9.4.2, here we cannot discern any clear semantic differences. Onthe contrary, the prepositions prove synonymous in the context:18. You may have heard people speak about the loss of brain cells as we get older.19. That evening Carter gave a warm toast to the Shah in which he spoke of thegreat importance of Iran’s relationship to the United States.20. (...) that he should not be prevented, by convention or otherwise, fromspeaking in Parliament on such a matter.The contextual synonymy of about and of seems very strong. Speak on seems tostand out somewhat – its occurrences are restricted to phrases with abstract nouns.Moreover, it seems to be marked stylistically (see Sect. 9.4.5).As in Sect. 9.4.2, the researched dependent does not have a fixed formal shape,which suggests there is no government. Yet, we have a very limited set of alternativepatterns. While this case may not meet the definitions of government inSect. 9.2.1, it fits the actual scholarly practice (cf. Lesz-Duk 1988: 19–24; Kosek1999). If we consider speak not to have government, it is clearly a weaker casethan hide behind. An alternative analysis would be that speak has a weak type ofgovernment and allows a fixed limited set of dependent forms instead of just one.Unlike the ones in Sect. 9.4.2, this set is not valid as a synonym set generally(though it is rather regular in the context of verbs of speech).Contextual synonymy abounds in the case studies. 14 Thus, alongside speak to wehave speak with, effective in next to effective for and effective at, attempt at andattempt of, reason for and reason behind. As far as the decisions about governmentgo, all of these are similar to speak of.Let us also consider ‘synonyms’ of blame for:13 This analysis may seem truistic, but the most frequent spatial preposition in a given context tendsto be wrongly assumed to be governed, e.g. Polish do in dochodzić do bramy is governed accordingto Gołąb (1967: 21–22). More instances can be found in Karolak (1999).14 NB the case studies discussed in the previous section also include synonyms, e.g. hide under andhide beneath, which are synonymous generally.

130 J. Gaszewski21. They have been replaced by a Labour leadership keen to blame many of theirproblems at the Government’s door.22. Mrs Smart, 41, hit the headlines in TODAY when, (...), she blamed the Toriesof betrayal after a lifetime of her support.23. (...) there was a lapse with that guy, don’t know who was to blame about thatthird goal but (...).In (21) blame for was undoubtedly mixed with lay the blame at sb’s door –astructure involving the noun blame and not the verb. (22) seems to feature aconfusion of the verbs blame and accuse and was thus labelled “irrelevant” becauseof its semantics. (23) cannot be dismissed in a similar way, but it differs quitesignificantly from the contextual synonyms mentioned earlier. The few examples ofspeak with, effective for, attempt of or reason behind in our sample can be supportedwith more instances in the whole BNC. These are infrequent but fairly regularpatterns existing in the language. Blame about appears in (23) and in one moreexample only, so we may treat it as marginal. 15 It does not undermine the governmentpattern blame for. Similar cases are conducive for and conducive of. Thesingle example of attempt on in our sample,24. If only he could manage to finish the Koran. He had made several attemptson it.is likely a transfer from the other meaning of the noun.Such isolated counterexamples should not count against government. In general,frequency of patterns relative to each other does matter. Speak of occurs onlyroughly twice as often as speak about, but attempt at is around thirty times morefrequent than attempt of. The former contextual synonym set defies governmentmore than the latter. In the case of reason for and reason behind there is aneven greater discrepancy, which cannot be ignored. We could view reason for asa borderline case of government.9.4.4 Pragmatic, Syntactic, Stylistic Factors: Defenceof GovernmentThe case study of blame for featured oddities like (21), (22) and (23). Yet, thecompeting pattern most significant numerically involves a bare NP instead of a PPwith for:25. The Trust blames the redundancies on the recession.15 Cf. Charzyńska-Wójcik’s (2010) discussion of a putative Old English ditransitive verb type.Despite having a single attested example of the type, she dismisses it as an active syntactic patternbecause the example is unique among all Old English ditransitive structures as well as among theoccurrences of the verb it includes (97).

9 Governed Prepositions in English: A Corpus-Based Study 131In passive voice it can be placed as a subject:26. A lot can be blamed on years of dictatorship and rotten planning – but not all.All examples of this kind also include a PP with on that has a different meaning.This does not necessarily contradict the government pattern blame for. (25) and(26) involve a re-shuffling of the thematic structure of a basic clause with blame for.The peripheral argument (for + NP) is raised to the role of direct object, while theoriginal direct object is demoted to a peripheral but obligatory argument (on +NP). 16We conclude that blame allows two mutually exclusive clause patterns, one ofwhich involves governed for.A large group of competing patterns involved dependents of clausal character:27. The Mills can be proud that all their hard work has paid off.28. I am afraid that one day I’ll come to hate even my brothers and sisters.29. All attempts to engineer a dialogue between the two have so far come tonothing (...).It seems that in such sentences the amount of information put into the dependentis rather unwieldy for a PP. Even though a preposition may be combined with agerund, which enables including the meaning of practically the whole clause in aPP, that-clauses and infinitives generally outnumber preposition + gerundsequences across the case studies. Clausal constituents simply handle that load ofinformation better and are the most natural way of rendering it. Thus, we shouldaccept structures alternative to the tested pattern if they are of considerably differentsyntactic character (cf. also Karolak (1999: 504) on infinitival alternants forgoverned PPs in Polish). This remark applies to most of the case studies.9.4.5 Variation in the DataThe data in the case studies showed considerable variation. Sections 9.4.1–9.4.4commented on the most important problems it poses. One also has to remember thata reference corpus presents an averaged version of different regional varieties,sociolects, registers and idiolects and this increases the variation of the extracteddata. Varieties of all these kinds are largely autonomous and may differ at any levelof language structure, e.g. in rules concerning governed prepositions. Variation inthe data may also reflect a diachronic change in progress.Such factors might motivate at least some of the competing patterns found. Forinstance, as mentioned above, speak on seems to belong to the formal register only.16 Similar alternations exist for other verbs in English, e.g. supply with vs. supply to. In languageswhere similar phenomena are more regular and grammaticalised, they are called applicative voice(see e.g. Dixon 2010: 169–70).

132 J. GaszewskiExamples in the sample are formal and written, but their low number does not allowfirm statements here. Exact assessment of these factors is beyond the limitationsof this paper.9.5 ConclusionsThis section sums up the observations made heretofore and attempts to model thedistinction between government and lack of it in general terms. The actual conceptof government used in a given research project depends to some extent on its needs.As is hoped to have been implied so far, the most encompassing analysis is toenvisage government as a gradable phenomenon, some kind of a continuum.However, if discrete distinctions are necessary, there are certain natural cut-offpoints in the continuum.As said before, percentages of the tested patterns taken separately do notenable to decide if a preposition is governed. Government and obligatoriness aredifferent things. There are obligatory, but non-governed dependents, e.g. expressionof location after put, and also governed but optional ones (almost all the examples ofgovernment in the paper).What is significant is the relation between the tested and competing patterns.When they form a coherent set of antonyms, this clearly indicates lack of government.There are also cases in which the competing pattern(s) are contextual synonyms of thetested pattern. This, on principle, points to lack of government. However, if a looserconcept of government is needed, all such contextually synonymous patterns could betreated as examples of variable government. If the less common options in a contextualsynonym set are of very low frequency relative to the most common pattern, theycould possibly be treated as marginal in the language. In that case the most frequentoption would be governed by more rigid standards as well.With some head words we can observe contextual antonymy, which could alsocount against government, e.g. blame for vs. blame on. However, along the lines ofargumentation in Sects. 9.2.1 and 9.4.4, such contrastive dependents should beregarded as separate arguments of the head. They do not conflict with each otherand can all be governed. A similar case is the pattern afraid for found in a handful ofexamples. It certainly contrasts with afraid of, but it was treated as a separateargument licensed by the head. Corpus examples with this structure do not renderthe researched dependent’s meaning and were labelled “zero”.The difference between government and non-government is best conceived of asa continuum. 17 Table 9.13 proposes such a continuum with case studies described inthe paper. A continuum may not fit some research purposes. If there is a need to17 The gradability of government is not a novel idea. Lesz-Duk (1988), for instance, gives anoverview of different concept of government, some of which distinguish “strong” and “weak”government (7, 9). The distinctions are defined in terms of syntactic functions and obligatoriness.Hentschel’s (2003) proposal of a continuum of grammaticalisation status of prepositions is closestto the analysis presented here.

9 Governed Prepositions in English: A Corpus-Based Study 133Table 9.13 Continuum of governmentNon-governmentpoleDescriptionTested and competing patternsform an antonym setTested and competing patternsare contextual synonymsCompeting patterns areacceptable on pragmaticor stylistic groundsRelative frequencyof competingpatterns(High)(Low)Tested patternsHide behindVote forSpeak ofEffective inSpeak toAttempt atReason forBlame forConducive toGovernment poleCompeting patterns (if any) areof syntactically discrepantcharacterAfraid ofProud ofdraw a line between government and non-government, the most advisable place forthis is above blame for in Table 9.13. For statistical reasons government may beextended to include reason for and possibly attempt at. If a loose concept ofgovernment is needed, the dividing line should be placed above speak of.ReferencesBiber, D., S. Johansson, G. Leech, S. Conrad and E. Finegan. 1999. Longman grammar of spokenand written English. Harlow: Longman.Chalker, S. 1992. Government. In The Oxford companion to the English language, ed. T. McArthur, 445. Oxford, NY: Oxford University Press.Charzyńska-Wójcik, M. 2010. The Syntax of Old English clausal ditransitive verbs. In: Verbstructures: Between phonology and morphosyntax, eds. E. Cyran and B. Szymanek, 75–109.Lublin: Wydawnictwo KUL.Chomsky, N. 1981. Lectures on government and binding: The Pisa lectures. Berlin: Mouton.Cook, V. J. and M. Newson. 1996. Chomsky’s Universal Grammar: An introduction. Oxford: BasilBlackwell.Crystal, D. 1997. A dictionary of linguistics and phonetics. Oxford: Blackwell.Dixon, R. M. W. 2010. Basic linguistic theory. Vol. 1. Methodology. Oxford: Oxford UniversityPress.Gołąb, Z. 1967. Próba klasyfikacji syntaktycznej czasowników polskich (na zasadzie konotacji).Biuletyn PTJ XXV: 3–43. Wrocław: Ossolineum.Gołąb, Z., A. Heinz and K. Polański. 1968. Słownik terminologii językoznawczej. Warszawa:PWN.Grybosiowa, A. 1979. Fraza nominalna przyimkowa w funkcji okolicznika przyczyny. Studia zfilologii polskiej i słowiańskiej XVIII: 67–79.Hentschel, G. 2003. Zur Klassifikation von Pr€apositionen im Vergleich zur Klassifikation vonKasus. In Pr€apositionen im Polnischen. Beitr€age zu einer gleichnamigen Tagung, Oldenburg,

134 J. Gaszewski8. bis 11. Februar 2000. (Studia Slavica Oldenburgensia 11), eds. G. Hentschel and Th.Menzel, 161–191. Oldenburg.Herbst, Th. 1999. English valency structures - A first sketch. Erfurt Electronic Studies in English(EESE) 6. http://webdoc.gwdg.de/edoc/ia/eese/artic99/herbst/6_99.html.Accessed30June2010.Huddleston, R. 2002. The clause: complements. In The Cambridge grammar of the Englishlanguage, eds. R. Huddleston and G. K. Pullum, 213–321. Cambridge: Cambridge UniversityPress.Karolak, S. 1999. Rząd. In Encyklopedia językoznawstwa ogólnego, ed. K. Polański, 503–505.Wrocław, Warszawa, Kraków: Ossolineum.Kosek, I. 1999. Przyczasownikowe frazy przyimkowo-nominalne w zdaniach współczasnegojęzyka polskiego. Olsztyn: UWM.Lesz-Duk, M. 1988. Czasowniki o składni przyimkowej w języku polskim. Częstochowa: WSP.Palmer, F. R. 1971. Grammar. Harmondsworth: Penguin.Pullum, G. K. and R. Huddleston. 2002. Prepositions and preposition phrases. In: The Cambridgegrammar of the English language, eds. R. Huddleston and G. K. Pullum, 597–661. Cambridge:Cambridge University Press.Richards, J., J. Platt and H. Platt. 1992. Longman dictionary of language teaching and appliedlinguistics. Harlow: Longman.Rusiecki, J. 2006. Badania korpusowe a ‘lingwistyka gabinetowa’: czy te dwa podejścia kiedyś sięspotkają? In: Korpusy w angielsko-polskim językoznawstwie kontrastywnym. Teoria i praktyka,eds. A. Duszak, E. Gajek and U. Okulska, 17–27. Kraków: Universitas.Saloni, Z. and M. Świdziński. 1998. Składnia współczesnego języka polskiego. Warszawa: PWN.

Part IIIHistorical Linguistics

Chapter 10Reflections on Structural Variationin Old English VerbsMagdalena Charzyńska-WójcikAbstract The paper shows the inapplicability of traditional descriptions of OldEnglish verbs (cf. Visser 1963–73; Mitchell 1995; Ogura 1995) construed in termsof categories and classes suitable for MnE verbs. Classifications based on notionssuch as “intransitive”, “transitive” and “ditransitive” verbs or “direct”, “indirect”and “quasi” objects fail to capture the inherent properties of the syntax of the OldEnglish verb. The data discussed in the paper reveal the inadequacy of the existingclassifications and show the necessity for a new approach to Old English verbs.On the basis of selected Old English verbs and their usages it is shown that the OldEnglish verb exhibits a very high degree of variation as far as the structures itappears in are concerned. This optionality implies that what is needed is a classificationresulting from the study of variation exhibited by a particular verb, since it isprecisely the available variation that should be seen as a defining factor in determiningOld English verb classes.10.1 IntroductionThe aim of the paper is to show that a description of Old English (henceforth OE)verbs (cf. Visser 1963–1973; Mitchell 1985; Ogura 1995) in terms of classificationswhich we apply to Modern English (henceforth MnE) verbs is not only inaccuratebut, more importantly, it allows the inherent properties of OE verbs to go unnoticed.Therefore, we will argue against an account of OE verbs in terms of verb classes,which divides verbs into transitive vs. intransitive and further subdivides the formergroup into monotransitive vs. ditransitive, where still finer distinctions obtainM. Charzyńska-Wójcik (*)John Paul II Catholic University of Lublin, Lublin, Polande-mail: mcwojcik@kul.plM. Pawlak and J. Bielak (eds.), New Perspectives in Language,Discourse and Translation Studies, Second Language Learning and Teaching,DOI 10.1007/978-3-642-20083-0_10, # Springer-Verlag Berlin Heidelberg 2011137

138 M. Charzyńska-Wójcikdepending on what the constituent elements are, i.e. NP(s), PP(s) or CLAUSE(s). Itwill be demonstrated that an OE verb allows such a degree of optionality as far asthe expression of its complements is concerned that a (relatively) straightforwardclassification can apply to a particular usage or structure rather than to a given OldEnglish verb as such. This, however, is only a necessary first step as it is preciselythe available variation exhibited by a particular verb that, we believe, should beseen as a defining factor in determining verb classes.The paper is divided into two major parts. The existing classifications of OldEnglish verbs together with the problems these traditional approaches induce arepresented in Sect. 10.2. Three major authors are discussed here: Visser(1963–1973) in Sect. 10.2.1, Mitchell (1985) in Sect. 10.2.2 and Ogura (1995) inSect. 10.2.3. These traditional classifications are tested against the actual linguisticdata in Sect. 10.3, which falls into two major parts: Sect. 10.3.1, devoted to verbswithout complement, and Sect. 10.3.2, where verbs with an object are discussed.The paper ends with concluding remarks given in Sect. 10.4, which brings togetherall the classification problems encountered in the course of the discussion andattempts at some tentative answers.10.2 Classifications of Old English VerbsTraditional descriptions of verbs (cf. Visser 1963–1973; Mitchell 1985; Ogura1995) which focus on the structure of a clause draw a dividing line betweenintransitive and transitive verbs. Transitive verbs are further subdivided intomonotransitive and ditransitive ones. Monotransitive verbs can be furtherdifferentiated into those which take an NP complement, a PP complement or aCLAUSAL one. The same distinctions can be applied to ditransitive verbs, onlyhere one needs to consider a number of combinations of the two elements: 2NPs,NP + PP, NP + CLAUSE or PP + CLAUSE. For a language with Case markingsuch as OE the number of these combinations for verbs with one or two objectsgrows as the objects can reveal a variety of Cases. In OE NP objects can bear ACC,DAC or GEN Case. Before discussing the actual verbs and verb types we need tohave a look at the available classifications of OE verbs.10.2.1 Visser (1963–1973)Visser’s classification of OE verbs is by necessity superficial – the author dealswith syntactic structures throughout the history of English so he imposes acommon panchronic classification and within these divisions shows furtherintricacies and details for particular periods. As a result, this classification doesnot focus on OE verbs as such but it merely encompasses them. Beforeembarking on a discussion of Visser’s classification of verbs, we need to clarify

10 Reflections on Structural Variation in Old English Verbs 139an important terminological point. Visser (1963–1973: }129) defines the termtransitive as referring to verbs which require a direct object, while the termintransitive is used to mean that a verb “does not take a direct object”, Visser(1963–1973: }418) explains further on that a direct object 1 refers to the accusativecomplementing the subject and verb group. In effect, transitive verbs are thosewhich take an accusative object (cf. Visser 1963–1973: }418), while intransitiveverbs are those which either do not take any object or take an indirect object, i.e. anobject in the dative, or a causative object, i.e. an object in the genitive. Theterminological problems aside, Visser draws a dividing line between “verbs withoutcomplement” and “verbs with complement”.Verbs which appear without complement are divided by Visser (1963–1973: }129) into four major groups presented in (1) below.1. (a) “Self-sufficient intransitive” verbs which never appear with any kind ofobject, either direct, indirect, prepositional or causative.(b) Intransitive verbs for which there exist etymologically related transitivehomonyms 2 .(c) Absolute use of “intransitive verbs that are usually construed with an indirectobject (in the dative) or with a causative object (in the genitive)” (Visser1963–1973: }129).(d) Transitive verbs which are used without the direct object they normallyrequire, i.e. the absolute use of transitive verbs. Visser (1963–1973: }129)remarks that “the suppression of the direct object occurs wherever it isso clear from the context that the utterance is intelligible without it”. 3However, as noted by Visser (1963–1973: }152), sometimes “the objectunderstood is so vague that the verb is virtually an intransitive one, especiallyin those combinations in which it has acquired a different meaningfrom the habitual one”.The second major type of verbs distinguished by Visser are verbs with complement,divided into copulas, where the verb is “semantically subordinated” to thecomplement – irrelevant for us here, and verbs which are not semanticallysubordinated to the complement, in which case Visser (}315) calls the complement1 Visser (1963–1973: }418) remarks that the expression is used “for want of a better one” and that ithas not been accurately described in terms of notional or logical relations despite the many efforts.2 Visser (1963–1973: }129) notes that verbs of this type are conventionally marked in grammarbooks and dictionaries as “intrans. + trans.”, a classification which in itself is self-contradictory.These, Visser remarks, should be viewed as separate verbs and concludes that “[f]or lack of abetter term, and for the sake of convenient reference (...) these verbs will be called doublefunctionedor amphibious verbs”.3 Interestingly, in apparent contradiction to this statement, in }}492–4 Visser discusses verbs whichare normally used with an object but appear with a dummy hit. This obviously raises the questionwhat the function of this dummy object was in view of what Visser says in }129 about the omissionof an object if the context made the meaning of the phrase clear.

140 M. Charzyńska-Wójcikan object. The first and obvious distinction within the latter group is drawn betweenverbs with a single object and verbs with two objects.As far as single objects are concerned, these are divided into four types shown in(2) below. 42. (a) Indirect object (DAT)(b) Causative object 5 (GEN)(c) Prepositional object(d) Direct object (ACC) 6Double objects naturally encompass the existing variations of the above typesand these fall into eight subtypes:3. (a) Indirect object + causative object (DAT + GEN)(b) Direct object + causative object (ACC + GEN)(c) Direct object + ablative object (ACC + DAT)(d) Indirect object + direct object (DAT + ACC)(e) Direct object + direct object (ACC + ACC)(f) Indirect object + prepositional object (DAT + PP)(g) Direct object + prepositional object (ACC + PP)(h) Two prepositional objects (PP + PP)This is all Visser has to say as far as the classification of OE verbs is concerned,which leaves a lot of questions not only unanswered but also unasked. Our firstmajor objection concerns the lack of any independent classification of verbs withclausal objects. In further parts of the book, dealing with other aspects of syntax,Visser discusses the modally marked form in object clauses (}}868–73), the infinitiveas object (}}924–5) and the form in -ing as object (}1001–8), which indicatesthat this omission in the classification of verbs is not a side effect of the way theauthor classifies the above mentioned structures but of the way he classifies theverbs. Secondly, Visser does not seem to be consistent in the use of the term“transitive”, as shown in his }152, where under the heading “transitive verbs usedabsolutely”, he lists verbs with dative and genitive objects as well. This obviouslyresults in confusion as far as the understanding of the basic term “transitive” isconcerned. Moreover, Visser’s discussion ignores possible omissions of an objectin the double object subtype and the variation that is recorded within the subgroups:4 Within these basic divisions Visser introduces semantic classifications to the effect that he listsverbs appearing in particular structures under headings such as: “verbs of rejoicing, mourning,sorrowing”, etc.5 Visser objects to the term “genitive object” for this type of relation, emphasising the fact thatwhatever the terminological difficulties, it should be borne in mind that verbs in this constructiontype are intransitive. For details see Visser (1963–1973: }370).6 The class is further subdivided into: impersonal verbs with direct object, objects of result, cognateobjects, instrumental objects, reflexive objects, reciprocal objects. These subdivisions, however,do not contribute to the general picture as they mostly reflect differences of functions not ofstructures.

10 Reflections on Structural Variation in Old English Verbs 141a ditransitive verb of one type can very well be used in a structure of anotherditransitive type, monotransitively, or absolutely.Visser’s classification of OE verbs despite its drawbacks is the most comprehensiveone in the literature. Mitchell (1985) and Ogura (1995), who also discuss OEverb types, draw heavily on it. Let us see now how the other two authors classifyOE verbs.10.2.2 Mitchell (1985)Mitchell (1985: }602) presents the traditional syntactic classification of verbs whichdistinguishes principal verbs from auxiliaries. The former embrace “transitive” and“reflexive” verbs and also “intransitive” and “impersonal”. After Pei and Gaynor(1954), as reported in Mitchell, transitive verbs are defined as capable of governinga direct object while “intransitive verbs do not require (and often cannot take) adirect object”. However, as Mitchell remarks, OE verbs pose certain problems forthis neat classification. One difficulty consists in the existence of etymologicallyrelated homonyms, one of which is transitive, the other intransitive. Here Mitchell(1985) refers the reader to Visser’s (1963–1973: i 97) discussion (cf. alsoSect. 10.2.1 above). Next, there is a problem with verbs which are classified astransitive in MnE and which in OE take an accusative, while others take a dative inOE (cf. Mitchell 1985: }602) as the dative was used in OE for the sole object ofmany intransitive verbs 7 whose modern cognates are regarded as transitive. This, inturn, naturally forces Mitchell to pose the question of how to classify verbs whichcan take both ACC and DAT, or verbs which can take ACC, DAT or GEN? Thesedifficulties can, according to Mitchell (1985: }604), be avoided by adapting theclassification used by Visser (1963–1973). Mitchell’s main criticism of Visserconsists in rejecting the use of the term “intransitive” in definition (1c) givenabove. Mitchell suggests avoiding the terms “transitive” and “intransitive” as faras possible and remarks that the terms should only be used in the sense “used(in)transitively”. Instead, according to Mitchell (1985: }604), the terms “object”and “complement” should be used in the traditional way (cf. Sect. 10.2.1 above),where verbs taking objects should not be understood automatically as transitive forthey “may take other kinds of objects than an accusative”. Mitchell (1985: }1565)declares that he generally avoids the terms “indirect object” and “direct object”.When necessary, however, he uses the term “indirect object” in “the traditionalway” and the term “direct object” to embrace ACC, GEN and DAT of verbs whichare regarded as transitive in MnE or which take a direct object not found in MnE.A few remarks are in place here. First of all, some OE verbs are viewed byMitchell from the modern perspective and, consequently, they are classified7 This view is presented in Quirk and Wrenn (1958: }107). To complicate the picture even more,Mitchell reports that these verbs are referred to by Ardern (1948) as intransitive verbs withQuasi-objects.

142 M. Charzyńska-Wójcikthrough their MnE reflexes. As a result, a verb accompanied by an object in ACC,DAT or GEN will be classified as transitive whenever its modern equivalent istransitive. On the other hand, when a verb takes an ACC object, it is always viewedas transitive, regardless of its MnE equivalent. In effect, the classification of OEverbs is partly dependent on MnE verbal system. Moreover, it is obvious that byadapting Visser’s division of verbs, Mitchell’s classification will naturally sufferfrom the same drawbacks as those presented above with reference to Visser’sgroupings. A final, indispensable comment is required here. For the sake ofdiscussing passivisation, Mitchell (1985: }1090) proposes his own classificationof verbs. This classification relies on cases rather than on grammatical relations,which makes it more descriptive and less prone to drawbacks. 8 Most importantly,within this classification he does distinguish clausal objects, which were ignored inhis general classification, and remarks that OE verbs tend to straddle groups, whichis an important characteristics of OE.10.2.3 Ogura (1995)Ogura’s (1995) classification of verbs is to a great extent based on Visser’s(1963–1973) and Mitchell’s (1985), as the author declares, hence her discussionof the terms “transitive” and “intransitive” does not contribute to the debate. Herown classification of verbs is primarily based on cases and the category of theobject. As a result, the author distinguishes the following types of verbs. 94. (a) “Impersonal” verbs(b) Verbs with genitive (including reflexive)(c) Verbs with dative (including “impersonal” and reflexive)(d) Verbs with accusative (including “impersonal” and reflexive)(e) Verbs with genitive and dative (including “impersonal” and reflexive)(f) Verbs with dative and accusative (including “impersonal” and reflexive)(g) Verbs with accusative and accusative (including “impersonal” and reflexive)(h) Verbs with accusative and adverbial dative(i) Verbs with prepositional objects(j) Verbs with infinitive(k) Verbs with dependent clauseEach verb type is supplied with a list of Old and Middle English verbs whichappeared in the given structure. In addition, within these groups, a semanticclassification is added (in a manner resembling Visser’s semantic groupings).8 This classification is not free from its own drawbacks, as is shown in Charzyńska-Wójcik (2008,in prep. a).9 As in the case of Visser’s classification, I ignore copulas here as they are irrelevant for the topicpursued in this paper.

10 Reflections on Structural Variation in Old English Verbs 143As a result, all available structures are mentioned with the list of verbs used in thesestructures under the relevant headings.Three comments, we feel, are in order here. First of all, the lists of verbs givenunder each verb type are by no means complete 10 – the index given at the end of thebook contains 1,343 verbs, which consitute a small proportion of the total numberof verbs in OE. 11 Moreover, some verbs are listed several times as they appeared inOE in a variety of structures. Ogura does not, however, concentrate on this cooccurrenceto the extent that she does not comment on any of these options in anyway. This is not in itself a criticism of Ogura’s classification as it was meant for adifferent purpose but a necessary comment from the perspective assumed for thispaper. Finally, Ogura’s classification does not encompass verbs without complementat all and, as a result, the non-expression of an object is not a valid parameterwithin her approach.In conclusion, we have presented three classifications of OE verbs togetherwith their drawbacks and shortcomings, both those formulated by their authorsand those raised in this paper. Most of these comments were theoretical andmethodological in nature and were apparentevenwithoutanydetailed analysis.In the next section we will show the applicability of these classifications to theactual OE data.10.3 Traditional Approach to Data vs. Data ThemselvesThe above classifications offered an overview of verbs from the perspective of astructure in which they were used. As a result, each verb was classified into a type.However, one and the same verb could be used in a variety of structures. The kindof structures a given verb co-occurred in and the type of structures that were neverrealised by one and the same verb is not captured in a principled way by any of theabove classifications. The only (partial) exception to this criticism is Visser’sclassification given in (1) above, which takes into account not only the actualstructure but also what (if any) NP the given verb co-occurred with in its other uses.The picture emerging from the data to be presented below shows that none of theclassifications discussed in Sect. 10.2 above reveals anything about the true natureof the OE verb. In this section we will show how the above classifications fare withrespect to the actual data. We will first present verbs without complement and thengo on to verbs with an object. Due to the limitations of space we will not be able todiscuss verbs with two objects here. This, however, does not impoverish theanalysis presented, as the conclusions following from the discussion of the formertwo types converge with the picture emerging from the analysis of ditransitives(Charzyńska-Wójcik in prep.b).10 They are mostly based on Visser’s lists.11 By comparison, the number of verbal entries in B&T is 6,711, to say nothing of the Supplementvolume.

144 M. Charzyńska-Wójcik10.3.1 Verbs Without ComplementLet us begin with verbs which take no complement, i.e. verbs traditionally referredto as “intransitive”.5. Intransitive(a)(b)The examples above show the verbs cidan ‘to chide, rebuke’ and arisan ‘to arise’,which have no complement. At first glance the verbs seem to represent the samestructure, i.e. both seem to be intransitive. However, as we have seen, Visserdivides verbs without a complement into four types so we need to look at theseverbs more closely to see what the two structures represent. A conclusion thattranspires from the examination of example (6a) is that the term “intransitive” isinapplicable to the verb cidan:6. (a)The example in (6a) above features the verb cidan accompanied by a DAT object,which contrasts with example (5a), where the same verb appears without complement.Applying Visser’s classification we could say that cidan in (6a) is a verb oftype (2a), i.e. a verb with a single indirect object, while in (5a) it represents a verb oftype (1c), i.e. a verb which can appear with an indirect object but is used absolutely.However, example (6b) given below renders this classification of (5a) wrong.6. (b)The object in (6a) was dative and in (6b) the same verb appears with a direct object,i.e. in the accusative case. In consequence, (6b) could be classified as (2d), i.e. a verbwith a direct object, while the verb in (5a), if compared with (6b), would be classifiedas (1d). However, as we need to view (5a) with respect to both (6a) and (6b), thereseems to be no available way of classifying the structure in (5a).

10 Reflections on Structural Variation in Old English Verbs 145Let us now move on to the verb arisan ‘to arise’ exemplified in (5b). This verb,unlike the one in (5a) cannot be accompanied with an object of any sort. In Visser’sclassification, then, arisan should be classified as (1a), i.e. a self-sufficient intransitiveverb which never appears with any kind of object. Note that Visser’s classificationis useful here in that it draws a dividing line between (5b), which is classifiedas (1a), and (5a). Where it seriously fails is that it is completely incapable ofclassifying (5a), unless we resort to Visser (1963–1973: }152), where he confuseshis own classification given in }129 and, under the heading “transitive verbs usedabsolutely”, he mixes the verbs which he originally kept apart as (1c) and (1d), i.e.verbs with dative, genitive and accusative objects (among others). It has to be bornein mind, however, that this is an effect of confusion rather than his intendedclassification.10.3.2 Verbs with One ObjectThe next verb type to be discussed are verbs with an object. The available structuraloptions as far as the category of the object is concerned are illustrated below.7. (a)(b)(c)(d)Considering the category of the object, the above monotransitive structures can beclassified in a straightforward way, as shown in Table 10.1 above.This seemingly innocent classification, as we intend to show, poses someinteresting questions. First of all, Visser’s classification does not recognise the

146 M. Charzyńska-WójcikTable 10.1 Classification of the structures in (7)Ex. No. Verb Structure Visser’s classification7a wiþsacan NP-ACC 2d‘to reject, deny, refuse’7b cidan PP 2c‘to chide, rebuke’7c tocnawan Finite CLAUSE –‘to distinguish, understand’7d teolian/tilian Non-finite CLAUSE –‘to strive, attempt’types given under (7c) and (7d) at all. Mitchell’s general classification does notrecognise clausal objects, either. In contrast, Ogura would propose the sameclassification as the one given above, i.e. based on the category and case of theinternal argument. However, if we look at more data, none of the classificationsseems suitable.Let us compare the structure in (7a) containing the verb wiþsacan ‘to reject, deny,refuse’ with the structures given in (8) below, which all contain the same verb:8.(a)(b)(c)The verb wiþsacan in (7a) is accompanied with an ACC object NP, but (8a) exhibitsa complex sentence which consists of three clauses with the same verb. Herewiþsacan is accompanied with DAT, ACC and GEN object NPs. 12 Thus, while(7a) would be classified by Visser as a transitive verb (2d), i.e. a verb with a directobject in the accusative, there is no suitable group to classify all the uses of the same12 This issue is reported by Mitchell (1985: }603), who remarks that such verbs pose a problem.

10 Reflections on Structural Variation in Old English Verbs 147verb shown in (8a), while classifying the three uses under three different types seemsto be missing the point. Moreover, if we consider (8b), we can see that the same verbcan also appear with a clausal object (no classification is available here in Visser’s orMitchell’s systems). In (8c) the verb is used with no object at all but which of theoptions proposed by Visser (given here in (1) above) captures the structure properly?(1c) subsumes the absolute use of verbs which are normally used with dative or genitiveobjects, while (1d) encompasses verbs where the accusative object is not expressed.This is a recurring problem, as shown in the discussion of (5a) and (6) above.In sum, whatever terminological problems there are with classifying the relevantstructures, the choice of the internal argument for the verb wiþsacan seems to bebetween no object at all or an NP object in any of the Cases available for an objectin OE or it can even be an object expressed by a clause – a wide array ofpossibilities. The question, rhetorical for the time being, is whether this range ispredictable in any way?Let us now move on to the next structure with one object exemplified in (7), i.e. aprepositional object given in (7b). The example contains the verb cidan, which hasalready been shown in examples (5a) and (6) above. In (7b), cidan is accompaniedwith a PP object, while in (5a) it is used with no object at all and in (6a, b) with DATand ACC objects respectively. If this is not confusing enough, consider example (9)below:9.In (9) the verb cidan appears with two objects: a PP and a clause. As a result, interms of a traditional description, as we have already shown, the verb is usedabsolutely in (5a) but we cannot classify it into a type (2c or 2d?), monotransitivewith a direct object in (6b), indirect object or quasi-object in (6a) (transitive orintrasitive?), prepositional object in (7b), i.e. (2c) and ditransitive with a prepositionaland clausal object in (9), for which no classification either in Visser’s or inMitchell’s system exist. Ogura fails here as well as she does not classify separatelyverbs with a prepositional object and a dependent clause. The data throw new lighton the matter, clearly showing the insufficiency of the existing classifications.Let us move on to another verb, tocnawan ‘to distinguish, understand’. The verbis used in (7c) with a clausal object. In (10) below, however, we can see it with anACC NP object.10.

148 M. Charzyńska-WójcikWhile the classification of (10) above is straightforward: a transitive verb with adirect object in the accusative case, no classification is available for the usageexhibited in (7c). Note, moreover, that one cannot ignore the fact that whilestructures without complement, as discussed by Visser (cf. (1) above), fall intoseveral very different subtypes depending on what other structure a given verb canbe found in, it seems right that the same be done for structures with a direct object.In other words, should a transitive verb with a direct object always be classified atits face value, unlike structures without complement, or should one rather considerwhat other structures the verb can appear in?The last verb quoted in (7) to be discussed here is teolian/tilian ‘to strive,attempt’. In (7d) it is complemented with a non-finite clause, so no classificationis available for it from Visser or Mitchell. In (11a) below the complement clause isfinite (still unclassified), while in (11b) the verb has no complement at all, which, inview of the fact that in (11c) it is accompanied with an ACC object (transitive verbwith a direct object, i.e. 2d), indicates that it should be classified as an absolute useof a transitive verb (1d). Interestingly, the information supplied for this verb byB&T is that accusative is only one of the three available Cases for the NP object ofthis verb, which renders the above classification of (11b) unsuitable.11.(a)(b)(c)In sum, we have just analysed four different types of verbs with one object,shown in (7) above. The discussion that followed indicates that classifying OEverbs with one object according to the category of the object, in the way shown inTable 10.1 above is erroneous. This conclusion becomes especially clear if weclassify other structures 13 recorded with the very same verbs, as shown inTable 10.2 above.13 It has to be emphasised, however, that the presentation of structures possible with the relevantverbs given in (8)–(11) is not exhaustive. It is merely intended to show that viewing OE verbs insuch “static” terms is incorrect.

10 Reflections on Structural Variation in Old English Verbs 149Table 10.2 Classification of other structures recorded with the verbs from (7)Verb Structure, Ex. No. Structure, Ex. No. Structure, Ex. No. Structure, Ex. No.wiþsacan NP-ACC (7a) NP-ACC/DAT/GEN finite CLAUSE no object (8c)(8a)(8b)cidan PP (7b) no object (5a) NP-DAT (6a) PP + finiteNP-ACC (6b) CLAUSE (9)tocnawan Finite CLAUSE NP-ACC (10)(7c)tilian Non-finiteCLAUSE (7d)finite CLAUSE (11a) no object (11b) NP-ACC (11)To further strengthen the point made above that OE verbs should be classifiedwith great caution, we would like to point out the fact that if one selectivelyconsidered only the italicised structures given in Table 10.2 above, the very sameverbs which are classified as different in Tables 10.1 and 10.2 would end up in thesame class: verbs with an ACC object. While partly true, this classification in itselfwould answer no more questions than the one given in Table 10.2, or, for thatmatter, any other classification that we can easily make up from the variety of thedata given in Table 10.2. One important conclusion transpires from this discussion:a classification of OE verbs requires a more holistic approach to the structuresavailable with a given verb.10.4 ConclusionIn sum, whatever traditional descriptions based on the actual usage of a given verbwe apply, we fail. If we take verbs with no complement, as the ones given in (5)above, we can see that some of them can never be used with an object, while otherscan appear with a wide range of object types, to the effect that one and the sameverb would have to be classified as intransitive, monotransitive or even ditransitive,to say nothing of the available subtypes. If we consider verbs with one object, as theones exemplified in (7), we can see that they can be found in absolute uses and inditransitive usages of various types as well. The same applies to ditransitives, whichhave not been discussed here: verbs which appear in ditransitive structures caneasily be shown in a variety of other constructions, ranging from intransitivethrough monotransitive to various types of ditransitives in the case of one and thesame verb (cf. Charzyńska-Wójcik in prep. b). One obvious conclusion that followsfrom the above discussion is that classifying verbs on the basis of particular usagesdoes not make sense for OE. In other words, OE requires a different approach toverb classes. It would be more accurate to classify particular structures, i.e. a set ofusages of a given verb with individual classifications for each usage. However, purecataloguing of the different structures available with a given verb (in Ogura’smanner) does not result in a broader picture of the syntax of a particular verb.What we need is a classification resulting from the study of variation allowed by a

150 M. Charzyńska-Wójcikparticular verb. The question that poses itself at this point is the following: is thereany guiding principle behind all this variation? In other words, is the variationpredictable in any way? Is the optionality coded in the syntax of OE, or is itlexically determined for individual verbs?Some attempts at accounting for this property of the syntax of OE verbs can beseen in Charzyńska-Wójcik (2002, 2008, 2010, in prep.a), where it is shown thatat least some variation can be reduced to a common denominator. In particular,Charzyńska-Wójcik (2002) explores the variation between nominal and ClausalThemes exhibited by OE Experiencer verbs. What transpires from the study ofthe co-occurrence of individual Experiencer verbs in various construction types isthat the expression of the Theme argument (as an NP or a CLAUSE) does nothave to be lexically specified for each verb. Instead, we note that the variation ispresent if the meaning of a verb is compatible with a clausal Theme. InCharzyńska-Wójcik (2008) it is shown that Clausal Themes of ditransitiveverbs with DAT Recipients are in variation with nominal Themes in a waywhich, again, suggests that the category of the Theme need not be pre-specified,i.e. the Theme can be realised as a CLAUSE or an NP by one and the same verbas long as the meaning of the verb is compatible with a clausal Theme. This lineof reasoning is continued in Charzyńska-Wójcik (2010), where it is argued thatthis approach is applicable to all types of clausal ditransitive verbs and hasimportant consequences for other types of verbs which allow a clausal Theme.The issue is further expanded in Charzyńska-Wójcik (in prep. a.), where oneobjectverbs and prepositional ditransitive verbs are shown to exhibit variation inthe expression of its arguments. The variation exhibited by the above verb typesis, again, largely predictable from the semantics of the verb. These works,however, only deal with variation between a CLAUSE and an NP rather thanstructural variation in general. All, however, point to the same conclusion as faras this aspect of variation is concerned: the co-occurrence of clausal and nominalconstituents is predictable from the semantics of the verb. The above conclusionsapply to only one type of variation, yet they certainly open up a new perspectiveon OE verbs.Answers to the questions posed in this paper require much further research: asever, much less work is necessary to show a particular concept – in this case theclassifications of OE verbs – to be deficient for a given language or constructionthan to propose a new accurate account. We hope, however, to have arguedconclusively in favour of a necessity for a new approach to OE verbs.ReferencesArdern, P.S. 1948. First readings in Old English. Wellington. (rev) Cambridge 1953.B&T ¼ Bosworth, J. and T. N. Toller. 1898. An Anglo-Saxon dictionary based on the manuscriptcollections of the Late Joseph Bosworth, edited and enlarged by T. N. Toller. London: OxfordUniversity Press. Available at: http://lexicon.ff.cuni.cz/texts/oe_bosworthtoller_about.html.

10 Reflections on Structural Variation in Old English Verbs 151BTs ¼ An Anglo-Saxon dictionary based on the manuscript collections of the Late JosephBosworth. Supplement, by T. N. Toller. London: Oxford University Press. Available at:http://lexicon.ff.cuni.cz/texts/oe_bosworthtoller_about.html.Charzyńska-Wójcik, M. 2002. The syntax of Old English Experiencer verbs. SKY Journalof Linguistics 15: 31–60.Charzyńska-Wójcik, M. 2008. Clausal ditransitive verbs in Old English. In Language encounters,eds. M. Charzyńska-Wójcik, A. Malicka-Kleparska and J. Wójcik, 103–114. Lublin:Towarzystwo Naukowe KUL.Charzyńska-Wójcik, M. 2010. The syntax of Old English clausal ditransitive verbs. Verbstructures: Between phonology and morphosyntax, eds. E. Cyran and B. Szymanek, 75–109Lublin: Wydawnictwo KUL.Charzyńska-Wójcik, M. in prep. a. The ca(u)se for a clause.Charzyńska-Wójcik, M. in prep. b. In-, mono- and ditransitive verbs in Old English and how theyall fit in.DOECEF ¼ Dictionary of Old English corpus in electronic form, ed. di Paolo Healey,A. Distributed by the Oxford Text Archive.Mitchell, B. 1985. Old English syntax. Oxford: Clarendon Press.Ogura, M. 1995. Verbs in Medieval English: Differences in verb choice in verse and prose. Berlin:Mouton de Gruyter.Pei, M. and F. Gaynor. 1954. A dictionary of linguistics. New York.Quirk, R. and C. L. Wrenn. 1958. An Old English grammar. (2 nd edition.) London.Visser. F.Th. 1963–1973. An historical syntax of the English language. Brill: Leiden.

Chapter 11The Semantic Analysis of Old English unnanAgnieszka WawrzyniakAbstract The aim of the present paper is to draw the semantic profile of unnan.According to the Anglo-Saxon Dictionary (BT, sv. unnan), unnan denoted thefollowing senses:(a) ‘to wish’,(b) ‘to be pleased with’,(c) ‘to grant’.One of the most crucial factors in the profile of unnan was the semantic path of theverb. Namely, the general development did not proceed from the concrete to theabstract but the process was reverse. One can observe a change from the abstract sense‘permission based on wish’ to the concrete one ‘to grant’. In other words, what takesplace is not the growth in the abstractness of meaning but progressive concretization.The paper focuses on the intermediate stages in the development from the formersense ‘permission based on wish’ to the latter one ‘to grant’ such as conceptual shift ofa subject, conventionalisation of conversational implicature, as well as defocusing ofthe divine referential object and its replacement by a human referential object.Furthermore, the analysis will also show morphological forms in which either thenew sense or the former one was central. Moreover, the study tries to account for thesemantic factors which contributed to the absence of epistemic senses in unnan.11.1 IntroductionAccording to the Anglo-Saxon Dictionary, unnan (BT, sv. unnan) is related to Icel.unna whose meaning was ‘to grant, allow, bestow’ as well as ‘to love, to bestowone’s love on a person’. Another cognate is Goth anst denoting joy, thanks, grace,favour. The common Indo-european root for unnan, unna and anst is ans, whichA. Wawrzyniak (*)Adam Mickiewicz University, Poznań, Polande-mail: agnieszka78kaga@op.plM. Pawlak and J. Bielak (eds.), New Perspectives in Language,Discourse and Translation Studies, Second Language Learning and Teaching,DOI 10.1007/978-3-642-20083-0_11, # Springer-Verlag Berlin Heidelberg 2011153

154 A. Wawrzyniakaccording to Pokorny (1959) (sv. ans) means ‘wohlgeneigt, gunstig sein’, i.e. ‘to befavourable’. Following the Anglo-Saxon Dictionary (BT, sv. unnan), unnandenoted the following senses:(a) ‘to grant’,(b) ‘to be pleased with’,(c) ‘to wish’.Before starting the analysis of individual morphological forms to indicatesemantic discrepancies between them, the first step will be to analyse the generalconcept of unnan and the mechanisms involved in the process of evolving the sense‘to grant’. Unlike most verbs where the path of a change proceeds from the concreteto the abstract (Goossens 1987; Traugott 1989; Sweetser 1990; Koivisto Alanko2000; Barcelona 2003; Radden 2003; Cruse 2004), unnan reflects a reverse shift,namely from the abstract to the concrete.To begin with, unnan exemplifies a concept whose initial meaning was notconcrete but abstract, i.e. not ‘to grant’ but ‘to wish/permit’ (permission based onsomeone’s wish). Hence, the semantic path did not proceed from the more concretesense (to grant) to a more abstract one (to wish/permit), but in the oppositedirection. The abstract meaning was the primary one and acted as a basis for thedevelopment of the concrete sense ‘to grant’, which resulted in the absence ofabstract senses in unnan. The reasons underlying this assumption are the following.The analysis of the Old English (henceforth OE) data shows that the sense ofunnan when rendered as permission/wish does not need another co-referential verb. Itis the verb itself that evokes such associations. When, however, unnan denotes theidea of granting, it is mostly in a set phrase with willan, whose meaning was ‘to bewilling’. The frequent connotations of willan and unnan to denote the sense ofgranting indicate that this sense was a later development as the verb itself could notrender this concept on its own. Moreover, the collocation willan + unnan evokesinitially the association that it is the speaker’s wish to grant. Though gravitatingtowards the independent status, the above phrase makes unnan contingent on anotherverb, willan, whose semantics evokes attributes which were initially denoted byunnan and later implied, namely ‘wish, favour, willingness’. Consequently, thesense ‘to grant’ is bound with the association it is the speaker’s wish to act so.Even though there were contexts where ann (first person singular present)appeared also without the co-referential verb willan, its concept remained unchanged.In other words, the subject granted some entity because he wished to do so. Granting,then, always evoked positive associations. There were no contexts recorded in theToronto Corpus where the subject granted some entity against his own will. In otherwords, the idea of a wish was entrenched in the semantics of unnan so that it nolonger needed another verb which evoked similar, pleasant associations.Nevertheless, at a certain point of time, when unnan started to develop a new sense,it often needed co-referential elements. Except for willan, another set phrase recordedin the analysed texts is ann mid unne,whereunne denoted will, wish, pleasure (BT, sv.unne). Such phrases, which were not conceived as tautologous by the speakers of OEwere indicative of two linguistic factors. Firstly, ann developed a new sense, which

11 The Semantic Analysis of Old English unnan 155was viewed as distinct from the noun unne. Secondly, it implied similar semanticattributes in the sense that granting was always contingent on the speaker’s wish to actso. The attributes of both interplayed and gave rise to semantically harmonicexpressions.Another argument that underlies the path of development from the abstract sense‘will’, ‘wish’, ‘permission’ to the concrete one ‘to grant’ is that the latter sense wasused as the only one in the first person singular present, and scarcely in the past. Inother forms, the earlier meaning ‘wish’ or ‘permission based on wish’ was expressed.One can trace the semantics of unnan on its way to the emergence of the sense ofgranting. It should be emphasised that the roles of both the speaker and theaddressee should be analysed in the path of the development. In the first stage,the primary sense was ‘to wish/permit’. There is a speaker and his subjective wish.The speaker voices his/her wish, or rather permission based on the wish.The idea of granting is conceptually more complex, though it could be conceived asoriginating from permission. While the concept of permission needs only an individualexpressing his/her wish, the concept of granting involves the individual’s wish andsome authority’s readiness/willingness to accept it in positive terms. Therefore,conceptually, the sense of granting evokes the individual and the addressee. Onecould even claim that the process of granting, though initiated by the individual, shiftsthe point of reference and puts a greater conceptual weight on the addressee while theinitial subject recedes into the background. Consequently, the change from the sense‘permission/wish’ to the sense ‘to grant’ should also be viewed in terms of a change ofthe conceptual subject, as it is no longer the individual who is in the centre, but theaddressee, who is willing to grant, thereby reflecting the speaker’s wish.The mechanism accounting for the shift of conceptual subjects, and hence for theshift in meaning, seems to be the mechanism of conventionalisation of conversationalimplicature. It should be emphasised that the discourse type accompanyingunnan was of a different nature. Namely, it did not take place between twointerlocutors but between the individual (speaker) and God. The speaker voiceshis/her wish. He cannot hear God and may only conceptually project his response.The speaker asks God to realise his/her wish. He strongly believes that God will actaccordingly. If the speaker says: if you are willing to permit, this may imply youwill do it. Therefore, the conventionalisation of conversational implicature that waspresent in the mind of the speaker each time he/she stated his/her wish could lead tothe gradual change in meaning.The transfer that takes place and leads to the gradual change in meaning ismetonymic in nature, where the result stands for the cause. Thus, the resultcorresponds to the final effect of a wish, hence the sense of granting, while the causecorresponds to a wish for the event to be fulfilled. Therefore, the emergence of thesense ‘to grant’ in the first person originated in discourse situations, from the speaker’swish of implementation of his/her requests, via the conventionalisation of conversationalimplicature based on a metonymic link - result for the cause. The consequence isthe shift of the conceptual subjects. Ann gradually acquires conceptual independence,yet such attributes as a wish or willingness are still present in its conceptual network,as the implications evoked by ann are always positive for both the person who

156 A. Wawrzyniakgrants as well as for the individual whose wish is fulfilled by having receivedthe gift.Moreover, another aspect that affected the shift in meaning is the shift in thedegree of abstractness. When God was conceptualised as a point of reference, theattributes He could give were abstract ones, such as happiness, life in eternity,peace, etc. Therefore, the interpretation of such instances gravitated towards thepermission sense, where God permits to live eternally rather than gives such a life.Such contexts presupposed humbleness of the speaker, the greatness ofthe Almighty God, and abstract, intangible ideas. The replacement, however, ofdivine object by a human being resulted in a subsequent shift in meaning from thesense ‘to permit’ to the sense ‘to grant’. At that point, abstract, intangible ideas weresubstituted for by concrete entities, such as lands and territories. When a human wasconceptualised as a source of gifts, the entities he could give were never abstract buthighly concrete. Furthermore, the range of gifts was narrow as it could be eitherland or legal rights. Therefore, one could say that the sense ‘to permit’ was relatedto God, as a source of a gift, and abstract soul-directed attributes, while the sense ‘togrant’ presupposes a human as a source of a gift, and concrete, tangible, entitieslimited in scope. The borderline between the two senses seems rather blurred as it isrelated to the level of abstractness. In other words, soul-directed attributes gave riseto the sense ‘permission’ rather than ‘give’, which is related to concrete entities.The present analysis is based on the Toronto Corpus compiled by Antonette diPaolo Healey (1986). The Corpus is an online database consisting of about 3 millionwords of OE. The study presents a quantitative approach based on detailed statisticsof particular senses. In order to achieve maximum accuracy, unnan was analysed inall the attested contexts in the Toronto Corpus. Nevertheless, one should take intoaccount the fact that however detailed and accurate the corpus is, it may not alwaysreflect the semantic reality of the distant past. The data can thus be distorted as notall the texts that were written down in the OE period are preserved in the corpora.Consequently, the semantic analysis of the profile of unnan may be more of acorpus artifact rather than a reflection of the actual semantics of unnan in theAnglo-Saxon period. Yet, corpora are valuable sources of information for theearly stages in the history of a language. Consequently, one can hope that itscontents reflect, to a certain extent, the semantic reality of unnan in the Anglo-Saxon period.The study concentrates on the three subperiods within the Anglo-Saxon periodand covers the following time span, which was semantically conditioned.– Stage I – 900– Stage II – the end of 900 and the beginning of 1,000– Stage III – 1,000It should be emphasised that the earlier studies (Campbell 1959; Lightfoot 1979;Mitchell 1985; Warner 1999) were based on either syntactic or morphological ratherthan semantic criteria. The present study is not based on the assumptions postulatedby linguists concentrating on syntax or morphology and thereby neglecting the

11 The Semantic Analysis of Old English unnan 157semantic study of lexical units. Instead, it will emphasise the role of meaning andthe gradual changes in the concept of unnan.11.2 The Semantic Analysis of the Infinitive unnanThe semantic analysis of unnan records three instances dated to the 10c and two tothe 11c. Yet, despite the time span of one century, there seems to be no progress inthe development of the conceptual content of the infinitive. All the analysedinstances are recorded in a set phrase with willan. Moreover, four of them arerelated to abstract attributes, i.e. they give rise to the sense of permission. However,there is one instance, where the direct object of unnan is a concrete entity (oars),which allows the sense ‘to give/grant’, yet the recorded instance is dated to the 10c.Consequently, such a statement should be viewed as a rare instance rather than astep in the development of the sense ‘to grant’. The prototypical meaning of theinfinitive was the permission sense, recorded in a set phrase with willan: permissionsense in a fixed phrase.1. He wæs æwicsmod, hyhtaleas, ac se halga wer ælda gehwylces unnan wolde þæthe blædes brucan moste, worulde lifes. (Guthlac: Krapp and Dobbie 1936,44–88)‘He was disgraced in mind, in despair, but the holy man, to everybody, wanted topermit that he would enjoy the glory of worldly life.’2. Ða antswarede heo þam engle and þus cwæþ, ic gyrnen wolde gif hit mægþ wæreþæt hit gyrnan durste þæt ure drihten me unnan wolde þæt ic þa halga rode iseonmoste. (Invention of the Cross: Napier 1894, 2–34)‘She answered to the angel: I would yearn, if it were allowed that I would dare toyearn so that Our God allowed me to see the holy Cross.’The Toronto Corpus records one context where unnan is juxtaposed with a concreteentity, thereby allowing the sense ‘to give/grant’.3. Him ondswarode engla þeoden: We eow aras ofer yþbold unnan willaþ.(Andreas: Krapp 1932a, 3–51)‘The highest of angels answered him: We are willing to give you oars across theshipside.’However, the analysis of the above sentence shows that the meaning ‘to give’ isthe consequence of the permission received thereby denoting further implicationsbesides the apparent surface meaning ‘to give’. Firstly, the speaker is an angelrather than a human being. Thus, the act of giving should be interpreted aspermission to survive, and hence to live. Consequently, the religious sense ‘togive’ evoked a different dimension where it was linked at some conceptual levelwith the permission sense. Contrariwise, the study of ann will exemplify a secularcontext where ann denotes the only sense – ‘to give/grant’.

158 A. Wawrzyniak11.3 The Semantic Analysis of Present Forms11.3.1 The Semantic Analysis of annThe semantic analysis of ann in the first person singular yields the following senses:11c– Grant (land) 12 (52%)– Grant (rights) 4 (17%)– Set phrases 7 (31%)The above analysis leads to the following conclusions. To begin with, ann in thefirst person is recorded only in the 11c. Moreover, its central meaning was ‘to give/grant’. However, the range of entities that could be the subject of granting wasnarrow. Namely, it was either land or rights that were linked with ann. The TorontoCorpus records 52% (12 cases) of contexts which convey the meaning ‘to grantland’, and 17% (4 cases) which denote ‘to grant rights’. Additionally, anncollocates frequently with willan, which indicates that the link with the formersense was preserved at some layer. The analysis documents 7 cases (31%) of fixedphrases, 6 of which are recorded with the verb willan (ic wille and ic ann), and onewith the phrase mid unne (ic ann mid filre unne), all of which are related to theproperty of land.Furthermore, the context for ann is always secular and appears only in legaldocuments, such as wills and transfers of property.4. And ic ann Æþelwolde bisceope þæs lændes. (Will of Ælgifu, Sawyer 1484:Whitelock 1930, no. 8)‘And I grant the land to Athelwold.’5. Ic ann Leofstane Leofwines brēþer þære land þe ic of his brēþer nam. (Will ofÆtheling Athelstan, Sawyer 1503: Whitelock 1930, no. 20)‘I grant to Leofstan Leofwine’s brother the land I took from his brother.’6. Ic ann heom þæt hie habben þer to fullne freodom saca and socna. (Writ of KingEdward, Coventry, Sawyer 1098: Harmer 1952, no. 45)‘I grant the full freedom of jurisdiction.’7. Ic ann mid fulre unne þa ilca gife. (Writ of King Edward, Coventry, Sawyer1098: Harmer 1952, no. 45)‘I grant with full consent the same gift.’8. Ic cyþþe eow þæt ic wille and ic ann þæt Sancte Peter and þa gebroþra habben þacotlif mid eallum þam landum. (Writ of King Edward, Coventry, Sawyer 1098:Harmer 1952, no. 46)‘I let you know that I am willing and I grant to Saint Peter and his brothers thedwelling with all lands.’Thus, the sense of granting land is conceptualised in sentences (4), (5), that ofrights in (6), while fixed expressions are exemplified in sentences (7), (8).

11 The Semantic Analysis of Old English unnan 159The semantic analysis of ann in the third person singular indicates that suchcontexts were scarcely recorded when compared with the first person singular. TheToronto Corpus records only three instances when the subject was the third personsingular. Moreover, the first instance for ann in the third person singular is recordedin the 10c, and the two others in the 11c. The statement from the 10c denotes thesense of permission as it evokes God as a point of reference, and two other contextsfrom the 11c record the sense ‘to give/grant’. At that point one can notice that it wasthe first person, rather than the third one that gave rise to the new sense. Firstly, therewere no contexts for ann in the first person before the 11c, while in the 11c its onlymeaning is ‘to grant’. Secondly, ann in the third person, though it was recorded onlyonce in the 10c, acquires the meaning of permission, where God is the source of agift. Moreover, ann in the third person is scarcely attested, when compared with annin the first person. The senses of ann (third person singular) were as follows.10c11c– To permit 1 (100%) –– To grant – 2 (100%)– To permit9. Ic nat hwy he ne sy þara landa wyrþe gyf him heora God ann. (ArchbishopDunstan to King Æþelred, Sawyer 1296: Napier and Stevenson 1895, no. 7)‘I do not know why he would not be worth of this land if God permits him tohave it.’10. Heo ann into ealdan mynster þæs landes. (Willof Ælgifu, Sawyer 1484:Whitelock 1930, no. 8)‘She grants the land to the old mynster.’11.3.2 The Semantic Analysis of unnonThe analysis of unnon (first person plural) records two instances which are dated tothe 11c. No contexts are documented before the 11c. The meaning of unnon (firstperson plural) is the same for the two instances, and denotes ‘to permit’.11. And we þæt nu swyþe eadmodlice unnon. (Letter to Bishop Denewulf, Sawyer1444: Birch 1885–1999, no. 619)‘And that is what we humbly permit.’12. Ðæt is þæt we hit unnon on Godes ast and Sancta Maria and þæs halagan were.(Ælfweard, Abbot of Evesham, to Æthelmær, Sawyer 1423: Robertson 1956,no. 81)‘That is what we permitted for the love of God and Mary and the holy man.’The analysis of unnon (third person plural) records only one instance dated to the11c with the sense “to wish”.13. Ða þe me yfeles unnon (Headings to Psalms 1–50: Thorpe 1835, 1–128)‘Those who wish me evil’

160 A. Wawrzyniak11.4 The Semantic Analysis of Past Forms11.4.1 The Semantic Analysis of uþeThe analysis of uþe related to the first person singular reveals only one instance,dated to the 11c, with the sense ‘to grant’.14. Ic uþe þara ylca lande Eadmund cyninge. (Writ of King Edward, Westminster,Sawyer 1150: Harmer 1952, no. 106)‘I granted land to king Edmund.’The analysis of uþe indicates that almost all contexts, except for the onementioned above, were recorded with the third person pronoun. One can thus noticethe discrepancy between uþe and ann. Ann was most frequently attested with thefirst person, while uþe with the third person, which entails semantic consequences.The Toronto corpus records the following senses of uþe (third person singular).10c 10/11c 11c– Give/permit 1 (100%) – 3 (18%)– Wish/please – 1 (50%) 4 (26%)– Fixed phrases (swa God uþe) – – 8 (50%)– Permission – 1 (50%) 1 (6%)The analysis shows that uþe was typically used with the third person pronoun.In the time span 10/11–11c uþe in the first person was recorded once, while uþe inthe third person 20 times.As the process of the development of the new sense ‘to grant’ started in the firstperson pronoun, from which it started to spread to other pronouns, the past form uþerecords more senses imbued with permission. Hence, one can notice a discrepancyin the development of the first and the third person pronouns.Furthermore, even though the sense ‘to grant’ is recorded in 18% of all contexts, theentities that are subject to granting are not as concrete as in the first person. Namely, nocontexts involve land, which was mostly documented in the first person. Instead, theentities recorded in the analysis which were subject to granting were legal rights,dwelling places and even one abstract attribute – help.15. Æfter his dæge he nanum menn sel ne uþe þonne me. (Will of King Alfred,Sawyer 1507: Harmer 1914, no. 11)‘After his days, he did not give better to anybody but me.’16. Uþe is swiþor ic hine sylfne geseon moste. (Saint Margaret: Herbst 1957,62–82)‘I would be more pleased if I could see him.’17. Isaias geseah on gesyhþe, swa him God uþe, hu þære þeode for heora synnumsceolde gelimpan. (Saint Giles: Dictionary of Old English transcript, CorpusChristi College, MS.303)

11 The Semantic Analysis of Old English unnan 161‘Isaias saw, the way God wished, how these people should repent for their sins.’permission18. God heom uþe lisse and miltse. (In Letania maiore: Tristram 1970, 430–37)‘God gave him/permitted him grace.’11.4.2 The Semantic Analysis of uþonThe semantic analysis of uþon records no instances in the first person and twoinstances in the third person. The analysed contexts are from the 11c and denote thesense ‘to wish/be pleased’ and ‘to grant’:19. Ða bead he heom feos and landes gif hie him þæs rices uþon. (Oxford, BodleianLibrary, MS. Land Misc. 636: Plummer 1892–9)‘He asked them for wealth and land if they could give him that.’20. Ða ne dorste se hym þæs wurnan ac cwæþon hy þæs wel uþon gyf sy cing andse biscop þæs geuþon. (The Community at Sherborne to Edmund Ætheling,Sawyer 1422: Robertson 1956, no. 74)‘He did not dare refuse him that and said that they would be much pleased if theking and the bishop permitted that.’11.5 ConclusionsTo conclude, the development of the sense ‘to grant’ was morphologicallyconditioned. The sense ‘to grant’ was the prototypical one in the first person presentand past indicative, while the sense ‘to permit’ and ‘to wish’ was retained as thecentral one in the third person present and past indicative. Moreover, the study hasindicated that unnan in the present tense was most frequently attested with the firstperson, while in the past tense in the third person. Finally, the newly developedconcrete sense ‘to grant’, which emerged in the 11c presupposed a narrower rangeof objects (land or rights) than the earlier meaning ‘to permit’, whose range ofobjects it could be semantically linked with was unlimited. Consequently, its newmeaning became more concrete and narrower.ReferencesBarcelona, A. 2003. On the possibility of claiming a metaphoric motivation for a conceptualmetaphor. In Metaphor and metonymy at the cross-roads. A cognitive perspective,ed. A. Barcelona, 31–58. Berlin: Mouton de Gruyter.Bosworth, J. and N. Toller. 1882. An Anglo-Saxon dictionary. London: Clarendon Press.Campbell, A. 1959. Old English grammar. Oxford: Clarendon Press.Cruse, D. A. 2004. Meaning in language. An introduction to semantics and pragmatics. Oxford:Oxford University Press.

162 A. WawrzyniakGoossens, L. 1987. The auxiliarisation of the English modals. In Historical development ofauxiliaries, ed. M. Harris and P. Ramat, 111–143. Berlin: Mouton de Gruyter.Healey, A. di Paolo ed. 1986. Dictionary of Old English corpus. Toronto: The University ofToronto Press.Koivisto Alanko, P. 2000. Abstract words in abstract worlds. Helsinki: Societe Neophilologique.Lightfoot, D. 1979. Principles of diachronic syntax. Cambridge: Cambridge University Press.Mitchel, B. 1985. Old English syntax. Oxford: Clarendon Press.Pokorny, J. 1959. Indogermanisches etymologisches worterbuch. Bern: Francke.Radden, G. 2003. How metonymic are metaphors. In Metaphor and metonymy at the cross-roads.A cognitive perspective, ed. A. Barcelona, 92–108. Berlin: Mouton de Gruyter.Sweetser, Eve. 1990. From etymology to pragmatics. Metaphorical and cultural aspects ofsemantic structure. Cambridge: Cambridge University Press.Traugott, E. C. 1989. On the rise of epistemic meaning in English: An example of subjectificationin semantic change. Language 12: 31–55.Warner, A. 1999. English auxiliaries: Structure and history. Cambridge: Cambridge UniversityPress.

Part IVPragmatics, Discourse Analysis andSociolinguistics

Chapter 12“When We Talk, It Never Materializes”:Functions of Off-Record Communicationin Conflict TalkJoanna BobinAbstract Drawing on the recent contribution to the studies on linguistic impoliteness(Bousfield 2008), this chapter aims at examining and describing the roleof “the inarticulate” in interpersonal conflict, with particular attention devotedto the dynamics of conflictive exchanges. “The inarticulate” is by all means anon-academic term referring to a range of off-record strategies such as silence,“wordy silence”, i.e. “a torrent of words that are not addressing the true issue”(Tannen 1990), white lies, compassionate untruths, concealments, speaking inquotes, etc. Understanding what is meant but not said involves a necessity towork out implicatures, but how easy is it to retrieve speaker intention in conflictsituations? In light of the study of mental context in interaction (Kopytko 2002),one may attempt to identify the impact of the cognitive, affective and conativestates and processes on the expression and perception of the above mentioned(non)verbal behaviors. With regard to conflict dynamics, the presentation willconsider the emerging doubts about the Cooperative Principle; namely: is italways necessary to keep talking? Given the role of the mental context, howuncooperative is “the inarticulate”? Following the recent interest in discoursestylistics, especially in the relatively uncharted analysis of dramatic dialogues,the author will select conflictive exchanges between family members frommodern American plays, e.g. by Eugene O’Neill, Tennessee Williams or ArthurMiller.J. Bobin (*)State School of Higher Professional Education, Gorzów Wielkopolski, Polande-mail: jbobin@pwsz.plM. Pawlak and J. Bielak (eds.), New Perspectives in Language,Discourse and Translation Studies, Second Language Learning and Teaching,DOI 10.1007/978-3-642-20083-0_12, # Springer-Verlag Berlin Heidelberg 2011165

166 J. Bobin12.1 IntroductionOff-record communication, or “the inarticulate” is used as a term covering variousforms of indirect communication. In a conversation, and in this case, in a conflictiveexchange, indirectness serves different functions and may be as meaningful asdirect expression. After all, sometimes what is not said is more important for aninteraction than what actually is said. What indirect expressions, or “inarticulacies”,often indicate in conflict is repressed emotional intensity rather than lack ofinvolvement and non-commitment.12.2 Grice’s Cooperative PrincipleThe most convenient framework for considering indirectness seems to be Grice’sCooperative Principle (the CP). Grice suggests four maxims and a number of submaximsthat the CP subsumes: Maxim of Quantity, Quality, Relevance and Manner(1975: 45–46). The first three categories are related to what people say, and thefourth – Maxim of Manner – to how they say it, therefore, this aspect of the CP willinclude non-verbal, prosodic and paralinguistic information (Bousfield 2008: 22).In brief, the CP assumes interactants to give exactly as much information as isnecessary, and make sure it is true, relevant and expressed efficiently. However, it isfairly obvious that people do not always (or, in fact, rarely) abide by the maxims,yet they are understood. Grice (1975: 49) lists four forms of transgression of themaxims, of which flouting – or an intentional, blatant failure to observe a maxim –has greatest significance for conflict analysis. Due to the assumption that the overallCP is nevertheless being observed (conversations reflect some rational structure,rather than being disconnected), and with the help of contextual clues, a floutgenerates conversational implicature and is a possible vehicle for the expressionof impolite beliefs (Bousfield 2008: 23). At the rational level, this assumption“allows the audience to detect what the speaker is conversationally implicatingbut not explicitly saying” (Lumsden 2008: 1897) under one condition that Griceclarifies: that the hearer must be able to work out the implicature.12.3 Flouting the Maxims: Off-Record CommunicationMooney (2004: 905) notes that maxims “help to understand interactions not onlywhen followed, but also when they are not”. Consequently, off-record communicationin the form of indirectness contributes as much meaning to an exchange asdirect communication (observing Gricean maxims in this case). Brumark (2006:1207) remarks that “from a Gricean perspective, indirect speech may be explainedas more or less deliberate non-observance of the maxims requesting one to be asinformative, brief, relevant and adequate as appropriate to a given situation”.

12 When We Talk, It Never Materializes 16712.3.1 Reasons for Going Off-Record: PolitenessOne reason why people resort to indirectness is the regard of face: e.g. the want tominimize negative face threat (cf. Brown and Levinson’s (1987) negative facewants), with strategies such as the following: be conventionally indirect, don’tpresume/assume (question, hedge), impersonalize (Brown and Levinson 1987:131). In the face framework, a face threatening act (FTA) may be performed offrecord, that is “typically through the deployment of an indirect illocutionary actwhich has more than one interpretation and, thus, allows for plausible deniability onthe part of the utterer if the intended recipient takes offence at the face threatinherent in the utterance” (Bousfield 2008: 58). So while face threat iscommunicated, the implicature may easily be cancelled, and even more, the FTAitself may not be recognized. In terms of Gricean maxims, off-record FTAs areperformed in the form of the following linguistic strategies: those which generateconversational implicatures, e.g. give hints, give association rules, presuppose(violating Relevance Maxim), understate, overstate, use tautologies (violatingQuantity Maxim), use contradictions, be ironic, use metaphors, use rhetoricalquestions (violating Quality Maxim); and those that violate the Maxim of Manner:be vague, be ambiguous, over-generalize, displace H, be incomplete, use ellipsis(ibid.; Locher 2004). 1 These strategies are mostly used in order to mitigate facethreat (although the function of e.g. irony is debatable, as it may also be faceaggravating).Off-record is the weakest strategy for performing FTAs and oftenconstitutes violating rather than flouting (unostentatious). Off-record FTAs, despiteconveying impolite beliefs, are “redressed” not by strategies of politeness but by thevery fact that their form aims to mitigate face threat.12.3.2 Reasons for Going Off-Record: ImpolitenessTherefore, off-record impoliteness will often be present in interpersonal conflict. It isdefined as a strategy in which “the FTA is performed by means of implicature but insuch a way that one attributable intention clearly outweighs any others” (Culpeper2005: 44); with a sub-strategy of mock politeness, an obviously insincere “politebehavior”, linguistically realized as e.g. sarcasm. Bousfield (2008: 95) presents aslightly modified definition of off-record impoliteness, adding “withholding politenesswhere it is expected” as a sub-strategy. The danger of not being understood(though still existent) is minimized, as impoliteness in general is a framework whichassumes intentionality of offense; thus, given the larger context, it should be possibleto reconstruct that intention. It is a useful framework considering the fact thatanother reason for going off-record may be to increase the force of implicature1 Bousfield (2008) in fact suggests that off-record FTAs should be assigned redressive strategies aswell, since they are equally face-directed as on-record utterances.

168 J. Bobin(Brumark 2006). In some contexts, the offense communicated indirectly mayincrease the force of the message. The strongest, unmistakably “attributable intention”carries an impolite belief and hostile intention, thus communicating impoliteness(superiority, scorn, disagreement, criticism etc.) may clearly be recognized forexample by a third party as well as the addressee. Alternatively, an indirect impoliteexpression may aim to reinforce its implicit message when ‘normal’, directutterances are ineffective. Brumark (2006: 1211) notes that this function seems“especially common in asymmetrical but intimate situations, such as familydiscourse”.12.4 Context and the Cooperative PrincipleThe interpretation of utterances that do not conform to the maxims may sometimesbe troublesome: what is as informative as required for one interactant does not haveto be the same for another; the same may happen with deciding whether what is saidis true, relevant and clear enough not to generate an implicature. Grice’s maximsare all relative, subject to how the speaker, hearer and analyst interpret the contextof a given interaction. The key issue for conflict analysis is that of speaker intention.As Culpeper et al. (2003: 1552) put it, people “do not wear their intentions on theirsleeves and one interlocutor does not have access to the internal states of otherinterlocutors”, but they also assume that given adequate evidence, plausible speakerintention can be reconstructed. That “adequate evidence” would include suchfeatures of context as e.g. social and discoursal roles, physical context, previousevents (encounters), affect, power, rights and obligations, co-text and activity type.Having considered these (and other) contextual clues, the hearer is in the position toattribute specific intentions to the speaker at a given point of the conversation.Therefore, since implicatures are generated from different non-observances of theCP, all maxims and their breaches should be considered within context. However,there is always risk that it will be misunderstood – indirectness, in fact, requiresmore effort from the speaker (who risks “misfiring”) and hearer (who must be in aposition to work out the implicature).The attribution of hostile intentions and offences, even if indirect, is of greatimportance for the study of conflict and its dynamics. In Grice’s framework, theoperation of mental context is crucial as it constitutes the background knowledgewhich facilitates the interpretation of utterances: “Gricean language users sharecognitive context, beliefs and background assumptions with their interlocutors andeven more significantly they interpret meaning with the help of shared knowledge”(Kopytko 2002: 166). Kopytko (2002: 34) claims that “language is used in itscognitive, affective and conative context”, where the cognitive context is connectedwith knowledge structures and their accessibility to interactants, rationality, thinking,perceiving, attention and information processing; the affective context influences

12 When We Talk, It Never Materializes 169interaction through language users’ personality traits, needs, attitudes; and the conativecontext is associated with the interlocutors’ motivations and goals regulating theirstrategic use of language (ibid.: 35). This integrated cognitive-affective-conativesystem constitutes the mental (internal) context of language use. It is the conativecontext that is viewed as the speaker’s set of goals and motivations, and it is wherespeaker intention seems to originate.12.5 The Cooperative Principle and “Cooperation”For the purpose of conflict analysis, two important, though mutually exclusiveinterpretations of the Cooperative Principle put forward by linguists and interdisciplinaryresearchers should be discussed: social goal sharing and linguistic goalsharing. From the point of view of conflict researchers, cooperation as social goalsharing is impossible in language, as “conversation should immediately cease, or atthe very least become highly problematical when ‘quarreling’ or other conflictive orimpolite discourse begins to occur” (Bousfield 2008: 28). For impoliteness to occur,it has to be communicated and recognized as intentional. For conflict talk, animpolite utterance has to be replied to. As a consequence, large portions ofeveryday interaction would be deemed as uncooperative, rejecting the essence ofGrice’s principle.Davies (2007: 2311) points to an ambiguity of presentation of the term cooperationin textbooks and papers, resulting in potentially problematic interpretations,where the general problem with these unclear uses of the term is ascribing the CP“qualities that are more appropriate to its non-technical sense: high levels of efforton the part of the speaker, perfect utterances, and avoidance of misunderstandings”(2311). Precisely because this is not the case, the CP ought to be taken as operatingin terms of linguistic goal sharing.The notion of linguistic cooperation, as opposed to the everyday, “folk” understandingof the term cooperation, legitimizes the production and understanding ofimpolite and aggressive utterances in interaction. The Cooperative Principle aslinguistic cooperation in fact renders conflictive exchanges quite cooperative.Bousfield (2008: 29) supports the view that the only goal of a given interaction isthe transmission of information, and there are no other aims between interactantsthan “establishing the speaker’s illocutionary intent and getting the hearer(s) tounderstand the proposition which is being expressed or implied”. This happensregardless of whether the content of the utterance is “cooperative” or “uncooperative”in the “folk” sense of the term. The CP does not specify that the propositionexpressed or implied must be polite, truthful or relevant; it only facilitates understandingof what is being expressed or implied, and has no connection with thespeaker’s good intentions (Bousfield 2008). As an intended meaning can be conveyeddirectly as well as indirectly, and through countless different utterances, it isthe role of context to help the hearer reconstruct the right one.

170 J. BobinActivity types are useful for analyses of interactions which are uncooperativefrom the perspective of social goal sharing, but which generate implicature and helpestablish plausible speaker intentions. Thomas (1995) suggests that an activity typecomprises the following features: the goals of the participants (of the individualsrather than the event), allowable contributions (linguistic options constrained bysocial or legal frames), the degree to which Gricean maxims are adhered to orsuspended (in some contexts, suspension of the maxims is more expected than inothers, this is true of e.g. army discourse), the degree to which the interpersonalmaxims are adhered to or suspended (some activity types anticipate inadherence toe.g. Leech’s Politeness Principle 2 ), turn-taking and topic control (as a means ofcontrolling and managing the interaction), and the manipulation of pragmaticparameters (or contextual variables, dependant on the participants’ discoursalroles) (Bousfield 2008: 172–173).To analyze conflictive exchanges in the framework of the Cooperative Principle,it should be assumed that they are instances of cooperation a rebours, wherepragmatic principles (such as the Politeness Principle) are suspended. 3 Thus,paradoxically, interactants cooperate in the violation of pragmatic principles,within the recognized activity type – conflict talk. Such exchanges are non-cooperativein that they do not lead to compromise or solution, but are destructive andhostile; however, interactants cooperate in conflict, so implicatures and intentionsare correctly recognized.12.6 Linguistic Output StrategiesIndirectness is a manner of communication that may flout different maxims.Different linguistic output strategies characteristic of conflict will count as breachesof different maxims. Some linguistic realizations of maxim non-observances are asfollows:– Maxim of quantity: silences and pauses; could be flouts or violations, dependingon the context; understatements, overstatements.– Maxim of quality: white lies, “compassionate untruths” (cf. Mandala 2007),these are, however, mostly instances of non-ostentatious violation (out of sympathy);rhetorical questions may count as flouts, depending on their presupposition;in terms of speech act theory, all acts where felicity conditions considering“proper thoughts and feelings” are not fulfilled, e.g. insincere apologies(cf. Lowe 1998).– Maxims of relation and quantity combined: “wordy silence”, i.e. torrents ofwords not addressing the issue, blatantly flouting the maxim (cf. Tannen 1990).2 The Politeness Principle: Minimize the expression of impolite beliefs, maximize the expression ofpolite beliefs. It is further divided into submaxims of: Tact, Generosity, Approbation, Modesty,Agreement, Sympathy.3 Kopytko, R. (personal communication, 2010).

12 When We Talk, It Never Materializes 171– Maxim of manner: irony and sarcasm (ambiguity); innuendo, speaking in quotes,abrupt changes of topic (obscurity), blatantly flouting the maxim.Each of these linguistic output strategies may indirectly communicates impolitebeliefs (when they are instances of flouting, since violating has been classified asunostentatious).12.6.1 SilenceSilence has traditionally been associated with a “void”: lack of communication, lackof rapport and involvement, absence of cooperation (in terms of social goal sharing),“absence of speech, and absence of meaning and intention” (Ephratt 2008: 1910).Tannen (1985), however, highlights the pragmatic functions of silence, claimingthat silence is “anything but” – that is, anything but void, lack, or absence. Ephratt(2008) examines the role of “eloquent silence” in discourse and finds that it suitsJakobson’s five functions of language: it may convey information, express emotions,perform speech acts, fulfill the poetic function, and be a means of maintaining contactand alliance. Eloquent silence, in Ephratt’s description, is an equivalent of speech.Eloquent silence is employed by the speaker to communicate a message and canappear in such circumstances as religious ceremonies, “moments of silence”,substituting taboo words, following rhetorical questions, terminating conversations,opting out of conversations; or can take the form of “the unsaid”: empty speech,metaphor, euphemism, substituting A for B, allusion (Ephratt 2008: 1912).Tannen (1990) concentrates on the role of silence as a tool for conflict management,and also finds that silence and speech (noise) can be functional equivalents,for example in managing strong, problematic emotions. Silences and pauses aretreated here as actions, not as transitions between utterances. Silence does notrepresent an end of turn, but rather a climax, often signifying the most intense,damaging part of interaction. Similarly, a pause marks a character’s reaction towhat was said, it shows that the character is feeling, thinking something but doesnot want to reveal what it is. Pauses and silences, Tannen observes, “prevent theconflict from exploding and destroying the possibility of continuing the relationship”(ibid.: 263). They mask strong, unstated feelings, but as the mask can be seen,it is at the same time revealing.12.6.2 UnderstatementUnderstatement, for example in the form of a minimal response where a moreelaborate one would be expected, is a flout of quantity maxim, and producesimplicature of possibly impolite speaker intention. It may also take the form ofunderstated criticisms or compliments, or deliberate overt downplaying the seriousnessof an issue. Understatement implies a gap between what the speaker says

172 J. Bobinand what is in fact to be taken as his point of view (Berntsen and Kennedy 1996).Understatement expresses an attitude: although expressed indirectly (as it is a figureof reduction), (3) below is a strong criticism.Example 1: Eugene O’Neill, Long Day’s Journey Into Night ([1955] 2002: 128)1. EDMUND Ablaze with electricity! One bulb! Hell, everyone keeps a light on inthe front hall until they go to bed.He rubs his knee.I damned near busted my knee on the hat stand.2. TYRONE The light from here shows in the hall. You could see your way wellenough if you were sober.3. EDMUND If I was sober? I like that!In this excerpt, Edmund’s indirect criticism is expressed by prosodic features: heemphasizes “I” drawing attention to the fact that Tyrone is just as drunk, not sayingit though. By merely stressing the pronoun, he retorts with an identical accusation.12.6.3 Rhetorical Questions/Fallacious QuestionsIn Douglas Walton’s (1999) model of fallacious questions such uncooperativequestions are in fact intended to trick the respondent. They are a tactic of entrapmentvery common in everyday arguments. They are not asked to obtain information;they’re a means of attack as they’re asked in such a way that they don’t giveoptions of answers. Needless to say, traps involve high face damage. Waltonillustrates this theory with a classic “Have you lost your horns?” where whicheverway you answer, you either admit that you have horns, or that you had them.It is so because the respondent is not committed to the presupposition of thequestion or to part of it. Presupposition of a question is defined as a proposition, animplicit assumption or a background belief that is presumed to be acceptable to therespondent when the question is asked. The respondent becomes committed to thisproposition when he gives any direct answer (Walton 1999).Example 2: Arthur Miller, Death of a Salesman ([1947] 2006: 2021)1. WILLY [turning away] Ah, you’re counting your chickens again.2. BIFF [starting left for the stairs] Oh, Jesus, I’m going to sleep!3. WILLY [calling after him] Don’t curse in this house!4. BIFF [turning] Since when did you get so clean?5. HAPPY [trying to stop them] Wait a...6. WILLY Don’t use that language to me! I won’t have it!12.6.4 Wordy SilenceAnother type of maxim flout is the so-called “wordy silence”. It is also definedas a torrent of words that stands out against otherwise short turns. It is a smoke

12 When We Talk, It Never Materializes 173screen – it does not address the issue, so it flouts both the maxim of quantity andrelation. The emotional load of such an outburst is directed at the hearer, but thecontents are irrelevant. Being irrelevant, such an utterance receives no emotionalresponse from the addressee, so there are no damaging consequences. Wordysilence as we hear it is an indication of the opposite – of what we do not hear(Tannen 1990).Example 3: Sam Shepard, Buried Child ([1977] 2006: 60)1. DODGE (to VINCE) You could get me a bottle. (Pointing off left.) There’smoney on the table.2. VINCE Grandpa, why don’t you lay down for a while?3. DODGE I don’t want to lay down for a while! Every time I lay down somethinghappens! (Whips off his cap, points at his head.) Look what happens! That’swhat happens! (Pulls his cap back on.) You go lay down and see what happens toyou! See how you like it! They’ll steal your bottle! They’ll cut your hair! They’llmurder your children! That’s what’ll happen. They’ll eat you alive.In this example, Dodge’s response is not directed at his grandson Vince, butrather at his sons who are also present in this scene. It is an indirect form of(cancellable) attack; Dodge seems to be voicing his resentment at the treatmenthe receives from his sons – something that Vince is not involved in.Therefore (3) here contributes to the development of conflict between Dodgeand his sons.12.6.5 Irony and SarcasmA relatively common pragmatic effect of maxim non-fulfillment is irony (sayingone thing and meaning the opposite) and sarcasm (similar to irony, but with anintent to hurt), which Brumark classifies as “non-conventional indirect” utterances,whose implicit meaning not only differs from what is their locutionary content/illocutionary force (as is the case with “conventional indirect” utterances), butalso is even less transparent and even “more off record”, making them hardlyunderstandable outside their particular context (2006: 1211). In family conflict,such off-record utterances – given their hostile speaker intention – may carryincreased offensive force hidden under ambiguity.Irony and sarcasm need not posit threat to face, or be recognized as such.Brumark suggests that “indirect or implicit utterances often arise as spontaneousirony or sarcasm in most kinds of communication”, and they may have the samestatus as jokes in that they “rely on socio-culturally accepted norms and beliefs”(ibid.) and serve to strengthen in-group social bonds. In a case like this, however,they would be treated as banter (mock impoliteness), which does not contribute tothe development of conflict (the context of strong positive ties betweeninterlocutors would exclude such treatment of irony/sarcasm from the analysis ofconflict).

174 J. Bobin12.6.6 InnuendoInnuendo is a pragmatic act – a term which Mey (1993) uses to refer to various formsof indirect expression: hints, prompts, clues, which are used by interactants toinfluence each other. Bell (1997: 41) suggests three main features of pragmatic acts:they are non-overt, context-dependent and may be denied or ignored. Like indirectspeech acts, innuendos have two layers of meaning: overt (or pseudo-overt, as it maycomprise direct and indirect meaning) and non-overt, or implied meaning (where,however, the speaker’s intent is not meant to be recognized). Compared to indirectspeech acts, innuendos are motivated by possible sanctions that may be broughtagainst the speaker rather than interests of face. Innuendo can be said to be a typeof conversational implicature (Bell 1997: 46), at least in the cases where the intent isrecognized. However, while conversational implicature can be cancelled, innuendocannot; it becomes more transparent in the act of cancellation. Many fallaciousquestions carry innuendos in their presuppositions. Yet, in dyadic exchanges, theaddressee of the innuendo is rarely the target; it seems important in conflict analysisalso as far as the topic is concerned. Using innuendo directed at something that theaddressee identifies with may cause the conflictive episode to continue.Example 4: Tennessee Williams, Cat on a Hot Tin Roof ([1955] 2004: 118–119)1. BIG DADDY [leaving a lot unspoken] –I seen all things and understood a lot ofthem, till 1910. Christ, the year that–I had worn my shoes through, hocked my–Ihopped off a yellow dog freight car half a mile down the road, slept in a wagon ofcotton outside the gin–Jack Straw an’ Peter Ochello took me in. Hired me tomanage this place which grew into this one.–When Jack Straw died–why, oldPeter Ochello quit eatin’ like a dog does when its master’s dead, and died, too!2. BRICK Christ!3. BIG DADDY I’m just saying I understand such–4. BRICK [violently] Skipper is dead. I have not quit eating!5. BIG DADDY No, but you started drinking.In this excerpt, Big Daddy’s utterance contains a weak accusation that his sonBrick is homosexual: by invoking the memory of a homosexual couple, Straw andOchello, Big Daddy suggests an analogy between Ochello’s and Brick’s behavior.He is, as if, sending a hidden message about his suspicions. In terms of context, BigDaddy is not in a position to accuse his adult, independent son of homosexuality(a taboo topic in the conservative 1950s), but the innuendo in his utterance is quiteexplicit.12.6.7 Speaking in QuotesSpeaking in quotes is another way of indirect expression; as Bigsby (2000: 18)notes, “to speak as another is for the moment to evade the self which can bevulnerable to pain”. It is fairly impersonal, and the implicature can be cancelled.

12 When We Talk, It Never Materializes 175Ephratt (2008: 1916) discusses the emotive function of eloquent silence: “[e]venthough ‘I’ (‘me’) changes from one speaker to another, pragmatically speaking, it isalways unequivocal, referring to the ‘I’ who speaks (except in cases of fictivespeech, such as quoting in general and actors on stage in particular ...)”.Example 5: Eugene O’Neill, Long Day’s Journey Into Night ([1955] 2002:32–33)1. TYRONE(...) After all the money I’d wasted on your education, and all you didwas get fired in disgrace from every college you went to!2. JAMIE Oh, for God’s sake, don’t drag up that ancient history!(...)3. TYRONE (stares at him puzzledly, then quotes mechanically) “Ingratitude, thevilest weed that grows!”4. JAMIE I could see that line coming! God, how many thousand times– ! (Hestops, bored with their quarrel, and shrugs his shoulders)12.7 ConclusionWhen pragmatic principles are suspended, and participants cooperate intransgressing maxims (other pragmatic principles), then off-record has variousfunctions equivalent to on-record expression. Speaker intention is recognized and“the unsaid” has an aggravating function. In terms of options that an interactant haswhen faced with an FTA, indirect impoliteness can be said to be an offensivecounter-response, as in the framework of impoliteness it is attributed an unmistakablespeaker intention. The analysis of context and pragmatic parameters revealsthat conflictive exchanges are in fact cooperative, but they are a mirror reflection of“cooperation”: the goals of such interaction as well as the outcome are negative –but they are achieved (e.g. negotiation of power and dominance, expression ofanger, voicing unresolved past grievances, etc.). If pragmatic principles are notsuspended, then the exchange becomes uncooperative and leads to breakdown ofcommunication – then the unsaid is a termination format, as it ends the conflict(mostly in a stand-off or withdrawal).ReferencesBell, D. M. 1997. Innuendo. Journal of Pragmatics 27: 35–59.Berntsen, D. and J. M. Kennedy. 1996. Unresolved contradictions specifying attitudes – inmetaphor, irony, understatement and tautology. Poetics 24: 13–29.Bigsby, C. W. E. 2000. Modern American drama, 1945 – 2000. Cambridge: Cambridge UniversityPress.Bousfield, D. 2008. Impoliteness in interaction. Amsterdam: John Benjamins PublishingCompany.Brown, P. and S. C. Levinson. 1987. Politeness: Some universals in language usage. Cambridge:Cambridge University Press.

176 J. BobinBrumark, A. 2006. Non-observance of Gricean maxims in family dinner table conversation.Journal of Pragmatics 38: 1206–1238.Culpeper, J. 2005. Impoliteness and entertainment in the television quiz show: The Weakest Link.Journal of Politeness Research 1: 35–72.Culpeper, J., D. Bousfield and A. Wichmann. 2003. Impoliteness revisited: with special referenceto dynamic and prosodic aspects. Journal of Pragmatics 35: 1545–1579.Davies, B. L. 2007. Grice’s Cooperative Principle: Meaning and rationality. Journal of Pragmatics39: 2308–2331.Ephratt, M. 2008. The functions of silence. Journal of Pragmatics 40: 1909–1938.Grice, H. P. 1975. Logic and conversation. In: Syntax and semantics 3: Speech Acts, eds. P. Coleand J. Morgan, 4–58. New York: Academic Press.Kopytko, R. 2002. The mental aspects of pragmatic theory. Poznań: Motivex.Locher, M. 2004. Power and politeness in action. Berlin: Walter de Gruyter.Lowe, V. 1998. ‘Unhappy’ confessions in The Crucible. A pragmatic explanation. In Exploringthe language of drama: From text to context, eds. P. Verdonk, J. Culpeper and M. Short,128–141. London and New York: Routledge.Lumsden, D. 2008. Kinds of conversational cooperation. Journal of Pragmatics 40: 1896–1908.Mandala, S. 2007. Twentieth-century drama dialogue as ordinary talk. Speaking between the lines.Aldershot: Ashgate.Mey, J. L. 1993. Pragmatics: An introduction. Oxford: Blackwell.Miller, A. [1947] 2006. Death of a Salesman. In: S. Barnet, W. Burto, and W. E. Cain.An introduction to literature: Fiction, poetry, and drama, 1564–1631. New York: PearsonLongman.Mooney, A. 2004. Co-operation, violations and making sense. Journal of Pragmatics 36: 899–920.O’Neill, E. [1955] 2002. Long Day’s Journey Into Night. New Haven and London: Yale UniversityPress.Shepard, S. [1977] 2006. Buried Child. New York: Vintage Books.Tannen, D. 1985. Silence: Anything but. In Perspectives on silence, eds. D. Tannen andM. Saville-Troike, 93–111. Norwood, NJ: Ablex.Tannen, D. 1990. Silence as conflict management in fiction and drama: Pinter’s Betrayal and ashort story, Great Wits. InConflict talk: Sociolinguistic investigations of arguments andconversations, ed. A. D. Grimshaw, 260–279. Cambridge: Cambridge University Press.Thomas, J. 1995. Meaning in interaction. London and New York: Longman.Walton, D. 1999. The fallacy of many questions: On the notions of complexity, loadedness andunfair entrapment in interrogative theory. Argumentation 13: 379–383.Williams, T. [1954] 2004. Cat on a Hot Tin Roof. New York: New Directions.

Chapter 13Territorialization in Political Discourse:A Pragma-Linguistic Study of Jerzy Buzek’sInaugural SpeechesKatarzyna Molek-KozakowskaAbstract The purpose of this study is to review some discursive strategies used to(de)territorialize the European public sphere by the newly elected President of theEuropean Parliament Jerzy Buzek. A corpus of his inaugural speeches (over 7,000words) is examined in order to identify salient pragma-linguistic devices, such asfor example high-frequency references, linguistic markers of identities, values andinterests, as well as metaphors and argumentative schemata. These are presumed tohave been used by Buzek to territorialize the presidential office: to position himselfas its leader, to establish his credibility, to become its agenda-setter. Additionally,the analysis focuses on the way Buzek constructs Europe and the EU rhetoricallyfor the purposes of political self-legitimization. In this respect, Europe is projectedas a fairly deteritorrialized space: a common, even homogenous, public sphere thatdepends on specific European institutions for administration.13.1 IntroductionAccording to Macgregor Wise (2008: 11), a territory is an area of influence one has.We tend to mark our territories with real and symbolic signs that change the spacesaround us. Sometimes we first need to deterritorialize that space by removing the signsimprinted there by others in order to reterritorialize it in our own fashion, as is the casewith redecorating a newly bought house. That is why some cultural territories tend tobe rather ephemeral configurations of physical and symbolic means of expression ofour diverse identities, affiliations, ideologies and interests, whereas others seem topersist through our traditions or memories – our habitus (Bourdieu 1990).Political institutions can also be thought of as symbolic territories, which isevident in a range of metaphors we use when describing them. For example, oneperforms politics through constant attempts to “appropriate public space”.K. Molek-Kozakowska (*)Opole University, Opole, Polande-mail: molekk@uni.opole.plM. Pawlak and J. Bielak (eds.), New Perspectives in Language,Discourse and Translation Studies, Second Language Learning and Teaching,DOI 10.1007/978-3-642-20083-0_13, # Springer-Verlag Berlin Heidelberg 2011177

178 K. Molek-KozakowskaOne exerts political influence by marking the “political arena” with one’s presence,by voicing one’s opinions and advancing one’s agenda, by installing a network ofunique signs that could “resonate” with the public (cf. Wodak 2009). As a result,territorialization is predicated on specific rhetorical resources that one employs to“refurbish” a political institution or “refashion” its discourse. On the other hand,one may sometimes attempt to “clear” the political arena of previous influences, ortake a political issue out of its institutional context, usually in order to rhetoricallyenhance one’s argument or position. In such cases, the discursive dimension ofdeterritorialization seems to be of strategic significance, despite the fact that itsmechanisms have been relatively under-researched so far.The purpose of this study is to review some discursive strategies used to (de)territorialize the European public sphere by the newly elected President of theEuropean Parliament Jerzy Buzek. A corpus of his inaugural speeches (over7,000 words) will be examined in order to identify salient pragma-linguisticdevices, such as for example high-frequency references, linguistic markers ofidentities, values and interests, as well as pervasive metaphors and argumentativeschemata. These are presumed to have been used by Buzek to territorialize thepresidential office: to position himself as its leader, to establish his credibility, tobecome its agenda-setter. Additionally, the analysis will focus on the way Buzekconstructs Europe and the European institutions rhetorically for the purposes ofpolitical self-legitimization. In this respect, it is assumed that Europe will beprojected as a fairly deterritorialized space: a common, even homogenous, publicsphere that depends on specific European institutions for administration. Thepresent study will be informed by the current theoretical models of the Europeanpublic sphere, and will apply selected methodological instruments of culturallinguistics, pragmatics and Critical Discourse Analysis that allow for the operationalizationof the concept of territorialization and for a critical analysis of Buzek’sspeeches.13.2 European Public Sphere as a Symbolic TerritoryPolitical communication studies, cultural studies and media studies are often citedas the disciplines that provide contextualization for the pragmatic analysis of verbaldata in political discourse. In like manner, this project attempts to operationalize thecultural construct of territorialization and apply it to a sample of textual material.This is because discursive territorialization, as described in the introductory section,seems to be one of the key mechanisms of “doing politics” (Wodak 2009) throughspecific uses of language. Such mechanisms, according to critical discourseanalysts, should be subjected not only to linguistic description but also to pragmaticevaluation and social explanation and critique (cf. Fairclough 1989). Althoughthe mechanism of territorialization has only begun to be studied and it is mainlyconsidered in the context of global popular culture and mass-media influence(cf. Macgregor Wise 2008), there is no reason to assume that it is somehow absent

13 Territorialization in Political Discourse: A Pragma-Linguistic Study 179from the discourses emanating from the common European public sphere. On thecontrary, because the European public sphere has not been clearly delineatedyet, national, international and transnational issues, interests and influences areroutinely played and displayed there. This may contribute to the fact that the EU’spolitical discourse still abounds in ideological inconsistencies and power struggles,in which territorialization can be used as a useful rhetorical strategy applied byvarious political actors and interest groups to “win more space” for themselves.The notion of the post-modern public sphere has already been extensivelytheorized (e.g. Habermas 1989; Thompson 1995), although it is most often usedto describe national political arenas of public debate. By extension, in this paper,the term European public sphere (EPS) will be used to denote a temporally andinstitutionally unified space for a geographically and ethnically dispersed Europeanpopulation to communicate on the issues of European governance. The developmentof EPS not only facilitates the everyday implementation of transnationalpolicies in Europe, but also, in further consequence, contributes to the sense ofbelonging to a larger European community. This model of EPS has at least twoimplications. Firstly and significantly for the present study, such understanding ofthe public sphere validates the role of research into the properties of communicativeactivities taking place in the European institutions (and the discursive strategies thatare routinely employed there), as well as the processes through which these aremediated to the populace by the European and national media (Trenz 2004).Secondly, such understanding is compatible with the idea that EPS is a “projected”political space, even a form of “imagined community”, to use Benedict Anderson’s(1984) term. This implies that it is the institutional and mass-mediated discourses ofthe European public sphere that largely construct the ways Europeans perceiveand experience their membership to the political, economic, cultural and socialcollectivity labeled Europe (and not just reflect a pre-conceived notion of Europe orEuropean-ness).So, what are these discourses like? And do European institutions and mediafacilitate the construction of a transnational, transcultural, truly European publicsphere through their discourses? John Downey and Thomas Koenig seem to havedoubts about it, observing that “while the European Union is regularly presented asthe leading example of cosmopolitan citizenship, it is also commonly asserted thatit contains a ‘democratic deficit,’ because system integration has greatly outpacedsocial integration” (2006: 166). This would suggest that the elimination of manypolitical, legal and economic boundaries at the institutional level has not yet led tosufficient cultural and social integration in Europe, which is a condition for theconstruction of European community in any meaningful sense. In addition, despitetransnational media’s connectivity and circulation, the development of commonEuropean identity through the media has been relatively difficult, if not to sayineffective, so far (Volkner 2008).The European Parliament (EP) might well be the institution that suffers from the“democratic deficit” the least, as it is the only one that emerges in the course ofdirect elections. The EP has important legislative prerogatives that impinge on theworkings of the European Commission, as well as on legislation of individual

180 K. Molek-KozakowskaMember States. The EP delegates are often proponents of the “European cause” asmuch as guardians of their nations’ interests. The office of the EP’s President,currently held by a Pole, Jerzy Buzek, is of specific political, representative andadministrative importance. The President’s main function seems to be to ensuresmooth parliamentary proceedings by facilitating co-operation and consensusseeking,which are central to the EU’s political style. However, he is also, to borrowa term from media studies, an important agenda-setter, who is responsible fordeciding not only which issues are communicated, but also how and why they arediscussed.13.3 Territorialization in DiscourseThe notion of (de)territorialization can be operationalized for the purposes ofpolitical discourse analysis; however, there are at least several ways in whichpolitical institutions can be discursively territorialized. At the technical level ofdiscursive territorialization, for example, it should be considered (1) where landmarkpolitical speeches are made (at the official headquarters or abroad, e.g. inhistoric places), (2) in which language they are delivered, and (3) into whichlanguages they have been translated, if their mediated/archive versions are availableonline. For example, in the case of Jerzy Buzek, most of his early speecheshave been delivered in Polish, and their online transcripts at www.ep-president.euare available in English, French, German and Polish, with the exception of theInaugural Address, which is posted in the majority of EU official languages.At the (socio)linguistic level of discursive territorialization, one should note thespeaker’s accent, jargon usage, terms of address or other sociolinguistic variables,which may either reflect particular local conventions and national culturalpreferences or bear unmarked, transcultural characteristics (e.g. “Euro-Englishbureaucratese”). In the last instance, the political speaker will often replicate thestyle of the institutional documents, effacing his/her language- and culture-specifichabits for the sake of contributing to a deterritorialized perception of the office. Thisis somewhat at odds with the official priorities of the European Union, whoselanguage policy is to foster diversity and preserve equal status of minority languages.Finally, at the symbolic level of discursive territorialization, there is a variety oficons, images and symbols available in political expression, with cultural, religiousand historical references being the most common devices in political rhetoric(Chruszczewski 2003). These, again, could be more or less nationalistic in character(Billig 1995), based on the delineation, even polarization, between “us” and “them”(van Dijk 1998). Sometimes, the accrual of symbolic signs leads to the constructionof myths and other narratives, such as those of “the clash of cultures” between theWest and the East, or “the common European heritage”. These can be invoked asnaturalized, non-reflexive argumentative schemata for persuasive purposes.In this study particular attention will be paid to the following territorializationcodings: (1) identities – created or projected by means of adjectives and nouns that

13 Territorialization in Political Discourse: A Pragma-Linguistic Study 181index human collectivities, such as ethnic/national references, group labels, membershipcategories, e.g. “citizens”, “East Europeans”, “women”, “Christians” or“immigrants”; (2) values – invoked mainly by means of abstract nouns, e.g.“freedom”, “democracy” or “progress”, or their derivatives, in order to be positivelyemotionally associated with focal issues and to justify specific policies; and(3) interests – delimited by means of various linguistic exponents, includingpredication and modality to establish priorities in political planning and administration,e.g. “we need to...”, “we should...”.Apart from these linguistic resources for symbolic territorialization in politicaldiscourse, the study aims to investigate the means by which “Europe” is constructedrhetorically. There seems to be a range of discursive means to actuallydeterritorialize Europe and project it as a homogenous community: as a sharedworld of meanings beyond diverse interests and affiliations. Looking into theso-called “European rhetoric” (Trenz 2004: 310) requires identifying stylisticdevices that construct Europe as a unity, for example through the strategic use ofthe pronoun “we” for inclusion, or of the adjective “European”, as in such phrasesas “European heritage” or “European unemployment”, which have little precisedenotation but a huge potential for a “bandwagon appeal”. In the same vein, the useof generalizations and comparative statements may be indicative of the rhetoricallyenhanced construction of Europe, rather than its objective representation. Slogansand catch-phrases may be used to engender “banal Europeanism” (Cram 2001),while presupposed pan-Europeanism may hide important distinctions anddivisions behind a veil of “common European identity”. Finally, critical attentionto metaphors in which Europe is conceptualized as a specifically evaluated entity(e.g. “as our common home”), or to framing devices (“European politics as basedon consensus rather than conflict”) that guide recipients to specific interpretations,helps to demystify the rhetorical uses of deterritorialized Europe applied forthe purposes of justifying the ever-growing European bureaucracy (cf. Molek-Kozakowska 2010).13.4 Territorialization in Jerzy Buzek’s Selected SpeechesThe material for the subsequent analysis of territorialization is a textual corpusconsisting of 7,009 words of the first four official speeches delivered by the newlyelected President of the European Parliament Jerzy Buzek. All the speeches underanalysis are in English (although three of them were originally made in Polish) andare available from www.ep-president.eu/view/en/press/speeches:1. Inaugural speech by Jerzy Buzek following his election as President of theEuropean Parliament, Strasbourg, European Parliament, 17/07/2009 (990 words);2. Speech by Jerzy Buzek at the ceremony commemorating the 70th anniversary ofthe outbreak of WWII, Westerplatte, Poland, 1/09/2009 (728 words);

182 K. Molek-Kozakowska3. Speech by Jerzy Buzek at the Institute of International and European Affairs:“The Lisbon Treaty, Ireland and the Future of the EU”, Dublin, Ireland, 8/09/2009 (1,763 words);4. Inaugural Address by Jerzy Buzek, President of the European Parliament,Strasbourg, Plenary sitting of the European Parliament, 15/09/2009 (3,528words).Although the choice of speeches is random (they have not been cherry-picked toprove a specific point), the fact that they are the first ones delivered by the newPresident makes them especially relevant to the purposes of the study: the inauguralspeeches may include important markers of territorialization as well as carry typicalfeatures of “European rhetoric”.13.4.1 Self-Positioning in the New Territory: Negotiatingbetween the Polish and European IdentityThroughout all the studied speeches, Jerzy Buzek attempts to position himselfwithin the political territory of the office of the EP President. On the one hand hemakes frequent references to his personal experiences and Polish roots, on the otherhand he reveals himself to be a strong proponent of further European integration,particularly by relentlessly campaigning for the Lisbon Treaty (at that time it wasstill not clear if it was to be adopted).As regards territorialization through identity/value coding, in his inauguralspeeches Jerzy Buzek describes himself as a former dissident political activist,who used to live behind the Iron Curtain (1: 6, 1: 1, 3: 4), 1 and who, like millions ofcitizens of Central and Eastern European countries, “refused to give in to a hatefulsystem” (1: 12). He regularly invokes and validates the role of the Solidarityworkers’ movement in Poland, of which he was a member and a leader (1: 13,1: 21, 2: 5), stressing its revolutionary drive to abolish the post-war ideologicaldivisions in Europe. These divisions have been overcome, according to Buzek, whoopenly admits his religious affiliation, due to the victory of ethical imperatives thatstem from “our Christian heritage” (3: 15–16, 2: 11). Also “the lessons of Pope JohnPaul II” are shown to have catalyzed political change in Europe (1: 6). Last but notleast, Buzek demonstrates his strong national anchoring: he makes use of Poland’shistory for parallels and exemplifications of his political points and sometimescolors his argumentation with references to Polish traditions (1: 6, 1: 11–13,1: 21, 2: 1, 2: 5, 3: 4, 3: 15–16). In some instances he “mythologizes” Poland’sWWII and Cold War sacrifice and its contribution to Europe’s present-day stability.For example, he draws attention to the Polish struggle against fascism and totalitarianism,which he reframes at Westerplatte as the fight “for the freedom and honor1 Numbers refer to speech number in the corpus (as listed in Sect. 13.4): paragraph number.

13 Territorialization in Political Discourse: A Pragma-Linguistic Study 183of Europe” (2: 1), and applauds Poland for its efforts in promoting democracy andEuropean integration in the last two decades (1: 11–12, 2: 5).It can be observed that Buzek’s relatively frequent references to his personalexperiences and political activities, to Christian values, and to Polish war heroismare mainly oriented towards establishing his credibility within this new Europeanterritory, and should not be interpreted as just “empty rhetoric”. This way of selfpositioningis congruent with the demands of the presidential office, as it is difficultto be a respected European leader without being a national leader. Moreover, suchan expressive projection of personal stature and national identity is key to Europeanpolitical discourse, which abounds in symbolism, as in Buzek’s recognition of thesignificance of “conferring this great responsibility [of EP presidency] on a representativeof a Central and Eastern European country” (1: 10). Hence, notwithstandinghis firm roots, Buzek projects himself as a truly European leader, for exampleby applauding the Lisbon Treaty or other founding EU documents in all hisspeeches, and devoting much space to illustrating why it is crucial to strengthenthe effectiveness of European institutions and, in this way, help to build a morestable and prosperous union (1: 15–17, 2: 9, 2: 12–14, most of 3 and 4). Indeed,a rough word frequency count reveals that Buzek uses the word “Europe” or“European” every 56 words (sometimes the modifier is left out when referring tothe Parliament, the Community or the Commission).13.4.2 Who Are “We”?An important aspect of discursive territorialization is how identities are constructedthrough linguistic means of inclusion and exclusion. This is why the key categorystudied here is the use of the pronoun “we”, which often indicates the speaker’simplicit categorization of political actors. In Jerzy Buzek’s speeches, “we” seems tohave various scopes depending on his rhetorical purpose. For example, in his firstspeech he makes a distinction between “us” – Central and Eastern Europeans and“you” – Western Europeans, only to invalidate that distinction:We on one side of the Iron Curtain struggled for freedom and democracy. You, on the otherside, helped us politically and through small, but extremely important, gestures of support(...) we have been working together to build a united Europe. There is no “us” and “you”.We can say loud and clear that this Europe belongs to us all. (1: 13–14).The maximally inclusive “we” is instrumental to Buzek’s endeavor to abolish thelong-lasting division into Western and Eastern Europe (also known as “old” and“new” Europe) still persisting in the EU discourse. In addition, this device functionsto enhance a frequent conceptual metaphor typical of all Buzek’s speeches, namely,that of a “border-less Europe”, a continent without real and mental “walls” and“barriers,” as in the following:Over the many years since that first Community was established, we have been knockingdown the walls left by the Second World War. We cannot now allow them ever to be raisedagain, through exploitation of energy resources or manipulation of historical facts. (2: 10)

184 K. Molek-KozakowskaIn the second speech, commemorating the outbreak of WWII in Europe andspecifically the attack on Westerplatte in Poland, Buzek is careful not to introducethe opposition between “us” – the victims and “them” – the invaders, as this wouldbe contrary to his integrationist agenda. Instead he makes systematic references to“us” – present-day Europeans, who are fairly united and integrated, and who needto “ensure that our nations are truly reconciled” (2: 14). Constructing this categoryof “us” implies the existence of “the others” – those who lived and fought, survived orperished in past military conflicts and who embody the past ideological divisions ofEurope (2: 5–6). This distinction is a contrast schema that argumentativelyunderpins Buzek’s warning to “remember the lessons of the past” and further“strengthen our European solidarity” in order to make future wars unthinkable(2: 13), as envisioned by the European Coal and Steel Community’s foundingfather Robert Schuman.The third speech, delivered in Dublin and devoted to promoting the Treaty ofLisbon before the Irish referendum, makes another strategic use of the pronoun“we”. Here, Buzek attempts to build an informal friendly relation with the Irishpublic, for example by casual references to his personal feelings and experiences:I am certainly not here to tell the Irish people how to vote (...) I believe that this is tooimportant a referendum for the luxury of a low turnout (...) I was very surprised to seecertain posters around the city put up by the NO campaign. (3: 4,7,8)Such disclaimers make an initial impression that he is not a party in thiscontroversy, which is not so. Then Buzek highlights some similarities betweenIreland and Poland (which has already completed its process of ratification of theTreaty) in order to dispel Irish concerns as to the purported loss of sovereignty onratifying the Lisbon Treaty. By insisting that Poland and Ireland have much incommon, Buzek develops a narrative in which he applies a schema of comparisonwhich is conducive to his persuasive purposes:We have similar traditions and a similar history: a history of occupation, immigration dueto poverty and political oppression. We are both now modern societies with a commitmentto free trade, a market-economy, export-driven growth. (3: 17)By claiming a common identity and noting the rise of similar concerns withrespect to e.g. taxation, minimum wage regulation and abortion in both countries,Buzek successfully places himself on “their” side of the debate to convince the Irishthat these issues are not to be regulated by Brussels. The members of the NOcampaign are then effectively marginalized and their fear-mongering and Euroskepticismexposed (3: 20).Finally, in his Inaugural Address, Buzek develops his vision of a united Europeby repeatedly constructing the common European identity invoked by the inclusive“we”. He shows his enthusiasm for further European integration in frequent sloganlikeinterjections in his political address, e.g. “Old and new Europe are no more.This is our Europe!” (4: 4), “The richness and strength of our institutions also derivefrom our differences (...) That is what makes Europe fascinating” (4: 65), “Becauseit is our Europe. A modern Europe. A strong Europe” (4: 84). As is common in thisgenre, Buzek also uses “we” to refer to his fellow Members of the European

13 Territorialization in Political Discourse: A Pragma-Linguistic Study 185Parliament (MEPs) and outlines the main tasks and challenges the assembly mustface under his guidance. The new President tends to eschew the use of “I” to sethis agenda, preferring the collective “we” to detail his policies, e.g. “Yet, regardlessof the Treaty, we feel the need for change. We feel the need for a moredynamic parliamentary dimension within our institution” (4: 48), “We must forgecloser ties with the Council of Ministers” (4: 59). Sometimes Buzek’s “we” isambiguous and oscillates between references to MEPs and all Europeans, as in thefollowing:As we [MEPs/Europeans] tackle this crisis, let us [MEPs] listen to the economists who saywe [MEPs/Europeans] should use this period to undertake a profound reform of theEuropean and world economy. Once we [Europeans] emerge from the present crisis, theenthusiasm for reform will be lost and we [MEPs] will not have safeguarded ourselves[Europeans/MEPs] against the next one. (4: 14)Such conflation of the scopes of “we” is quite common in political discourse,since it is a useful rhetorical strategy to project the interests of the institution as theinterests of the society at large. This type of rhetoric seems to validate the parliamentarydecision-making process as truly democratic and representative of allnations, parties and groups. In its generality and abstractness, it also results indeterritorializing the European political arena of nation-specific problems andnuances of policy.13.4.3 “Common” European Identities, Values, InterestsAs shown above, the European identity, especially in the Inaugural Address, seemsto be almost all-inclusive, which is intensified with Buzek’s occasional overgeneralizationsto advance the argument for integration, e.g. (emphases mine) “Thecohesion policy must remain a priority in the next Community budget, if we wantto achieve full integration of our reunited continent” (4: 17), “[l]et us explain to ourcitizens why Europe is good, and why the Community method benefits allEuropeans” (4: 19), “[i]mmigration has always brought Europe benefits” (4: 28).The most commonly used group membership categories are “Europeans” and“citizens”, which are large collectives invoked to forge common identity. Theseare complemented with other collective labels, e.g. “Member States”, “women”,“immigrants” or “foreign partners”, when specific policies are discussed.The projection of uniform European identity is consolidated through referencesto common values and ideals, which constitute “the basis” or “the foundations”(4: 7) (note the BUILDING metaphor) of European institutions. By quotingsnippets from various philosophical, historical and religious sources, rangingfrom Aristotle (4: 8) to Hannah Arendt (4: 9), form the French revolutionaries(1: 8) to Pope Benedict XVI (3: 16), from Jean Monnet (4: 36) to BronisławGeremek (4: 78), Buzek aims to acknowledge the variety of influences that areconstitutive of the present-day EU ethos. The frequency of such words as

186 K. Molek-Kozakowska“democracy”, “peace and stability”, “integration”, “freedom and well-being”, aswell as “change and reform” is, on the one hand, indicative of Buzek’s politicalprinciples and priorities, and, on the other, likely to resonate with general Europeanpublic: who would refuse to cherish such values as the “glittering generalities”listed above?Out of these values and identities stem the European “common interests”, assuccinctly put by Buzek in his first official speech as the EP President:On the big issues, we need to stand in defense of democracy. On the bread-and-butterissues, the people of Europe expect us politicians to resolve this [economic] crisis, and wemust set about doing so immediately. People want jobs (...). Energy security is crucial(...). We have to tackle climate change. (1: 15)In addition, solving the demographic crisis, ensuring equal rights for women andimmigrants, developing effective common foreign policy, campaigning for humanrights and reducing the “democratic deficit” of European institutions seem to beother “must-do”, “should-do” and “need-to-do” entries of Buzek’s presidency.Since many interests are expressed by means of bare assertions (and many nounsare preceded by the definite article “the”, which presupposes previous knowledge),this projects a particular version of European political reality and constructs thatversion as universally shared and undisputable.What might be noticed in Buzek’s speeches is that very often all the Europeancountries seem to have the same interests and preoccupations, and that allMember States do their utmost to cooperate in ensuring that those challengesare met. This is because the main political ethos of the unified Europe, accordingto Buzek, is “to resolve conflicts and opposing interests (...) through debate andargument” (4: 8). Here, the attempt at deterritorializing Europe of nationalinterests works rhetorically to advance an idealistic vision of the EuropeanUnion’s key integrative and mediating role. As deterritorialization backgroundsparticular national differences and priorities, it can be effectively used to justifythe ever-growing European bureaucracy. That is why it can be stated that Buzek’sinaugural speeches largely conform to the characteristics of “European rhetoric”,as outlined in Sect. 13.3.13.5 ConclusionsAlthough the understanding of a nation as an “imagined community” (cf. Anderson1984) is widely embraced, and the functioning of national public spheres, even intheir banal forms (cf. Billig 1995), has been relatively well-researched, this cannotbe said of the European public sphere (EPS), which is still a highly contestedterritory. Hence, taking into consideration the imagined status of European identity,its discursive constructedness, and the widespread perception of the “democraticdeficit” in European institutions (cf. Downey and Koenig 2006), the present studyhas used the notion of territorialization to investigate some salient properties of EPS

13 Territorialization in Political Discourse: A Pragma-Linguistic Study 187discourse, as exemplified by Jerzy Buzek, the current President of the EuropeanParliament.Firstly, the notion of territorialization has been operationalized for the purposesof political discourse analysis, and then the uses of “territorializing” and“deterritorializing” pragma-linguistic devices have been identified and analyzed.In the course of a critical analysis of a corpus of Buzek’s inaugural speeches, it hasbeen illustrated that (de)territorialization codings are deployed strategically toenhance his political arguments. This partly confirms Downey and Koenig’s(2006) thesis of the relative shortage of “truly transnational, pan-European publicsphere”. The patterns of territorialization found in Buzek’s speeches seem toindicate that the core territorial identification concept in EU discourse is still thatof the nation and that the emphasis on Member State sovereignty is still central tothe EPS. In a qualitative study of recurrent keywords, as well as applications ofsalient metaphors, argumentative schemata and markers of identities, values andinterests, it has been demonstrated that territorialization constitutes a powerfulrhetorical strategy and can be used to build political credibility and to advance anagenda. Likewise, the use of “deterritorialized” Europe – conceived of as a homogenousspace – can be in many situations an effective rhetorical device. As a slogan,a presupposition, an overgeneralization or a projection, “Europe” is often a constructused for the legitimization of certain political visions and interests rather thanfor the genuine expression of common European identity.ReferencesAnderson, B. 1984. Imagined community: Reflections on the origin and spread of nationalism.London: Verso.Billig, M. 1995. Banal nationalism. London: Sage.Bourdieu, P. 1990. In other words: Essays towards reflexive sociology, trans. Matthew Adamson.Stanford: Stanford University Press.Chruszczewski, P. 2003. American political discourse analysis. Berlin: Logos Verlag.Cram, L. 2001. Imagining the union: The case of banal Europeanism? In Whose Europe:Interlocking dimensions of European integration, ed. H. Wallace, 343–362. London:Macmillan.Downey, J. and T. Koenig. 2006. Is there a European public sphere? European Journal ofCommunication 21: 165–187.Fairclough, N. 1989. Language and power. Harlow: Longman.Habermas, J. 1989. The structural transformation of the public sphere. Cambridge, MA: The MITPress.Macgregor Wise, J. 2008. Cultural globalization. Oxford: Blackwell.Molek-Kozakowska, K. 2010. The rhetoric of space in political discourse: Spatial metaphors inpolitical speeches of the European Commission. In Exploring space: Spatial notions incultural, literary and language studies, ed. A. Ciuk and K. Molek-Kozakowska, 93–103.Newcastle: Cambridge Scholars Publishing.Thompson, J. 1995. The media and modernity: A social theory of the media. Cambridge: PolityPress.

188 K. Molek-KozakowskaTrenz, H. 2004. Media coverage on European governance: Exploring the European public spherein national quality newspapers. European Journal of Communication 19: 291–319.Van Dijk, T. A. 1998. Ideology. A multidisciplinary approach. London: Sage.Volkner, I. 2008. Satellite cultures in Europe: Between national spheres and a globalized space.Global Media and Communications 4: 231–244.Wodak, R. 2009. The discourse of politics in action: Politics as usual. Basingstoke: PalgraveMacmillan.

Chapter 14From a Complaint through Therapyto Recovery: Patient Indexicalityin Medical Case ReportsMagdalena MurawskaAbstract In the present paper, the issue of patient indexicality in professionalmedical texts has been addressed. To this aim a corpus of medical case reports hasbeen compiled to examine both direct and indirect references to the patientsdescribed there. The studied tokens have been investigated from two perspectives.First, the focus has fallen on patient textual presence/absence as conditioned by theaims of the respective text-parts. Second, the analysis of patient reference in thesections of the case reports has been discussed with respect to some of the factsfrom the history of the development of medicine. The form and content of thetexts under study may also be influenced by the currently practiced model ofmedicine, i.e. the biomedical model. Furthermore, the analysis has drawn on thehierarchical levels of medical description as well as on two models of diseasepresentation which also help to explain the choice of modes of writing aboutpatients and diseases they suffer from. The study reveals that as the texts progress,they become more patient-evacuated and focus on his/her progressively smallerbody parts. In other words, patient reference changes from direct indexicality toindirect references to his/her body parts or to the textual absence of the treated. Thiseffect is achieved not only by the type of information imparted but also by thelexical and grammatical resources used to describe it. Consequently, the modeof writing as testified in the case reports at hand contributes to the presentation ofmental/bodily experience, disease and treatment in abstraction from the patient.14.1 IntroductionThe language medical professionals use in order to document their academicactivities has been widely researched in recent years within the framework ofspecialised discourse analysis. It may seem that the bulk of the studies on writtenM. Murawska (*)Adam Mickiewicz University, Poznań, Polande-mail: mmurawska@ifa.amu.edu.plM. Pawlak and J. Bielak (eds.), New Perspectives in Language,Discourse and Translation Studies, Second Language Learning and Teaching,DOI 10.1007/978-3-642-20083-0_14, # Springer-Verlag Berlin Heidelberg 2011189

190 M. Murawskamedical discourse constitute the quantitative investigations into specific lexical andgrammatical features (Gotti and Salager-Meyer 2006) and their respectivefunctions as well as the organisation and presentation of ideas in specific textparts (Myers 1990). Another substantial body of research has examined linguisticfeatures of texts and the effect they produce. These are studies devoted toimpersonality (Hyland 2001), authorial identity (the KIAP project), metaphors(Van Rijn-Van Tongeren 1997), and the presentation of patients and diseases(Anspach 1988; Grice and Kramer-Dahl 1992; Kenny and Beagan 2004). In thispaper, I investigate the modes of patient imaging in medical case reports asconditioned by the context of their production as well as by the aims of therespective text-parts. I assume that, depending on the macro-structural functionsof particular sections of the texts, the authors can allocate communicative accents ina sentence in a variety of ways by adopting different perspectives and the mode ofpatient reference. I will begin with a theoretical background for the study. Next, thedata and the methods applied in the study will be described. Finally, the results ofthe analysis will be discussed.14.2 Theoretical BackgroundIn this section, the theoretical background for the study will be presented.According to Bazerman (1988: 47), scientific discourses are shaped by givendisciplines. It follows that the ways in which academics inform about their scientificactivities are influenced by modes of reasoning, methodologies, objectives, etc. of agiven area of study (cf. Atkinson 2001; Taavitsainen and Pahta 2000). In otherwords, how researchers argue in scientific papers and the theories and methods theychoose depend on a particular model which is practiced at a particular moment in aparticular discipline. Following this line of reasoning, features of medical textsmight be conditioned by the nature of medicine both as an area of study and ofpractice.As regards medical practice, the framework that has outlined the premises ofhow medicine has been practiced in western societies since the mid nineteenthcentury is the biomedical model. It views illness as a direct consequence of thediseased body and patients as mere recipients of treatment (cf. Wade and Halligan2004: 1398). Therefore, the model is reductionist because it limits the understandingof disease only to its biological manifestations excluding social and psychologicalaspects. Relevant to the present study is also the perspective on the biomedicalmodel offered by the sociology of medicine. This discipline approaches medicinecritically, which means that it does not treat medical knowledge as given but as theproduct of social and cultural practices (cf. Atkinson 1995: 25). Consequently, thesociology of medicine provides a number of notions and distinctions which prove tobe instrumental in examining how medical discourse reflects the current status ofmedicine, namely the biomedical model.

14 From a Complaint through Therapy to Recovery 191Firstly, the sociology of medicine distinguishes between disease and illness.Disease is a concept of a state conditioned by the presence or absence of themanifestations indicating a given pathological change. Illness, on the other hand,is defined in terms of its subjective perception by the patient (cf. Bond and Bond1986: 200). It is a conceptual differentiation between what the doctor sees and whatthe patient feels and it goes in line with the biomedical model in which only theabnormal states within the body are treated. This way, the model centres around thepatient’s body and its biological processes abstracting from the patient as a wholeand social and psychological aspects of his/her illness.Secondly, according to the sociologists of medicine, the biomedical modelconceptualises diseases as “it”, i.e. as an “isolatable entity” (Blois 1984), whichmanifests itself unchangingly in all patients. Associated with Plato, whose aim wasto classify diseases, this mode of disease presentation is referred to as the nominalistmode. It allows to describe a disease as a purely abstract concept and as separatefrom its context, i.e. the patient (cf. Blois 1984: 92). For instance, doctors andpatients may refer differently to the same medical condition. While the patient maytalk about a stomachache, the doctor may refer to it as gallstone colic, thus reducingthe experience of a pathological state to an entity carrying a particular meaning inmedical discourse (cf. Nijhof 1998: 739). This entity can be enumerated andreferred to in abstraction from the patient, as opposed to particular sensations, i.e.“attributes that constitute his illness” (Blois 1984: 94). The following sentenceexemplifies this mode:1. Fifteen months after the patient’s injury, staff members reported possible legflexion and eye closure on two separate occasions in response to command, butthe responses were rare and inconsistent during the next 2 months (Childs andMercer 1996).Although example (1) includes a direct reference to the whole patient, his/hersymptoms and reactions are enumerated as if they were entities which are notpart of the patient’s experience of illness.An alternative is the psychological mode, which is attributed to Hippocrates andviews illness as a collection of changes that are experienced by a particularindividual.14.2.1 Hierarchical Levels of Medical Description (Blois 1984)The hierarchical levels of medical description as proposed by Blois (1984: 113)help to explain the choice of modes of writing about patients and their diseases:Level 0: Patient as a whole.Level 1: Major patient part: e.g. chest, abdomen, head.Level 2: Physiologic system: e.g. cardiovascular system, respiratory system.Level 3: System part, or organ: e.g. heart, major vessels, lungs.

192 M. MurawskaLevel 4: Organ part, or tissue: e.g. myocardium, bone marrow.Level 5: Cell: e.g. epithelial cell, fibroblast, lymphocyte.Level 6: Cell part: e.g. cell membrane, organelles, nucleus.Level 7: Macromolecule: e.g. enzyme, structural protein, nucleic acid.Level 8: Micromolecule: e.g. glucose, ascorbic acid.Level 9: Atoms or ions: e.g. sodium ion.Following this description, the departures in the patient’s condition are featuresof diseases which can be considered with reference to the patient in general or to thespecific parts of his/her body affected by these changes. On this reasoning, whileanxiety, fever or anorexia pertain to the whole patient’s condition, the pain whichoccurs in angina pectoris can be felt in given body-parts (Blois 1984:112). Thelevels above start with the whole body, go through systems and organs where bodilysensations of various sorts can be experienced, and end with cellular, molecular andatomic levels at which biological processes can be inspected. These processes maynot always cause directly observable or directly felt signs, yet very often certainbodily or mental reactions are the consequences of the abnormalities at these lowerlevels. Therefore, while the description at levels 0–4 may refer to specificsymptoms or reactions which patients have, the other levels describe changes notdirectly affecting patients’ experience (levels from 5 to 9). Consequently, a diseasecan be presented at various hierarchical levels of medical description referring todifferent body-parts or their constituents, which affects patient indexicality.14.2.2 Container MetaphorOne of the techniques of patient imaging is the metaphor of a container. Metaphorcan be defined as “the use of language to refer to something other than what it wasoriginally applied to, or what it ‘literally’ means, in order to suggest some resemblanceor to make a connection between the two things” (Knowles and Moon 2006:1). In other words, metaphor describes a thing in terms of another thing establishinga common ground between the two. Introduced by Lakoff and Johnson (1980), thecontainer metaphor presents objects or notions as having an inside and outside andas being capable of holding something. As Lakoff and Johnson (1980: 28) explain,“[w]e are physical beings, bounded and set off from the rest of the world by thesurface of our skins, and we experience the rest of the world as outside us. Each ofus is a container, with a bounding surface and an in-out orientation”. In medicaldiscourse, the concept of disease in the patient is utilised to describe medicalprocedures or to give an account of medical facts. From this perspective, thepatient’s body tends to be viewed as a container in which diseases are localisedand particular treatment is performed. Language-wise, this effect is achieved byplacing patient referents in the positions of complements of prepositional phraseswith the meaning of location.

14 From a Complaint through Therapy to Recovery 19314.2.3 Sentence from Three PerspectivesAs patient imaging will be examined on the basis of exemplary sentences, a fewterms regarding the structure and function of sentential elements need to bediscussed. Traditionally and from a syntactic point of view, a sentence is thoughtto consist of two parts, namely the subject and the predicate, the latter comprised ofthe object, complement and adverbial. As regards the subject and the object,position-wise, they precede and follow the verb respectively (in the case of adeclarative sentence) (Greenbaum and Quirk 1990: 207). Functionally, the subjectis “prominent positionally” in that it is a “perspectival centre” and “the starting ofthe communication of a sentence” (Smith 2003: 192–193). Supporting evidence forthe prominence of the subject can also be found in psychological studies (Brennan1995; Paivio 1979). Chafe (1976) compares the structure of a sentence to a packageof information which is “unwrapped” step by step by a reader. Although it consistsof many elements, “knowledge directly attached to the subject may be mostimmediately accessible” (Chafe 1976: 44). Semantically, given a prototypicalActive Voice sentence, the Agent, “the ‘doer’, or instigator of the action denotedby the predicate” (Aarts 1997:88), is “the most topical participant”, “the apex of thetopic hierarchy” (Givon 1990: 566), being the first candidate for the subjectposition. In contrast, in a prototypical Passive Voice sentence it is the Patient,“the ‘undergoer’ of the action or event denoted by the predicate” (Aarts 1997: 88),who/that is granted the title of “the most topical participant” (Givon 1990: 566) andthe position of the subject. Indeed, if a given situation with participants involved isdescribed, “they rank on a scale according to their importance” (Givon 1984: 137).Consequently, the subject and the direct object are primary and secondary clausaltopics respectively (Givon 1984: 138; cf. Smith 2003:193). Clearly then, grammaticalrelations have their pragmatic consequences for the prominence of specificparticipants with their thematic roles (cf. Van Dijk 1980: 95–96). Underlying thisreasoning is the approach to the sentence introduced by the linguists of the PragueSchool – Functional Sentence Perspective – in which semantic, grammatical andfunctional levels are considered (cf. Firbas 1974).14.3 Data and MethodsThe corpus for this study comprises 56 case reports taken from four internationalmedical journals aimed at health professionals – The Lancet (15), The Journal ofAmerican Medical Association (13), The New England Journal of Medicine (16)and The British Medical Journal (12). In their study of case reports from adiachronic perspective, Taavitsainen and Pahta (2000: 60) define this genre in thefollowing way: “[i]n its typical form, the case report records the course of a patient’sdisease from the onset of symptoms to the outcome, usually either recovery or death.The background and a commentary on the disease are also given, but their scope

194 M. Murawskamay vary. Often a limited review of the literature is added and the number of knowncases stated”. Generally, case reports present new diseases or diseases that arealready known but which have unusual manifestations.As regards their structure, case reports are “the briefest and simplest category ofarticle” (Adams Smith 1984: 27) and usually consist of three parts: “a shortIntroduction, a more detailed Case Report body, and a brief Comment or Discussionsection” (Rowley-Jolivet 2007: 185). Rhetorically, the presentation of informationin case reports renders them “descriptive and expository” (Salager-Meyer et al.1989: 155), which is in keeping with the primary aim of this genre.In the analysis, each article was carefully read in search of any words thatreferred to the patients described there. Next, the examples containing referencesto the patients were isolated by means of WordSmith 5 and further examined. Theexamples which did not refer to the patients directly, yet concerned variousaspects of their treatment, were also taken into consideration. In the followingdiscussion of the results, it will be demonstrated how various grammatical andlexical configurations of the texts in the examined corpus allow the authors to focuson various aspects discussed in different sections of the case reports. The presentationof the results of the study as well as their discussion will follow the order of thesections in a case report with a view to showing how patient imaging changes as thetexts progress.14.4 Results and Discussion14.4.1 Introduction SectionA typical case report commences with an Introduction, which explains the reason ofthe patient’s presentation.2. A 64-year-old woman presented to the emergency department with a stiff painfuljaw (Lindley-Jones et al. 2004).In this section, the description is restricted to the levels considering conditions ofthe whole body or its parts and patients are presented as experiencers of theseconditions (cf. the adjective painful referring to certain bodily experience, whichmeans that the psychological mode is used here). As such words as painful denotesubjective perception, what is narrated here is the account of an illness as experiencedby an individual, in contrast to a disease which is objectively observed by aphysician. The floor is given to the patient who is granted the agency of complainingor reporting. Yet, it is achieved only symbolically through the authorial persona, asthe patient’s account is in the third person.Language-wise, the words referring to patients in this section are personalpronouns or nouns and they are located in the subject position. As a result, patientsare imaged as whole persons (cf. Wade and Halligan 2004: 1400), being the primarytopics and content of the clauses (Givon 1990: 137–138; Halliday 1994: 75).

14 From a Complaint through Therapy to Recovery 19514.4.2 The Case Report BodyThe case report body follows Introduction and consists of three parts: History,Examination/Tests and Treatment. As regards History, it informs about previousdiseases that the patient underwent.3. He was otherwise asymptomatic and had no history of drug use (White et al.2004).4. The patient’s history was characterised by poor orthostatic tolerance and aninability to stand upright for more than 2 minutes without fainting (Robertsonet al. 2005).5. There was no history of trauma (Kong et al. 2005).In (3), the word referring to the patient occupies the subject position. (4) and (5)are quite the opposite. Here the history is underscored, yet, while in (4) it is stillpresented as belonging to the patient, in (5) there is no mention of him/her. As aresult, the patient’s textual prominence decreases from (3) and (4) to (5), with (5)being abstracted from him/her. (4) is also an example of the nominalist mode ofdisease presentation, i.e. one based on the enumeration of symptoms, reactions,conditions, etc. and contributing to the perception of a disease as an entity, i.e. “it”(Blois 1984:97). Consequently, this segment of a case report serves the purpose ofimparting information about the diseases that the patient underwent withoutconcentrating on their character or course. Also, what can be observed here is agradual shift in focus from the patient to disease issues, which announces a moredisease-oriented text that is to follow.Examination/Tests addresses the assessment of the patient’s condition, whichusually takes place at two levels. The first part of the diagnostic procedure deliversthe external evidence based on the doctor’s observation of the patient’s body and itsreactions that are interpreted accordingly.6. She was afebrile and growth was on the 50th centile (Carroll et al. 2005).7. On examination, she had a large, firm, tender mass in the left lower abdomenwhich she said she had first noticed a year and a half previously (James 2005).8. On examination, we found large venous ulcers on both legs, and bilateral ankleoedema (figure) (Sheridan et al. 2004).These observations are restricted to the sensorially perceivable phenomena,hence the level of description reaches only the whole body, its parts and systems.In (6) and (7), the central sentential positions are held by the patients. In (8), it is thecondition that is described in abstraction from the patient. Referring to patients asbeing in a particular condition (cf. 6) contributes to viewing the state as experiencedby him/her. “Having” diseases, on the other hand, presents them as “objects”,separate from the patient’s experience (cf. 7 above; Fleischman 1999; Staiano1986). Furthermore, due to the observable character of the phenomena describedhere, they are manifested linguistically by the adjectives pertaining to the senses ofe.g. sight and feeling, i.e. tender in (7) and bilateral in (8) respectively.

196 M. MurawskaPhysical examination is usually followed by a series of medical tests. Theirmodes of presentation are as in the following:9. Her 24-h urinary freecortisol was high at 31,000 nmol/24 h (normal 270), 900 hplasma adrenocorticotropic hormone (ACTH) was high at 204.5 ng/L (normal50 ng/L) (Keenan et al. 2006).10. Cystoscopy showed an inflamed bladder that bled on distension, and we sentbiopsy samples for analysis (Lo et al. 2004).This subsection is dominated by the quantitative and qualitative results ofvarious medical tests. The dominance of such material reflects the widespreadapplication of modern diagnostic equipment which allows doctors to measureand/or observe every function or element of the human body respectively (Ashcroft2000). Such a type of material entails the objective account of a disease, i.e.“technology as the agent” (Anspach 1988), as exemplified in (10). It is a modewhere diagnostic equipment shows particular results. The fact that the agent isdeleted from the text contributes to rendering the information objective and independentfrom human involvement (“data primacy”, cf. Potter 1996: 153).Furthermore, as this part contains information revealed by laboratory tests,medical description goes to the bottom levels of the hierarchical scale, consideringbiological processes, usually cellular or molecular in nature. Changes at these levelsare not directly perceivable, hence the mode of presenting the results of tests as in(9). (10) exemplifies the focus on body-parts/organs. Here the history of medicinemight be helpful in explaining this linguistic phenomenon. According to Virchow(1880), whose work in autopsy and pathological anatomy underlies the biomedicalmodel, all diseases stem from the dysfunction of tissues. Following this medicalpremise, organs and tissues claimed centrality in medical case writing (Nowell-Smith 1995:52) as they began to be perceived as the location of diseases. In (10),the isolated organ represents the patient only metonymically.Treatment section is devoted to medical procedures performed on patients with aview to restoring their health.11. She was treated with intravenous lorazepam for presumed alcohol withdrawal,receiving a total of 432 mg over 10 h (Tuohy et al. 2003).12. Initial management was conservative but biochemical hyperthyroidism,hepatosplenomegaly, and irritability persisted (Carroll et al. 2005).13. Painful muscular spasms continued for weeks after extubation and were controlledby supranormal magnesium levels for further 9 days and subsequentlybaclofen (Sheridan et al. 2004).Generally, this part considers treatment that patients undergo, so there is noquestion of their agency. Yet, although in (11) the patient is not an active participantbut the one to whom medical procedures apply, she is textually prominent thanks tothe subject position of the personal pronoun. (11) and (12) can be contrasted, withthe latter not mentioning the patient to whom these procedures apply. In (12), ratherthan to patients themselves, the readers’ attention is drawn to the treatmentperformed and its specific execution, the so called “medical techniques and

14 From a Complaint through Therapy to Recovery 197therapeutics” (Ashcroft 2000: 288). (13) presents the patient’s experience of aparticular symptom as “painful”, yet without mentioning the experiencer. Consequently,it may be argued that the aim of this section is to give technical details of achosen therapy and patients are those who undergo it. Nevertheless, the descriptionstend to omit the very subject of medical procedures and relegate it to the object towhich they apply.14.4.3 Comment/DiscussionComment/Discussion discusses the examined case of a disease with reference to thealready available research and draws more general conclusions.14. Her lack of cardiovascular instability may reflect the chronicity of thehypovolaemia and activation of the renin angiotensin system (Tuohy et al. 2003).15. Although vasculitis remains a possible cause of stroke in our patient, themultiple territorial ischaemic lesions in the setting of atrial fibrillation andcardiomyopathy make embolism more likely (Libman et al. 2005).16. Many of the reported cases are children and only two cases have survived(White et al. 2004).In this section, the authors refer to a particular aspect of the patient’s condition/treatment rather than to the whole person (cf. 14). In (15), the patient referent doesnot occur either in the subject or object position but in a prepositional phrase withthe meaning of location, which images him/her as a container (cf. Lakoff andJohnson 1980). In (16), the word case does not refer to an occurrence of a particulardisease but to the patient. As a result, patient imaging in the final section of a casereport refers not to the patient as the “whole self” (Wade and Halligan 2004: 1400)but to a single aspect of his/her condition/treatment under examination.14.5 ConclusionIf all three parts are considered, the patient’s textual prominence decreases fromIntroduction (his/her account of an illness), through Examination/Tests and Treatment(numerical or graphic values and methods of a chosen therapy taking priority)to Comment/Discussion where the information about a disease must be abstractedfrom a particular patient. In other words, while the beginning describes anindividual’s experience of illness, the rest of the report is constructed as a scientificallyobjective report about a disease. In this case, when the account of treatment isgiven, the patient’s mental/bodily reactions as well as body-parts and biologicalprocesses are rendered separate from him/her.Nowell-Smith (1995) concludes that “[m]edical cases are stories not of painand fear but of medical intervention, and the leading characters are doctors,

198 M. Murawskainstruments, and organs” (1995:64). Linguistically, it is achieved by choosingmedical procedures and treated body-parts/organs for the subject position in asentence or the sentential absence of the patient. Consequently, although the goalof a particular section has bearing on patient imaging, in many cases, the option ofthe depersonalised mode is chosen irrespective of the type of information imparted.ReferencesAarts, B. 1997. English syntax and argumentation. Houndmills: Macmillan Press.Adams Smith, D. E. 1984. Medical discourse: Aspects of author’s comment. English for SpecificPurposes 3: 25–36.Anspach, R. R. 1988. Notes on the sociology of medical discourse: The language of casepresentation. Journal of Health and Social Behaviour 29: 357–375.Ashcroft, R. E. 2000. Teaching for patient-centred ethics. Medicine, Health Care and Philosophy3: 287–295.Atkinson, D. 2001. Scientific discourse across history: A combined multi-dimensional/rhetoricalanalysis of the Philosophical Transactions of the Royal Society of London. In: Variation inEnglish: Multi-dimensional studies, eds. S. Conrad and D. Biber, 45–46. Harlow: Longman.Atkinson, P. 1995. Medical talk and medical work. London: Sage Publications.Bazerman, Ch. 1988. Shaping written knowledge. The genre and activity of the experimentalarticle in science. Madison: The University of Wisconsin Press.Blois, M. S. 1984. Information and medicine. Berkeley: University of California Press.Bond, J. and S. Bond. 1986. Sociology and health care. Edinburgh: Churchill Livingstone.Brennan, S. E. 1995. Centring attention in discourse. Language & Cognitive Processes 10:137–167.Carroll, D. N., P. Kamath and L. Stewart. 2005. Congenital viral infection? The Lancet 365: 1110.Chafe, W. 1976. Giveness, contrastiveness, definiteness, subject, topic and point of view. In:Subject and topic, ed. C. N. Lee, 25–55. New York: Academic Press.Childs, N. L. and W. N. Mercer. 1996. Brief report: Late improvement in consciousness after posttraumaticvegetative state. The New England Journal of Medicine 334: 24–25.Firbas, J. 1974. Some aspects of the Czechoslovak approach to problems of functional sentenceperspective. In: Papers on functional sentence perspective, ed. F. Danes, 11–37. Prague:Academia.Fleischman, S. 1999. Iam..., I have..., I suffer from...: A linguist reflects on the language ofillness and disease. Journal of Medical Humanities 20: 3–32.Givon, T. 1984. Syntax. A functional-typological introduction. Vol. 1. Amsterdam: JohnBenjamins Publishing Company.Givon, T. 1990. Syntax. A functional-typological introduction. Vol. 2. Amsterdam: JohnBenjamins Publishing Company.Gotti, M. and F. Salager-Meyer eds. 2006. Advances in medical discourse analysis – oral andwritten contexts. Bern: Peter Lang.Greenbaum, S. and R. Quirk. 1990. A student’s grammar of the English language. Harlow:Longman.Grice, F. and A. Kramer-Dahl. 1992. Grammaticalising the medical case history. In: Language,text and context: Essays in stylistics, ed. M. Toolan, 56–90. London: Routledge.Halliday, M.A.K. 1994. An introduction to functional grammar. London: Edward Arnold.Hyland, K. 2001. Humble servants of the discipline? Self-mention in research articles. English forSpecific Purposes 20: 207–226.James, R. F. 2005. Rectus sheath haematoma. The Lancet 365: 1824.

14 From a Complaint through Therapy to Recovery 199Keenan, N., W. S. Dhillo, G. R. Williams and J. F. Todd. 2006. Unexpected shortness of breath in apatient with Cushing’s syndrome. The Lancet 367: 446.Kenny, N. P. and B. L. Beagan. 2004. The patient as text – a challenge for problem-based learning.Medical Education 38: 1071–1079.KIAP – Cultural Identity in Academic Prose: National vs. discipline-specific. 2006. http://www.uib.no/kiap/index-e.htm. Accessed 1 July 2008.Knowles, M. and R. Moon. 2006. Introducing metaphor. London: Routledge.Kong, M.-F., R. Jogia, S. Jackson, M. Quinn, P. McNally and M. Davies. 2005. Malignantmelanoma presenting as a foot ulcer. The Lancet 366: 1750.Lakoff, G. and M. Johnson. 1980. Metaphors we live by. Chicago: The University of ChicagoPress.Libman, R. B., B. L. Menna and S. Gulati. 2005. Consequences of ephedra use in an athlete.The Lancet 366: 522.Lindley-Jones, M., D. Lewis and J. L. Southgate. 2004. Recurrent tetanus. The Lancet 363: 2048.Lo, S., J. Noble, I. Bowler and B. Angus. 2004. Dysuria and a headache. The Lancet 364: 1554.Myers, G. 1990. Writing biology: The social construction of popular science. Madison, WI:University of Wisconsin Press.Nijhof, G. 1998. Naming as naturalisation in medical discourse. Journal of Pragmatics 30:735–753.Nowell-Smith, H. 1995. Nineteenth-century narrative case histories: An inquiry into stylistics andhistory. CBMH/BCHM 12: 47–67.Paivio, A. 1979. Imagery and verbal processes. Hillsdale: Lawrence Erlbaum.Potter, J. 1996. Representing reality: Discourse, rhetoric and social construction. London: Sage.Robertson, D., E. M. Garland, S. R. Raj and N. Demartinis. 2005. Marathon runner with severeautonomic failure. The Lancet 366: 513.Rowley-Jolivet, E. 2007. A genre study of If in medical discourse. In: Language and disciplineperspectives on academic discourse, ed. K. Fløttum, 176–201. Newcastle: Cambridge ScholarsPublishing.Salager-Meyer, F., G. Defives, C. Jensen, and M. de Filipis. 1989. Communicative function andgrammatical variations in medical English scholarly papers: A genre analysis study. In: Speciallanguage: From humans thinking to thinking machines, eds. C. Laurén and M. Nordman,151-160. Philadelphia: Multilingual Matters.Sheridan, E. A., J. Cepeda, R. De Palma, M. M. Brett and K. Nagendran. 2004. A drug user with asore throat. The Lancet 364: 1286.Smith, C. S. 2003. Modes of discourse. The local structure of texts. Cambridge: CambridgeUniversity Press.Staiano, K. V. 1986. Interpreting signs of illness. A case study in medical semiotics. Berlin & NewYork: Mouton de Gruyter.Taavitsainen, I. and P. Pahta. 2000. Conventions of professional writing: The medical case reportin a historical perspective. Journal of English Linguistics 28: 60–76.Tuohy, K. A., W. J. Nicholson and F. Schiffman. 2003. Agitation by sedation. The Lancet 361:308.Van Dijk, T. A. 1980. Macrostructures. An interdisciplinary study of global structures indiscourse, interaction, and cognition. Hillsdale: Lawrence Erlbaum Associates, Publishers.Van Rijn-Van Tongeren, G. W. 1997. Metaphors in medical texts. Amsterdam: Rodopi.Virchow, R. 1880. Post-mortem examinations: With especial reference to medico-legal practice.(Translated by T. Smith.) Philadelphia: Presley Blackiston.Wade, D. T. and P. W. Halligan. 2004. Do biomedical models of illness make for good healthcaresystems? British Medical Journal 329: 1398–1401.White, J. M., R. D. Barker, J. R. Salisbury, A. J. Fife, S. B. Lucas, D. C. Warhurst andE. M. Higgins. 2004. Granulomatous amoebic encephalitis. The Lancet 364: 220.

Chapter 15Beyond and within Standard English:Categories, Category Boundaries and FuzzinessMaciej RatajAbstract Standard dialects are frequently thought of as fixed and stable entitieswhich possess clear boundaries and a set of criteria that allow us to recognise easilywhich texts or utterances are standard and which are not. This view, as has beenproved by sociolinguists, is an oversimplification and idealisation since real-worldstandard dialects are best viewed both as fuzzy categories and as members of a largecategory centred around a prototype. The paper attempts to present a view ofstandard dialects inspired by the prototype theory and present-day sociolinguisticsand exemplify this view by means of Standard British English. The analysispresents the notion of category in Standard English at three levels. Firstly, particularfragments of texts and utterances cannot be shown to be doubtlessly standard ornon-standard, this issue being further complicated by the frequent confusion oftypically spoken and typically written norms. Secondly, Standard English cannot betreated as a fixed entity because of its optional variability, its diversity in terms ofstyle and register and the considerable number of local standard varieties of Englishthroughout the English-speaking world. Finally, considering Standard English aspossessing a long history of continued development towards minimal variation inform and maximal variation in function is also an idealisation of what has actuallybeen a more complex process, in particular because, as linguists agree, languagestandardisation is not a fact but a process. All this leads to the conclusionthat standard dialects can be analysed in terms of cognitive categorisation andprototypicality.M. Rataj (*)University of Gdańsk, Gdańsk, Polande-mail: rataj.maciej109@gmail.comM. Pawlak and J. Bielak (eds.), New Perspectives in Language,Discourse and Translation Studies, Second Language Learning and Teaching,DOI 10.1007/978-3-642-20083-0_15, # Springer-Verlag Berlin Heidelberg 2011201

202 M. Rataj15.1 IntroductionStandard dialects 1 are frequently conceived of as fixed and stable entities withnegligible variation in form and a finite set of criteria which clearly distinguish themfrom non-standard dialects and facilitate recognising whether or not a given sample ofspeech or writing represents a given standard dialect. This oversimplified view wasprevalent before the dawn of modern linguistics, i.e. before the acrolects (prestigiousvarieties) of some languages became known as their standard dialects, and is still heldby some linguists, in particular those who are in favour of prescriptivism. One instanceof standard-nonstandard dichotomy is Stewart’s (1968) typology of languagevarieties, in which he uses a set of binary features, namely “standardisation”, “autonomy”,“historicity” and “vitality” (Stewart 1968: 534–536), to classify several types ofwhat he calls “linguistic systems”. In his set (see Stewart 1968: 537) the standardpossesses all the four attributes, and even though elsewhere in the text Stewartmentions multimodal standardisation in such languages as English, includingendonormative and exonormative standardisation (i.e. standardisation focusing onlocal and foreign norms respectively), his standard linguistic system category failsto account for the complexity and diversity of the real-world standard dialects and theprocesses of standardisation they undergo. The most noticeable problem here, as maybe argued, involves regarding standardisation not as a process which is never completebut as an attribute that is either present or absent.In 1962 Ferguson, as quoted by Haugen ([1966] 1974: 107), made an attempt atconsidering language standardisation as a gradable phenomenon, labellinglanguages with multimodal standardisation as St. 1 and giving St. 2 to languageswith a “single, widely accepted norm which is felt to be appropriate with only minormodifications or variations for all purposes for which the language is used”.Together with the “utilisation in writing” scale ranging from W 0 to W 3, the St.degrees describe standardised languages somewhat more accurately. Nonetheless,it is felt that a different approach is needed to reflect the complexity of standarddialects.Other questionable opinions include the belief that standard dialects are formedby single people or organisations (see Deumert and Vandenbussche 2003: 455) orthat standard dialects must not be allowed to change in terms of grammar,spelling, word meaning or pronunciation, as all language change equals decline(a stance discussed by Aitchison 2001: 4–7). Needless to say, every standarddialect is embedded in the community of its users and is therefore associated witha wealth of extralinguistic factors, an issue that will not be discussed in detail inthis paper.1 The term “standard dialect” is used here instead of “standard language” for two reasons: firstly, ina natural language a standard dialect is but a language variety and not a complete language and it isonly in the case of artificial languages that the term “standard language” is fully justified; secondly,we must not mistake standard dialects for standardised languages and the term “standard language”may lead to such confusion. For a discussion of the issue see Hudson (1980: 31–32, 34),Crystal (1994: 109–114) and Trudgill (1999: 118).

15 Beyond and within Standard English 203To proceed, if the standard dialect is considered as a category, the abovementionedclaims put forward by Stewart (1968), Ferguson (1962) and othersmake it a category consistent with the logical or classical view of categorisation,as discussed by Ungerer and Schmid (1996) amongst others. Firstly, Stewart’s“standardisation”, “autonomy”, “historicity” and “vitality” or Haugen’s (1974:97–111) “selection”, “codification”, “elaboration of function”, and “acceptance”serve as necessary and/or sufficient conditions which allow one to draw theboundaries of the standard dialect category and classify particular dialects asstandard or non-standard, all of them being equal members of the category (seeFig. 15.1).Secondly, various levels of linguistic structure (morphology, phonology, syntax,etc.) contain rules and units which may be regarded as the necessary or sufficientattributes of particular standard dialects, making it possible to decide whether or nota text or utterance represents a given standard variety (see Table 15.1).Finally, the logical view enables the linguist to imagine language standardisationas a continued development towards “minimal variation in form” and “maximalvariation in function”, as used by Haugen (1974: 107) (see Fig. 15.2).STANDARDDIALECTStandard English,Standard French,Standard Russian,etc.VERNA- CLASSICALCULARLANGUAGEAfrican AmericanVernacular,Greek, etc.Latin, Classicaletc.Fig. 15.1 The standard dialect category and some other linguistic systems according to the logicalviewTable 15.1 The standardness of language samples according to the logical viewStandardness(Standard BritishFeaturesEnglish)TextHe goes to school every day Conforms to the rules of Standard Englishgrammar, punctuation, vocabulary, etc. +He going to shool – lah. Contains non-standard features of grammar,vocabulary, spelling, etc.Utterance/hi gəʊz tə sku:1 ‘evri deɪ/ Conforms to the rules of RP, StandardEnglish grammar, vocabulary, etc. +/xi ‘gɔwiŋk tʊʃʊ:1 1ɑ:/ Contains non-standard features of grammar,vocabulary, pronunciation, etc.

204 M. Rataj3025formal variationformal variation201510501 2 3 4 5timeFig. 15.2 Variation in form in the standardisation process according to the logical viewThe problem with the logical view, when applied to the issue under discussion, isthat it misrepresents the nature of standard dialects, for they are not categories withhomogeneous structure or clear boundaries; nor is any one standard dialect a regularset of units, e.g. sentences, which are its equally good members. Finally, livinglanguages do not easily submit to rigid codification and they invariably retain someoptional variability, the extent of which may be difficult to assess or describe.Hence it may be argued that a different approach is needed, one based oncategorisation as viewed by cognitive linguistics, with categories which haveprototypes, central and marginal members and fuzzy boundaries. Standard English,in particular Standard British English, will be used throughout the analysis to serveas an instance of such a category.15.2 The Prototype Structure of the Standard Dialect CategoryAs has been mentioned, despite the generally traditional approach they adopt,Stewart (1968) and Ferguson (1962) realise that not all standard dialects and notall standardised languages are the same, as the former differ in terms of the degreeof codification, elaboration of function, historicity and numerous other factors (seealso Hudson 1980: 34). Furthermore, many of these attributes cannot be expressedas scalar quantities. For instance, while it is possible to ask a cross-section of agiven speech community whether or not they consider the standard dialect of theirmother tongue prestigious and thus obtain a certain percentage value, doing so willfail to take into account the complexity of prestige in language, in particular itshistorical, sociolinguistic and psycholinguistic background. Needless to say, prestigeis just one of many factors concerning standard dialects which are difficultto put in numbers. If, however, we imagine a prototypical standard dialect that

15 Beyond and within Standard English 205possesses a number of attributes which linguists and laypeople alike may expect itto have and put it in the centre of a category with a heterogeneous structure, we willobtain a more truthful picture of the standard dialects of natural languages (seeFig. 15.3).Fig. 15.3 The standarddialect category according tothe cognitive viewSTANDARDDIALECTprototypeIf we now attempt to place Standard English somewhere in this category, we willencounter two major difficulties. First of all, English possesses not one standarddialect but several of them; secondly, even the number of standard Englishes isdisputable (e.g. Microsoft Office 2007 lists 19 national varieties of English forspelling and grammar correction purposes). One viable alternative is to decide thatthere exists one standard English in each English-speaking country, and reduce ourexemplifying discussion to Standard British English (SBE), rejecting the theory ofmultiple standards within British English.To proceed to Standard British English, we may now decide how close to theprototype it lies. The present discussion does not allow a detailed analysis; however,one may easily observe that although SBE is without doubt a standard dialectwhich has been thoroughly described and prescribed for generations, it differs fromthe image of a typical standard dialect in several respects. First, it is not the onlystandard dialect of the language in question; consider the notion of Englishes, asused in a number of papers: for instance for Kachru and Smith (2009: 3) worldEnglishes are the result of “the acculturation and nativisation of English in variousparts of the world”. Second, while it has been codified to a considerable extent, it isnot codified so rigidly as Standard Metropolitan French (see Hudson 1980: 34) andlacks a standardising body such as the authoritative French Academy or theadvisory Council for the Polish Language (attempts to form such an academymade in the eighteenth century failed). Besides, as will be mentioned below,Standard British English allows a certain measure of optional variability. Fourth,some linguists have questioned the status of Received Pronunciation (RP) as aninherent part of SBE (e.g. Trudgill 1999: 118–119, 123). As regards SBE and RP inparticular, they are used by a minority of the British, and RP, with perhaps fewerthan two per cent of the British population using it in the early twenty-first century(see Crystal 2006: 184), is regarded as a sign of snobbery or conceit by certainnative speakers of British English (see Crystal 2006: 183–184). All this and manyother factors not mentioned in the present discussion justify the placement of SBEat a certain distance from the centrally located prototype (see Fig. 15.4), althoughnot very close to the category boundary.

206 M. RatajFig. 15.4 Standard BritishEnglish as an element of thestandard dialect categoryaccording to the cognitiveviewSTANDARDDIALECTSBEprototypeIn cognitive linguistics, categories possess fuzzy boundaries, often partly mergewith neighbouring categories and may even form continua. This is the case withregional dialects (consider the Dutch-German continuum between the Netherlandsand Austria) and the standard dialect category as well, for non-linguists may havedifficulties deciding if a given language variety is standard and even linguists maytreat Classical Latin, Esperanto or the written Chancery Standard of WilliamCaxton (see Burchfield 1986: 22, 147; Crystal 2006: 15–17) as peripheral examplesof standard dialects which at the same time also belong to other categories.One may conclude this part by deciding which standard dialect is more prototypicalthan SBE. Apparently, Standard Polish might be suggested as a case inpoint, as it is the only standardised dialect of Polish, it is, generally speaking,prestigious among Poles and it has only one standard accent associated with it.Owing to the demographic changes following World War II, numerous Poles, inparticular educated city dwellers, are native speakers of either Standard Polish or alevelled dialect which is not considerably divergent from the standard. Furthermore,the codification of Standard Polish has been supervised by the Council forthe Polish Language, and earlier by other more or less authoritative panels ofexperts. Last but not least, the teaching of Standard Polish has never aroused asmuch controversy as the great debate about the National Curriculum forEngland and Wales of the 1990s, discussed at length by Cameron (1995) and byCarter (1994).15.3 The Prototype Structure of Particular Standard Dialectsas CategoriesIn the previous section particular standard dialects are considered to be elementsof a large standard dialect category centred around a prototype. These elements,however, may themselves be regarded as categories. In such categories units, e.g.words, and features, e.g. syntactic, phonological and morphological rules, functionas attributes. Consequently, the texts and utterances which consist of standardunits and comply with standard rules are prototypical and those which clearly donot are not part of the category. Certainly, there also exists a wide spectrum ofmore or less prototypical instances of standard dialect texts and utterances,

15 Beyond and within Standard English 207including borderline cases. The same holds true for smaller units such as spoken orwritten sentences.We may now ask how to decide whether a given text or utterance is indeed partof a given standard dialect. Unfortunately, three problems that complicate the issuepresent themselves here. To begin with, it is difficult to establish what percentage ofnon-standard features or errors related to one’s language competence or performanceis permissible in a text or utterance (in other words, at which point thestandardness test is failed) and which features are more salient, i.e. more nonstandard,than others. In the case of SBE, is the use of renumeration instead ofrenumeration a less serious problem than multiple negation or a dangling participle?And if so, who is to decide? Next, even in the languages which have been welldescribed and analysed the standardness of certain norms and units remains uncertain.As regards Standard English, while the word ain’t and the form I seen insteadof I saw or I have seen are clearly non-standard, using a preposition at the end of arelative clause and splitting infinitives have long been a bone of contention, beingstigmatised by some and regarded as standard, though perhaps informal, by others.All this is connected with a variety of linguistic and extralinguistic phenomena,including language change, which naturally affects standard dialects, but alsoprescriptivism and the complaint tradition present in English, in Britain and elsewhere,as well as in other languages (as discussed by Milroy and Milroy 1998:24–46; McWhorter 1998: 7–8). Finally, numerous linguists postulate that in everystandard dialect we are dealing not with one but with two sets of norms. In Polishlinguistics, for example, many researchers differentiate between the model norm(norma wzorcowa), which is associated with diligent and conscious use of languagein official contexts, and the common or colloquial norm (norma potoczna), whichsuits informal contexts (see Podlawska and Świątek-Brzezińska 2008: 4–5). InAnglophone linguistics and the description of English many experts now tend torecognise a set of norms typical of spoken usage, which permit more formalvariation than the written language, with Carter going so far as to claim that e.g.standalone relative clauses and double main verbs are features of spoken StandardEnglish (see Carter 1994: 11–12).Having briefly considered the difficulties pertinent to the classification of languagesamples as standard or non-standard, we may claim that the only viablesolution is to ask native speakers of a particular language or perhaps only nativespeakers of a given standard dialect to judge the standardness of particular unitssuch as sentences, features such as distinct phonemes or norms such as word orderrules. Fig. 15.5 depicts standardness in SBE from the viewpoint of categorisationand is based on fragments of an unpublished small-scale survey of languageprototypicality judgements conducted by the present author in 2006 as part of hisMaster’s thesis research among about twenty native speakers of British English.Naturally, it makes no claims to exhaustiveness but instead it serves to exemplifyhow a more complete study of the issue could be conducted. From the figure it maybe inferred that the respondents are likely to regard language standardness as agradable phenomenon rather than a dichotomous feature and that it is in borderlinecases that the greatest differences in opinion among them tend to occur. Although

208 M. Ratajthe survey cannot be said to represent the language attitudes of a cross-section ofBritish society, conducted as it was mostly among young members of forums onlanguage and English usage, its results show certain patterns which may beanalysed in greater detail and lead to general conclusions concerning StandardEnglish if a large-scale survey of a similar kind is undertaken in the future.S B E(1)(3)(2)(4)prototypeAverage score from 0 (typical ofSentencesSBE) to 5 (non-standard)1. John does not work here. 02. Jane has recently moved to a quaint three-rooms cottage. 1.73. How R U 2day? 44. Hi did not si nothink spezial thear. 4.8Average score from 0 (typical ofPhonetic featuresSBE) to 5 (non-standard)1. Distinguishing between /ɑɪ/ and /eɪ/ 0.12. The long vowel /ɔ:/ instead of /ʊə/ 1.93. Stressing the final syllable in every polysyllabic word 2 3.7Fig. 15.5 Standard British English as a category according to the cognitive view15.4 The Standardisation Process form the Viewpointof Cognitive LinguisticsAs was stated in the introduction, natural languages do not easily yield to rigidcodification. Thus even in standard dialects there is invariably a margin of optionalvariability which must not be underestimated. If we also take account of theaforementioned norms and units whose standardness is uncertain, we obtain animage of a category with fuzzy boundaries not only in synchronic but also indiachronic depiction of the standardisation process.2 None of the phonetic features included in the survey was considered to be entirely non-standardby the respondents.

15 Beyond and within Standard English 209“Minimal variation in form” and “maximal variation in function” (Haugen 1974:107) may be the ultimate objectives of linguistic standardisation but they arepractically unattainable and perhaps also not altogether desirable. They areunachievable because every natural language, including its standard dialect ordialects, is invariably subject to the human factor, e.g. idiolects and performanceerrors by an individual despite his or her fluency in the standard, and multiplechanges in progress taking place at any moment of its history. It has been shownthat attempts at reducing variation to a minimum or stopping language change aredoomed to failure, as is the case with the efforts on the part of the French Academy(L’Académie Française) to perfect Standard Metropolitan French, or as Leith(1997: 50) puts it, “to fix the unfixable”. The supposedly ideal state of codificationand functional elaboration is not a desirable state of affairs either, for if we imaginea standard dialect which is fixed, arrested in its formal development and possessingits unique means to express every possible concept related to human experience, itis bound to be impoverished in terms of style and register and practically dead, as nolanguage users will be able to meet its strict formal requirements sufficiently well tobe considered its native or habitual speakers. Instead, what we need to take intoconsideration is the standard dialect in a state of flux with fuzzy boundaries in itsdevelopment. These fuzzy boundaries relate both to codification, as optionalvariability never disappears, and to elaboration of function, as arguably eachstandard dialect occasionally resorts to calque and borrowing to supplement itslexical resources. Figs. 15.6 and 15.7 are an attempt at depicting the standardisationprocess with the aforementioned issues taken into account, with lighter colours nearthe edges indicating optional variability in Fig. 15.6 and borrowed or copied meansof expressing certain concepts in Fig. 15.7. The fact that any clear boundarywhatsoever is shown facilitates the viewing of the graph instead of indicating thatsuch a boundary is objectively present. It also needs to be explained that the3025formal variationformal variation201510501 2 3time4 5Fig. 15.6 Variation in form in the standardisation process according to the cognitive view

210 M. Rataj3025functional variationfunctional variation201510501 23 4 5timeFig. 15.7 Variation in function in the standardisation process according to the cognitive viewnumbers on the vertical scale are abstract, as standardness is immeasurable, as wasstated in the introduction.15.5 ConclusionsTo summarise, the cognitive view of categorization applied to classifying standarddialects appears to be a good model on the whole. Several arguments in its favour maybe enumerated. Firstly, to use an expression from the business jargon, it allows us tothink outside the box, i.e. to consider the standard dialect category and particularstandard dialects not as separate entities characterised by dichotomous categorymembership and homogeneous structure but as categories with prototypes and membershipas a gradable feature. The latter entails fuzzy category boundaries andsometimes even the partial merging of categories, e.g. Standard British English andStandard American English in the case of what is known as Mid-Atlantic English. 3Secondly, teaching this view to students of languages, in particular trainee teachersand translators of foreign languages, allows them to see the complexity involved instandard dialects and helps them to avoid prejudice involving e.g. categorisinglanguage and its users as good and cultured as opposed to bad and uncultured. Thirdly,realising how standardisation and language change work may prevent people to whomCrystal (2006) and McWhorter (1998) refer as “language pundits” from complainingabout the supposed decline of standard dialects and the accompanying moral decayof their users. These of course are somewhat indirect consequences of applying3 Although some people consider Mid-Atlantic English to be an accent, it may be claimed to be afull dialect which is levelled on the levels of vocabulary and grammar in addition to pronunciation.This in turn is related to the concepts of World Standard English or International English, whichare not analysed in the present discussion.

15 Beyond and within Standard English 211the cognitive model to the issue in question and contemporary sociolinguists, e.g.Deborah Cameron, David Crystal, Lesley and James Milroy, and Peter Trudgill havedealt without it; however, we may argue that the cognitive view presents some ideas ofpresent-day sociolinguistics sufficiently well to be applied for the purposes of illustrationof these theories. This usefulness is even more visible where sociolinguistsoccasionally refer to concepts related to cognitive linguistics, e.g. McWhorter(1998: 52–54) talks about “fuzzy logic” when referring to dialect continua. Since, asdiscussed above, standard dialects may also be considered as continua, McWhorter’sterm fits in well in the present discussion.We should nevertheless bear in mind that the cognitive view of standard dialectsis not devoid of problems, as it does not allow us to use a rigid taxonomy oflinguistic systems or clearly decide what constitutes an error or an unacceptableform. To put it differently, at times prescriptivism is indispensable, e.g. in foreignlanguage teaching, as instead of discussing the suitability or standardness of acertain form of structure it is far more suitable to provide clear signals as tolanguage correctness or what is known as good usage. In a similar vein, in mothertongue instruction to young learners, providing clear rules is also less confusing thatexplaining optional variability and differences between regional and social dialectsor styles and registers to uncomprehending pupils. To conclude, in certain contextswe may use the logical model without hesitation; we should not forget, however,that other ways of looking at languages are just as important.ReferencesAitchison, J. 2001. Language change: Progress or decay? (third edition). Cambridge: CambridgeUniversity Press.Burchfield, R. 1986. The English language. Oxford: Oxford University Press.Cameron, D. 1995. Verbal hygiene. London: Routledge.Carter, R. 1994. Standard Englishes in teaching and learning. In Who owns English?, eds.M. Hayhoe and S. Parker, 10–23. Buckingham: Open University Press.Crystal, D. 1994. Which English – or English Which? InWho owns English?, eds. M. Hayhoeand S. Parker, 109–114. Buckingham: Open University Press.Crystal, D. 2006. The fight for English: How language pundits ate, shot and left. Oxford: OxfordUniversity Press.Deumert, A. and V. Vandenbussche. 2003. Research directions in the study of languagestandardization. In Germanic standardizations: Past to present, eds. A. Deumert andV. Vandenbussche, 455–470. Amsterdam: John Benjamins.Ferguson, C. 1962. The language factor in national development. Anthropological Linguistics4: 23–27.Haugen, E. [1966] 1974. Dialect, language, nation. In Sociolinguistics: Selected readings, eds.J. B. Pride and J. Holmes. 97–111. Harmondsworth: Penguin.Hudson, R. 1980. Sociolinguistics. Cambridge: Cambridge University Press.Kachru, Y. and L. E. Smith. 2009. The Karmic cycle of world Englishes: Some futuristicconstructs. World Englishes 28: 1–14.McWhorter, J. 1998. Word on the street: Debunking the myth of “pure” standard English.New York: Basic Books.

212 M. RatajMilroy, J. and L. Milroy. 1998. Authority in language: Investigating language prescription andstandardization. (third edition). London: Routledge and Kegan Paul.Podlawska, D. and M. Świątek-Brzezińska. 2008. Słownik poprawnej polszczyzny. Warsaw: PWN.Stewart, W. 1968. A sociolinguistic typology for describing national multilingualism. In Readingsin the sociology of language, ed. J. Fishman, 531–545. The Hague: Mouton.Trudgill, P. 1999. Standard English: What it isn’t. In Standard English: The widening debate,ed. T. Bex and R. J. Watts, 117–128. London: Routledge.Ungerer, F. and H.-J. Schmid. 1996. An introduction to cognitive linguistics. London: Longman.

Part VTranslation

Chapter 16Construction Grammar as a Frameworkfor Describing Translation: A ProlegomenonIzabela SzymańskaAbstract This paper explores Construction Grammar as a potential framework todescribe translation, arguing that it is a model of language capable of addressing inan integrated way a wide range of problems recognized by translation studies,without drawing a sharp distinction between the linguistic and the cultural issues.Construction Grammar is a non-modular, non-derivational and inherently functionalmodel of language. The knowledge of the language user is represented as a network ofconstructions. A construction is understood as a language-specific pairing of form andmeaning/function, a cluster of properties cutting across the traditional modules oflanguage description, which may include phonetic, morphological, syntactic, semantic,pragmatic and discourse properties. Constructions thus understood include morphemes,lexical items, phrasal patterns, sentence patterns, idioms, intonation patterns, discoursepatterns, etc. Construction Grammar declares the aim of accounting within an integratedframework for all the conventions that the speaker of a language uses in communication.It is argued that the constructional view of language and the scope of theconstructional model are highly convergent with the needs of translation studiesand can offer a range of tools to address various aspects of the translation process.This article focuses on the possibility of describing the lack of exact equivalencebetween the constructions of the source and the target language, and the possibilityof identifying very precisely the so-called gains and losses and through them thetranslator’s priorities in decision-making.16.1 IntroductionThis paper is intended as a general outline of some premises of my research(Szymańska 2011) which explores the applicability of the linguistic frameworkknown as Construction Grammar (henceforth CxG) to translation studies. The CxGI. Szymańska (*)University of Warsaw, Warsaw, Polande-mail: i.szymanska@uw.edu.pllM. Pawlak and J. Bielak (eds.), New Perspectives in Language,Discourse and Translation Studies, Second Language Learning and Teaching,DOI 10.1007/978-3-642-20083-0_16, # Springer-Verlag Berlin Heidelberg 2011215

216 I. Szymańskaframework has not yet been explored from this perspective, so the work mentionedattempts to “stake out a new territory” for translation studies. Since the territory inquestion is obviously vast, this paper can only address a few selected issues. Its aimis to show some features of CxG which make it a model of language capable ofaddressing in an integrated way a wide range of problems recognized by translationstudies.By way of introduction, some comments need to be made about translationstudies. The development of this discipline in recent decades has been marked byseveral “turns”, as it is put by Snell-Hornby (2006), signalling significant changesof focus. One important turn is the shift of interest from lexico-syntactic problemsin translation to the pragmatic and textual aspects, which has led to exploring theimportance of discourse, register, text-types and usage-related factors in the shapingof the translated text. Another vital development is the so-called cultural turn,whose proponents focus on the extra-linguistic determinants of the process oftranslation, such as the purpose assigned to the translated text by its commissioner,the aesthetic and ideological requirements of the target culture, power relations, thestatus of particular texts and genres in the target culture, the accepted norms oftranslation practice, the profile and the cultural background of the targetedaddressee, and many others which may be assumed to influence the decisionstaken by the translator and thus shape the target text. The chief postulate of thecultural turn is to abandon “painstaking comparisons between originals andtranslations” and move “beyond” the text (Lefevere and Bassnett 1990: 4), whichsometimes results in juxtaposing the act of translation viewed as “linguistictranscoding” and as “cultural transfer” (Snell-Hornby 1990), and in suggestingthat translation is primarily cultural transfer, not linguistic transcoding, sincelanguage cannot be separated from culture. Actually, however, the juxtapositionis somewhat artificial, since – reversing the perspective – we can also say thatwhatever extra-linguistic/cultural factors influence the translator, their impact hasto be traced in the text and, as is pointed out by Tabakowska (1997: 29), all texts aremade of language. Acknowledging the insights of the cultural-turn approaches,when theorizing about translation we would like to balance the two perspectives,for instance to find tools to describe how the undeniable extra-linguistic influence isexpressed/realized in language, i.e. to address the micro-level of target text production,together with the well-known problem of the lack of exact equivalencebetween the signs of two languages and the influence of the available resourcesof the target language on the production of the target text.If we accept the inherent link between the linguistic and the extra-linguistic intranslation, linguistic approaches, somewhat overshadowed by the cultural turn,may turn out to be “new research avenues to explore for further advancement oftranslation studies” (Anderman 2007: 62). However, the gradual extension of thescope of interest in translation studies signals that a fruitful linguistic approach totranslation should be based on a framework that can at least to some extentaccommodate the insights of both the pragmatic turn and the cultural turn, that isit has to provide tools for describing individual expressions from the point ofview of the language system, but also of their contribution to the text and of their

16 Construction Grammar as a Framework for Describing Translation 217usage, i.e. include pragmatics, and it has to assume that there is no sharp divisionbetween the “linguistic” and the “extra-linguistic” knowledge. We would like toolsthat can describe both the motivations of translation choices and their mechanism,taking into consideration that the translation process, in accordance with the abovementionedfindings, can be envisaged as taking place on two planes – on the serialplane of interpreting ST expressions and producing corresponding TT expressions,as well as on the text-structural plane which involves pragmatic, inter-textual andsocio-cultural knowledge necessary in interpreting and producing texts (Holmes2006: 418). The following discussion will present CxG as a linguistic model whosescope is largely convergent with so-defined needs of translation studies, and whichhas a potential to address a range of translation issues within an integrated system.16.2 Construction GrammarCxG was initiated by Charles Fillmore in the late 1980s, and it is in a way acontinuation of his previous ideas, Case Grammar (1968) and Frame Semantics(1977, 1985, 1987, among others), underlain by the interest in the exploration of thesemantic, pragmatic and functional motivation of linguistic patterns. Sincethe publication of Fillmore’s early works on CxG (1988, Fillmore et al. 1988) theframework has been developed in various directions, producing several trends, ofwhich the best known is probably the approach represented by Adele Goldberg(1995, 2003, 2006), which she herself calls Cognitive Construction Grammar(CCxG). 1 This paper will only concentrate on some of the general premises ofthe constructional view of language which are shared by all of them and which canbe considered important for addressing problems of translation.Within CxG it is assumed that that the linguistic competence of a language useris best represented in terms of constructions (Fried and Östman 2004: 23), i.e.conventional learned pairings of form and meaning. Form may include phonetic,morphological and syntactic properties, while meaning is understood very broadly,bordering on function and including semantic, pragmatic and discourse (informationstructure) properties (Croft 2007: 472). A construction can be envisaged as abuilding block that the user of a language has in his/her repertoire, a cluster ofproperties from various traditionally differentiated layers or domains of language(Fillmore 1988: 35; Fillmore et al. 1988: 534), as is shown in (1). CxG is a nonmodularmodel; it does not split language into modules whose relations have to bedescribed; the relations between various aspects of the linguistic sign are internal tothat sign (Croft 2007: 472), which stresses the inseparability of form and function.1 See Croft (2007) or Goldberg (2006: 213–226) for an exposition of different constructionalapproaches.

218 I. Szymańska1.DiscoursePragmaticSemanticSyntacticMorphologicalPhonetic and prosodicConstructions are abstract blueprints, generalizations; they may be of variousdegrees of schematicity, so in some constructions some of those domains will not berepresented. Constructions are elements of the language system, templates whichlicense the use of actual expressions and are recognized by the language user aswholes in the process of interpretation. Actual expressions used in communicationare referred to as constructs, i.e. instantiations of constructions. A construct, e.g. anactual sentence, is usually an instantiation of several constructions.Constructions include morphemes, lexical items, phrasal patterns, sentencepatterns, idioms, intonation patterns, discourse patterns, etc. Any language patternthat is conventionalized and unpredictable and has some functional properties (semanticand/or pragmatic) can be considered a construction. CxG is based on the postulateof the full coverage of linguistic data, i.e. of describing all the conventions that thespeaker can use in communication (Fried and Östman 2004: 23–24). Consequently,it does not assume a sharp distinction between the “core” (regular/productive) and“peripheral” (idiomatic/unproductive) elements of the system, envisaging them as acline (Goldberg and Jackendoff 2004: 532) and paying much attention to the“periphery” end of the cline, for instance idioms and patterns of limited productivity,pointing out that they are an important part of communication which cannot beneglected in the description of language (Fried and Östman 2004: 15–16).One of the semantic properties of constructions is that they are underlain by andmay activate conceptual structures which schematize human experience, termedinterpretive frames (or scenes). This idea is drawn from Frame Semantics (sometimesalso called “scenes-and-frames semantics”), Fillmore’s earlier proposal,which many construction grammarians assume to be the semantic complement ofCxG (e.g. Östman and Fried 2004: 5). Although the notions and goals of FrameSemantics have been evolving over the decades (e.g. Fillmore 1977, 1985, 1987;Gawron 2008) and there are differences in focus between various constructiongrammarians as to technical details, in general it may be assumed that CxG(Goldberg 1995: 25–31) shares the general vision of the “semantics of understanding”advanced by Fillmore to account for the relationship between texts and thelanguage users’ interpretation of texts in contexts (Fillmore 1985: 231). Fillmoredescribes text interpretation as relying on the activation of conceptual structures byactually used linguistic material and applying general cognitive operations (e.g.filling blanks in schemas or comparing real situations with prototypical ones)to those structures, which leads to constructing a coherent mental representationconsistent with the language user’s knowledge (Fillmore 1977: 63–66).

16 Construction Grammar as a Framework for Describing Translation 219The conceptual/interpretive structures involved, labelled frames, are organized insuch a way that introducing one element to the text makes all the other elementsavailable to the interpreter, so they create a context necessary for interpretation,allowing the text recipient to reach beyond what is explicitly said by activatingcomplex networks of knowledge (Fillmore 1985: 232; Gawron 2008: 4). They aredescribed as schematizing human beliefs, actions or imaginings, and envisaged asrooted in cultural experience (Fillmore 1987: 33). The notion of frame can alsoaccount for the schematization of the recipient’s expectations concerning text typesand genres, which play an important role in the interpretation process (Östman2005: 128–135). In general then, the idea of frame activation is a very powerful toolfor constructions to account for text interpretation and the role of extra-linguisticknowledge and cultural background in that process.Constructions are envisaged as basically language-specific, which is obvious inthe case of lexical items, but may be less so in the case of syntactic patterns. Thelanguage-specificity of constructions implies that patterns in two languages, even ifthey are traditionally called “the same”, e.g. the passive, need not necessarilycorrespond to each other with respect to all their properties (Goldberg 2003:222). Goldberg (2006: 16) stresses that CxG anticipates considerable variabilitybetween languages, despite the fact that humans have common communicativeneeds and that languages need to express similar messages.Another feature of CxG that should be stressed is its usage-based character. It isassumed that generalizations about language which are stated in terms ofconstructions emerge from usage, which in turn is determined by the need tocommunicate in a socially acceptable way. 2 Consequently, it is assumed, especiallyin CCxG (Goldberg 2006: 64), that some facts about usage, going beyond thetraditionally understood knowledge of the language system and accounting forthe “native-like selection” from the various options conforming to the grammaticalpatterns of the language, for instance intuitions about the frequency of some items,should also be included into the constructional description of the language user’sknowledge (Goldberg 2006: 54–55).16.3 The Applicability of Construction Grammarto Describing TranslationEven this sketchy and highly selective outline of CxG points to its very wide scopeof interest, which is in fact crucial when considering a linguistic model capable ofaddressing the variety of issues involved in the process of translation. CxG is in fact2 The usage-based nature of the constructional approach is strongly related to its account oflanguage acquisition, which is of secondary importance for the purposes of this paper and cannotbe elaborated on here, but which figures importantly in the constructional agenda in general (e.g.Goldberg 2006: Chaps. 4–6; Tomasello 2006).

220 I. Szymańskaa “maximalist” model of language (Fried and Östman 2004: 24), attempting toaccount for all the communicative conventions within one system. Although thispostulate is far from being realized in practice as yet, from the theoretical point ofview such a scope is what is needed to address translation, which involves all kindsof communicative conventions in actual use. The intended maximalism of CxG andabandoning the distinction between the core and the periphery in language description,although they seem to be very theoretical aspects of a language model, in factmake it a very “practical” framework for addressing translation. Let us note that intranslation we have to recognize and represent in the target text any pattern that ismeaningful and functional, and the status of language phenomena in terms ofproductivity has no impact on translational choices. Interestingly, it is often thenon-core patterns of low productivity that cause problems in translation and are thusmost interesting, since they often involve complex configurations of properties,including pragmatic ones, which are not clustered together in the target language.Assuming that CxG adopts the notions of Frame Semantics, including thecontinuity of the linguistic and the extra-linguistic knowledge, and thatconstructions are capable of evoking frames of knowledge, which are rooted inlanguage use and cultural experience, the constructional framework has a potentialto address the cultural conditioning of translational decisions as realized throughthe principled choice from the resources of the target language. It also has a way ofaddressing text interpretation, which can be seen as the recognition of constructionsand frames and the integration of the interpretive clues that are provided by thosestructures. Therefore, the constructional view of language can accounts for the textstructuralplane of translation and capture the widely understood cultural barriersand cultural conditioning of translation as differences in the framing systems thatunderlie two languages. It should be pointed out that the potential of framesemanticconcepts in that respect has been noticed previously (e.g. Rojo López2002; Snell-Hornby 2005), but CxG combines it with a complex view of thelanguage sign, which offers non-negligible descriptive opportunities for explainingthe determinants of the translation process.CxG has a way of accounting for the relationship between the language systemand language use, which are both involved in translation, by clearly stating therelationship between constructions as blueprints/templates and constructs as actualexpressions. Being usage-based and including the pragmatic aspects of language,CxG can also account for the influence of subtle usage-related factors on thetranslator’s choices, such as the frequency of certain items and collocability,which contribute to the “native-like quality” which is usually expected of (althoughnot necessarily actually exhibited by) a translated text.In very general terms, constructional thinking about language is highly relevantto translation since it presents text interpretation and production as relying onrecognizing functional wholes rather than assembling small elements. It is wellknownthat in translation the failure to recognize functional wholes often leads todisastrous effects. An obvious example of that is the translation of idioms, which, aswas already stressed, are of great interest for CxG, but we can also notice this in the

16 Construction Grammar as a Framework for Describing Translation 221case of syntactic patterns, which as wholes have information structure functions,the overlooking of which often leads to incoherence or at least awkwardness of thetarget text. The notion of construction is based on the observation that the meaningof a complex expression need not be predictable from the meanings of its parts.Analogically, in translation, the global functional equivalence between texts orfragments of texts need not follow from local equivalences between their segments.The crucial features of the constructional approach to language from the point ofview of describing translation are its non-modularity, the inseparability of form andmeaning, which underlies the notion of construction, and the language-specificityof constructions. When considered jointly, they highlight and explain the fact thatin translating we are often restricted by the resources of the target language, whichmay cluster various properties differently than the resources of the source language,which were used in the source text for certain communicative purposes. Looking atconstructs, which are used in text creation as instantiations of a number ofconstructions, highlights the fact that they are complex clusters of properties andthat various aspects of the utterance, including its phonetics, morphosyntax, semantics,information structure, register and other pragmatic properties, are realizedsimultaneously by the same linguistic “material”. Such a view of language signsaccounts for the text being created simultaneously on the linear and text-structuralplane, in the sense that each choice in the process of the linear production of the textsimultaneously contributes to the creation of its overall structure.This fact is particularly important in translation, when the target text, as well asits parts, are usually expected to be equivalent in sense and function to the sourcetext, as well as to resemble the source text in structure, linguistic organization andstyle (e.g. the target text is usually not to be a summary of the source text). What weconstantly experience in translation, however, are difficulties in finding exactequivalents of the source text constructs in the target language, i.e. expressionsthat would match them in all the functional and formal properties. Therefore, it iscustomary to speak about losses and gains in translation. CxG explains whyequivalence between constructs in two languages is rarely complete, and howgains and losses originate, by assuming that constructions are language-specificclusters of properties. If in translating we cannot find a target language constructmatching the source language construct in all respects, we have to choose thepriority, that is the dimension of equivalence that is most important in the particularsituation of translating. However, the inseparability of forms and functions inconstructions also entails that sometimes the construction(s) chosen because ofthat priority will bring into the text some properties absent form the original,because we cannot filter out or add properties to the constructions of the targetlanguage to suit our translational purposes, we can only use the resources that areoffered by the target language. Since constructions are described as clusters ofproperties, CxG offers very precise tools to capture the issue of losses and gains intranslation and the problem of equivalence being approximate. The explanation ofthose phenomena follows form the every architecture of the constructional view oflanguage.

222 I. Szymańska16.4 Gains, Losses and Priorities in Translation: An ExampleTo illustrate the last point made above, this section will show how the mechanismof losses and gains in translation can be captured in a constructional framework.The construction chosen for the purpose of this exemplification is one of theEnglish periphrastic causative constructions, the pattern make sb do sth. Thesuggestions made below are based on the examination of a number of examplesof its translation into Polish in novels and film subtitles, aimed at demonstrating theconstructional grounding of the mechanism which is at work in translation, ratherthan at arriving at any closed repertoire of translational options.In constructional terms the pattern can be characterised as follows: like othercausative and resultative constructions (Goldberg and Jackendoff 2004: 538–539),the one in question can be assumed to combine two subevents – the causative oneidentified by make and the resultant one, identified by the other verb. As to theparticipants of the event, the construction involves a Causer and a Causee (Gilquin2006: 2). Other possible participant(s) will depend on the verb in the secondsubevent. The subevents are linked through the assumption that the Causee exertssome influence on the Causer, which indirectly brings about the second subevent(Wierzbicka 1988: 240–242, 244–245; Gilquin 2006: 9), but the nature of thisinfluence is not specified. A corpus-based analysis suggests that the register specificationshould be “neutral/informal” (Gilquin 2006: 7). Those properties can berepresented as in (2):2.Prag Register Neutral/informalSem [Causer CAUSE [Causee resultant subevent]]Syn NP Subj make NP Obj V-bare infinitvePolish does not have a construction corresponding to the English one withrespect to all the properties and this is reflected in translations, in which variousconstructions are combined to produce constructs matching the original in somerespects. When the indirect causation is rendered without specifying the manner ofthe causative influence, as in translations through the verb sprawić że/żeby/aby‘cause that’ or spowodować że/żeby/aby ‘cause that’ (both require a finite clause tofollow) the register is clearly higher, as in (3a). This is a kind of “gain” notnecessarily welcome in every context. On the other hand, translations which keepthe register introduce other little changes; for example, they “erase” the causativesubevent, loosing the element of causality, as in (3b), which may change thepresentation of the situation and of the participants’ involvement in it, or theyapply one of a range of “coercive” verbs (kazać ‘order’, skłonić ‘persuade’,namówić ‘talk into’, zmusić ‘force’, etc.), which specify the nature of the causativeinfluence, unlike the original (e.g. (3c)).

16 Construction Grammar as a Framework for Describing Translation 2233. (a) “And have you worked out how to use it?” he said to her. “No. Least, I canmake the three short hands point to different pictures, but I can’t do anythingwith the long one”. (Pullman 2003:111)– Czy dowiedziałaś się już, jak go używać? – spytał Lyrę.– Nie. Potrafię sprawić, aby trzy krótkie wskazówki pokazały obrazki, jednaknie umiem poruszyć tej długiej. (Pullman 2004:143)[back translation: And have you got to know how to use it? No. I can causethat the three short hands point to pictures, but I cannot move the long one.](b) We had to make it [the situation] look like a casual social encounter, talklightly about matters of life and death. (Lodge 2002: 304)Musieliśmy udawać zwykłych znajomych, rozmawiać lekko o sprawachżycia i śmierci. (Lodge 2001: 304)[back translation: We had to pose as casual acquaintances, to talk lightlyabout matters of life and death.](c) Why make people go through hell before you help them out of it? (Lodge2002: 283)Dlaczego zmuszamy ludzi, żeby przechodzili przez piekło? (Lodge2001: 283)[back translation: Why do we force people to go through hell?]Leaving aside the evaluation of the influence of such micro-changes on thetranslated text and concentrating on the descriptive perspective, the constructionalview of language signs as clusters of properties from various domains, including thepragmatic one, highlights the approximate nature of equivalence and helps trace thetranslators’ priorities (e.g. the priority of register over the semantic precision inpresenting the event in (3b) and (3c)).16.5 ConclusionTo sum up, it has been argued that Construction Grammar is highly convergent withthe needs of translation studies defined by the extending scope of this disciplineindicated in Sect. 16.1. CxG is inherently functional and usage-based, beingmaximalist in the aim of describing the totality of the communicative conventionsavailable to the language user. It has a potential to address the issues of pragmatics,text structure, text interpretation and the influence of culture-bound knowledge onthe use of language, which makes it capable of acknowledging the insights of thepragmatic turn and the cultural turn in translation studies. On the other hand, itcombines this potential with a very detailed approach to the description of languagesigns and to functionalizing structures, which offers an opportunity of a preciseanalysis of the translator’s linguistic decisions following from the differences

224 I. Szymańskabetween the two systems involved in the process of translating. Finally, being nonmodularand including discourse and pragmatic properties in the description ofconstructions, it highlights the fact that the micro-level of linguistic decisions andthe macro-level of text are inherently connected.ReferencesAnderman, G. 2007. Linguistics and translation. In The companion to translation studies, eds.P. Kuhiwczak and K. Littau, 45-62. Clevedon: Multilingual Matters.Croft, W. 2007. Construction grammar. In The Oxford handbook of cognitive linguistics, eds.D. Geeraerts and H. Cuyckens, 463-508. Oxford: Oxford University Press.Fillmore, C. J. 1968. The case for case. In Universals in linguistic theory, ed. E. Bach and R. T.Harms, 1-88. New York: Holt, Rinehart and Winston.Fillmore, C. J. 1977. Scenes-and-frames semantics. In Linguistic structures processing,ed. A. Zampolli, 55-81. Amsterdam: North Holland.Fillmore, C. J. 1985. Frames and the semantics of understanding. Quaderni di semantica 6: 222-254.Fillmore, C. J. 1987. A private history of the concept ‘frame’. In Concepts of case, eds. R. Dirvenand G. Radden, 28-36. T€ubingen: Gunter Narr Verlag.Fillmore, C. J. 1988. The mechanisms of Construction Grammar. Proceedings of the AnnualMeetings of Berkeley Linguistics Society 14: 35-55.Fillmore, C. J., P. Kay and M. C. O’Connor. 1988. Regularity and idiomaticity in grammaticalconstructions: The case of let alone. Language 64: 501-538.Fried, M. and J.-O. Östman. 2004. Construction Grammar: A thumbnail sketch. In ConstructionGrammar in a cross-language perspective, eds. M. Fried and J.-O. Östman, 11-86. Amsterdam/Philadelphia: John Benjamins.Gawron, J. M. 2008. Frame semantics. http://www.hf.uib.no/forskerskole/new_frames_intro.pdf.Accessed 14 April 2009.Gilquin, G. 2006. The verb slot in causative constructions. Finding the best fit. Constructions SV1-9/2006. http://www.constructions-online.de. Accessed 10 June 2009.Goldberg, A. E. 1995. Constructions. A Construction Grammar approach to argument structure.Chicago: University of Chicago Press.Goldberg, A. E. 2003. Constructions: A new theoretical approach to language. Trends in cognitivesciences 7: 219-224.Goldberg, A. E. 2006. Constructions at work. The nature of generalizations in language. Oxford:Oxford University Press.Goldberg, A. E and R. Jackendoff. 2004. The English resultative as a family of constructions.Language 80: 532-568.Holmes, J. S. 2006 [1978]. Describing literary translations: Methods and models. In Translation –theory and practice. A historical reader, eds. D. Weissbort and A. Eysteinsson, 416-422.Oxford: Oxford University Press.Lefevere, A. and S. Bassnett. 1990. Proust’s grandmother and the Thousand and One Nights. The‘cultural turn’ in translation studies. In Translation, history and culture, eds. S. Bassnett andA. Lefevere, 1-13. London: Pinter Publishers.Lodge, D. 2001. Myśląc...Translated by Z. Naczyńska. Poznań: Rebis.Lodge, D. [2001] 2002. Thinks...New York: Penguin Books.Östman, J.-O. 2005. Construction discourse. A prolegomenon. In Construction grammars. Cognitivegrounding and theoretical extensions, eds. J.-O. Östman and M. Fried, 121-144.Amsterdam/Philadelphia: John Benjamins.

16 Construction Grammar as a Framework for Describing Translation 225Östman, J.-O. and M. Fried. 2004. Historical and intellectual background of Construction Grammar.In Construction Grammar in a cross-language perspective, eds. M. Fried and J.-O.Östman, 1-10. Amsterdam/Philadelphia: John Benjamins.Pullman, P. 2003. The golden compass [first published 1995 as Northern lights]. New York:Random House.Pullman, P. 2004. Zorza północna. Translated by E. Wojtczak. Warszawa: Albatros.Rojo López, A. M. 2002. Applying Frame Semantics to translation: A practical example. MetaXLVII: 311-350.Snell-Hornby, M. 1990. Linguistic transcoding or cultural transfer? A critique of translation theoryin Germany. In Translation, history and culture, eds. S. Bassnett and A. Lefevere, 79-86.London: Pinter Publishers.Snell-Hornby, M. 2005. Of catfish and blue bananas: scenes-and frames semantics as a contrastiveknowledge system for translation. In Knowledge systems and translation, eds. H. V. Dam,J. Engberg and H. Gerzymisch-Arbogast, 193-206. Berlin/New York: Mouton de Gruyter.Snell-Hornby, M. 2006. The turns of translation studies. Amsterdam/Philadelphia: JohnBenjamins.Szymańska, I. 2011. Mosaics. A Construction-Grammar-based approach to translation.Warszawa: Semper.Tabakowska, E. 1997. O tłumaczeniu wiersza – perspektywa językoznawcy. In Międzyoryginałem a przekładem III, ed. M. Filipowicz-Rudek, J. Konieczna-Twardzikowa andM. Stoch, 29-43. Kraków: Universitas.Tomasello, M. 2006. Construction Grammar for kids. Constructions SV1-11/2006. http://www.constructions-online.de. Accessed 20 March 2010.Wierzbicka, A. 1988. The semantics of grammar. Amsterdam: John Benjamins.

Chapter 17Crossing the Frontiers of LinguisticTypology: Lexical Differences and TranslationPatterns in English and Russian Lolitaby Vladimir NabokovŁukasz GrabowskiAbstract This article presents the results of the corpus-driven comparison betweenthe English-original (1955) and Russian auto-translation (1967) of the novel Lolitaby Vladimir Nabokov. The aim of the study, which was facilitated by the computerprogram WordSmith Tools 4.0, was to answer the question whether the differencesattested between the English and Russian parallel texts arise from translationstrategies [Nabokov was an ardent advocate of literal translation as the only strategyof truly transposing the original text (Beaujour 1995: 716; Grayson 1977: 13–15)],or whether they are due to typological differences between the English and Russianlanguages. This corpus-driven study consists of two parts. The first part aims at acomparison of lexical wordlists (i.e. top-frequency lexical words) generated forEnglish and Russian Lolita. The analysis revealed, among others, that Nabokovused synonymy as a frequent translation strategy (particularly in the case of Englishreporting verbs), which indicates that repetitions are regarded as a bad style inRussian texts. Moreover, the analysis highlighted a conspicuous typological differencebetween the two languages whereby Russian is more explicit semantically (i.e.words have more specific meaning distinctions) than English, which in turn is moreambiguous and vague in its surface forms (Comrie 1981: 31–79). The second partaims at an examination of translation strategies used by Nabokov while translatingcreative, author-specific hapax legomena, following a similar study of English andGerman prose conducted by Kenny (2001). The analysis revealed that Nabokovexhibited a strong tendency towards lexical normalization while translating creativehapax legomena into Russian. All in all, the corpus-driven analysis revealedthat although translators are free to use multifarious translation strategies whileŁ. Grabowski (*)Opole University, Opole, Polande-mail: lukasz@uni.opole.plM. Pawlak and J. Bielak (eds.), New Perspectives in Language,Discourse and Translation Studies, Second Language Learning and Teaching,DOI 10.1007/978-3-642-20083-0_17, # Springer-Verlag Berlin Heidelberg 2011227

228 Ł. Grabowskitransposing original texts, they are still at the mercy of typological differencesbetween the relevant languages.17.1 IntroductionLolita (1955) is one of the best known novels by Vladimir Nabokov, which firmlyestablished him as an outstanding American novelist, brought him fame, liberatedhim from his teaching post at Cornell University (1959), and created both interestand the market for English translations of his Russian works. Due to controversial,at that time, subject matter (both a narrator and protagonist, middle-aged HumbertHumbert, becomes obsessed and sexually involved with a 12-year-old girl, DoloresHaze), Nabokov was unable to find a publisher of the novel in the United States, andinstead the book appeared under the imprint of Olympia Press Publishing House inParis in 1955 (Boyd 1993: 356–357). The novel is written in a highly artistic,masterful and precise style which made Nabokov one of the most brilliant andidiosyncratic stylists of English (Stiller 1991: 421).If Lolita was ever going to be translated into Russian, Nabokov wanted to ensurecontrol over the outcome. When his siblings offered to divide the translation taskbetween them, Nabokov was initially delighted, but at the same time alertlycautious. Their sample translations, however, left him aghast (Trzeciak 2005: 10).In one of the interviews given in 1964, Nabokov expressed both dissatisfaction withand concern about the future prospect of the translation, which prompted him tocommence the self-translation of Lolita himself (quoted in Osimo 1999: 230):I trained my inner telescope upon that particular point in the distant future and I saw thatevery paragraph, pockmarked as it is with pitfalls, could lend itself to hideous mistranslation.In the hands of a harmful drudge, the Russian version of Lolita would be entirelydegraded and botched by vulgar paraphrases or blunders. So I decided to translate it myself.However, as in the case of his translation into English of the novel Omчaяниe/Despair, 1 a rationale behind his self-translation of Lolita (this time in the oppositedirection, i.e. from English into Russian) was Nabokov’s displeasure at the resultsof other translators’ non-conformity to his views and guidelines for successfultranslation. In the postscript to Russian Lolita, Nabokov frequently envisionedabysmal translator blundering of his novel while translating it into Russian(Nabokov 1989: 361).1 Beaujour (1995: 720) claims that it provides excellent examples of both the problems inherent instrict and literal self-translation and instances of advantageous employment of “unfaithful” selftranslation.Obviously enough, his first translation into English was completed in 1936 andconstituted the first attempt to use English for artistic purposes, the experience Nabokov referredto as disagreeable (ibid.) Having translated Lolita into Russian in 1965, Nabokov capitalized on hisyears of experience and undertook his second attempt at translating Omчaяниe into English in afreer fashion. Later on Nabokov claimed that he would have been “pleased and excited” if he hadhad a chance to read his 1965 version of Despair at the end of 1930s (ibid.).

17 Crossing the Frontiers of Linguistic Typology 229When publishing Lolita in Russian, I have a very simple aim in mind: I want my bestEnglish book – or let’s put it modestly, one of the best of my English books, to be correctlytranslated into my native language 2 [translation mine, ŁG].Another important factor was the lack of hope that Lolita will be published in theSoviet Union in Nabokov’s lifetime (Nabokov 1989: 360; Trzeciak 2005: 10).It is difficult even to imagine any regime in my stiff fatherland, be it liberal or totalitarian,which would enable the censorship to launch Lolita into the market 3 [translationmine, Ł. G.].Holding no hope for the future, the writer decided to translate the novel himself.He started this task in winter of 1963 and in March 1965 the Russianized Lolita wasrevised and completed (Boyd 1995: xlvii; Trzeciak 2005: 11). Nevertheless, hispostscript to the Russian edition of Lolita contains passages which illustrateNabokov’s skepticism and dissatisfaction with his self-translation, which he calls“the story of disappointment” (Nabokov 1989: 358):My private tragedy, which cannot, and indeed should not, be anybody’s concern, is that Ihad to abandon my natural idiom, my untrammeled, rich, and infinitely docile Russiantongue for a second-rate brand of English, devoid of any of those apparatuses — thebaffling mirror, the black velvet backdrop, the implied associations and traditions —which the native illusionist, frac-tails flying, can magically use to transcend the heritagein his own way. 4Even though Nabokov frequently complained about the rusty condition of hisRussian, most of the critics emphasized the role of his translation as a giant step inthe development of the Russian literary language (Grabowski 2008: 129). Anémigré critic, Vladimir Weidle, claimed that in the long term Nabokov’s translationof Lolita will be “fertile for Russian literature and an even greater phenomenonthere than the English Lolita in English literature” (Boyd 1993: 490). Nevertheless,there were also negative opinions about the Russianized novel. According toProffer (1970: 258), even Nabokov’s most ardent fans dislike the Russian translationof Lolita and the majority of Russians she had met tend to find Nabokov’stranslation clumsy and even ungrammatical. According to Grayson (1977: 193), thestyle of Russian Lolita “bears such strong traces of English constructions that it cannot safely be treated as an autonomous piece of Russian”. Furthermore, Proffer(1970: 259) quotes an opinion of an anonymous Russian linguist who read the2 The original reads: Издaвaя “ЛoлиTу” пo-pуccки, япpecлeдую oчeнь пpocTую цeль:xoчу, чToбы мoя лучшaя aнглийcкaя книгa –или, cкaжeм eщe cкpoмнee, oднa излучшиx мoиx aнглийcкиx книг – былa пpaвильнo пepeвeдeнa нa мoй poднoй язык.3 The original reads: Mнe Tpуднo пpeдcTaвиTь ceбe peжим, либepaльный ли илиToTaлиTapный, в чoпopнoй мoeй oTчизнe, пpи кoTopoм цeнзуpa пpoпуcTилa бы“ЛoлиTу”.4 The original reads: “Увы, ToT ‘дивный pуccкий язык’, кoTopый,cдaвaлocьмнe, вce ждeTмeня гдe-To, цвeTeT, кaк вepнaя вecнa зa нaглуxo зaпepTыми вopoTaми, oT кoTopыxcToлькo лeT xpaнилcя умeня ключ, oкaзaлcя нecущecTвующим, изa вopoTaми нeTничeгo, кpoмe oбуглeнныx пнeй иoceннeй бeзнaдeжнoй дaли, aключ в pукe cкopeeпoxoж нaoTмычку”.

230 Ł. GrabowskiRussianized Lolita and found it fascinating because it had been written in a kind ofdead language nobody used in the Soviet Union of the 1960s. Robert Stiller, whoproduced the first full translation of the novel into Polish in 1991, represents theopinion that the auto-translation is awkward and clumsy, riddled with vulgarity,which is not so omnipresent in the English original (Stiller 1991: 428). EvenNabokov himself was far from being satisfied with the quality of his auto-translation;in the postscript to Russian Lolita he opined that many fragments of histranslation are “clumsy, loquacious and frequently disgusting in terms of styleand rhythm” (Nabokov 1989: 358). 5All in all, taking into consideration the inside story of the Russianization ofLolita, including Nabokov’s attachment to literal translation (Beaujour 1995: 716)as well as the reception of the Russian Lolita by the critics and the author andtranslator himself (Ginter 2003:15), the aim of this article is to identify majorlexical differences between the English and Russian texts of the novel (henceforth“parallel texts”). In particular, the comparison undertaken herein attempts to determinewhether the lexical differences between the parallel texts arise from translationstrategies or whether they are due to typological differences between Englishand Russian. Is it justifiable to claim that although translators are free to usemultifarious translation strategies while translating original texts, they are still atthe mercy of typological differences between relevant languages?17.2 Methodology, Tools and Stages of the AnalysisIn order to provide answers to the aforementioned study questions, the corpusdrivenmethodology was applied. In contrast to the corpus-based approach, whichalways works within commonly accepted frameworks of theories of language(which implies prior classification of linguistic data), the parallel texts were notadjusted to fit any predefined categories or theoretical schemata, and the studyquestions were addressed through the analysis of frequency distributions of wordsand recurrent patterns of language use as found in English and Russian Lolita.Asaresult, the parallel texts were compared through bottom-up observation of empiricallinguistic data, which were presented in quantitative terms and, where necessary,supplemented with qualitative observations.Both novels, i.e. the English-original (Nabokov 2000, henceforth ENL) as well asthe Russian Lolita (Nabokov 1989, henceforth RUL), were purchased in bookstoresin paper format and they were further converted into the machine-readable formatsupported by the software used throughout the study. To that aim, the parallel texts5 The original reads: (...) нo cToль cвoйcTвeнныe aнглийcкoму Toнкиe нeдoгoвopeннocTи,пoэзия мыcли, мгнoвeннaя пepeкличкa мeжду oTвлeчeннeйшими пoняTиями, poeниeoднocлoжныx эпиTeToв – вce эTo, a Taкжe вce oTнocящeecя к Texникe, мoдaм, cпopTу,ecTecTвeнным нaукaм и пpoTивoecTecTвeнным cTpacTям – cTaнoвиTcя пo-pуccкиToпopным, мнoгocлoвным и чacTo oTвpaTиTeльным в cмыcлe cTиля и pиTмa”.

17 Crossing the Frontiers of Linguistic Typology 231were manually scanned and subjected to the OCR procedure. The scanned texts weresubjected to repeated proofreading in order to ensure spelling accuracy, and they werefurther verified against the paper format versions. At that stage, any cases ofmisrecognition of characters were edited and corrected using a spellchecker, or asearch-and-replace facility of a word processor. Finally, the parallel texts were savedin two files in a plain text format.The corpus-driven analysis conducted in this study was facilitated by the use ofthe computer software WordSmith Tools 4.0 developed by Scott (2004), which is asuite of programs custom-designed for text analysis. The study was broken downinto two successive stages. First, top-frequency lexical words and theirdistributions, as found in ENL and RUL, were compared. As a result, the mostconspicuous lexical differences between the two texts were identified, and selectedtranslation strategies used by Nabokov were determined through qualitative interpretationof word frequency lists. In the second stage of the study, translationstrategies used by Nabokov while translating creative, author-specific hapaxlegomena were examined. This part of the study was modelled on the studyconducted by Kenny (2001), who studied lexical normalization in the translationfrom German into English of creative hapax legomena attested in a customdesignedcorpus of literary texts. Finally, the analyses conducted in stage one andtwo highlighted conspicuous typological differences between the English andRussian languages, which were presented in greater detail.17.3 Comparison of WordlistsAccording to Scott (2004), wordlists, or more specifically – word frequency lists,can be used, among other purposes, to study type, range and distribution ofvocabulary used in a text, identify the most common word clusters, comparefrequencies of cognate words or translation equivalents across source and targetlanguage texts. The Wordlist facility of WordSmith Tools 4.0 generated thewordlists for ENL and RUL. Table 17.1 shows 25 top-frequency word typespresented in decreasing frequency order together with the cumulative percentage oftotal word count (%W) in each of the texts.Apart from only a few translation equivalents obtained in the wordlists, such asconjunctions and – и (the most frequently used co-ordinating conjunction), thepersonal pronoun I – я (sign of the 1st person narration), etc., there are no translationequivalents for the majority of top frequency words, which shows that we arecomparing very different language systems. Where the English language employsprepositions or articles (which are separate running words in a text), Russianemploys endings, prefixes and suffixes (which are bound with other words). Itexplains why the top English words are much more frequent than the top Russianwords – if one adds the percentage of the total word count for 25 top-frequencywords, it occurs that in ENL they account for 35.07% of the overall number ofrunning words, whereas in RUL – 23.32%. Having determined that, I aimed to find

232 Ł. GrabowskiTable 17.1 Wordlist with 25 top-frequency word types in ENL and RULENLRULRank Word Frequency %W Rank Word Frequency %W1 THE 5,165 4.59 1 И 3,394 3.302 AND 3,712 7.89 2 B 3,174 6.393 A 3,579 11.07 3 Я 2,217 8.554 OF 3,179 13.90 4 HA 1,476 9.985 I 2,983 16.55 5 ЧTO 1,397 11.346 TO 2,446 18.73 6 HE 1,377 12.687 IN 2,105 20.60 7 C 1,343 13.998 HER 1,613 22.03 8 КAК 915 14.889 WAS 1,486 23.36 9 OHA 876 15.7310 MY 1,419 24.62 10 EE 766 16.4811 WITH 1,223 25.71 11 MEHЯ 606 17.0712 HAD 1,197 26.77 12 MHE 603 17.6513 THAT 1,178 27.82 13 К 545 18.1814 SHE 1,078 28.78 14 HO 517 18.6915 ME 809 29.50 15 TO 512 19.1916 ON 779 30.19 16 ПO 486 19.6617 IT 738 30.84 17 ИЗ 481 20.1318 AS 721 31.49 18 У 456 20.5719 AT 700 32.11 19 БЫ 454 21.0120 FOR 657 32.69 20 OH 433 21.4321 NOT 654 33.27 21 BCE 421 21.8422 YOU 566 33.78 22 ЭTO 394 22.2323 BUT 536 34.25 23 ЗA 390 22.6124 HE 462 34.66 24 A 377 22.9725 WOULD 459 35.07 25 EГO 360 23.32a word rank at which the total word count reaches 50% in ENL and RUL. It wasrevealed that in the case of top-frequency words the lexical variety (in terms of thenumber of repetitions of the most frequent words) attested in both ENL and RULdiffers more than considerably – word ranks at which the word count reaches 50%are 105 and 593 in ENL and RUL, respectively. These results are presented inTable 17.2 below.Table 17.2 shows that in ENL 105 word types account for 50% of the total wordcount, whereas in the case of RUL – 593 word types. As a result, there are 488 moreword types in RUL which account for 50% of the total word count. This text-specificdata can be further generalized as the difference between distribution of the topfrequencywords in English and Russian texts. English texts have more repetitionsand lower lexical variety in terms of top-frequency words, whereas their Russiantranslations (and Russian texts in general) have fewer repetitions and higher lexicalvariety among top frequency words. As a result, raw frequencies of top-frequencywords are considerably higher in ENL than in RUL (the word type ranked 105th(never) in ENL has the frequency of 114, whereas the word type ranked 105th in RUL

17 Crossing the Frontiers of Linguistic Typology 233Table 17.2 Proportion of top-frequency words in the total word count in ENL and RULENLRULRank %W (cumulative) Rank %W (cumulative)1 4.593316 1 3.303210 24.62249 10 16.4819120 32.69747 20 21.4386530 36.97419 30 25.0026840 40.03877 40 27.4854350 42.43993 50 29.3511460 44.36174 60 30.8450770 45.95539 70 32.0976480 47.26624 80 33.1935190 48.4686 90 34.16578100 49.54556 100 35.04949105 50.05869 593 50.01022(oпяmь ‘again’) has the frequency of 85, and the word type ranked 593rd (мoгуm‘(they) can’) has the frequency of 15).However, a major problem with information extracted from frequency lists is thefact that raw frequencies highlight very common grammatical words, such as the, a,and, of revealed above, and their occurrence is unlikely to provide evidence of anyspecifically used vocabulary across source and target language texts. To remedythis inconvenience, all grammatical words were deleted from the top-frequencyitems, and the most common content (lexical) words used in ENL and RUL,respectively, were isolated. Such a filtered-out frequency list with 25 top-frequencycontent words is presented in Table 17.3.Apart from certain regularities which result from typological differences betweenthe two language systems, Table 17.3 also reveals some characteristic features of thetranslation. As regards the differences between the language systems, high frequencyof the verb forms was and have in ENL results from their functioning notonly as inflectional forms of the verbs to be and to have, but also from functioning asauxiliary verbs used in multiple grammatical tenses (which explains their higherfrequency as compared with the aggregated frequency of the equivalent verb formsбылo, был and былa (was/were) in Russian). Another observation refers to the highrank and frequency of the reporting verb said, with 344 occurrences in ENL, whichis more than the aggregate frequency of the corresponding Russian reportingverbs cкaзaл (masc.) and cкaзaлa (fem.), with 211 occurrences of both. Thisobservation shows that repetitions in Russian texts are regarded as bad style, andtherefore Nabokov used either synonymous reporting verbs while translating theEnglish reporting verb said into Russian 6 (e.g. пoдcкaзaлa, paccкaзaлa,выcкaзaл, зaгoвopилa, гoвopилa, пpoизнec/лa, пpoгoвopил/a, зaявилa,гoвopил, пoгoвopил) or employed more specific and concrete verbs, such as6 The data presented below come from the analysis of parallel concordances, i.e. fragments ofEnglish Lolita aligned with the corresponding fragments of Russian Lolita.

234 Ł. GrabowskiTable 17.3 25 top-frequency content (lexical) words in ENL and RULENLRULRank Word Frequency Rank Word Frequency9 WAS 1,486 31 БЫЛO 32230 HAVE 388 32 БЫЛ 30837 SAID 344 38 TOЛЬКO 22239 WERE 305 41 БЫЛA 21442 LITTLE 287 43 ЛOЛИTA 20251 LO 236 58 БЫЛИ 14361 OLD 197 62 MOЖET 13762 LOLITA 193 64 MOГ 13263 TWO 192 74 CКAЗAЛ 11581 KNOW 139 87 BPEMЯ 9884 HAZE 137 88 CКAЗAЛA 9687 WAY 135 95 ДOЛЛИ 9189 CHILD 132 101 ДBA 8793 ROOM 121 102 ЛET 8695 GIRL 120 103 ГУMБEPT 85100 AM 117 106 ГEЙЗ 82101 CAR 117 108 COBEPШEHHO 79102 GOOD 116 111 ДEHЬ 77103 HUMBERT 115 117 ЛO 74108 EYES 109 122 ДOBOЛЬHO 70113 HAND 104 127 ECTЬ 67114 MADE 104 131 ДOЛOPEC 64115 DAY 103 134 OДИH 63116 FIRST 103 139 BPEMEHИ 60120 LET 98 141 ЛOЛИTЫ 60выpaзилcя ‘expressed/voiced’, omвemил ‘answered/replied’, oбъяcнил/a‘explained’, oбeщaлa ‘promised’, кpичaл ‘screamed’, пoдxвamилa ‘picked up’,oбъявилa ‘announced’, oбpamилacь [кo мнe] ‘turned [to me]’, глacилa 7 ‘read’,пpocилa ‘asked’, пpoцeдил ‘muttered/mumbled’, вocкликнулa ‘shouted’,шeпчem ‘whisper’. Alternatively, he resorted to structural changes, omissions orcondensing, which is revealed in the passages presented in Table 17.4.Using synonyms or employing various syntactic, semantic or pragmaticstrategies when translating reporting verbs shows that the neutral reporting verbsaid is rendered with more emotionally charged synonyms or equivalent expressionswith slightly different meanings (e.g. screamed, shouted, explained, answered,asked, announced, etc.). As a result, the English verb said is not semanticallydetermined and it is only context which allows for the specification of its exactmeaning. This observation refers to one of the specific problems of translation7 Picture of the Week, said the legend. ! Haдпиcь глacилa: ЗaмeчaTeльнeйшaя зaHeдeлюФoToгpaфия.

17 Crossing the Frontiers of Linguistic Typology 235Table 17.4 Examples of translation of the reporting verb said from English (in ENL) into Russian(in RUL)ENLRUL“Sure,” she said.“Кoнeчнo, пoпpoбуйTe”. [OMISSION]Now, at twelve, she was a regular pest, saidHaze.She had a message for me, she said; and,topping my automatic thanks with a kindly“you’re welcome,” good Louise left anunstamped, curiously clean-looking letter inmy shaking hand.The muse of invention handed me a rifle and Ishot a white bear who sat down and said:Ah!Hынe, вдвeнaдцaTь лeT, эTo пpямo бичБoжий, пo cлoвaм Гeйзиxи.[STRUCTURAL CHANGE – according toHaze/in accordance with Haze’s words]Унee былo, пo eecлoвaм, пopучeниe кoмнe, иувeнчaв мoe aвToмaTичecкoe“cпacибo” paдушным “нe зa чTo”,дoбpaя Луизa ocTaвилa cTpaннo-чиcToe,бeз мapки и бeз пoмapки, пиcьмo вмoeй Tpяcущeйcя pукe. [STRUCTURALCHANGE – according to her]Mузa вымыcлa пpoTянулa мнe винToвку, иявыcTpeлил в бeлoгo мeдвeдя,кoTopый ceл иoxнул. [CONDENSINGsaid: Ah! into one verb, which literallymeans to ah]between English and other Slavic languages, i.e. the fact that Russian (and otherSlavic languages) is more explicit semantically (i.e. the words have more specificmeaning distinctions) than English, which in turn is more ambiguous and vague inits surface forms (Comrie 1981: 31–79). Hence, the English language is moredependant on pragmatic and contextual information in specifying the exact interpretationof its linguistic forms (e.g. the past tense verb said), which are broad inmeaning. According to Piotrowski (1994: 95–96), users of English tend to chooselexemes which are broad in meaning rather than specific (even though English hasboth broad and specific word types), whereas users of Russian (and probably otherSlavic languages) tend to choose specific lexemes, and therefore they regard textswith multiple repetitions as plain, simple, or even as having bad style. The sideeffectof that is that a proper translation of English reporting verbs (or broad Englishword types in general) requires that more lexical (content) words be used in order toproduce a natural and acceptable text or translation into Russian.Apart from the general observations adduced above, the 25 top-frequencycontent words also provide evidence of certain stylistic choices made by Nabokovwhile translating English Lolita into Russian. The frequencies of abbreviated namesLo and Лo (236 vs. 74 occurrences in ENL and RUL, respectively) show thatinstead of Lo Nabokov employed different means of addressing the eponymousprotagonist, such as the full name Лoлиma (Lolita), personal pronoun oнa (she), ordiminutive form Лoлиmoчкa (little Lolita).The Russian equivalent of the English noun girl, referring to Lolita, is also notrepresented among the top-frequency content words in RUL. The analysis of itstranslation equivalents shows that Nabokov used different semantic and pragmaticstrategies while translating it into Russian [occasionally he used direct loans when

236 Ł. Grabowskireferring to culture-specific American concepts, e.g. college girl (кoллeджгэpл),scout girl (гэpл-cкaуm)]. His choices include, among others, дeвoчкa, дeвушкa(synonyms) ‘girl’, пoдpугa ‘girlfriend’, шкoльницa ‘pupil’, гимнaзиcmкa ‘highschool student’, дeвчуpкa diminutive of ‘girl’, aмepикaнoчкa ‘American girl’,пoдpocmoк, пoдpужкa diminutive of ‘girlfriend’, дeвчoнoчкa diminutive of‘girl’, etc.An interesting finding, however, is the fact that compound nouns which includethe noun girl, such as girl friend, English girl, American girl, girl children, girlchild, school girl are rendered in RUL with single nouns, such as пoдpугa,aнгличaнкa, aмepикaнoчкa, дeвoчки, гимнaзиcmкa and шкoльницa. Itshowsthat the meaning of English compound nouns is conflated into single Russian nouns.Finally, an interesting example of the pragmatic strategy of explicitation isrevealed in the translation of the compound noun career girls, occurring in ENLtwice, which is rendered in RUL with the coordinative compound ceкpemapши икoнmopщицы (literally: “secretaries/PAs and office ladies”), and the noun phraseкoнmopcкaя дeвицa (literally “an office lady”), which are both more explicit inmeaning (as those ladies make their careers in the corporate office world).17.4 Translation of Creative Hapax LegomenaThe comparison of wordlists presented above focuses on top-frequency words, asfound in ENL and RUL. However, creative and author-specific words tend to occurin a text with low frequency. As a result, wordlists featuring top-frequency wordsare not suitable to capture creative or author-specific terms, which can be foundamong bottom-frequency words. Thus, in order to compare lexical variety amongbottom-frequency words, I focused on words which occur in a text only once, whichare referred to as hapax legomena. The analysis of the frequency lists generated forENL and RUL, respectively, shows that the former text contains 6,984 hapaxlegomena (they account for 49.91% of the overall vocabulary and 6.22% of theoverall number of running words in ENL), whereas the latter text contains 20,368hapax legomena, which account for 68.69% of the overall vocabulary and 19.86%of the overall number of running words in RUL). 8 Due to the lack of lemmatizationof the Russian wordlist, it is impossible to determine whether this discrepancyresults from multiple inflectional forms of the same lemmas treated as separateword types (and hence also as hapax legomena), or whether it is due to morecreative (i.e. non-normalized) translation of English expressions into Russian. Thelatter observation applies also to hapax legomena as it is possible to either translatethem creatively (a hapax legomenon in a source text is rendered with a hapax8 This difference is due to the fact that each inflectional form of a particular word type (e.g.genitive, accusative or locative case of the verb, in either singular or plural, feminine or masculine)is treated as a single occurrence of a type, which accounts for a typical problem of operating withnon-lemmatized types and tokens in highly inflectional languages such as Russian.

17 Crossing the Frontiers of Linguistic Typology 237legomenon in a target text), or to normalize them (a hapax legomenon in asource text is rendered with typical or conventional expressions in a target textin an attempt to solve problems posed by translation of author-specific orunusual source-text features).Kenny (2001), who conducted a study of lexical normalization in the translationfrom German into English of creative linguistic expressions 9 attested in the customdesignedcorpus of literary language, revealed that normalization occurs in 44% ofcases where translators have to deal with creative hapax legomena in source texts(ibid., 187). Hence, there is a choice between creativity and normalization, andadherence to one or the other may further influence the differences between thenumbers of hapax legomena as found in source and target texts.Therefore, this case study aimed to determine whether selected creative hapaxlegomena attested in the English original version of Lolita were rendered witheither creative target language forms, or conventionalized (normalized) ones. Next,it was determined whether the translated expressions account for hapax legomenain RUL or not. In other words, it was eventually aimed to determine whattendencies were exhibited and what translation strategies Nabokov had employedwhen translating creative, author-specific hapax legomena into Russian.As the English-original version of Lolita has 6,984 hapaxes, 10 a statisticallysignificant number of hapaxes is 3,575, which was determined by applying the chisquareformula proposed by Berber-Sardinha (1996: 1–6). As an analysis of such anumber of hapaxes falls beyond the scope of this study, 15 creative hapax legomenawere selected, which account for 0.21% 11 of the total number of hapax legomena inENL. By creative hapax legomena I understand the word types which occurred inthe text only once, constitute author-specific expressions, are not found in thereference dictionary of English (Longman Dictionary of Contemporary English,2003). These are the following: instantenousness, fountainous, crenullated, cobwebby,beasthood, charshaf, alembics, antiphony, axillary, barroomette, barelimbed,biscuity, cesspoolful, semitranslucent and libidream.The analysis of the patterns of translation revealed that 7 out of 15 creative hapaxlegomena in ENL preserved their unique status in RUL, i.e. they occurred in RULonly once, and these were translated either literally (e.g. libidream – либидocoн,antiphony – aнmифoния, biscuity – биcквиmный), or were rendered with thecorresponding Russian synonyms (e.g. instantenousness – мгнoвeннocmь,barroomette – бapчик, crenullated – зубчamый, fountainous – вoдoмemный).Further, it was determined how many of the said 7 hapaxes both constitute authorspecificexpressions and are not recorded in Great Reference Dictionary of Russian12 (Ozhegov and Shvedova 2000). It was revealed that only 2 hapax legomena9 Kenny (2001) used selected hapax legomena deemed as creative expressions, which constituted0.2% of the total number of hapax legomena in a source text.10 Such a plural form (hapaxes) of “hapax legomena” was used, among others, by Kenny (2001).11 This figure approximates the one proposed by Kenny (2001), which was 0.2%.12 Oжeгoв, C., Швeдoвa H. (2000). Toлкoвый Cлoвapь pуccкoгo языкa. 4-e издaниe,дoпoлнeннoe. Mocквa.

238 Ł. Grabowskiare not recorded in the said dictionary [those are вoдoмemный (equivalent offountainous) and либидocoн (equivalent of libidream)]. As a result, only 2 Russiantranslations (13%) of the 15 creative hapax legomena in ENL constitute creativehapax legomena in RUL.Furthermore, 8 out of 15 creative hapax legomena in English did not preservetheir unique status of hapax legomena in RUL. Five of them were translated withdescriptive expressions, which implies the translation strategy of explicitation (e.g.cobwebby – пaуmинoй зapocшee; beasthood – cнacmь camиpa; alembics –вoлxoвcкиe иcчиcлeниe; axillary – пoд мышкoй; cesspoolful – цeлoйвыгpeбнoй ямoй, пoлнoй). The remaining three hapax legomena in ENL wererendered with non-creative synonyms, which were used in RUL more than once,e.g. charshaf – вуaль, semitranslucent – кpужeвнoй, barelimbed – гoлoнoгий.Summing up, the results presented above show that Nabokov exhibited a strongtendency towards normalization while translating creative hapax legomena intoRussian. On the other hand, the preference for explicitation (and thus for morewords) when translating 5 creative hapax legomena into Russian again providesevidence of the fact that the English language is more ambiguous and vague in itssurface forms (Comrie 1981: 31–79), which are broad in meaning, whereas Russianis more explicit semantically. It again follows that more words are required inRussian to render apt and faithful translation of broad-meaning English words.17.5 ConclusionsThe aim of this corpus-driven comparison of English and Russian Lolita byVladimir Nabokov was to identify major lexical differences obtaining betweenthe two parallel texts. To that end, the study focused on a quantitative comparisonof the wordlists generated for both texts as well as on a qualitative examination ofthe translation strategies used by Nabokov while translating creative hapaxlegomena from English into Russian. At the same time, the study revealed majortypological differences between English and Russian, which exerted a profoundinfluence on the characteristics of the Russian translation.The analysis of word frequency lists revealed that top-frequency words in ENLaccount for a higher percentage of the overall number of running words in the saidtext, as compared with RUL. As a result, ENL has more repetitions and lowerlexical variety in terms of top-frequency words, while RUL (and Russian texts ingeneral) has fewer repetitions and higher lexical variety among top-frequencywords. This phenomenon results not from the subjective influence of thetranslator’s activity, but it is due to the typological differences between Russianand English morphology – where English employs articles, prepositions orparticles, Russian employs inflectional endings or affixes, which are bound withthe base forms of lexical words and thus constitute individual word types featuredin the frequency list in lower positions.

17 Crossing the Frontiers of Linguistic Typology 239The analysis of the lexical wordlists (with top-frequency grammatical wordsdeleted) revealed more lexical and stylistic features of the translation. Whiletranslating reporting verbs, Nabokov used either synonymous reporting verbs,employed more specific and concrete verbs, or resorted to structural changes,omissions or condensing. Using more concrete synonyms indicates that theseEnglish reporting verbs are not semantically determined and it is only contextwhich allows for the specification of their exact meaning. This phenomenon isdue to another typological difference whereby the Russian language is moreexplicit semantically (i.e. words have more specific meaning distinctions) thanEnglish, which in turn is more ambiguous and vague in its surface forms (Comrie1981: 31–79). As a result, a proper translation of English reporting verbs (or broadEnglish lexemes in general) requires that more lexical (content) words be used inorder to produce a natural and acceptable text in Russian.The analysis of translation of English creative hapax legomena into Russianrevealed that Nabokov exhibited a strong tendency towards lexical normalization.More specifically, only 2 Russian translations of 15 creative hapax legomena inENL constitute creative hapax legomena in RUL. Furthermore, the study revealedthe preference for explicitation (and thus using more words) when translatingcreative hapax legomena into Russian, which again proves that the English languageis more ambiguous and vague in its surface forms (which are broad in meaning andundetermined without context), while Russian is more explicit semantically.All in all, the study revealed that lexical differences between English andRussian Lolita result form both objective (typological differences between thelanguage systems) and subjective factors (translation strategies). As a result,although translators are free to use multifarious lexical, syntactic or pragmatictranslation strategies, they are still to a large extent dependant on the typologicalcharacteristics of source and target languages.ReferencesBeaujour, E. 1995. Translation and self-translation. In The Garland companion to VladimirNabokov, ed. V. Alexandrov, 714–724. New York: Garland Publishing Inc.Berber-Sardinha, T. 1996. Applications of WordSmith KeyWords. Liverpool Working Papers inApplied Linguistics (LWPAL) 2: 1–9.Boyd, B. 1993. Vladimir Nabokov. The American years. Princeton: Princeton University Press.Boyd, B. 1995. Chronology of Nabokov’s life and works. In The Garland companion to VladimirNabokov, ed. V. Alexandrov, 37–43. New York: Garland Publishing Inc.Comrie, B. 1981. Language universals and linguistic typology: Syntax and morphology. Oxford:Basil Blackwell.Ginter, A. 2003. S´wiat za słowami Vladimira Nabokova. Zabawy słowne i ich przekład. Łódź:Wydawnictwo Uniwersytetu Łódzkiego.Grabowski, Ł. 2008. Nabokov’s theory of translation and its influence on the Russianization of‘Lolita’. Studia i szkice slawistyczne IX: 123–132.Grayson, J. 1977. Nabokov translated: A comparison of Nabokov’s Russian and English prose.Oxford: Oxford University Press.

240 Ł. GrabowskiKenny, D. 2001. Lexis and creativity in translation. A corpus-based study. Manchester: St. JeromePublishing.Nabokov, V. 1955. Lolita. Paris: Olympia Press.Nabokov, V. 1989. Lolita: A novel. Moscow: Izvestia [Haбoкoв, B. B. 1989. Лoлиma.Poмaн.Mocквa: Извecmия].Nabokov, V. 2000. Lolita. London: Penguin Books.Osimo, B. 1999. Nabokov’s self-translations: Interpretations, problems and solutions in Lolita’sRussian version. Sign System Studies 27: 213-233.Ozhegov, S. and Shvedova, N. 2000. Great reference dictionary of Russian. Moscow: Nauka[Oжeгoв, C., Швeдoвa H. 2000. Toлкoвый Cлoвapь pуccкoгo языкa. 4-e издaниe,дoпoлнeннoe. Mocквa: Haукa].Piotrowski, T. 1994. Problems in bilingual lexicography. Wrocław: Wydawnictwo UniwersytetuWrocławskiego.Proffer, E. 1970. Nabokov’s Russian readers. In Nabokov: Criticism, reminiscences, translationsand tributes, ed. A. Appel and Ch. Newman, 266-283. New York: Clarion.Scott, M. 2004. Wordsmith Tools, version 4. Oxford: Oxford University Press.Stiller, R. 1991. Lolita jako gra i paradoks. In Lolita: powieść, V. Nabokov, 419-439. Warszawa:Państwowy Instytut Wydawniczy.Trzeciak, J. 2005. Nabokov as self-translating author. Unpublished PhD dissertation, Universityof Chicago.