The Processing of singular and plural nouns in French and English ...
The Processing of singular and plural nouns in French and English ...
The Processing of singular and plural nouns in French and English ...
- No tags were found...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 1<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong>Boris New 1 , Marc Brysbaert 1 , Juan Segui 2 , Ludovic Ferr<strong>and</strong> 2 , Kathy Rastle 12 CNRS <strong>and</strong> Université René Descartes, Paris, France1 Royal Holloway, University <strong>of</strong> LondonThis research was supported by a post-doctoral grant form the Fondation Fyssen to the firstauthor <strong>and</strong> a British Academy Grant to the second one. Correspondence should be addressedto B. New, Department <strong>of</strong> Psychology, Royal Holloway, Egham Surrey, TW20 0EX (e-mail:boris.new@rhul.ac.uk).
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 2AbstractContradictory data have been obta<strong>in</strong>ed about the process<strong>in</strong>g <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong>Dutch <strong>and</strong> <strong>English</strong>. Whereas the Dutch f<strong>in</strong>d<strong>in</strong>gs po<strong>in</strong>ted to an <strong>in</strong>fluence <strong>of</strong> the cumulativefrequency <strong>of</strong> the <strong>s<strong>in</strong>gular</strong> <strong>and</strong> the <strong>plural</strong> word forms on lexical decision times, the <strong>English</strong> datafound evidence for an <strong>in</strong>fluence <strong>of</strong> the frequency <strong>of</strong> the presented word form only. <strong>The</strong> Dutchf<strong>in</strong>d<strong>in</strong>gs have been <strong>in</strong>terpreted with<strong>in</strong> a dual-route framework, accord<strong>in</strong>g to whichmorphologically complex words are processed <strong>in</strong> parallel on the basis <strong>of</strong> whole-wordrecognition <strong>and</strong> decomposition. <strong>The</strong> <strong>English</strong> data have been <strong>in</strong>terpreted as evidence for a fullstoragemodel, which states that morphologically complex words are always recognized onthe basis <strong>of</strong> whole-word recognition, even when the morphological decomposition <strong>of</strong> thestimulus <strong>in</strong>volves a transparent <strong>and</strong> productive rule. To settle the contradiction, we firstexam<strong>in</strong>ed the issue <strong>in</strong> the <strong>French</strong> language, <strong>and</strong> then we reassessed the <strong>English</strong> evidence. Onthe basis <strong>of</strong> our f<strong>in</strong>d<strong>in</strong>gs, we conclude that the similarities among the languages are greaterthan the differences, <strong>and</strong> that the data are more <strong>in</strong> l<strong>in</strong>e with the Dutch pattern than with thepreviously reported <strong>English</strong> pattern. This rules out the full-storage model as a viable account<strong>of</strong> how <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> noun forms are recognized.Keywords: Word Recognition, Lexical Decision, Inflectional morphology, Dual-route model,Number
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 3<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong>Many words used <strong>in</strong> daily life are variants <strong>of</strong> other words, either obta<strong>in</strong>ed by acomb<strong>in</strong>ation <strong>of</strong> two words (compound words; e.g., blackberry, snowman), or by add<strong>in</strong>g anaffix (a prefix or a suffix) to a previously unaffixed word (e.g., unclean, distrust; cleaner,trusty). An important question <strong>in</strong> the theory <strong>of</strong> read<strong>in</strong>g is how such words are recognized.Traditionally, the theoretical discussion <strong>of</strong> this issue has been framed <strong>in</strong> terms <strong>of</strong> either ‘fullstorage’ or ‘decomposition’ theories: Morphologically-complex words are analyzed as wholeforms (Butterworth, 1983), or they are decomposed <strong>in</strong>to their constituent morphemes (e.g.,un+clean; clean+er; Taft, 1979; Taft & Forster, 1975).Recently, it has been proposed that neither <strong>of</strong> these theoretical possibilities describesadequately how readers recognize morphologically-complex words. Rather, numerous authorshave argued that such words are recognized through a comb<strong>in</strong>ation <strong>of</strong> both whole-word <strong>and</strong>decomposition procedures (see e.g., Baayen, Dijkstra, & Schreuder, 1997; Caramazza,Laudanna, & Romani, 1988; S<strong>and</strong>ra, 1994; Schreuder & Baayen, 1995), with a number <strong>of</strong>factors determ<strong>in</strong><strong>in</strong>g the relative contribution <strong>of</strong> each pathway to the recognition <strong>of</strong> a particularword. For <strong>in</strong>stance, Bertram, Schreuder, <strong>and</strong> Baayen (2000) presented a taxonomy for theprocess<strong>in</strong>g <strong>of</strong> suffixed words based on three factors: word formation type, suffix productivity,<strong>and</strong> whether or not the same suffix is used <strong>in</strong> more than one type <strong>of</strong> derivation or <strong>in</strong>flection.<strong>The</strong> word formation type variable refers to the mean<strong>in</strong>g relationship between themorphologically-complex word <strong>and</strong> the base word, <strong>and</strong> is considered to reside on acont<strong>in</strong>uum. At one extreme, there are morphologically complex words that do not alter themean<strong>in</strong>g <strong>of</strong> the root word. Examples <strong>of</strong> these are person <strong>and</strong> number mark<strong>in</strong>gs <strong>of</strong> verbs (e.g.,eats) <strong>and</strong> case mark<strong>in</strong>gs <strong>of</strong> <strong>nouns</strong> <strong>in</strong> languages such as Italian <strong>and</strong> German. At the otherextreme, there are morphologically complex words with a substantially different mean<strong>in</strong>gfrom 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 mean<strong>in</strong>g has been added by thesecond morpheme (e.g., darker, shoes). <strong>The</strong> productivity <strong>of</strong> a suffix refers to the number <strong>of</strong>complex words that exist with the suffix, <strong>and</strong> how easy it is to underst<strong>and</strong> new words formedwith this suffix. An example <strong>of</strong> a productive suffix <strong>in</strong> <strong>English</strong> is “adjective + -ness”(alertness, bluntness, cautiousness, …, “scanableness”); an example <strong>of</strong> an unproductivesuffix is “adjective + -th” (warmth, …, “scanableth?”). F<strong>in</strong>ally, the balance <strong>of</strong> whole-wordversus decomposition procedures also depends on whether a certa<strong>in</strong> suffix is used <strong>in</strong> more
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 4than one type <strong>of</strong> derivation/<strong>in</strong>flection. For example <strong>in</strong> <strong>English</strong>, the end –er is used both tomake a noun from a verb (digger, looker) <strong>and</strong> to make a comparative form <strong>of</strong> short adjectives(larger, smaller). Bertram et al. (2000) hypothesized that decomposition is the most importantprocedure for words with a productive, mean<strong>in</strong>g-<strong>in</strong>variant suffix that does not have aproductive rival use (e.g. “verb + -ed” <strong>in</strong> <strong>English</strong>). On the other h<strong>and</strong>, 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). <strong>The</strong> importance <strong>of</strong> whole-word recognition also <strong>in</strong>creases ifthe mean<strong>in</strong>g <strong>of</strong> the morphologically-complex word deviates from that <strong>of</strong> its root, even if thosecomplex words comprise productive suffixes without rival uses.<strong>The</strong> importance <strong>of</strong> each <strong>of</strong> the process<strong>in</strong>g pathways is <strong>in</strong>vestigated by manipulat<strong>in</strong>g thesurface frequencies <strong>and</strong> the base frequencies <strong>of</strong> the words. <strong>The</strong> surface frequency <strong>of</strong> a wordform is the token frequency (per million) with which this particular word form appears <strong>in</strong> arepresentative corpus. <strong>The</strong> base frequency is the sum <strong>of</strong> the frequencies <strong>of</strong> all the <strong>in</strong>flections<strong>of</strong> a word (e.g., for a verb, it is the sum <strong>of</strong> all the forms <strong>in</strong> which the verb can be written). <strong>The</strong>general idea is that if a morphologically complex word is processed entirely on the basis <strong>of</strong> astored representation, then only the surface frequency <strong>of</strong> that word will matter. On the otherh<strong>and</strong>, if the complex word is processed only through a process <strong>of</strong> pars<strong>in</strong>g, then the frequency<strong>of</strong> the base word should be the most important. For <strong>in</strong>stance, Bertram et al. (2000) looked atthe suffix –te <strong>in</strong> Dutch. In a few <strong>in</strong>stances, this suffix is added to an adjective to form a noun(e.g., warm – warmte [warmth]). However, the predom<strong>in</strong>ant use <strong>of</strong> the suffix is to form thepast tense <strong>s<strong>in</strong>gular</strong> <strong>of</strong> verbs (e.g., blaf-te [bark-ed]). For the first type <strong>of</strong> words (warmte),Bertram et al. (2000) observed an effect <strong>of</strong> surface frequency only; there was no difference <strong>in</strong>word process<strong>in</strong>g times due to the base frequency <strong>of</strong> the word warm. In contrast, lexicaldecision times to the second type <strong>of</strong> words (blafte) depended entirely on the base frequency <strong>of</strong>the verb stem <strong>and</strong> not on the surface frequency <strong>of</strong> the verb form.For most suffixed words, Bertram <strong>and</strong> colleagues postulated a contribution <strong>of</strong> bothsurface frequency <strong>and</strong> base frequency, because the storage <strong>and</strong> the decomposition route work<strong>in</strong> parallel <strong>and</strong> overlap <strong>in</strong> time. This prediction is largely based on the work <strong>of</strong> Baayen et al.(1997), who <strong>in</strong>vestigated the process<strong>in</strong>g <strong>of</strong> Dutch <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> noun forms. In a firstexperiment, they kept the base frequency constant <strong>and</strong> manipulated the surface frequency.Half <strong>of</strong> the words had high-frequency s<strong>in</strong>gle forms <strong>and</strong> low-frequency <strong>plural</strong> forms, becausethe <strong>in</strong>stances to which they referred usually occur alone (e.g. bruid – bruiden [bride – brides].
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 5<strong>The</strong> other half <strong>of</strong> the words had low-frequency s<strong>in</strong>gle forms <strong>and</strong> high-frequency <strong>plural</strong> formsbecause the <strong>in</strong>stances to which they referred usually are encountered <strong>in</strong> multiples (e.g., wolk –wolken [cloud – clouds]). Baayen et al. observed that for the first type <strong>of</strong> words, lexicaldecision times took much longer to the low frequency <strong>plural</strong>s than to the high-frequency<strong>s<strong>in</strong>gular</strong>s. In contrast, for the second type <strong>of</strong> words, lexical decision times took equally longfor the low-frequency <strong>s<strong>in</strong>gular</strong>s as for the high-frequency <strong>plural</strong>s, <strong>and</strong> were equally fast asthose to the <strong>s<strong>in</strong>gular</strong> forms <strong>of</strong> the <strong>s<strong>in</strong>gular</strong>-dom<strong>in</strong>ant words (see Figure 1 below). In a secondexperiment, Baayen et al. kept the surface frequency <strong>of</strong> the <strong>s<strong>in</strong>gular</strong> <strong>nouns</strong> constant, butmanipulated the frequency <strong>of</strong> the <strong>plural</strong> forms (<strong>and</strong>, hence, the base frequency). <strong>The</strong>yobserved a significant effect <strong>of</strong> the base frequency on the lexical decision times to the <strong>s<strong>in</strong>gular</strong>word forms.To expla<strong>in</strong> their f<strong>in</strong>d<strong>in</strong>gs, Baayen et al. (1997) proposed a dual-route model (see alsoBaayen, Schreuder, & Sproat, 2000; Schreuder & Baayen, 1995). This model roughly consists<strong>of</strong> three stages. In the first stage, the visual <strong>in</strong>put activates a number <strong>of</strong> stored representations<strong>in</strong> long-term memory. Usually 1 , these <strong>in</strong>clude the word as a whole, but also, <strong>in</strong> parallel, thesegments with<strong>in</strong> the stimulus word that form mean<strong>in</strong>gful units. So, a stimulus word like dogsnot only activates the long-term memory representation <strong>of</strong> dogs, but also <strong>of</strong> do, dog, <strong>and</strong> -s.Representations that exceed a threshold value <strong>of</strong> activation are entered <strong>in</strong>to a morphologicalshort-term memory buffer, which forms the basis <strong>of</strong> the second stage. In this stage, a process<strong>of</strong> licens<strong>in</strong>g takes place for those segments that are shorter than the stimulus word. <strong>The</strong>licens<strong>in</strong>g process ensures that the selected comb<strong>in</strong>ations <strong>of</strong> segments are as long as theorig<strong>in</strong>al stimulus word (exclud<strong>in</strong>g comb<strong>in</strong>ations like do+s), <strong>and</strong> that the comb<strong>in</strong>ation <strong>of</strong>selected morphemes is grammatically allowed (exclud<strong>in</strong>g comb<strong>in</strong>ations like ear+th, becauseear is not an adjective). F<strong>in</strong>ally, <strong>in</strong> the last stage the syntactic <strong>and</strong> semantic features <strong>of</strong> thelicensed segments are activated. For comb<strong>in</strong>ations <strong>of</strong> sub-word segments, this <strong>in</strong>volves thecomputation <strong>of</strong> mean<strong>in</strong>g on the basis <strong>of</strong> the constitut<strong>in</strong>g segments.Notice that Baayen <strong>and</strong> Schreuder’s model <strong>in</strong>corporates both a whole-word “route”<strong>and</strong> a decomposition “route”. 2 <strong>The</strong> speed <strong>of</strong> the routes depends on the frequency <strong>of</strong> the wholeword on the one h<strong>and</strong>, <strong>and</strong> on the frequency <strong>of</strong> the segments on the other h<strong>and</strong>. <strong>The</strong> whole-1 Unless the stimulus word is <strong>of</strong> very low frequency.2 In the first versions <strong>of</strong> the model (Schreuder & Baayen, 1995 ; Baayen et al., 1997), the routes operated<strong>in</strong>dependently; <strong>in</strong> more recent versions (Baayen et al., 2000), the routes are no longer functionally separated.<strong>The</strong>y make use <strong>of</strong> the same "mach<strong>in</strong>ery", as has been described here.
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 6word route will be faster for a high-frequency <strong>plural</strong> like “clouds” than for a low-frequencyroute like “brides”. <strong>The</strong> speed <strong>of</strong> the decomposition route depends on the frequency <strong>of</strong> theconstitut<strong>in</strong>g segments (e.g., cloud <strong>and</strong> -s, bride <strong>and</strong> -s) <strong>and</strong> on the time costs for segmentation,licens<strong>in</strong>g, <strong>and</strong> composition. Also, notice that the lexical representation <strong>of</strong> a <strong>s<strong>in</strong>gular</strong> noun isactivated not only when the <strong>s<strong>in</strong>gular</strong> word form is presented, but also <strong>in</strong> the process<strong>in</strong>g <strong>of</strong> the<strong>plural</strong> word form (because the process<strong>in</strong>g <strong>of</strong> clouds entails the activation <strong>of</strong> the cloud+s).<strong>The</strong>se two pr<strong>in</strong>ciples expla<strong>in</strong>, accord<strong>in</strong>g to Baayen <strong>and</strong> colleagues, their pattern <strong>of</strong> resultsobta<strong>in</strong>ed for <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> noun forms (<strong>and</strong> by extension for all other <strong>in</strong>flected <strong>and</strong>derived words). Because <strong>s<strong>in</strong>gular</strong> forms are activated upon see<strong>in</strong>g both the <strong>s<strong>in</strong>gular</strong> <strong>and</strong> the<strong>plural</strong> form, their process<strong>in</strong>g times should be a function <strong>of</strong> the base frequency. In contrast,process<strong>in</strong>g times <strong>of</strong> <strong>plural</strong> forms should differ as a function <strong>of</strong> the base frequency, the surfacefrequency, <strong>and</strong> the pars<strong>in</strong>g cost for the affix. High-frequency <strong>plural</strong>s are easier to retrieve viathe storage route, <strong>and</strong> therefore their process<strong>in</strong>g times will largely be sensitive to the surfacefrequency. In contrast, low-frequency <strong>plural</strong>s have more chances <strong>of</strong> be<strong>in</strong>g recognized via thedecomposition route <strong>and</strong>, thus, the RTs to them will be more sensitive to the base frequency +the pars<strong>in</strong>g time.<strong>The</strong> work by Bertram et al. (2000) <strong>and</strong> Baayen et al. (1997) is nearlyexclusively based on Dutch <strong>and</strong> F<strong>in</strong>nish f<strong>in</strong>d<strong>in</strong>gs (Baayen, Burani, & Schreuder, 1996,presented some data on Italian). This may be a problem, because the only <strong>English</strong> study onthe process<strong>in</strong>g <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> seems to contradict both the taxonomy proposedby Bertram et al. (2000) <strong>and</strong> Baayen et al.'s (1997) dual-route model. This study waspublished by Sereno <strong>and</strong> Jongman (1997) <strong>and</strong> conta<strong>in</strong>ed Baayen et al.’s (1997) two basicexperiments. In the first experiment, Sereno <strong>and</strong> Jongman presented words that were muchmore frequent <strong>in</strong> the <strong>s<strong>in</strong>gular</strong> form than <strong>in</strong> the <strong>plural</strong> form (e.g., isl<strong>and</strong>) <strong>and</strong> words that weremuch more frequent <strong>in</strong> the <strong>plural</strong> from than <strong>in</strong> the <strong>s<strong>in</strong>gular</strong> form (e.g., product). <strong>The</strong>y failed t<strong>of</strong><strong>in</strong>d an effect <strong>of</strong> the base frequency, but <strong>in</strong>stead reported an effect <strong>of</strong> surface frequency. Thatis, lexical decision times to high-frequency <strong>s<strong>in</strong>gular</strong>s were faster than lexical decision times tolow-frequency <strong>s<strong>in</strong>gular</strong>s, <strong>and</strong> lexical decisions to high-frequency <strong>plural</strong>s were faster thanlexical decision times to low-frequency <strong>plural</strong>s. In the second experiment, Sereno <strong>and</strong>Jongman used words with a constant frequency <strong>in</strong> <strong>s<strong>in</strong>gular</strong> but different frequencies <strong>in</strong> <strong>plural</strong>.Contrary to Baayen et al. (1997), they failed to obta<strong>in</strong> an effect <strong>of</strong> the <strong>plural</strong> frequency (<strong>and</strong>hence the base frequency) on the lexical decision times, <strong>and</strong> concluded that their data wereevidence aga<strong>in</strong>st a dual-route model but fully <strong>in</strong> l<strong>in</strong>e with a full storage model, accord<strong>in</strong>g towhich lexical decision times to <strong>plural</strong>s <strong>in</strong> <strong>English</strong> are based on the retrieval <strong>of</strong> holisticrepresentations. <strong>The</strong>se f<strong>in</strong>d<strong>in</strong>gs contradict Bertram et al.’s (2000) taxonomy for suffixed
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 7words, because add<strong>in</strong>g the suffix –s to a noun is an extremely productive way <strong>of</strong> form<strong>in</strong>g<strong>plural</strong>s <strong>in</strong> <strong>English</strong>, does not dramatically change the mean<strong>in</strong>g <strong>of</strong> the word, <strong>and</strong> does not haveto compete with a higher frequency alternative use <strong>of</strong> the suffix (the other productive use <strong>of</strong> –s is limited to the second form present <strong>of</strong> verbs). So, accord<strong>in</strong>g to the taxonomy, <strong>English</strong><strong>plural</strong>s should be processed predom<strong>in</strong>antly by pars<strong>in</strong>g, not by word form retrieval. Sereno <strong>and</strong>Jongman’s f<strong>in</strong>d<strong>in</strong>gs 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 <strong>s<strong>in</strong>gular</strong>s do not seem to depend on the base frequency (which they should if the <strong>s<strong>in</strong>gular</strong>form is co-activated upon see<strong>in</strong>g the <strong>plural</strong> form).Below, we address the contradiction between the Dutch <strong>and</strong> the <strong>English</strong> data by firstlook<strong>in</strong>g at the effects <strong>in</strong> the <strong>French</strong> language (Experiments 1 <strong>and</strong> 2), to see which pattern <strong>of</strong>results generalizes better to a new language. Because our <strong>French</strong> f<strong>in</strong>d<strong>in</strong>gs were <strong>in</strong> l<strong>in</strong>e withthose <strong>of</strong> the Dutch language, we then re-assessed the evidence <strong>in</strong> <strong>English</strong> by repeat<strong>in</strong>g Sereno<strong>and</strong> Jongman us<strong>in</strong>g better stimuli <strong>and</strong> more adequate research designs (Experiments 3 <strong>and</strong> 4).EXPERIMENT 1Given the results obta<strong>in</strong>ed <strong>in</strong> Dutch by Baayen, Dijkstra, <strong>and</strong> Schreuder (1997) <strong>and</strong> theconflict<strong>in</strong>g results <strong>in</strong> <strong>English</strong> reported by Sereno <strong>and</strong> Jongman (1997), we designedExperiment 1 to assess the importance <strong>of</strong> surface frequency <strong>in</strong> the <strong>French</strong> language.Specifically, Experiment 1 exam<strong>in</strong>ed the contribution <strong>of</strong> the surface frequency to theprocess<strong>in</strong>g <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> noun forms. We composed two lists <strong>of</strong> words with the samebase frequency but with different surface frequencies for the <strong>s<strong>in</strong>gular</strong> <strong>and</strong> the <strong>plural</strong> forms.Half <strong>of</strong> the stimuli were <strong>s<strong>in</strong>gular</strong> dom<strong>in</strong>ant, mean<strong>in</strong>g that the frequency <strong>of</strong> the <strong>s<strong>in</strong>gular</strong> formwas higher than that <strong>of</strong> the <strong>plural</strong> form; the other half <strong>of</strong> the stimuli were <strong>plural</strong> dom<strong>in</strong>ant.<strong>The</strong> stimuli were presented to the participants either <strong>in</strong> the un<strong>in</strong>flected <strong>s<strong>in</strong>gular</strong> form or <strong>in</strong> the<strong>in</strong>flected <strong>plural</strong> form. <strong>The</strong> experimental task was lexical decision.<strong>French</strong> <strong>plural</strong>s have the follow<strong>in</strong>g characteristics. <strong>The</strong> end morpheme –s is extremelyregular <strong>and</strong> productive (more than 98% <strong>of</strong> <strong>plural</strong> adjectives <strong>and</strong> names end <strong>in</strong> –s). It has acontender <strong>in</strong> some verb end<strong>in</strong>gs (<strong>in</strong> particular the second person <strong>s<strong>in</strong>gular</strong>; tu manges [you
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 8eat]), but the frequency <strong>of</strong> this rival is much lower. Accord<strong>in</strong>g to Bertram et al. (2000), thesecharacteristics imply that <strong>French</strong> <strong>plural</strong>s should predom<strong>in</strong>antly be computed on-l<strong>in</strong>e ratherthan retrieved as a whole from memory. Another particularity <strong>of</strong> <strong>French</strong> <strong>plural</strong>s is that theyare not phonologically marked (i.e., <strong>s<strong>in</strong>gular</strong>s <strong>and</strong> <strong>plural</strong>s are homophones), contrary to<strong>English</strong> <strong>and</strong> Dutch.METHODParticipantsThirty-two students from the Université René Descartes, Paris V, took part <strong>in</strong> the experiment<strong>in</strong> return for course credits. <strong>The</strong>y were all native <strong>French</strong> speakers <strong>and</strong> had normal orcorrected-to-normal vision.Stimulus Materials<strong>The</strong> stimuli were 48 <strong>nouns</strong> drawn from the database Lexique 3 (New, Pallier, Ferr<strong>and</strong>, & Matos; 2001), which is a newly created database <strong>of</strong> <strong>French</strong> word forms with accompany<strong>in</strong>gfrequencies based on a corpus <strong>of</strong> written texts (31 million word tokens). Infectional studies <strong>in</strong>the <strong>French</strong> language were not possible before the release <strong>of</strong> this database, because the exist<strong>in</strong>gdatabases lacked frequencies for <strong>in</strong>flected forms. In this <strong>and</strong> all subsequent experiments,frequency is reported as the number <strong>of</strong> appearances per million. Special care was taken toselect only those words for which the <strong>s<strong>in</strong>gular</strong> <strong>and</strong> the <strong>plural</strong> did not exist as other word forms(e.g., as <strong>in</strong>flections <strong>of</strong> a verb, as <strong>in</strong> danse [dance]), <strong>and</strong> for which the <strong>s<strong>in</strong>gular</strong> <strong>and</strong> the <strong>plural</strong>were the only possible realizations (e.g., some <strong>nouns</strong> exist <strong>in</strong> a male <strong>and</strong> female form, as <strong>in</strong>chien, chienne [dog]). In addition, for each word the <strong>plural</strong> consisted <strong>of</strong> the orthographic form<strong>of</strong> the <strong>s<strong>in</strong>gular</strong> with the end-morpheme –s (e.g. nuage-nuages [cloud-s]).<strong>The</strong> first list consisted <strong>of</strong> 24 words <strong>of</strong> which the <strong>s<strong>in</strong>gular</strong> form was more frequent than the<strong>plural</strong> form (hence called <strong>s<strong>in</strong>gular</strong> dom<strong>in</strong>ant items). <strong>The</strong> mean frequency <strong>of</strong> the <strong>s<strong>in</strong>gular</strong> <strong>and</strong>the <strong>plural</strong> forms were respectively 47 <strong>and</strong> 15 per million. <strong>The</strong> second list <strong>of</strong> 24 wordsconsisted <strong>of</strong> <strong>plural</strong> dom<strong>in</strong>ant items, with an average frequency <strong>of</strong> 15 for the <strong>s<strong>in</strong>gular</strong> form <strong>and</strong>41 for the <strong>plural</strong> form. <strong>The</strong> base frequencies (i.e., the cumulative frequency <strong>of</strong> the two forms)3 This database is available at the follow<strong>in</strong>g website: http://www.lexique.org
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 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 <strong>of</strong> letters (6.6 <strong>and</strong> 6.6) <strong>and</strong> the number <strong>of</strong> syllables (1.8 <strong>and</strong>1.8). A complete list <strong>of</strong> the stimuli is presented <strong>in</strong> Appendix A. Two versions <strong>of</strong> the word listswere made. Half <strong>of</strong> the words had their <strong>s<strong>in</strong>gular</strong> form <strong>in</strong> one version <strong>and</strong> the <strong>plural</strong> <strong>in</strong> theother; for the other half the assignment was reversed.In addition, 48 nonword stimuli were created from <strong>French</strong> words by replac<strong>in</strong>g a s<strong>in</strong>gleconsonant with another consonant, or a s<strong>in</strong>gle vowel with another vowel. <strong>The</strong>se nonwordswere phonotactically legal, <strong>and</strong> were matched to the word stimuli <strong>in</strong> terms <strong>of</strong> number <strong>of</strong>letters (6.6) <strong>and</strong> number <strong>of</strong> syllables (1.8). Half <strong>of</strong> the nonwords term<strong>in</strong>ated on –s to match the<strong>plural</strong> word forms that were presented.ProcedureParticipants were tested <strong>in</strong>dividually <strong>in</strong> a soundpro<strong>of</strong> room. <strong>The</strong>y were asked to <strong>in</strong>dicate asquickly <strong>and</strong> accurately as possible whether the presented letter str<strong>in</strong>g formed an exist<strong>in</strong>g<strong>French</strong> word or not. <strong>The</strong>y did so by press<strong>in</strong>g one <strong>of</strong> two buttons <strong>of</strong> a joypad "LogitechW<strong>in</strong>gman Extreme". Each trial began with a 200ms fixation cross (a plus sign <strong>in</strong> the center <strong>of</strong>the screen), followed by the stimulus which rema<strong>in</strong>ed visible until the participant responded(with a maximum time period <strong>of</strong> 4 s). Between trials, there was a 1s black screen <strong>in</strong>terval.Each participant saw one <strong>of</strong> the two word list versions (counterbalanced across participants).<strong>The</strong> stimuli were r<strong>and</strong>omized anew for each participant <strong>and</strong> presented with the use <strong>of</strong> DMDX(Forster et Forster, 2003) on a Pentium 166. <strong>The</strong> test items were preceded by twenty practicetrials.RESULTSTable 1 shows the mean reaction times <strong>and</strong> percentages <strong>of</strong> errors, as a function <strong>of</strong> word type(<strong>s<strong>in</strong>gular</strong> dom<strong>in</strong>ant vs. <strong>plural</strong> dom<strong>in</strong>ant) <strong>and</strong> as a function <strong>of</strong> the word form presented(<strong>s<strong>in</strong>gular</strong> vs. <strong>plural</strong>). Response times <strong>of</strong> more than two st<strong>and</strong>ard deviations above or below themean were discarded as outliers. Thus 6.2% <strong>of</strong> the data <strong>in</strong> the subjects analysis as well as5.2% <strong>in</strong> the item analysis were discarded. Because the error rates were low <strong>and</strong> fully <strong>in</strong> l<strong>in</strong>ewith the RTs, they were not analyzed separately. A conventional significance level <strong>of</strong> .05 wasused unless otherwise <strong>in</strong>dicated.
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 10ANOVAs on the RTs <strong>of</strong> the correct responses returned a significant ma<strong>in</strong> effect <strong>of</strong> word form(<strong>s<strong>in</strong>gular</strong> vs. <strong>plural</strong>; F1(1,31)=7.48, MSe = 732.84; F2(1,46)=6.24, MSe = 981.2767), <strong>and</strong> asignificant <strong>in</strong>teraction between word type <strong>and</strong> 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 <strong>in</strong>teraction was dueto the longer RTs <strong>in</strong> the condition where participants had to respond to the <strong>plural</strong> form <strong>of</strong> a<strong>s<strong>in</strong>gular</strong> dom<strong>in</strong>ant noun.DISCUSSION<strong>The</strong> basic question addressed by Experiment 1 was to what extent lexical decisiontimes to <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> noun forms are determ<strong>in</strong>ed by the surface frequency <strong>of</strong> the formwhen the base frequency is controlled . Table 1 shows that the data are completely <strong>in</strong> l<strong>in</strong>e withBaayen et al.’s (1997). For <strong>s<strong>in</strong>gular</strong> dom<strong>in</strong>ant items, a reliable difference was observedbetween the <strong>s<strong>in</strong>gular</strong> <strong>and</strong> the <strong>plural</strong> forms, whereas for <strong>plural</strong> dom<strong>in</strong>ant items, no significantdifference was obta<strong>in</strong>ed. In addition, the RTs to <strong>s<strong>in</strong>gular</strong> <strong>nouns</strong> did not differ as a function <strong>of</strong>the word type (<strong>s<strong>in</strong>gular</strong> dom<strong>in</strong>ant or <strong>plural</strong> dom<strong>in</strong>ant). <strong>The</strong>se f<strong>in</strong>d<strong>in</strong>gs are <strong>in</strong> l<strong>in</strong>e with thehypothesis that RTs to <strong>s<strong>in</strong>gular</strong> <strong>nouns</strong> are a function <strong>of</strong> the base frequency <strong>of</strong> the noun,whereas RTs to <strong>plural</strong> <strong>nouns</strong> partly depend on the surface frequency <strong>of</strong> the word form. In theGeneral Discussion section, we will assess more <strong>in</strong> detail the relative importance <strong>of</strong> thedecomposition <strong>and</strong> the storage route with<strong>in</strong> Baayen et al's (1997) dual-route framework. First,however, <strong>in</strong> Experiment 2 we <strong>in</strong>vestigated whether reaction times to <strong>s<strong>in</strong>gular</strong> <strong>nouns</strong> <strong>in</strong> the<strong>French</strong> language are <strong>in</strong>fluenced by the base frequency <strong>of</strong> the <strong>nouns</strong>.EXPERIMENT 2After hav<strong>in</strong>g <strong>in</strong>vestigated the effects <strong>of</strong> surface frequency <strong>in</strong> our first experiment, weassessed the contribution <strong>of</strong> the cumulative frequency on the process<strong>in</strong>g <strong>of</strong> <strong>s<strong>in</strong>gular</strong> noun
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 11forms. <strong>The</strong>refore, we looked at the lexical decision times to <strong>s<strong>in</strong>gular</strong> <strong>French</strong> <strong>nouns</strong> that hadthe same surface frequency but different base frequencies (because the frequency <strong>of</strong> the <strong>plural</strong>form was high or low).METHODParticipantsFifteen new students from the Université René Descartes, Paris V, took part <strong>in</strong> the experiment<strong>in</strong> return for course credits. <strong>The</strong>y were native <strong>French</strong>-speakers <strong>and</strong> had normal or correctedto-normalvision.Stimulus MaterialsForty-four words were selected from Lexique <strong>and</strong> 44 match<strong>in</strong>g nonwords were constructed.Selection <strong>and</strong> construction criteria were the same as <strong>in</strong> Experiment 1, except for thefrequencies <strong>of</strong> the word forms. One list <strong>of</strong> 22 words had a <strong>s<strong>in</strong>gular</strong> frequency <strong>of</strong> 16, <strong>and</strong> a<strong>plural</strong> frequency <strong>of</strong> 43; the other list had a <strong>s<strong>in</strong>gular</strong> frequency <strong>of</strong> 16, <strong>and</strong> a <strong>plural</strong> frequency <strong>of</strong>4. <strong>The</strong> two lists <strong>of</strong> words were matched on the number <strong>of</strong> letters (6.4 <strong>and</strong> 6.3) <strong>and</strong> the number<strong>of</strong> syllables (1.7 <strong>and</strong> 1.7). A complete list <strong>of</strong> the words is given <strong>in</strong> Appendix B.Procedure<strong>The</strong> procedure was identical to that described <strong>in</strong> Experiment 1, except that <strong>in</strong> this experimentonly the <strong>s<strong>in</strong>gular</strong> word forms were presented. Because <strong>of</strong> this, no non-word ended <strong>in</strong> –s either.RESULTSTable 2 shows the mean reaction times <strong>and</strong> percentages <strong>of</strong> errors. Extreme reaction timeswere removed accord<strong>in</strong>g to the procedure described <strong>in</strong> Experiment 1. Thus 5.5% <strong>of</strong> the data <strong>in</strong>the subjects analysis as well as 4.7% <strong>in</strong> the item analysis were discarded. ANOVAs with onerepeated measure revealed a ma<strong>in</strong> effect <strong>of</strong> the frequency <strong>of</strong> the <strong>plural</strong> form both <strong>in</strong> theanalysis over participants (F1(1,14)=24.09, Mse = 946.06) <strong>and</strong> <strong>in</strong> the analysis over itemsF2(1,21)=19.57, Mse = 1371.50). Participants responded faster to <strong>s<strong>in</strong>gular</strong> word forms withhigh-frequency <strong>plural</strong>s than to <strong>s<strong>in</strong>gular</strong> word forms with low-frequency <strong>plural</strong>s.
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 12DISCUSSION<strong>The</strong> ma<strong>in</strong> f<strong>in</strong>d<strong>in</strong>g <strong>of</strong> experiment 2 was the presence <strong>of</strong> a base frequency effect whenthe <strong>s<strong>in</strong>gular</strong> forms were matched <strong>in</strong> surface frequency. When two <strong>s<strong>in</strong>gular</strong> forms have thesame surface frequency but differ <strong>in</strong> the frequency <strong>of</strong> their <strong>plural</strong> forms, the <strong>s<strong>in</strong>gular</strong> with themore frequent <strong>plural</strong> is processed faster. This result agrees with Baayen et al.’s f<strong>in</strong>d<strong>in</strong>gs <strong>in</strong>Dutch, but deviates from Sereno <strong>and</strong> Jongman's f<strong>in</strong>d<strong>in</strong>gs <strong>in</strong> <strong>English</strong>.So, on the basis <strong>of</strong> the two experiments reported thus far, it seems that Dutch <strong>and</strong><strong>French</strong> <strong>plural</strong>s are processed <strong>in</strong> the same way, <strong>and</strong> both differ significantly from the f<strong>in</strong>d<strong>in</strong>gs<strong>in</strong> <strong>English</strong>. In addition, there is some suggestive evidence that the Dutch/<strong>French</strong> pattern couldalso be present <strong>in</strong> Italian (Baayen et al., 1996) <strong>and</strong> <strong>in</strong> Spanish (Dom<strong>in</strong>guez, Cuetos, & Segui,1999), mak<strong>in</strong>g the <strong>English</strong> f<strong>in</strong>d<strong>in</strong>g even more isolated. <strong>The</strong>refore, we decided to repeat theSereno <strong>and</strong> Jongman experiments.EXPERIMENT 3A closer look at Sereno <strong>and</strong> Jongman (1997) revealed a number <strong>of</strong> methodologicaldifferences between that study <strong>and</strong> all the other studies. For a start, Sereno <strong>and</strong> Jongmanpresented their <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> stimuli <strong>in</strong> different experiments (Experiments 2a <strong>and</strong> 2b).This blocked presentation may have encouraged participants to ignore the end –s <strong>in</strong> theexperiment with the <strong>plural</strong> stimuli. Another problem is that Sereno <strong>and</strong> Jongman usednonwords that were quite dissimilar to words. F<strong>in</strong>ally, their word frequencies were based onthe Brown corpus which only <strong>in</strong>cludes one million words. This is a quite limited corpus if wecompare it to the <strong>French</strong> corpus used <strong>in</strong> Lexique (31 millions <strong>of</strong> tokens) <strong>and</strong> the <strong>English</strong>corpus used <strong>in</strong> Celex (16.6 millions <strong>of</strong> tokens). For these reasons, we decided to repeat theSereno <strong>and</strong> Jongman experiments, follow<strong>in</strong>g the same procedures used <strong>in</strong> our <strong>French</strong> studies(<strong>and</strong> <strong>in</strong> the Dutch studies).
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 13METHODParticipantsThirty-four students from Royal Holloway, University <strong>of</strong> London, took part <strong>in</strong> the experiment<strong>in</strong> return for course credits. <strong>The</strong>y were all native <strong>English</strong>-speakers <strong>and</strong> had normal orcorrected-to-normal vision.Stimulus Materials.<strong>The</strong> word stimuli were two lists <strong>of</strong> 25 <strong>nouns</strong> drawn from the database Celex (Baayen,Piepenbrock, & van Rijn 1993), based on a corpus <strong>of</strong> 16.6 million word tokens <strong>of</strong> written<strong>English</strong>. <strong>The</strong> first list consisted <strong>of</strong> <strong>s<strong>in</strong>gular</strong> dom<strong>in</strong>ant items, with an average frequency <strong>of</strong> 39per million for the <strong>s<strong>in</strong>gular</strong> forms <strong>and</strong> 14 for the <strong>plural</strong> forms. <strong>The</strong> second list consisted <strong>of</strong><strong>plural</strong> dom<strong>in</strong>ant items with average frequencies <strong>of</strong> 15 <strong>and</strong> 39 respectively. <strong>The</strong> stemfrequencies (53 vs. 54) did not differ between the lists. <strong>The</strong> stimuli were further matched onthe number <strong>of</strong> letters (6 <strong>and</strong> 6) <strong>and</strong> the number <strong>of</strong> syllables (2 <strong>and</strong> 2). A complete list <strong>of</strong> thestimuli is presented <strong>in</strong> Appendix C. As <strong>in</strong> Experiment 1, two versions <strong>of</strong> the word list werecreated, so that each participants saw only one form <strong>of</strong> a word.In addition, 50 nonword stimuli were created from <strong>English</strong> words by replac<strong>in</strong>g a s<strong>in</strong>gleconsonant with another consonant, or a s<strong>in</strong>gle vowel with another vowel. <strong>The</strong> nonwords werephonotactically legal, <strong>and</strong> were matched to the word stimuli <strong>in</strong> terms <strong>of</strong> mean number <strong>of</strong>letters (6.6) <strong>and</strong> number <strong>of</strong> syllables (1.8). Half <strong>of</strong> the nonwords ended on –s.ProcedureStimulus presentation was the same as <strong>in</strong> Experiment 1, except that an external buttonresponse box was used for response collection.RESULTSTable 3 shows the mean reaction times <strong>and</strong> percentages <strong>of</strong> errors. Extreme reaction timeswere removed by the procedure followed <strong>in</strong> the first experiment. Thus 5.5% <strong>of</strong> the data <strong>in</strong> thesubjects analysis as well as 5.3% <strong>in</strong> the item analysis were discarded. We also removed oneitem <strong>in</strong> each list that led to extremely long RTs. (<strong>in</strong>stitution <strong>and</strong> exam<strong>in</strong>ation). Two-wayANOVAs revealed a ma<strong>in</strong> effect <strong>of</strong> word form <strong>in</strong> the analysis by items (F1(1,33)=2.75, MSe
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 14= 4996.97; F2(1,46)=14.32, MSe = 904.27), <strong>and</strong> a significant <strong>in</strong>teraction between word type<strong>and</strong> word form (F1(1,33)=8.76, MSe = 1319.45; F2(1,46)=12, MSe = 904.27). No ma<strong>in</strong> effect<strong>of</strong> word type (<strong>s<strong>in</strong>gular</strong> dom<strong>in</strong>ant vs. <strong>plural</strong> dom<strong>in</strong>ant) was found (F1(1,33)=2.38, MSe =1465.50; F2(1,46)= 0.03, MSe = 6912.34). Planned comparisons <strong>in</strong>dicated a significantdifference <strong>in</strong> the RTs to the <strong>s<strong>in</strong>gular</strong> noun forms between the <strong>s<strong>in</strong>gular</strong> dom<strong>in</strong>ant <strong>and</strong> the <strong>plural</strong>dom<strong>in</strong>ant words <strong>in</strong> 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). <strong>The</strong>re was asignificant difference between the <strong>s<strong>in</strong>gular</strong> <strong>and</strong> the <strong>plural</strong> forms for the <strong>s<strong>in</strong>gular</strong> dom<strong>in</strong>antitems (F1(1,33)=10.08, MSe = 2506.61; F2(1,23)=25.04, MSe = 948.82) but not for the <strong>plural</strong>dom<strong>in</strong>ant items (F1(1,33)
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 15EXPERIMENT 4Sereno <strong>and</strong> Jongman (Experiments 3a <strong>and</strong> 3b) failed to f<strong>in</strong>d an effect <strong>of</strong> basefrequency on lexical decision times to <strong>s<strong>in</strong>gular</strong> <strong>nouns</strong> <strong>in</strong> <strong>English</strong>. However, a closer look attheir stimuli reveals a possible difficulty. <strong>The</strong>y used <strong>s<strong>in</strong>gular</strong> forms with high frequencies (onaverage 95 occurrences per million). This contrasts with the Dutch <strong>and</strong> the <strong>French</strong> studies,which were based on medium frequency items (10 to 15 occurrences per million). It ispossible that Sereno <strong>and</strong> Jongman did not f<strong>in</strong>d a base frequency effect, because their highfrequency <strong>s<strong>in</strong>gular</strong> items were already close to the ceil<strong>in</strong>g level <strong>and</strong> could not pr<strong>of</strong>it verymuch from the additional activation due to the <strong>plural</strong>s. <strong>The</strong>refore, we selected two lists <strong>of</strong><strong>s<strong>in</strong>gular</strong> words with a medium surface frequency <strong>and</strong> with highly different <strong>plural</strong> frequencies.METHODParticipantsN<strong>in</strong>eteen new students from Royal Holloway, University <strong>of</strong> London, took part <strong>in</strong> thisexperiment. <strong>The</strong>y were paid £5 for their participation. All participants were native <strong>English</strong>speakers <strong>and</strong> had normal or corrected vision.Materials<strong>The</strong> stimuli were 48 <strong>nouns</strong> drawn from Celex, based on a corpus <strong>of</strong> 16.6 million word tokens<strong>of</strong> written <strong>English</strong> (Baayen, Piepenbrock, & van Rijn 1993). <strong>The</strong> first list consisted <strong>of</strong> 24<strong>s<strong>in</strong>gular</strong> <strong>nouns</strong> with a high frequency <strong>plural</strong> (frequencies <strong>of</strong> 15 <strong>and</strong> 39 for <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong>respectively). <strong>The</strong>se items except one were the same as those used as <strong>in</strong> the <strong>s<strong>in</strong>gular</strong> dom<strong>in</strong>antcondition <strong>of</strong> Experiment 3. <strong>The</strong> second list consisted <strong>of</strong> 24 <strong>s<strong>in</strong>gular</strong> <strong>nouns</strong> with a lowfrequency <strong>plural</strong> (frequencies <strong>of</strong> 16 <strong>and</strong> 1.9). <strong>The</strong> lists were matched for number <strong>of</strong> letters (6<strong>and</strong> 6.2) <strong>and</strong> number <strong>of</strong> syllables (2 <strong>and</strong> 1.9). A complete list <strong>of</strong> the stimuli is presented <strong>in</strong>Appendix D. Aga<strong>in</strong>, a list <strong>of</strong> 48 match<strong>in</strong>g nonwords was created along the criteria outl<strong>in</strong>ed <strong>in</strong>the previous experiments.
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 16Procedure<strong>The</strong> procedure was identical to that described <strong>in</strong> Experiment 2.RESULTSTable 4 shows the mean reaction times <strong>and</strong> percentages <strong>of</strong> errors. Extreme reactiontimes were removed by the procedure followed <strong>in</strong> the Experiment 1. A one-way ANOVAwith repeated measures revealed a ma<strong>in</strong> effect <strong>of</strong> the frequency <strong>of</strong> the <strong>plural</strong> form(F1(1,18)=13.39, MSe = 479.61; F2(1,23)=6.67, MSe = 1350.82). Subjects reacted 26 msfaster to <strong>s<strong>in</strong>gular</strong> forms with high-frequency <strong>plural</strong>s than to <strong>s<strong>in</strong>gular</strong> forms with lowfrequency<strong>plural</strong>s. Subject <strong>and</strong> item analyses were also conducted for the error data. Nosignificant differences were found for the effect <strong>of</strong> the <strong>plural</strong>.DISCUSSIONIn this experiment we showed that <strong>in</strong> <strong>English</strong>, lexical decision times to <strong>s<strong>in</strong>gular</strong> <strong>nouns</strong>are affected by the frequency <strong>of</strong> the <strong>plural</strong> forms, as previously shown <strong>in</strong> Dutch <strong>and</strong> <strong>in</strong> <strong>French</strong>.This adds credit to our reservation about Sereno <strong>and</strong> Jongman's f<strong>in</strong>d<strong>in</strong>gs, which were based onhigh-frequency <strong>s<strong>in</strong>gular</strong> <strong>nouns</strong>. On the other h<strong>and</strong>, a comparison <strong>of</strong> Tables 2 <strong>and</strong> 4 suggeststhat the effect <strong>of</strong> base frequency may be stronger <strong>in</strong> <strong>French</strong> (56 ms) than <strong>in</strong> <strong>English</strong> (26 ms),given that the frequency ranges <strong>of</strong> the stimuli <strong>in</strong> both languages were nearly the same. On theother h<strong>and</strong>, because it is based on a between-items <strong>and</strong> between-participants analysis, thisf<strong>in</strong>d<strong>in</strong>g should be treated with extreme caution. Another <strong>in</strong>terest<strong>in</strong>g aspect <strong>of</strong> Experiment 4 isthat the decision latencies to the low-frequency items with high frequency <strong>plural</strong>s (522 ms)were considerably faster than the latencies <strong>in</strong> Experiment 3 (555 ms) to the same items. Thisadds further credit to our assertion <strong>in</strong> Experiment 3 that the difference <strong>in</strong> the <strong>s<strong>in</strong>gular</strong>conditions between the <strong>s<strong>in</strong>gular</strong>-dom<strong>in</strong>ant <strong>and</strong> <strong>plural</strong>-dom<strong>in</strong>ant words is likely not to besignificant.
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 17GENERAL DISCUSSION<strong>The</strong> present study was set up to further <strong>in</strong>vestigate how pr<strong>in</strong>ted <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong><strong>nouns</strong> are recognized. Previous research <strong>in</strong> Dutch (Baayen et al., 1997) suggested that lexicaldecision times to <strong>s<strong>in</strong>gular</strong> <strong>nouns</strong> depended on the comb<strong>in</strong>ed frequencies <strong>of</strong> the <strong>s<strong>in</strong>gular</strong> <strong>and</strong><strong>plural</strong> word forms (i.e., the base frequency). In contrast, lexical decision times to <strong>plural</strong> nounforms depended on the surface frequency <strong>of</strong> the <strong>plural</strong> forms only, although Baayen et al.(1997) favored a dual-route account, with parallel retrieval <strong>of</strong> whole forms <strong>and</strong> computationon the basis <strong>of</strong> the <strong>s<strong>in</strong>gular</strong> form <strong>and</strong> the suffix (see the Introduction). <strong>The</strong> Dutch f<strong>in</strong>d<strong>in</strong>gsseemed <strong>in</strong> l<strong>in</strong>e with data obta<strong>in</strong>ed <strong>in</strong> Italian (Baayen et al., 1996) <strong>and</strong> Spanish (Dom<strong>in</strong>guez etal., 1999), but not with data obta<strong>in</strong>ed <strong>in</strong> <strong>English</strong> (Sereno & Jongman, 1997). For thislanguage, Sereno & Jongman said that only surface frequency seemed to matter, <strong>in</strong> l<strong>in</strong>e with afull storage model (Butterworth, 1983).<strong>The</strong> f<strong>in</strong>d<strong>in</strong>gs reported <strong>in</strong> the present article considerably clarify the empirical evidence.First, <strong>in</strong> <strong>English</strong> <strong>and</strong> <strong>French</strong>, like <strong>in</strong> Dutch, lexical decisions to <strong>s<strong>in</strong>gular</strong> word forms were<strong>in</strong>fluenced by the frequencies <strong>of</strong> the <strong>plural</strong> forms (Experiments 2 <strong>and</strong> 4). Second, <strong>in</strong> all threelanguages, reaction times to the <strong>plural</strong> forms were slower than those to the <strong>s<strong>in</strong>gular</strong> formswhen the <strong>nouns</strong> were <strong>s<strong>in</strong>gular</strong> dom<strong>in</strong>ant (i.e., had a higher frequency <strong>in</strong> <strong>s<strong>in</strong>gular</strong> than <strong>in</strong><strong>plural</strong>; Experiments 1 <strong>and</strong> 3). Third, <strong>in</strong> all languages, reaction times to the <strong>plural</strong> forms werenot significantly different from those to the <strong>s<strong>in</strong>gular</strong> forms when the <strong>nouns</strong> were <strong>plural</strong>dom<strong>in</strong>ant (Experiments 1 <strong>and</strong> 3).<strong>The</strong> data <strong>of</strong> Experiments 1 <strong>and</strong> 3 are depicted <strong>in</strong> Figure 1, together with those <strong>of</strong>Baayen et al. (1997; <strong>in</strong> Dutch) <strong>and</strong> Jongman <strong>and</strong> Sereno (1997; <strong>in</strong> <strong>English</strong>). In each part <strong>of</strong> thefigure, we see the same pattern emerg<strong>in</strong>g. For the <strong>s<strong>in</strong>gular</strong>-dom<strong>in</strong>ant words, there is asubstantial difference <strong>in</strong> decision times between the <strong>s<strong>in</strong>gular</strong> <strong>and</strong> the <strong>plural</strong> forms. In contrast,for the <strong>plural</strong>-dom<strong>in</strong>ant <strong>nouns</strong>, there is no difference. <strong>The</strong> only deviation between <strong>English</strong> onthe one h<strong>and</strong>, <strong>and</strong> <strong>French</strong> <strong>and</strong> Dutch on the other h<strong>and</strong>, is the position <strong>of</strong> the <strong>plural</strong>-dom<strong>in</strong>antwords relative to that <strong>of</strong> the <strong>s<strong>in</strong>gular</strong>-dom<strong>in</strong>ant words. Whereas <strong>in</strong> Dutch <strong>and</strong> <strong>French</strong>, reaction
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 18times to the <strong>plural</strong>-dom<strong>in</strong>ant words are as fast as those to the <strong>s<strong>in</strong>gular</strong>s <strong>of</strong> the <strong>s<strong>in</strong>gular</strong>dom<strong>in</strong>antwords, <strong>in</strong> <strong>English</strong> the RTs are slightly elevated, so that the reaction times to the<strong>plural</strong>-dom<strong>in</strong>ant words fall <strong>in</strong>-between those to the <strong>s<strong>in</strong>gular</strong> forms <strong>and</strong> the <strong>plural</strong> forms <strong>of</strong> the<strong>s<strong>in</strong>gular</strong>-dom<strong>in</strong>ant words. As we <strong>in</strong>dicated <strong>in</strong> the discussion section <strong>of</strong> Experiment 3, for the<strong>in</strong>terpretation <strong>of</strong> this f<strong>in</strong>d<strong>in</strong>g, it is important to keep <strong>in</strong> m<strong>in</strong>d that the relative position <strong>of</strong> thetwo l<strong>in</strong>es <strong>in</strong> each panel <strong>of</strong> Figure 1 is the weakest aspect <strong>of</strong> the experimental design, becauseit is based on a between-stimuli comparison. This became apparent <strong>in</strong> our own experiment,where the difference between the <strong>s<strong>in</strong>gular</strong> forms failed to reach significant <strong>in</strong> the analysis overstimuli, despite the fact that it is the largest difference found <strong>in</strong> the four studies summarized <strong>in</strong>Figure 1.Hav<strong>in</strong>g cleared the empirical <strong>in</strong>consistencies, we are now <strong>in</strong> a better position to look atthe theoretical implications. Because our data are <strong>in</strong> l<strong>in</strong>e with those <strong>of</strong> Baayen et al. (1997),they can easily be accounted for by the dual-route model proposed <strong>in</strong> that article (see alsoBaayen et al., 2000; Schreuder & Baayen, 1995). Accord<strong>in</strong>g to this model, <strong>s<strong>in</strong>gular</strong> <strong>nouns</strong> arealways recognized by the whole-word recognition route, <strong>and</strong> lexical decision times to themare a function <strong>of</strong> the base frequency (i.e., the cumulative frequencies <strong>of</strong> the <strong>s<strong>in</strong>gular</strong> <strong>and</strong> the<strong>plural</strong> forms). <strong>The</strong> lexical decision times to the <strong>plural</strong> forms are determ<strong>in</strong>ed by the faster <strong>of</strong>two possible routes 4 . <strong>The</strong> first route is the decomposition route. In this route, reaction timesequal the reaction time to the <strong>s<strong>in</strong>gular</strong> form, <strong>in</strong>creased by a time constant needed to segmentthe stimulus <strong>in</strong>put, license the comb<strong>in</strong>ation <strong>of</strong> segments, <strong>and</strong> compute the mean<strong>in</strong>g on thebasis <strong>of</strong> the <strong>s<strong>in</strong>gular</strong> <strong>and</strong> the suffix (together summarized under the term "pars<strong>in</strong>g cost"). <strong>The</strong>second route is the whole-word recognition route. Here, reaction times (RTs) depend on thesurface frequency <strong>of</strong> the <strong>plural</strong> word form. An important question with<strong>in</strong> the dual-route modelis how much each route contributes to the recognition <strong>of</strong> <strong>plural</strong> <strong>nouns</strong>. This can easily be4 A problem <strong>in</strong> reviews <strong>of</strong> models <strong>of</strong> visual word recognition, is that <strong>in</strong> recent years a transition is happen<strong>in</strong>gfrom horse-race models to activation-based models. In horse-race models, the faster route determ<strong>in</strong>es the output.In activation-based models, both routes always contribute to the output, because one route is not faster than theother (both make use <strong>of</strong> the same process<strong>in</strong>g cycle). In these models, the contribution <strong>of</strong> a route depends on theamount <strong>of</strong> activation it adds to the output units per process<strong>in</strong>g cycle. A similar transition is tak<strong>in</strong>g place <strong>in</strong>Baayen <strong>and</strong> Schreuder's th<strong>in</strong>k<strong>in</strong>g (e.g., compare Baayen et al., 2000, to Baayen et al., 1997). However, becausethe model consists <strong>of</strong> three, largely serial, stages, the horse-race model can still be used as a roughapproximation.
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 19estimated, as we will show for the data <strong>of</strong> Experiment 1 (<strong>French</strong> f<strong>in</strong>d<strong>in</strong>gs).If we first look at the whole-word recognition route, the model says that (1) RTs to<strong>s<strong>in</strong>gular</strong> forms will be the same for the <strong>s<strong>in</strong>gular</strong> dom<strong>in</strong>ant <strong>nouns</strong> <strong>and</strong> the <strong>s<strong>in</strong>gular</strong> <strong>plural</strong>dom<strong>in</strong>ant <strong>nouns</strong> (because their base frequencies were matched; both around 59 per million),(2) RTs to the high-frequency <strong>plural</strong>s <strong>of</strong> the <strong>plural</strong> dom<strong>in</strong>ant <strong>nouns</strong> will be slightly longer thanthose to the <strong>s<strong>in</strong>gular</strong> forms, because the average surface frequency <strong>of</strong> the <strong>plural</strong> forms (41) isslightly lower than the base frequency, <strong>and</strong> (3) RTs to the low-frequency <strong>plural</strong>s <strong>of</strong> the<strong>s<strong>in</strong>gular</strong> dom<strong>in</strong>ant <strong>nouns</strong> will be substantially longer than those to the <strong>s<strong>in</strong>gular</strong> forms, becausetheir average surface frequency (15) is much lower than the base frequency.<strong>The</strong> estimates <strong>of</strong> the whole-route RTs are easy for the <strong>s<strong>in</strong>gular</strong> <strong>nouns</strong>, because theseRTs are assumed to be due to the storage route alone. Table 1 <strong>in</strong>forms us that these RTs (for<strong>s<strong>in</strong>gular</strong> <strong>nouns</strong> with a base frequency <strong>of</strong> 59) approximately form a normal distribution with amean <strong>of</strong> 547 ms <strong>and</strong> a st<strong>and</strong>ard deviation <strong>of</strong> 40 ms. This makes that most <strong>of</strong> the data will fallbetween 467 ms (mean m<strong>in</strong>us two st<strong>and</strong>ard deviations) <strong>and</strong> 627 ms (mean plus two st<strong>and</strong>arddeviations). <strong>The</strong> estimates <strong>of</strong> the whole-route RTs for the <strong>plural</strong> forms are slightly moredifficult to obta<strong>in</strong>, because the data <strong>in</strong> Table 1 are a mixture <strong>of</strong> whole-word recognition <strong>and</strong>decomposition. <strong>The</strong>refore, we cannot use these data to get a reasonable estimate <strong>of</strong> thefrequency effect due to the storage route alone. Such <strong>in</strong>formation, however, can be obta<strong>in</strong>edfrom Table 2 (Experiment 2). Here we see that RTs to <strong>s<strong>in</strong>gular</strong> <strong>nouns</strong> with a base frequency<strong>of</strong> 20 (596 ms) are 56 ms longer than the RTs to <strong>s<strong>in</strong>gular</strong> <strong>nouns</strong> with a base frequency <strong>of</strong> 59(540 ms). Because the RTs <strong>in</strong> Table 2 are based on <strong>s<strong>in</strong>gular</strong> word forms, they are completelydue to the storage route, so that the time difference <strong>of</strong> 56 ms can be considered as a reasonableestimate <strong>of</strong> the frequency effect <strong>in</strong> the whole-word recognition route (at least for a frequencydifference between 59 <strong>and</strong> 20 per million). So, by comb<strong>in</strong><strong>in</strong>g the results <strong>of</strong> Experiments 2 <strong>and</strong>1, <strong>and</strong> by assum<strong>in</strong>g that the effects <strong>of</strong> base frequency are the same as those <strong>of</strong> surfacefrequency (as Baayen <strong>and</strong> colleagues do), we can conclude that if <strong>in</strong> Experiment 1 we hadpresented <strong>plural</strong> <strong>nouns</strong> with a surface frequency <strong>of</strong> 20 per million, we would have expectedthe storage route to result <strong>in</strong> a normal distribution <strong>of</strong> RTs with a mean <strong>of</strong> 547 ms + 56 ms =603 ms, <strong>and</strong> a st<strong>and</strong>ard deviation <strong>of</strong> 40 ms. <strong>The</strong> s<strong>in</strong>gle next step to make then, is to rescale thefrequency effect from the low frequency <strong>of</strong> 20 used <strong>in</strong> Experiment 2, to the low frequencies <strong>of</strong>15 <strong>and</strong> 41 used <strong>in</strong> Experiment 1. Assum<strong>in</strong>g a logarithmic frequency function, this gives thefollow<strong>in</strong>g average values: For the <strong>plural</strong> forms with a surface frequency <strong>of</strong> 15, we get an
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 20estimate <strong>of</strong> 547 ms + 71 ms 5 ; <strong>and</strong> for the <strong>plural</strong> forms with a surface frequency <strong>of</strong> 41, we getan estimate <strong>of</strong> 547 ms + 19 ms. Assum<strong>in</strong>g equal st<strong>and</strong>ard deviations <strong>in</strong> all conditions 6 , we getthe RT distributions shown <strong>in</strong> Table 5.<strong>The</strong> predictions <strong>in</strong> Table 5 can be compared with the obta<strong>in</strong>ed data <strong>of</strong> 546, 548, 574,<strong>and</strong> 546 ms respectively. In the dual-route model, the differences between the predicted <strong>and</strong>the obta<strong>in</strong>ed values for the <strong>plural</strong> forms come from the second, decomposition route, whichroughly will result <strong>in</strong> a normal distribution <strong>of</strong> RTs with mean equal to 547 + total pars<strong>in</strong>gcost, <strong>and</strong> a st<strong>and</strong>ard deviation <strong>of</strong> 40 as well. With very small values <strong>of</strong> the estimated pars<strong>in</strong>gcost, the RT distribution <strong>of</strong> the decomposition route will be nearly the same as the one for the<strong>s<strong>in</strong>gular</strong> forms. With very high values <strong>of</strong> the estimated pars<strong>in</strong>g time, the RT distribution <strong>of</strong> thedecomposition route will be so high that it will never be faster than the storage route. Simplesimulations allow us to search for a value <strong>of</strong> the estimated pars<strong>in</strong>g cost that is <strong>in</strong> l<strong>in</strong>e with theempirical data. Table 6 shows the effects <strong>of</strong> different values on the estimates RTs. <strong>The</strong>y arebased on 10,000 r<strong>and</strong>om values from the normal distributions def<strong>in</strong>ed for each route.As can be seen <strong>in</strong> Table 6, the empirical data <strong>of</strong> Experiment 1 are capturedbetter when <strong>in</strong> addition to the whole-word recognition route, the model <strong>in</strong>cludes adecomposition route with a pars<strong>in</strong>g cost <strong>of</strong> some 25-30 ms. <strong>The</strong> decomposition route is faster<strong>in</strong> 80% <strong>of</strong> the <strong>in</strong>stances for <strong>nouns</strong> with a low-frequency <strong>plural</strong>, <strong>and</strong> <strong>in</strong> 45% <strong>of</strong> the cases for<strong>nouns</strong> with a high-frequency <strong>plural</strong>. <strong>The</strong> strong impact <strong>of</strong> the decomposition route agrees withthe fact that the –s morpheme is a productive morpheme to <strong>plural</strong>ize <strong>nouns</strong> <strong>in</strong> <strong>French</strong>, withouta higher-frequency competitor. 75 <strong>The</strong> added time due to the lower frequency is estimated with the equation:log( 59)− log( 15)added _ time =x56log( 59)− log( 20)6 <strong>The</strong> constant value <strong>of</strong> SD is clearly a simplify<strong>in</strong>g assumption, because <strong>in</strong> RT data higher means are alwaysaccompanied by higher SDs, as can easily be verified <strong>in</strong> Tables 1-4.7 At the same time, Table 6 reveals a weakness <strong>in</strong> Baayen et al.'s (1997) model based on the horse-racemetaphor. When pars<strong>in</strong>g costs are low, RTs to <strong>plural</strong> forms tend to be faster than those to <strong>s<strong>in</strong>gular</strong> forms <strong>and</strong> lessvariable. This is because <strong>s<strong>in</strong>gular</strong> forms are supposed to be processed by a s<strong>in</strong>gle route only, whereas <strong>plural</strong>forms are processed by the faster <strong>of</strong> two parallel routes.
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 22<strong>in</strong> morphological process<strong>in</strong>g: <strong>The</strong> role <strong>of</strong> word formation type, affixal homonymy <strong>and</strong>productivity. Journal <strong>of</strong> Experimental Psychology: Learn<strong>in</strong>g, Memory, <strong>and</strong> Cognition, 26,489-511.Butterworth, B. (1983) Lexical representation. In B. Butterworth (Ed.), Language ProductionVolume 2 : Development, Writ<strong>in</strong>g <strong>and</strong> Other Language Processes. (pp. 257-294). London :Academic Press.Caramazza, A., Laudanna, A. et Romani, C. (1988). Lexical access <strong>and</strong> <strong>in</strong>flectionalmorphology. Cognition, 28, 297-332.Dom<strong>in</strong>guez, A., Cuetos, F. & Segui, J. (1999). <strong>The</strong> process<strong>in</strong>g <strong>of</strong> grammatical gender <strong>and</strong>number <strong>in</strong> Spanish. Journal <strong>of</strong> Psychol<strong>in</strong>guistic Research, 28, 485-498.Forster, K.I. & Forster, J.C. (2003). DMDX: A w<strong>in</strong>dows display program with millisecondaccuracy. Behavior Research Methods, Instruments, & Computers. 35, 116-124.New, B., Pallier C., Ferr<strong>and</strong> L., & Matos, R. (2001) Une base de données lexicales du françaiscontempora<strong>in</strong> sur <strong>in</strong>ternet : LEXIQUE, L'Année Psychologique, 101, 447-462.S<strong>and</strong>ra, D. (1994). <strong>The</strong> morphology <strong>of</strong> the mental lexicon Internal word structure viewedfrom a psychol<strong>in</strong>guistic perspective. Langage & Cognitive Processes, 9, 227-269.Schreuder, R. & Baayen, R.H. (1995). Model<strong>in</strong>g morphological process<strong>in</strong>g. Feldman, LaurieBeth (Ed). Morphological aspects <strong>of</strong> language process<strong>in</strong>g. (pp. 131-154). Hillsdale, NJ,Engl<strong>and</strong> : Lawrence Erlbaum Associates.Sereno, J.A. & Jongman, A. (1997). <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>English</strong> <strong>in</strong>flectional morphology. Memory<strong>and</strong> Cognition, 25, 425-437.Taft, M. (1979). Recognition <strong>of</strong> affixed words <strong>and</strong> the word frequency effect. Memory <strong>and</strong>Cognition, 7, 263-272.Taft, M. & Forster, K.I. (1975). Lexical storage <strong>and</strong> retrieval <strong>of</strong> prefixed words. Journal <strong>of</strong>Verbal Learn<strong>in</strong>g <strong>and</strong> Verbal Behavior, 14, 638-647.
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 23APPENDIX
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 24Word (<strong>in</strong> <strong>English</strong>)Meanreaction timeTable A1Materials used <strong>in</strong> Experiment 1S<strong>in</strong>gularSt<strong>and</strong>ardDeviationFrequencyMeanreaction timePluralSt<strong>and</strong>ardDeviationFrequencyS<strong>in</strong>gular High Frequencypa<strong>in</strong> [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 [spr<strong>in</strong>g] 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 [ro<strong>of</strong>] 508 96 29.58 539 83 3.39contrat [contract] 541 66 21.87 558 130 10.52écriva<strong>in</strong> [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 [build<strong>in</strong>g] 589 169 28.32 548 62 16.16fauteuil [armchair] 523 99 43.94 538 117 14.32faiblesse [weakness] 575 95 22.9 555 83 6m<strong>in</strong>istère [governmentdepartment] 535 106 41.23 699 109 9.68commissaire[super<strong>in</strong>tendent] 545 115 28.16 613 145 5.97Plural High Frequencydent [tooth] 475 76 9 553 105 70.87doigt [f<strong>in</strong>ger] 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 31.19march<strong>and</strong>ise [goods] 607 114 8 596 161 18.94
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 25Table A2Word (<strong>in</strong> <strong>English</strong>)Materials used <strong>in</strong> Experiment 2S<strong>in</strong>gularMean reaction St<strong>and</strong>ardPluralFrequenctime DeviationFrequencyyS<strong>in</strong>gular with low frequency <strong>plural</strong>doigt [f<strong>in</strong>ger] 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 31.19march<strong>and</strong>ise [good] 624 81 8 18.94S<strong>in</strong>gular with high frequency <strong>plural</strong>neige [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 0.87bass<strong>in</strong> [bas<strong>in</strong>] 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 1.77ch<strong>and</strong>ail [pullover] 665 109 5.74 1.48cercueil [c<strong>of</strong>f<strong>in</strong>] 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
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 26Table A3Materials used <strong>in</strong> Experiment 3S<strong>in</strong>gularPluralMeanMeanWord (<strong>in</strong>St<strong>and</strong>ard FrequencSt<strong>and</strong>ard Frequencreactionreaction<strong>English</strong>)Deviation yDeviation ytimetimeS<strong>in</strong>gulararticle 571 98 41 562 63 21bag 457 57 61 534 70 19beast 536 87 17 605 127 11cab<strong>in</strong> 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 23curta<strong>in</strong> 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
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 27Word (<strong>in</strong><strong>English</strong>)Table A4Materials used <strong>in</strong> Experiment 4Meanreaction timeSt<strong>and</strong>ardDeviationS<strong>in</strong>gularFrequencyPluralFrequencyS<strong>in</strong>gular with low frequency <strong>plural</strong>earl 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 1fl<strong>in</strong>t 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 1S<strong>in</strong>gular with high frequency <strong>plural</strong>acre 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 23curta<strong>in</strong> 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 18<strong>in</strong>stitution 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
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 28TablesTable 1. Mean Reaction Times (<strong>in</strong> ms), St<strong>and</strong>ard Deviations, <strong>and</strong> Percentages <strong>of</strong> Errors <strong>in</strong>Experiment 1S<strong>in</strong>gular Dom<strong>in</strong>antEx: Plafond [Ceil<strong>in</strong>g]Plural Dom<strong>in</strong>antEx: Nuages [Cloud]Presented form:<strong>s<strong>in</strong>gular</strong>Presented form:<strong>plural</strong>M 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 (<strong>in</strong> ms), st<strong>and</strong>ard deviations <strong>and</strong> percentages <strong>of</strong> Error <strong>in</strong>Experiment 2Frequency <strong>of</strong> thecomplementary formHigh Frequency PluralEx: Ongle [Nail]Low Frequency PluralEx: Frère [Brother]Presented forms: <strong>s<strong>in</strong>gular</strong>M SD %ER540 50 2.1%596 63 3.6%Table 3. Mean Reaction Times (<strong>in</strong> ms), st<strong>and</strong>ard deviations <strong>and</strong> percentage or Errors <strong>in</strong>Experiment 3S<strong>in</strong>gularDom<strong>in</strong>antEx: bagPlural Dom<strong>in</strong>antEx: lipPresented form: <strong>s<strong>in</strong>gular</strong> Presented form: <strong>plural</strong>M SD %ER M SD %ER527 63 2.1 565 59 4.4555 67 2.7 557 60 4.2
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 29Table 4. Mean Reaction Times (<strong>in</strong> ms), st<strong>and</strong>ard deviations <strong>and</strong> percentage or Errors <strong>in</strong>Experiment 4Frequency <strong>of</strong> thecomplementary formHigh Frequency PluralEx: HeelLow Frequency PluralEx: Fl<strong>in</strong>tPresented forms: <strong>s<strong>in</strong>gular</strong>M SD %ER522 56 5.4548 70 7.2Table 5: Predicted RTs us<strong>in</strong>g Baayen et al.'s storage route for Experiment 1 items.S<strong>in</strong>gularDom<strong>in</strong>antPresented form: <strong>s<strong>in</strong>gular</strong>Presented form: <strong>plural</strong>Frequency RT Range Frequency RT Range59 547 467-627 15 618 538-698Plural Dom<strong>in</strong>ant 59 547 467-627 41 566 486-646Table 6 : Simulated data for decision latencies to <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> word forms when adecomposition route is added to the storage route with different values <strong>of</strong> the pars<strong>in</strong>g time(PT). First, the average RT is reported; then the st<strong>and</strong>ard deviation (between brackets). For the<strong>plural</strong> forms, we also <strong>in</strong>dicate how <strong>of</strong>ten the decomposition route was faster than the wholewordrecognition route.Pars<strong>in</strong>g Time S<strong>in</strong>gular PluralS<strong>in</strong>gDomPlurDomPT = 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
<strong>The</strong> <strong>Process<strong>in</strong>g</strong> <strong>of</strong> <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> <strong>nouns</strong> <strong>in</strong> <strong>French</strong> <strong>and</strong> <strong>English</strong> 30FIGURES570560550540530520S<strong>in</strong>gular<strong>English</strong> (us)PluralS<strong>in</strong>gDom(jail)PlurDom(boot)580570560550540530S<strong>in</strong>gular<strong>French</strong>PluralS<strong>in</strong>gDom(ceil<strong>in</strong>g)PlurDom(cloud)680660640620600580S<strong>in</strong>gular<strong>English</strong> (Sereno)PluralS<strong>in</strong>gDom(jail)PlurDom(boot)620600580560540520500S<strong>in</strong>gularDutchPluralS<strong>in</strong>gDom(Jail)PlurDom(Boot)Figure LegendsFigure 1 : Lexical decision to <strong>s<strong>in</strong>gular</strong> <strong>and</strong> <strong>plural</strong> forms <strong>of</strong> <strong>s<strong>in</strong>gular</strong>-dom<strong>in</strong>ant <strong>and</strong> <strong>plural</strong>dom<strong>in</strong>ant<strong>nouns</strong> <strong>in</strong> <strong>English</strong> (present study, Experiment 3; Sereno & Jongman, 1997,Experiments 2a <strong>and</strong> 2b), Dutch (Baayen et al., 1997; Experiment 1), <strong>and</strong> <strong>French</strong> (present study,Experiment 1).