The Processing of singular and plural nouns in French and English ...

The Processing of singular and plural nouns in French and English 1The Processing of singular and plural nouns in French and EnglishBoris New 1 , Marc Brysbaert 1 , Juan Segui 2 , Ludovic Ferrand 2 , Kathy Rastle 12 CNRS and Université René Descartes, Paris, France1 Royal Holloway, University of LondonThis research was supported by a post-doctoral grant form the Fondation Fyssen to the firstauthor and a British Academy Grant to the second one. Correspondence should be addressedto B. New, Department of Psychology, Royal Holloway, Egham Surrey, TW20 0EX (e-mail:boris.new@rhul.ac.uk).

The Processing of singular and plural nouns in French and English 2AbstractContradictory data have been obtained about the processing of singular and plural nouns inDutch and English. Whereas the Dutch findings pointed to an influence of the cumulativefrequency of the singular and the plural word forms on lexical decision times, the English datafound evidence for an influence of the frequency of the presented word form only. The Dutchfindings have been interpreted within a dual-route framework, according to whichmorphologically complex words are processed in parallel on the basis of whole-wordrecognition and decomposition. The English data have been interpreted as evidence for a fullstoragemodel, which states that morphologically complex words are always recognized onthe basis of whole-word recognition, even when the morphological decomposition of thestimulus involves a transparent and productive rule. To settle the contradiction, we firstexamined the issue in the French language, and then we reassessed the English evidence. Onthe basis of our findings, we conclude that the similarities among the languages are greaterthan the differences, and that the data are more in line with the Dutch pattern than with thepreviously reported English pattern. This rules out the full-storage model as a viable accountof how singular and plural noun forms are recognized.Keywords: Word Recognition, Lexical Decision, Inflectional morphology, Dual-route model,Number

The Processing of singular and plural nouns in French and English 3The Processing of singular and plural nouns in French and EnglishMany words used in daily life are variants of other words, either obtained by acombination of two words (compound words; e.g., blackberry, snowman), or by adding anaffix (a prefix or a suffix) to a previously unaffixed word (e.g., unclean, distrust; cleaner,trusty). An important question in the theory of reading is how such words are recognized.Traditionally, the theoretical discussion of this issue has been framed in terms of either ‘fullstorage’ or ‘decomposition’ theories: Morphologically-complex words are analyzed as wholeforms (Butterworth, 1983), or they are decomposed into their constituent morphemes (e.g.,un+clean; clean+er; Taft, 1979; Taft & Forster, 1975).Recently, it has been proposed that neither of these theoretical possibilities describesadequately how readers recognize morphologically-complex words. Rather, numerous authorshave argued that such words are recognized through a combination of both whole-word anddecomposition procedures (see e.g., Baayen, Dijkstra, & Schreuder, 1997; Caramazza,Laudanna, & Romani, 1988; Sandra, 1994; Schreuder & Baayen, 1995), with a number offactors determining the relative contribution of each pathway to the recognition of a particularword. For instance, Bertram, Schreuder, and Baayen (2000) presented a taxonomy for theprocessing of suffixed words based on three factors: word formation type, suffix productivity,and whether or not the same suffix is used in more than one type of derivation or inflection.The word formation type variable refers to the meaning relationship between themorphologically-complex word and the base word, and is considered to reside on acontinuum. At one extreme, there are morphologically complex words that do not alter themeaning of the root word. Examples of these are person and number markings of verbs (e.g.,eats) and case markings of nouns in languages such as Italian and German. At the otherextreme, there are morphologically complex words with a substantially different meaningfrom the ground word (e.g., fruitful). In-between are the morphologically complex words thatlargely mean the same as the base word, but to which some meaning has been added by thesecond morpheme (e.g., darker, shoes). The productivity of a suffix refers to the number ofcomplex words that exist with the suffix, and how easy it is to understand new words formedwith this suffix. An example of a productive suffix in English is “adjective + -ness”(alertness, bluntness, cautiousness, …, “scanableness”); an example of an unproductivesuffix is “adjective + -th” (warmth, …, “scanableth?”). Finally, the balance of whole-wordversus decomposition procedures also depends on whether a certain suffix is used in more

The Processing of singular and plural nouns in French and English 4than one type of derivation/inflection. For example in English, the end –er is used both tomake a noun from a verb (digger, looker) and to make a comparative form of short adjectives(larger, smaller). Bertram et al. (2000) hypothesized that decomposition is the most importantprocedure for words with a productive, meaning-invariant suffix that does not have aproductive rival use (e.g. “verb + -ed” in English). On the other hand, the whole-wordrecognition procedure would make the greatest contribution for suffixes that are notproductive (‘-th’; warmth), or ones that have a more frequent rival with a different semanticfunction (‘-er’; smaller, builder). The importance of whole-word recognition also increases ifthe meaning of the morphologically-complex word deviates from that of its root, even if thosecomplex words comprise productive suffixes without rival uses.The importance of each of the processing pathways is investigated by manipulating thesurface frequencies and the base frequencies of the words. The surface frequency of a wordform is the token frequency (per million) with which this particular word form appears in arepresentative corpus. The base frequency is the sum of the frequencies of all the inflectionsof a word (e.g., for a verb, it is the sum of all the forms in which the verb can be written). Thegeneral idea is that if a morphologically complex word is processed entirely on the basis of astored representation, then only the surface frequency of that word will matter. On the otherhand, if the complex word is processed only through a process of parsing, then the frequencyof the base word should be the most important. For instance, Bertram et al. (2000) looked atthe suffix –te in Dutch. In a few instances, this suffix is added to an adjective to form a noun(e.g., warm – warmte [warmth]). However, the predominant use of the suffix is to form thepast tense singular of verbs (e.g., blaf-te [bark-ed]). For the first type of words (warmte),Bertram et al. (2000) observed an effect of surface frequency only; there was no difference inword processing times due to the base frequency of the word warm. In contrast, lexicaldecision times to the second type of words (blafte) depended entirely on the base frequency ofthe verb stem and not on the surface frequency of the verb form.For most suffixed words, Bertram and colleagues postulated a contribution of bothsurface frequency and base frequency, because the storage and the decomposition route workin parallel and overlap in time. This prediction is largely based on the work of Baayen et al.(1997), who investigated the processing of Dutch singular and plural noun forms. In a firstexperiment, they kept the base frequency constant and manipulated the surface frequency.Half of the words had high-frequency single forms and low-frequency plural forms, becausethe instances to which they referred usually occur alone (e.g. bruid – bruiden [bride – brides].

The Processing of singular and plural nouns in French and English 5The other half of the words had low-frequency single forms and high-frequency plural formsbecause the instances to which they referred usually are encountered in multiples (e.g., wolk –wolken [cloud – clouds]). Baayen et al. observed that for the first type of words, lexicaldecision times took much longer to the low frequency plurals than to the high-frequencysingulars. In contrast, for the second type of words, lexical decision times took equally longfor the low-frequency singulars as for the high-frequency plurals, and were equally fast asthose to the singular forms of the singular-dominant words (see Figure 1 below). In a secondexperiment, Baayen et al. kept the surface frequency of the singular nouns constant, butmanipulated the frequency of the plural forms (and, hence, the base frequency). Theyobserved a significant effect of the base frequency on the lexical decision times to the singularword forms.To explain their findings, Baayen et al. (1997) proposed a dual-route model (see alsoBaayen, Schreuder, & Sproat, 2000; Schreuder & Baayen, 1995). This model roughly consistsof three stages. In the first stage, the visual input activates a number of stored representationsin long-term memory. Usually 1 , these include the word as a whole, but also, in parallel, thesegments within the stimulus word that form meaningful units. So, a stimulus word like dogsnot only activates the long-term memory representation of dogs, but also of do, dog, and -s.Representations that exceed a threshold value of activation are entered into a morphologicalshort-term memory buffer, which forms the basis of the second stage. In this stage, a processof licensing takes place for those segments that are shorter than the stimulus word. Thelicensing process ensures that the selected combinations of segments are as long as theoriginal stimulus word (excluding combinations like do+s), and that the combination ofselected morphemes is grammatically allowed (excluding combinations like ear+th, becauseear is not an adjective). Finally, in the last stage the syntactic and semantic features of thelicensed segments are activated. For combinations of sub-word segments, this involves thecomputation of meaning on the basis of the constituting segments.Notice that Baayen and Schreuder’s model incorporates both a whole-word “route”and a decomposition “route”. 2 The speed of the routes depends on the frequency of the wholeword on the one hand, and on the frequency of the segments on the other hand. The whole-1 Unless the stimulus word is of very low frequency.2 In the first versions of the model (Schreuder & Baayen, 1995 ; Baayen et al., 1997), the routes operatedindependently; in more recent versions (Baayen et al., 2000), the routes are no longer functionally separated.They make use of the same "machinery", as has been described here.

The Processing of singular and plural nouns in French and English 6word route will be faster for a high-frequency plural like “clouds” than for a low-frequencyroute like “brides”. The speed of the decomposition route depends on the frequency of theconstituting segments (e.g., cloud and -s, bride and -s) and on the time costs for segmentation,licensing, and composition. Also, notice that the lexical representation of a singular noun isactivated not only when the singular word form is presented, but also in the processing of theplural word form (because the processing of clouds entails the activation of the cloud+s).These two principles explain, according to Baayen and colleagues, their pattern of resultsobtained for singular and plural noun forms (and by extension for all other inflected andderived words). Because singular forms are activated upon seeing both the singular and theplural form, their processing times should be a function of the base frequency. In contrast,processing times of plural forms should differ as a function of the base frequency, the surfacefrequency, and the parsing cost for the affix. High-frequency plurals are easier to retrieve viathe storage route, and therefore their processing times will largely be sensitive to the surfacefrequency. In contrast, low-frequency plurals have more chances of being recognized via thedecomposition route and, thus, the RTs to them will be more sensitive to the base frequency +the parsing time.The work by Bertram et al. (2000) and Baayen et al. (1997) is nearlyexclusively based on Dutch and Finnish findings (Baayen, Burani, & Schreuder, 1996,presented some data on Italian). This may be a problem, because the only English study onthe processing of singular and plural nouns seems to contradict both the taxonomy proposedby Bertram et al. (2000) and Baayen et al.'s (1997) dual-route model. This study waspublished by Sereno and Jongman (1997) and contained Baayen et al.’s (1997) two basicexperiments. In the first experiment, Sereno and Jongman presented words that were muchmore frequent in the singular form than in the plural form (e.g., island) and words that weremuch more frequent in the plural from than in the singular form (e.g., product). They failed tofind an effect of the base frequency, but instead reported an effect of surface frequency. Thatis, lexical decision times to high-frequency singulars were faster than lexical decision times tolow-frequency singulars, and lexical decisions to high-frequency plurals were faster thanlexical decision times to low-frequency plurals. In the second experiment, Sereno andJongman used words with a constant frequency in singular but different frequencies in plural.Contrary to Baayen et al. (1997), they failed to obtain an effect of the plural frequency (andhence the base frequency) on the lexical decision times, and concluded that their data wereevidence against a dual-route model but fully in line with a full storage model, according towhich lexical decision times to plurals in English are based on the retrieval of holisticrepresentations. These findings contradict Bertram et al.’s (2000) taxonomy for suffixed

The Processing of singular and plural nouns in French and English 7words, because adding the suffix –s to a noun is an extremely productive way of formingplurals in English, does not dramatically change the meaning of the word, and does not haveto compete with a higher frequency alternative use of the suffix (the other productive use of –s is limited to the second form present of verbs). So, according to the taxonomy, Englishplurals should be processed predominantly by parsing, not by word form retrieval. Sereno andJongman’s findings also question Baayen et al.’s (1997) dual-route model, not only becausethere is little evidence for a decomposition route, but also because the lexical decision timesto singulars do not seem to depend on the base frequency (which they should if the singularform is co-activated upon seeing the plural form).Below, we address the contradiction between the Dutch and the English data by firstlooking at the effects in the French language (Experiments 1 and 2), to see which pattern ofresults generalizes better to a new language. Because our French findings were in line withthose of the Dutch language, we then re-assessed the evidence in English by repeating Serenoand Jongman using better stimuli and more adequate research designs (Experiments 3 and 4).EXPERIMENT 1Given the results obtained in Dutch by Baayen, Dijkstra, and Schreuder (1997) and theconflicting results in English reported by Sereno and Jongman (1997), we designedExperiment 1 to assess the importance of surface frequency in the French language.Specifically, Experiment 1 examined the contribution of the surface frequency to theprocessing of singular and plural noun forms. We composed two lists of words with the samebase frequency but with different surface frequencies for the singular and the plural forms.Half of the stimuli were singular dominant, meaning that the frequency of the singular formwas higher than that of the plural form; the other half of the stimuli were plural dominant.The stimuli were presented to the participants either in the uninflected singular form or in theinflected plural form. The experimental task was lexical decision.French plurals have the following characteristics. The end morpheme –s is extremelyregular and productive (more than 98% of plural adjectives and names end in –s). It has acontender in some verb endings (in particular the second person singular; tu manges [you

The Processing of singular and plural nouns in French and English 8eat]), but the frequency of this rival is much lower. According to Bertram et al. (2000), thesecharacteristics imply that French plurals should predominantly be computed on-line ratherthan retrieved as a whole from memory. Another particularity of French plurals is that theyare not phonologically marked (i.e., singulars and plurals are homophones), contrary toEnglish and Dutch.METHODParticipantsThirty-two students from the Université René Descartes, Paris V, took part in the experimentin return for course credits. They were all native French speakers and had normal orcorrected-to-normal vision.Stimulus MaterialsThe stimuli were 48 nouns drawn from the database Lexique 3 (New, Pallier, Ferrand, & Matos; 2001), which is a newly created database of French word forms with accompanyingfrequencies based on a corpus of written texts (31 million word tokens). Infectional studies inthe French language were not possible before the release of this database, because the existingdatabases lacked frequencies for inflected forms. In this and all subsequent experiments,frequency is reported as the number of appearances per million. Special care was taken toselect only those words for which the singular and the plural did not exist as other word forms(e.g., as inflections of a verb, as in danse [dance]), and for which the singular and the pluralwere the only possible realizations (e.g., some nouns exist in a male and female form, as inchien, chienne [dog]). In addition, for each word the plural consisted of the orthographic formof the singular with the end-morpheme –s (e.g. nuage-nuages [cloud-s]).The first list consisted of 24 words of which the singular form was more frequent than theplural form (hence called singular dominant items). The mean frequency of the singular andthe plural forms were respectively 47 and 15 per million. The second list of 24 wordsconsisted of plural dominant items, with an average frequency of 15 for the singular form and41 for the plural form. The base frequencies (i.e., the cumulative frequency of the two forms)3 This database is available at the following website: http://www.lexique.org

The Processing of singular and plural nouns in French and English 9did not differ significantly between the lists (List 1 = 62, List 2 = 56; t=0.62; p>0.1). Stimuliwere also matched for the number of letters (6.6 and 6.6) and the number of syllables (1.8 and1.8). A complete list of the stimuli is presented in Appendix A. Two versions of the word listswere made. Half of the words had their singular form in one version and the plural in theother; for the other half the assignment was reversed.In addition, 48 nonword stimuli were created from French words by replacing a singleconsonant with another consonant, or a single vowel with another vowel. These nonwordswere phonotactically legal, and were matched to the word stimuli in terms of number ofletters (6.6) and number of syllables (1.8). Half of the nonwords terminated on –s to match theplural word forms that were presented.ProcedureParticipants were tested individually in a soundproof room. They were asked to indicate asquickly and accurately as possible whether the presented letter string formed an existingFrench word or not. They did so by pressing one of two buttons of a joypad "LogitechWingman Extreme". Each trial began with a 200ms fixation cross (a plus sign in the center ofthe screen), followed by the stimulus which remained visible until the participant responded(with a maximum time period of 4 s). Between trials, there was a 1s black screen interval.Each participant saw one of the two word list versions (counterbalanced across participants).The stimuli were randomized anew for each participant and presented with the use of DMDX(Forster et Forster, 2003) on a Pentium 166. The test items were preceded by twenty practicetrials.RESULTSTable 1 shows the mean reaction times and percentages of errors, as a function of word type(singular dominant vs. plural dominant) and as a function of the word form presented(singular vs. plural). Response times of more than two standard deviations above or below themean were discarded as outliers. Thus 6.2% of the data in the subjects analysis as well as5.2% in the item analysis were discarded. Because the error rates were low and fully in linewith the RTs, they were not analyzed separately. A conventional significance level of .05 wasused unless otherwise indicated.

The Processing of singular and plural nouns in French and English 10ANOVAs on the RTs of the correct responses returned a significant main effect of word form(singular vs. plural; F1(1,31)=7.48, MSe = 732.84; F2(1,46)=6.24, MSe = 981.2767), and asignificant interaction between word type and word form (F1(1,31) = 5.46, MSe = 981.27;F2(1,46)=6.57, MSe = 1339.22). Statistics are not needed to see that this interaction was dueto the longer RTs in the condition where participants had to respond to the plural form of asingular dominant noun.DISCUSSIONThe basic question addressed by Experiment 1 was to what extent lexical decisiontimes to singular and plural noun forms are determined by the surface frequency of the formwhen the base frequency is controlled . Table 1 shows that the data are completely in line withBaayen et al.’s (1997). For singular dominant items, a reliable difference was observedbetween the singular and the plural forms, whereas for plural dominant items, no significantdifference was obtained. In addition, the RTs to singular nouns did not differ as a function ofthe word type (singular dominant or plural dominant). These findings are in line with thehypothesis that RTs to singular nouns are a function of the base frequency of the noun,whereas RTs to plural nouns partly depend on the surface frequency of the word form. In theGeneral Discussion section, we will assess more in detail the relative importance of thedecomposition and the storage route within Baayen et al's (1997) dual-route framework. First,however, in Experiment 2 we investigated whether reaction times to singular nouns in theFrench language are influenced by the base frequency of the nouns.EXPERIMENT 2After having investigated the effects of surface frequency in our first experiment, weassessed the contribution of the cumulative frequency on the processing of singular noun

The Processing of singular and plural nouns in French and English 11forms. Therefore, we looked at the lexical decision times to singular French nouns that hadthe same surface frequency but different base frequencies (because the frequency of the pluralform was high or low).METHODParticipantsFifteen new students from the Université René Descartes, Paris V, took part in the experimentin return for course credits. They were native French-speakers and had normal or correctedto-normalvision.Stimulus MaterialsForty-four words were selected from Lexique and 44 matching nonwords were constructed.Selection and construction criteria were the same as in Experiment 1, except for thefrequencies of the word forms. One list of 22 words had a singular frequency of 16, and aplural frequency of 43; the other list had a singular frequency of 16, and a plural frequency of4. The two lists of words were matched on the number of letters (6.4 and 6.3) and the numberof syllables (1.7 and 1.7). A complete list of the words is given in Appendix B.ProcedureThe procedure was identical to that described in Experiment 1, except that in this experimentonly the singular word forms were presented. Because of this, no non-word ended in –s either.RESULTSTable 2 shows the mean reaction times and percentages of errors. Extreme reaction timeswere removed according to the procedure described in Experiment 1. Thus 5.5% of the data inthe subjects analysis as well as 4.7% in the item analysis were discarded. ANOVAs with onerepeated measure revealed a main effect of the frequency of the plural form both in theanalysis over participants (F1(1,14)=24.09, Mse = 946.06) and in the analysis over itemsF2(1,21)=19.57, Mse = 1371.50). Participants responded faster to singular word forms withhigh-frequency plurals than to singular word forms with low-frequency plurals.

The Processing of singular and plural nouns in French and English 12DISCUSSIONThe main finding of experiment 2 was the presence of a base frequency effect whenthe singular forms were matched in surface frequency. When two singular forms have thesame surface frequency but differ in the frequency of their plural forms, the singular with themore frequent plural is processed faster. This result agrees with Baayen et al.’s findings inDutch, but deviates from Sereno and Jongman's findings in English.So, on the basis of the two experiments reported thus far, it seems that Dutch andFrench plurals are processed in the same way, and both differ significantly from the findingsin English. In addition, there is some suggestive evidence that the Dutch/French pattern couldalso be present in Italian (Baayen et al., 1996) and in Spanish (Dominguez, Cuetos, & Segui,1999), making the English finding even more isolated. Therefore, we decided to repeat theSereno and Jongman experiments.EXPERIMENT 3A closer look at Sereno and Jongman (1997) revealed a number of methodologicaldifferences between that study and all the other studies. For a start, Sereno and Jongmanpresented their singular and plural stimuli in different experiments (Experiments 2a and 2b).This blocked presentation may have encouraged participants to ignore the end –s in theexperiment with the plural stimuli. Another problem is that Sereno and Jongman usednonwords that were quite dissimilar to words. Finally, their word frequencies were based onthe Brown corpus which only includes one million words. This is a quite limited corpus if wecompare it to the French corpus used in Lexique (31 millions of tokens) and the Englishcorpus used in Celex (16.6 millions of tokens). For these reasons, we decided to repeat theSereno and Jongman experiments, following the same procedures used in our French studies(and in the Dutch studies).

The Processing of singular and plural nouns in French and English 13METHODParticipantsThirty-four students from Royal Holloway, University of London, took part in the experimentin return for course credits. They were all native English-speakers and had normal orcorrected-to-normal vision.Stimulus Materials.The word stimuli were two lists of 25 nouns drawn from the database Celex (Baayen,Piepenbrock, & van Rijn 1993), based on a corpus of 16.6 million word tokens of writtenEnglish. The first list consisted of singular dominant items, with an average frequency of 39per million for the singular forms and 14 for the plural forms. The second list consisted ofplural dominant items with average frequencies of 15 and 39 respectively. The stemfrequencies (53 vs. 54) did not differ between the lists. The stimuli were further matched onthe number of letters (6 and 6) and the number of syllables (2 and 2). A complete list of thestimuli is presented in Appendix C. As in Experiment 1, two versions of the word list werecreated, so that each participants saw only one form of a word.In addition, 50 nonword stimuli were created from English words by replacing a singleconsonant with another consonant, or a single vowel with another vowel. The nonwords werephonotactically legal, and were matched to the word stimuli in terms of mean number ofletters (6.6) and number of syllables (1.8). Half of the nonwords ended on –s.ProcedureStimulus presentation was the same as in Experiment 1, except that an external buttonresponse box was used for response collection.RESULTSTable 3 shows the mean reaction times and percentages of errors. Extreme reaction timeswere removed by the procedure followed in the first experiment. Thus 5.5% of the data in thesubjects analysis as well as 5.3% in the item analysis were discarded. We also removed oneitem in each list that led to extremely long RTs. (institution and examination). Two-wayANOVAs revealed a main effect of word form in the analysis by items (F1(1,33)=2.75, MSe

The Processing of singular and plural nouns in French and English 14= 4996.97; F2(1,46)=14.32, MSe = 904.27), and a significant interaction between word typeand word form (F1(1,33)=8.76, MSe = 1319.45; F2(1,46)=12, MSe = 904.27). No main effectof word type (singular dominant vs. plural dominant) was found (F1(1,33)=2.38, MSe =1465.50; F2(1,46)= 0.03, MSe = 6912.34). Planned comparisons indicated a significantdifference in the RTs to the singular noun forms between the singular dominant and the pluraldominant words in the analysis over participants only (F1(1,33)=8.29, MSe = 1674.27). Thisdifference was not reliable over items (F2(1,46)=2.23, MSe = 3247.41). There was asignificant difference between the singular and the plural forms for the singular dominantitems (F1(1,33)=10.08, MSe = 2506.61; F2(1,23)=25.04, MSe = 948.82) but not for the pluraldominant items (F1(1,33)

The Processing of singular and plural nouns in French and English 15EXPERIMENT 4Sereno and Jongman (Experiments 3a and 3b) failed to find an effect of basefrequency on lexical decision times to singular nouns in English. However, a closer look attheir stimuli reveals a possible difficulty. They used singular forms with high frequencies (onaverage 95 occurrences per million). This contrasts with the Dutch and the French studies,which were based on medium frequency items (10 to 15 occurrences per million). It ispossible that Sereno and Jongman did not find a base frequency effect, because their highfrequency singular items were already close to the ceiling level and could not profit verymuch from the additional activation due to the plurals. Therefore, we selected two lists ofsingular words with a medium surface frequency and with highly different plural frequencies.METHODParticipantsNineteen new students from Royal Holloway, University of London, took part in thisexperiment. They were paid £5 for their participation. All participants were native Englishspeakers and had normal or corrected vision.MaterialsThe stimuli were 48 nouns drawn from Celex, based on a corpus of 16.6 million word tokensof written English (Baayen, Piepenbrock, & van Rijn 1993). The first list consisted of 24singular nouns with a high frequency plural (frequencies of 15 and 39 for singular and pluralrespectively). These items except one were the same as those used as in the singular dominantcondition of Experiment 3. The second list consisted of 24 singular nouns with a lowfrequency plural (frequencies of 16 and 1.9). The lists were matched for number of letters (6and 6.2) and number of syllables (2 and 1.9). A complete list of the stimuli is presented inAppendix D. Again, a list of 48 matching nonwords was created along the criteria outlined inthe previous experiments.

The Processing of singular and plural nouns in French and English 16ProcedureThe procedure was identical to that described in Experiment 2.RESULTSTable 4 shows the mean reaction times and percentages of errors. Extreme reactiontimes were removed by the procedure followed in the Experiment 1. A one-way ANOVAwith repeated measures revealed a main effect of the frequency of the plural form(F1(1,18)=13.39, MSe = 479.61; F2(1,23)=6.67, MSe = 1350.82). Subjects reacted 26 msfaster to singular forms with high-frequency plurals than to singular forms with lowfrequencyplurals. Subject and item analyses were also conducted for the error data. Nosignificant differences were found for the effect of the plural.DISCUSSIONIn this experiment we showed that in English, lexical decision times to singular nounsare affected by the frequency of the plural forms, as previously shown in Dutch and in French.This adds credit to our reservation about Sereno and Jongman's findings, which were based onhigh-frequency singular nouns. On the other hand, a comparison of Tables 2 and 4 suggeststhat the effect of base frequency may be stronger in French (56 ms) than in English (26 ms),given that the frequency ranges of the stimuli in both languages were nearly the same. On theother hand, because it is based on a between-items and between-participants analysis, thisfinding should be treated with extreme caution. Another interesting aspect of Experiment 4 isthat the decision latencies to the low-frequency items with high frequency plurals (522 ms)were considerably faster than the latencies in Experiment 3 (555 ms) to the same items. Thisadds further credit to our assertion in Experiment 3 that the difference in the singularconditions between the singular-dominant and plural-dominant words is likely not to besignificant.

The Processing of singular and plural nouns in French and English 17GENERAL DISCUSSIONThe present study was set up to further investigate how printed singular and pluralnouns are recognized. Previous research in Dutch (Baayen et al., 1997) suggested that lexicaldecision times to singular nouns depended on the combined frequencies of the singular andplural word forms (i.e., the base frequency). In contrast, lexical decision times to plural nounforms depended on the surface frequency of the plural forms only, although Baayen et al.(1997) favored a dual-route account, with parallel retrieval of whole forms and computationon the basis of the singular form and the suffix (see the Introduction). The Dutch findingsseemed in line with data obtained in Italian (Baayen et al., 1996) and Spanish (Dominguez etal., 1999), but not with data obtained in English (Sereno & Jongman, 1997). For thislanguage, Sereno & Jongman said that only surface frequency seemed to matter, in line with afull storage model (Butterworth, 1983).The findings reported in the present article considerably clarify the empirical evidence.First, in English and French, like in Dutch, lexical decisions to singular word forms wereinfluenced by the frequencies of the plural forms (Experiments 2 and 4). Second, in all threelanguages, reaction times to the plural forms were slower than those to the singular formswhen the nouns were singular dominant (i.e., had a higher frequency in singular than inplural; Experiments 1 and 3). Third, in all languages, reaction times to the plural forms werenot significantly different from those to the singular forms when the nouns were pluraldominant (Experiments 1 and 3).The data of Experiments 1 and 3 are depicted in Figure 1, together with those ofBaayen et al. (1997; in Dutch) and Jongman and Sereno (1997; in English). In each part of thefigure, we see the same pattern emerging. For the singular-dominant words, there is asubstantial difference in decision times between the singular and the plural forms. In contrast,for the plural-dominant nouns, there is no difference. The only deviation between English onthe one hand, and French and Dutch on the other hand, is the position of the plural-dominantwords relative to that of the singular-dominant words. Whereas in Dutch and French, reaction

The Processing of singular and plural nouns in French and English 18times to the plural-dominant words are as fast as those to the singulars of the singulardominantwords, in English the RTs are slightly elevated, so that the reaction times to theplural-dominant words fall in-between those to the singular forms and the plural forms of thesingular-dominant words. As we indicated in the discussion section of Experiment 3, for theinterpretation of this finding, it is important to keep in mind that the relative position of thetwo lines in each panel of Figure 1 is the weakest aspect of the experimental design, becauseit is based on a between-stimuli comparison. This became apparent in our own experiment,where the difference between the singular forms failed to reach significant in the analysis overstimuli, despite the fact that it is the largest difference found in the four studies summarized inFigure 1.Having cleared the empirical inconsistencies, we are now in a better position to look atthe theoretical implications. Because our data are in line with those of Baayen et al. (1997),they can easily be accounted for by the dual-route model proposed in that article (see alsoBaayen et al., 2000; Schreuder & Baayen, 1995). According to this model, singular nouns arealways recognized by the whole-word recognition route, and lexical decision times to themare a function of the base frequency (i.e., the cumulative frequencies of the singular and theplural forms). The lexical decision times to the plural forms are determined by the faster oftwo possible routes 4 . The first route is the decomposition route. In this route, reaction timesequal the reaction time to the singular form, increased by a time constant needed to segmentthe stimulus input, license the combination of segments, and compute the meaning on thebasis of the singular and the suffix (together summarized under the term "parsing cost"). Thesecond route is the whole-word recognition route. Here, reaction times (RTs) depend on thesurface frequency of the plural word form. An important question within the dual-route modelis how much each route contributes to the recognition of plural nouns. This can easily be4 A problem in reviews of models of visual word recognition, is that in recent years a transition is happeningfrom horse-race models to activation-based models. In horse-race models, the faster route determines the output.In activation-based models, both routes always contribute to the output, because one route is not faster than theother (both make use of the same processing cycle). In these models, the contribution of a route depends on theamount of activation it adds to the output units per processing cycle. A similar transition is taking place inBaayen and Schreuder's thinking (e.g., compare Baayen et al., 2000, to Baayen et al., 1997). However, becausethe model consists of three, largely serial, stages, the horse-race model can still be used as a roughapproximation.

The Processing of singular and plural nouns in French and English 19estimated, as we will show for the data of Experiment 1 (French findings).If we first look at the whole-word recognition route, the model says that (1) RTs tosingular forms will be the same for the singular dominant nouns and the singular pluraldominant nouns (because their base frequencies were matched; both around 59 per million),(2) RTs to the high-frequency plurals of the plural dominant nouns will be slightly longer thanthose to the singular forms, because the average surface frequency of the plural forms (41) isslightly lower than the base frequency, and (3) RTs to the low-frequency plurals of thesingular dominant nouns will be substantially longer than those to the singular forms, becausetheir average surface frequency (15) is much lower than the base frequency.The estimates of the whole-route RTs are easy for the singular nouns, because theseRTs are assumed to be due to the storage route alone. Table 1 informs us that these RTs (forsingular nouns with a base frequency of 59) approximately form a normal distribution with amean of 547 ms and a standard deviation of 40 ms. This makes that most of the data will fallbetween 467 ms (mean minus two standard deviations) and 627 ms (mean plus two standarddeviations). The estimates of the whole-route RTs for the plural forms are slightly moredifficult to obtain, because the data in Table 1 are a mixture of whole-word recognition anddecomposition. Therefore, we cannot use these data to get a reasonable estimate of thefrequency effect due to the storage route alone. Such information, however, can be obtainedfrom Table 2 (Experiment 2). Here we see that RTs to singular nouns with a base frequencyof 20 (596 ms) are 56 ms longer than the RTs to singular nouns with a base frequency of 59(540 ms). Because the RTs in Table 2 are based on singular word forms, they are completelydue to the storage route, so that the time difference of 56 ms can be considered as a reasonableestimate of the frequency effect in the whole-word recognition route (at least for a frequencydifference between 59 and 20 per million). So, by combining the results of Experiments 2 and1, and by assuming that the effects of base frequency are the same as those of surfacefrequency (as Baayen and colleagues do), we can conclude that if in Experiment 1 we hadpresented plural nouns with a surface frequency of 20 per million, we would have expectedthe storage route to result in a normal distribution of RTs with a mean of 547 ms + 56 ms =603 ms, and a standard deviation of 40 ms. The single next step to make then, is to rescale thefrequency effect from the low frequency of 20 used in Experiment 2, to the low frequencies of15 and 41 used in Experiment 1. Assuming a logarithmic frequency function, this gives thefollowing average values: For the plural forms with a surface frequency of 15, we get an

The Processing of singular and plural nouns in French and English 20estimate of 547 ms + 71 ms 5 ; and for the plural forms with a surface frequency of 41, we getan estimate of 547 ms + 19 ms. Assuming equal standard deviations in all conditions 6 , we getthe RT distributions shown in Table 5.The predictions in Table 5 can be compared with the obtained data of 546, 548, 574,and 546 ms respectively. In the dual-route model, the differences between the predicted andthe obtained values for the plural forms come from the second, decomposition route, whichroughly will result in a normal distribution of RTs with mean equal to 547 + total parsingcost, and a standard deviation of 40 as well. With very small values of the estimated parsingcost, the RT distribution of the decomposition route will be nearly the same as the one for thesingular forms. With very high values of the estimated parsing time, the RT distribution of thedecomposition route will be so high that it will never be faster than the storage route. Simplesimulations allow us to search for a value of the estimated parsing cost that is in line with theempirical data. Table 6 shows the effects of different values on the estimates RTs. They arebased on 10,000 random values from the normal distributions defined for each route.As can be seen in Table 6, the empirical data of Experiment 1 are capturedbetter when in addition to the whole-word recognition route, the model includes adecomposition route with a parsing cost of some 25-30 ms. The decomposition route is fasterin 80% of the instances for nouns with a low-frequency plural, and in 45% of the cases fornouns with a high-frequency plural. The strong impact of the decomposition route agrees withthe fact that the –s morpheme is a productive morpheme to pluralize nouns in French, withouta higher-frequency competitor. 75 The added time due to the lower frequency is estimated with the equation:log( 59)− log( 15)added _ time =x56log( 59)− log( 20)6 The constant value of SD is clearly a simplifying assumption, because in RT data higher means are alwaysaccompanied by higher SDs, as can easily be verified in Tables 1-4.7 At the same time, Table 6 reveals a weakness in Baayen et al.'s (1997) model based on the horse-racemetaphor. When parsing costs are low, RTs to plural forms tend to be faster than those to singular forms and lessvariable. This is because singular forms are supposed to be processed by a single route only, whereas pluralforms are processed by the faster of two parallel routes.

The Processing of singular and plural nouns in French and English 22in morphological processing: The role of word formation type, affixal homonymy andproductivity. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26,489-511.Butterworth, B. (1983) Lexical representation. In B. Butterworth (Ed.), Language ProductionVolume 2 : Development, Writing and Other Language Processes. (pp. 257-294). London :Academic Press.Caramazza, A., Laudanna, A. et Romani, C. (1988). Lexical access and inflectionalmorphology. Cognition, 28, 297-332.Dominguez, A., Cuetos, F. & Segui, J. (1999). The processing of grammatical gender andnumber in Spanish. Journal of Psycholinguistic Research, 28, 485-498.Forster, K.I. & Forster, J.C. (2003). DMDX: A windows display program with millisecondaccuracy. Behavior Research Methods, Instruments, & Computers. 35, 116-124.New, B., Pallier C., Ferrand L., & Matos, R. (2001) Une base de données lexicales du françaiscontemporain sur internet : LEXIQUE, L'Année Psychologique, 101, 447-462.Sandra, D. (1994). The morphology of the mental lexicon Internal word structure viewedfrom a psycholinguistic perspective. Langage & Cognitive Processes, 9, 227-269.Schreuder, R. & Baayen, R.H. (1995). Modeling morphological processing. Feldman, LaurieBeth (Ed). Morphological aspects of language processing. (pp. 131-154). Hillsdale, NJ,England : Lawrence Erlbaum Associates.Sereno, J.A. & Jongman, A. (1997). Processing of English inflectional morphology. Memoryand Cognition, 25, 425-437.Taft, M. (1979). Recognition of affixed words and the word frequency effect. Memory andCognition, 7, 263-272.Taft, M. & Forster, K.I. (1975). Lexical storage and retrieval of prefixed words. Journal ofVerbal Learning and Verbal Behavior, 14, 638-647.

The Processing of singular and plural nouns in French and English 23APPENDIX

The Processing of singular and plural nouns in French and English 24Word (in English)Meanreaction timeTable A1Materials used in Experiment 1SingularStandardDeviationFrequencyMeanreaction timePluralStandardDeviationFrequencySingular High Frequencypain [bread] 574 51 62.94 544 101 3.39frère [brother] 500 71 99.58 506 72 48.9poète [poet] 544 67 40.97 515 79 14.26hôtel [hotel] 544 111 83.77 577 144 14.71verre [glass] 516 74 115.29 510 56 33.94boîte [box] 481 92 58.77 527 67 27.39salle [room] 505 79 126.52 586 117 18.97bière [beer] 518 81 22.55 585 132 4.19orage [storm] 516 42 19.94 554 150 5.03auteur [author] 486 47 63.42 599 206 38.26manche [sleeve] 551 91 29.84 552 34 14.03source [spring] 515 115 55.1 549 67 29.23armoire [cupboard] 522 68 23.42 540 107 5.55artiste [artist] 521 77 40.45 521 54 23.68plafond [roof] 508 96 29.58 539 83 3.39contrat [contract] 541 66 21.87 558 130 10.52écrivain [writer] 547 161 33.97 556 70 17.65réussite [success] 542 117 21.19 574 94 6.13comptoir [bar] 562 93 20.94 593 50 1.81immeuble [building] 589 169 28.32 548 62 16.16fauteuil [armchair] 523 99 43.94 538 117 14.32faiblesse [weakness] 575 95 22.9 555 83 6ministère [governmentdepartment] 535 106 41.23 699 109 9.68commissaire[superintendent] 545 115 28.16 613 145 5.97Plural High Frequencydent [tooth] 475 76 9 553 105 70.87doigt [finger] 498 56 46.87 495 85 99.84nuage [cloud] 533 71 19.29 533 56 39.13fleur [flower] 473 77 32.97 517 69 83.42fruit [fruit] 529 63 25.84 491 46 50.16lèvre [lip] 535 98 11.32 513 80 107.35ongle [nail] 529 67 6.06 523 52 19.58volet [shutter] 515 91 5.03 555 67 19.65organe [organ] 646 136 20.71 534 62 36.16soldat [soldier] 503 79 26.42 531 66 46.74cuisse [thigh] 501 72 11.9 505 118 25.81troupe [troop] 580 81 24.29 566 92 44.87sourcil [eyebrow] 525 54 4.45 535 60 19.87facteur [postman] 517 57 32.13 516 99 45.29lunette [glasses] 530 82 5.68 502 83 36.61recette [receipt] 527 100 7.9 609 147 19.94soulier [shoe] 549 94 2.87 554 76 17.55document [document] 520 67 18.48 509 63 43.58paupière [eyelid] 538 79 4.35 597 150 29.97touriste [tourist] 552 91 6.03 532 112 16.16vêtement [cloth] 482 65 10.84 535 76 44.13chaussure [shoe] 605 100 5 551 82 26.03particule [particle] 615 61 9.94 621 139 31.19marchandise [goods] 607 114 8 596 161 18.94

The Processing of singular and plural nouns in French and English 25Table A2Word (in English)Materials used in Experiment 2SingularMean reaction StandardPluralFrequenctime DeviationFrequencyySingular with low frequency pluraldoigt [finger] 491 52 46.87 99.84nuage [cloud] 483 58 19.29 39.13fleur [flower] 502 69 32.97 83.42fruit [fruit] 509 78 25.84 50.16lèvre [lip] 555 88 11.32 107.35gant [glower] 668 207 4.39 16.77volet [shutter] 568 87 5.03 19.65organe [organ] 549 61 20.71 36.16soldat [soldier] 545 86 26.42 46.74troupe [troop] 631 78 24.29 44.87nerf [nerve] 555 93 11.23 21.32facteur [postman] 509 33 32.13 45.29meuble [furniture] 544 88 19.23 37.32recette [recipe] 526 78 7.9 19.94larme [tear] 492 42 6.48 69.32document[document] 515 64 18.48 43.58paupière [eyelip] 599 108 4.35 29.97touriste [tourist] 538 71 6.03 16.16vêtement [cloth] 525 100 10.84 44.13chaussure [shoe] 503 74 5 26.03particule [particule] 630 136 9.94 31.19marchandise [good] 624 81 8 18.94Singular with high frequency pluralneige [snow] 507 65 48.23 4.19orage [storm] 554 107 19.94 5.03huile [oil] 532 70 32.74 3.16nuque [nape] 589 114 25.81 1.23torse [chest] 618 113 12.35 1.77mare [pool] 686 125 5.65 2.03duvet [down] 589 93 4.71 0.87bassin [basin] 572 72 20.87 6.13vallée [valley] 520 76 25.81 6.45grange [barn] 629 143 20.97 3.13tige [stem] 578 84 11.52 7.77rivière [river] 584 107 34.45 7.97prêtre [priest] 635 178 19.97 12.77impasse [dead end] 612 107 7.9 1.42larme [tear] 550 86 6.48 69.32pavillon [bungalow] 491 71 17.06 4.29perruque [wig] 536 58 4.45 1.77chandail [pullover] 665 109 5.74 1.48cercueil [coffin] 660 181 10.74 1.71sculpteur [sculptor] 625 95 5.39 2.87partition [partition] 730 137 9.94 2.97camionnette [van] 686 110 8 1.65

The Processing of singular and plural nouns in French and English 26Table A3Materials used in Experiment 3SingularPluralMeanMeanWord (inStandard FrequencStandard FrequencreactionreactionEnglish)Deviation yDeviation ytimetimeSingulararticle 571 98 41 562 63 21bag 457 57 61 534 70 19beast 536 87 17 605 127 11cabin 512 78 27 580 55 3camera 514 79 24 564 135 12college 537 77 77 566 174 26commitment 658 192 36 742 142 6crime 487 62 49 530 57 16decade 573 128 51 709 266 27explanation 606 88 42 731 223 11jail 468 49 30 578 94 2restaurant 565 121 32 618 138 14rope 491 69 31 548 45 11row 518 51 26 497 72 20salad 542 106 16 522 56 4sister 517 69 82 515 77 32studio 487 48 22 519 80 6sum 500 81 32 532 110 17symbol 507 50 23 526 67 13talent 512 66 24 534 65 12task 519 98 65 507 65 17tension 527 69 34 557 77 10trial 496 66 55 574 92 11tribe 544 106 23 563 63 15Plural High Frequencyacre 606 76 15 595 101 23ancestor 606 80 6 674 63 22boot 465 51 8 480 86 30colleague 649 170 12 636 130 39critic 569 88 12 600 87 23curtain 523 75 19 574 69 24customer 530 103 14 576 108 24dollar 534 62 15 490 58 53fee 610 90 13 572 128 19heel 538 42 11 540 96 18knee 486 70 29 522 68 54lip 520 92 17 496 69 61metre 661 104 8 538 54 27pig 465 64 18 485 44 26politician 670 48 14 724 149 41prisoner 574 111 16 575 99 32pupil 504 52 14 494 53 34resource 639 129 14 615 115 80scientist 618 91 16 624 96 46shoe 465 38 14 481 59 65sock 528 100 3 478 54 16soldier 471 54 26 495 63 57tool 466 57 16 495 58 29weapon 538 58 24 522 91 79

The Processing of singular and plural nouns in French and English 27Word (inEnglish)Table A4Materials used in Experiment 4Meanreaction timeStandardDeviationSingularFrequencyPluralFrequencySingular with low frequency pluralearl 609.69 112.68 15 1enclosure 584.56 108.32 6 1tub 534.19 121.91 8 1staircase 517.54 80.71 12 2refuge 563.77 88.24 12 1outcome 503.83 124.74 19 2receiver 552.42 104.35 14 1mayor 515.27 61.34 15 1foe 642.34 117.88 14 1flint 582.74 134.14 12 1expenditure 646.40 162.20 28 4aunt 513.78 100.42 30 4deck 508.24 74.19 19 2haven 575.39 78.30 8 1lid 550.59 139.81 14 5federation 682.79 145.61 15 1reception 519.94 90.46 18 1pencil 465.21 64.86 15 3promotion 562.96 99.88 15 2moustache 538.06 108.34 16 2oak 465.48 63.08 14 3supper 511.19 96.09 27 1stove 618.09 169.06 16 4rubber 506.06 105.83 25 1Singular with high frequency pluralacre 584.27 73.39 15 23ancestor 595.00 81.91 6 22boot 467.87 64.83 8 30colleague 569.49 92.86 12 39critic 546.88 75.66 12 23curtain 500.25 77.77 19 24customer 495.67 85.37 14 24dollar 486.65 63.38 15 53fee 539.92 105.81 13 19heel 501.66 103.59 11 18institution 697.93 178.38 25 56knee 475.45 37.23 29 54lip 477.80 91.99 17 61metre 641.46 131.64 8 27pig 486.64 68.33 18 26politician 584.51 93.08 14 41prisoner 473.61 70.85 16 32pupil 453.18 67.27 14 34resource 561.62 86.10 14 80scientist 564.47 158.32 16 46shoe 451.48 62.57 14 65soldier 505.25 89.77 26 57tool 461.29 60.78 16 29weapon 489.87 103.02 24 79

The Processing of singular and plural nouns in French and English 28TablesTable 1. Mean Reaction Times (in ms), Standard Deviations, and Percentages of Errors inExperiment 1Singular DominantEx: Plafond [Ceiling]Plural DominantEx: Nuages [Cloud]Presented form:singularPresented form:pluralM SD %ER M SD %ER546 26 2.3% 574 45 3.9%548 37 2.0% 546 32 2.9%Table 2. Mean Reaction Times (in ms), standard deviations and percentages of Error inExperiment 2Frequency of thecomplementary formHigh Frequency PluralEx: Ongle [Nail]Low Frequency PluralEx: Frère [Brother]Presented forms: singularM SD %ER540 50 2.1%596 63 3.6%Table 3. Mean Reaction Times (in ms), standard deviations and percentage or Errors inExperiment 3SingularDominantEx: bagPlural DominantEx: lipPresented form: singular Presented form: pluralM SD %ER M SD %ER527 63 2.1 565 59 4.4555 67 2.7 557 60 4.2

The Processing of singular and plural nouns in French and English 29Table 4. Mean Reaction Times (in ms), standard deviations and percentage or Errors inExperiment 4Frequency of thecomplementary formHigh Frequency PluralEx: HeelLow Frequency PluralEx: FlintPresented forms: singularM SD %ER522 56 5.4548 70 7.2Table 5: Predicted RTs using Baayen et al.'s storage route for Experiment 1 items.SingularDominantPresented form: singularPresented form: pluralFrequency RT Range Frequency RT Range59 547 467-627 15 618 538-698Plural Dominant 59 547 467-627 41 566 486-646Table 6 : Simulated data for decision latencies to singular and plural word forms when adecomposition route is added to the storage route with different values of the parsing time(PT). First, the average RT is reported; then the standard deviation (between brackets). For theplural forms, we also indicate how often the decomposition route was faster than the wholewordrecognition route.Parsing Time Singular PluralSingDomPlurDomPT = 0 ms 547 (40) 543 (37) 89% 533 (33) 63%PT = 25 ms 547 (40) 565 (35) 80% 546 (33) 47%PT = 50 ms 547 (40) 584 (34) 65% 556 (34) 30%PT = 100 ms 547 (40) 607 (34) 31% 564 (38) 8%PT = 150 ms 547 (40) 615 (39) 8% 565 (40) 1%PT = ∞ 547 (40) 618 (40) 0% 566 (40) 0%Observed 547 574 546

The Processing of singular and plural nouns in French and English 30FIGURES570560550540530520SingularEnglish (us)PluralSingDom(jail)PlurDom(boot)580570560550540530SingularFrenchPluralSingDom(ceiling)PlurDom(cloud)680660640620600580SingularEnglish (Sereno)PluralSingDom(jail)PlurDom(boot)620600580560540520500SingularDutchPluralSingDom(Jail)PlurDom(Boot)Figure LegendsFigure 1 : Lexical decision to singular and plural forms of singular-dominant and pluraldominantnouns in English (present study, Experiment 3; Sereno & Jongman, 1997,Experiments 2a and 2b), Dutch (Baayen et al., 1997; Experiment 1), and French (present study,Experiment 1).