This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/authorsrights

Author's personal copy

An ER-fMRI study of Russian inflectional morphology

Natalia Slioussar a,b,⇑, Maxim V. Kireev c, Tatiana V. Chernigovskaya b, Galina V. Kataeva c,Alexander D. Korotkov c, Svyatoslav V. Medvedev c

a Utrecht Institute of Linguistics OTS, Trans 10, Utrecht 3512JK, The Netherlandsb Department of Liberal Arts and Sciences, St. Petersburg State University, Galernaya Street 58/60, St. Petersburg 190000, Russiac N.P. Bechtereva Institute of the Human Brain, Russian Academy of Sciences, Akademika Pavlova Street 9, St. Petersburg 197376, Russia

a r t i c l e i n f o

Article history:Accepted 23 January 2014Available online 25 February 2014

Keywords:fMRIRussianMorphologyMental lexiconDual-system theoriesSingle-system theoriesProcessing load

a b s t r a c t

The generation of regular and irregular past tense verbs has long been a testing ground for different mod-els of inflection in the mental lexicon. Behavioral studies examined a variety of languages, but neuroim-aging studies rely almost exclusively on English and German data. In our fMRI experiment, participantsinflected Russian verbs and nouns of different types and corresponding nonce stimuli. Irregular real andnonce verbs activated inferior frontal and inferior parietal regions more than regular verbs did, while noareas were more activated in the opposite comparison. We explain this activation pattern by increasingprocessing load: a parametric contrast revealed that these regions are also more activated for nonce stim-uli compared to real stimuli. A very similar pattern is found for nouns. Unlike most previously obtainedresults, our findings are more readily compatible with the single-system approach to inflection, whichdoes not postulate a categorical difference between regular and irregular forms.

� 2014 Elsevier Inc. All rights reserved.

1. Introduction

Inflectional morphology is at the center of an important debatein cognitive science, concerning the general principles according towhich the mental lexicon is organized. So-called ‘‘Dual-system’’(DS) approach distinguishes regular and irregular morphologicalforms. The former are computed by rules, the latter are stored inthe memory. In the alternative ‘‘Single-system’’ (SS) approach, allforms are generated and processed by a single integrated system.

Initially, English past tense morphology was the testing groundfor both approaches. According to the ‘‘Words and Rules’’ model, aversion of DS approach proposed by Pinker (1991, 1999), regular pasttense forms are generated and processed by a symbolic rule that ispart of the productive, combinatorial system of grammar. Irregularforms are learned by rote and stored in the lexicon, from where theycan be retrieved through associative memory mechanisms. The DSapproach was also advocated e.g. in (Marslen-Wilson & Tyler,1997; Pinker & Prince, 1988; Ullman, 2004). On the contrary, a con-nectionist network model from (Rumelhart & McClelland, 1986) rep-resents a single system without any symbolic rules. All past tense

forms are generated and processed by associative mechanisms thattake into account phonological similarity and token and typefrequencies of different elements. The SS approach was furtherdeveloped e.g. in (MacWhinney & Leinbach, 1991; McClelland &Patterson, 2002; Plunkett & Marchman, 1993). The range of dataused to test SS and DS theories has been very diverse: behavioraland neurophysiological experiments where participants generatedforms from various real and nonce verbs, language acquisition andlanguage deficit studies, and computer simulations. The results havealways been controversial.

However, English past tense morphology is exceptionally sim-ple: there is only one productive class that includes the vast major-ity of verbs and a small number of irregular verbs. So variousauthors investigated verb and noun inflection in other languageswhere the situation is more complex. German, Icelandic, Norwe-gian, Italian, Spanish, Arabic and Hebrew were among them (e.g.Berent, Pinker, & Shimron, 1999; Clahsen, 1999; Clahsen, Aveledo,& Roca, 2002; Hahn & Nakisa, 2000; Orsolini, Fanari, & Bowles,1998; Orsolini & Marslen-Wilson, 1997; Plunkett & Nakisa, 1997;Ragnasdóttir, Simonsen, & Plunkett, 1999). Studies on Russian arediscussed in Section 1.3. The findings offered new challenges forboth approaches, and some of them cannot be easily accountedfor by either approach. We illustrate this on the example of Russianbelow.

Thus, widening the pool of languages was extremely importantfor the SS vs. DS debate. For this reason, it appears to be

http://dx.doi.org/10.1016/j.bandl.2014.01.0060093-934X/� 2014 Elsevier Inc. All rights reserved.

⇑ Corresponding author at: Utrecht Institute of Linguistics OTS, Trans 10, Utrecht3512JK, The Netherlands.

E-mail addresses: slioussar@gmail.com (N. Slioussar), max@ihb.spb.ru(M.V. Kireev), tatiana.chernigovskaya@gmail.com (T.V. Chernigovskaya), kataeva@ihb.spb.ru (G.V. Kataeva), korotkov@ihb.spb.ru (A.D. Korotkov), medvedev@ihb.spb.ru (S.V. Medvedev).

Brain & Language 130 (2014) 33–41

Contents lists available at ScienceDirect

Brain & Language

journal homepage: www.elsevier .com/locate /b&l

Author's personal copy

problematic that existing functional imaging studies of past tensegeneration rely only on English, German and, in one case, Spanishdata (Beretta et al., 2003; de Diego-Balaguer et al., 2006; Desai,Conant, Waldron, & Binder, 2006; Dhond, Marinkovic, Dale, Witzel,& Halgren, 2003; Indefrey et al., 1997; Jaeger et al., 1996; Joanisse& Seidenberg, 2005; Oh, Tan, Ng, Berne, & Graham, 2011; Sach,Seitz, & Indefrey, 2004; Sahin, Pinker, & Halgren, 2006; Ullman,Bergida, & O’Craven, 1997a). In most studies, irregulars are associ-ated with a larger number of activated regions, but the list of theseregions, as well as proposed explanations differ greatly. There arealso several electrophysiological studies dedicated to past tenseformation in English and German (e.g. Lavric, Pizzagalli, Forstmeier,& Rippon, 2001; Marslen-Wilson & Tyler, 1998; Münte, Say, Clah-sen, Schiltz, & Kutas, 1999; Newman, Izvorski, Davis, Neville, & Ull-man, 1999; Newman, Ullman, Pancheva, Waligura, & Neville,2007), which show variable results.

The present paper aims to fill this gap. We conducted an fMRIstudy based on Russian language where participants were askedto generate present tense forms from different real and nonce(nonword) verbs and to pluralize real and nonce nouns. We triedto avoid numerous pitfalls identified by the critics of the previousstudies. We used two tasks (inflecting verbs and nouns) in a ran-dom order, which minimizes the risk of priming and strategy ef-fects, and large sets of stimuli matched for word frequency andphonological complexity.

1.1. A brief description of the Russian verb system

Russian verbs have two stems: the present/future tense stemand the past tense stem. Correlation between them determinesthe verb class. Out of several existing approaches, we will rely onthe one developed by Jakobson and his followers (Davidson, Gor,& Lekic, 1996; Jakobson, 1948; Townsend, 1975), according towhich Russian has 11 verb classes and several so-called anomalousverbs. Ten classes are identified by their suffixes (verbal classifi-ers). The eleventh class has a zero suffix, and is subdivided intosubclasses depending on the quality of the root-final consonant(Jakobson and Townsend counted them as 13 separate classes). Itincludes many conjugational patterns and contains well under100 basic stems.

Conjugational patterns of different classes include truncationsor additions of the final consonant or vowel and may also includestress shifts, suffix alternations, alternations of stem vowels andstem-final consonants. Russian has two conjugation types in thepresent and future tense, i.e. two different sets of endings, andto which one a verb belongs is determined by its class. Impor-tantly, the verb class is often unrecoverable from a particularform of the verb. For example, citát’ ‘to read’ belongs to the AJclass, and its 3Pl present tense form is citá-j-ut (-j- suffix isadded, first conjugation type). Pisát’ ‘to write’ belongs to the Aclass, and its 3Pl present tense form is píš-ut (-a- suffix is trun-cated, first conjugation type, final consonant alternation, stressshift). Drozát’ ‘to tremble’ belongs to the ZHA class, and its 3Plpresent tense form is droz-át (-a- suffix is truncated, second con-jugation type).

Thus, Russian verb system is very complex. And, crucially,there is no obvious division into regular and irregular verbs. Un-like in English, there is no single productive pattern that can beapplied to any stem irrespective of its phonological characteris-tics. Five verb classes are productive, but dramatically differ intype frequency. The Grammatical Dictionary of the Russian Lan-guage (Zaliznyak, 1977) contains 27,409 verbs. We counted thenumber of verbs in these five classes: 11,735 in the AJ class,6875 in the I class, 2815 in the OVA class, 1377 in the NU classand 638 in the EJ class.

1.2. A brief description of the Russian noun system

Russian noun system is much less complex than Russian verbsystem. Nouns are inflected for number and case and are classifiedinto different declensions depending on their gender and on the setof their number and case endings. There are three main declen-sions, the forth declension with adjectival endings, several excep-tional nouns and a number of uninflected nouns. First andsecond declensions usually have a choice of two endings for a par-ticular form depending on the last consonant of the stem (all stemsin the third declension end in palatal or sibilant consonants anduse one set of endings). In addition to that, inside every declensionthere are small groups of nouns with various irregularities: minorstem changes or unusual endings in some forms. The majority ofRussian nouns do not change their stem.

Some endings are unique for a particular declension, but mostof them are shared by two or even three main declensions. In par-ticular, Nom.Pl forms, which we looked at in this study, can havethe following endings: �i (used with palatal, sibilant and velarstems of masculine and feminine nouns in all three declensions),�y (used with the other stems of masculine and feminine nounsin the first and second declension), �ja (used with palatal, sibilantand velar stems of neuter and some masculine nouns in the firstdeclension), �a (used with the other stems of neuter and somemasculine nouns in the first declension), �e (used in a very smallgroup of animate masculine nouns in the first declension).

1.3. Previous studies testing the SS and DS approaches on Russian

The predictions of the SS and DS theories were tested innumerous experiments with Russian verbs (e.g. Chernigovskaya,Tkachenko, Dalbi, & Svistunova, 2007; Gor, 2003, 2010; Gor &Chernigovskaya, 2001; Gor & Chernigovskaya, 2003; Gor &Chernigovskaya, 2005; Gor & Jackson, 2013; Gor, Svistunova,Petrova, Khrakovskaya, & Chernigovskaya, 2009; Svistunova,2008; Tkachenko & Chernigovskaya, 2010). Adult native speakers,L1 and L2 learners and subjects with various neurological anddevelopmental deficits were examined. In the majority of theseexperiments, participants were provided with infinitives or pasttense forms of real or nonce verbs and prompted to generate 1Sgand 3Pl present tense forms.

Healthy adult native speakers showed a strong tendency toovergeneralize the AJ class pattern (AJ class is the most frequent).In particular, they applied it to the nonce verbs ending in �ili (onlytwo real verbs and their derivates have this conjugational pattern,all the others belong to the productive and highly frequent I class)and to the ones ending in �yli (no real verbs have this conjuga-tional pattern). Thus, despite the fact that Russian has several pro-ductive and highly frequent classes, one conjugational pattern isused as the default one. This is in conflict with the SS theory.

Four-year-old children also heavily rely on the AJ class pattern.But gradually, other patterns become more active. For example,around the age of five children stop making mistakes with OVAclass verbs and actively overgeneralize this pattern. Overgeneral-izations that do not respect phonological properties of the stemare a hallmark of a rule in the DS approach, and several rules thathave different potential to be overgeneralized depending on vari-ous frequency-related and phonological factors contradict its veryessence. The generalizations made in the studies of English-speak-ing subjects with SLI (specific language impairment), aphasiacdeficits and Alzheimer disease (e.g. Ullman et al., 1997b) that sup-ported the DS approach also were not borne out in Russian. Thegroup of authors working on Russian argues that Yang’s (2002)model relying on multiple rules of different status may be bettersuited to account for their findings. A similar model for Russianis developed in (Gor, 2003).

34 N. Slioussar et al. / Brain & Language 130 (2014) 33–41

Author's personal copy

1.4. Previous imaging studies

The data from most previous functional imaging studies wereinterpreted in favor of the DS theory (Beretta et al., 2003; Dhondet al., 2003; Indefrey et al., 1997; Jaeger et al., 1996; Oh et al.,2011; Sahin et al., 2006; Ullman, Bergida, & O’Craven, 1997a). DeDiego-Balaguer et al. (2006) appeal both to DS and to SS ap-proaches. In all these studies, regular and irregular verbs wereassociated with different sets of activated regions: Jaeger et al.(1996) compared past tense generation with a baseline of reading,while in all subsequent studies, direct regular vs. irregular compar-isons were made. But these sets did not coincide across studies.One can only say that irregulars usually produced more extensiveactivation.

According to the version of the DS approach developed byUllman (2004), memory-based processing of irregular verbsdepends on medial-temporal and temporo-parietal regions, whilerule-based processing of regular verbs depends on the basal gan-glia, Broca’s area, and neighboring anterior regions. No other prin-cipally different model was proposed, although many authorsdisagree with Ullman in details. However, only two above-men-tioned studies found more activation in Broca’s area for regularsthan for irregulars (Dhond et al., 2003; Oh et al., 2011). In additionto that, increased left IFG activation for regulars was observed in anfMRI study where processing of spoken regular and irregular formswas compared in a same-different judgment task (Tyler, Stamata-kis, Post, Randall, & Marslen-Wilson, 2005). Several other studiesobserved the reverse pattern: Broca’s area was more activated byirregulars (Beretta et al., 2003; de Diego-Balaguer et al., 2006;Sahin et al., 2006). Sahin et al. (2006) suggest that this can beexplained by conflict monitoring between the regular rule and anirregular form or by inhibition of the regular rule application.

In several studies the SS approach was preferred. Sach et al.(2004) observed only frequency effects, but no difference betweenregular and irregular verbs. Joanisse and Seidenberg (2005) foundgreater bilateral IFG activation for regulars than for irregulars,but then demonstrated that irregulars that were phonologicallysimilar to regulars (e.g., slept, fled, sold) produced the same levelof activation as regulars did, and significantly more activation thanirregulars that were not phonologically similar to regulars did (e.g.,took, gave). They conclude that observed activation patterns arebetter predicted by phonological properties of stimuli than by theirregularity.

Desai et al. (2006) found that, when word frequency andphonological complexity are controlled for, the ‘regular > irregular’comparison revealed no activated regions, while the ‘irregu-lar > regular’ comparison was associated with greater bilateral acti-vation of the posterior IFG (BA 44), the precentral gyrus, theanterior insula, the intraparietal sulcus (IPS), the basal ganglia, aswell as some other small foci of activation. The authors note thatthese areas are commonly associated with executive control andattentional processes and are also activated by regular past tensegeneration compared to reading. They conclude that the observedactivation differences reflect greater processing load posed byirregulars, which rely on less frequent inflectional patterns thanregular verbs and therefore have greater attentional and responseselection demands.

1.5. Present study

In the present study, real and nonce verbs and nouns were usedas stimuli (see Appendix). Verbs were visually presented in theinfinitive form, and nouns were presented in the Nom.Sg form.Subjects were asked to generate aloud as fast as possible the 1Sgpresent tense form if they see a verb or the Nom.Pl form if theysee a noun. All oral responses were recorded simultaneously with

fMRI data acquisition. Their correctness was assessed offline. Whena response was no longer appropriate to the target’s category, thecorresponding trial was discarded in the subsequent fMRI analyses.

The first group of 35 verbs belonged to the AJ class, the secondgroup contained 35 verbs from several small non-productive clas-ses. For the sake of brevity, we will further call the latter verbsirregular. The AJ class is productive and the most frequent. So wereasoned that if any differences between these two groups werefound, we could further compare them to other groups, and ifnot, no other differences would be expected because these arethe two poles of the Russian verb system (i.e. in the latter case,we will not be able to further exploit the diversity of this system,but this will become clear only when we have the results of thepresent study, because no similar fMRI experiments on morpho-logically rich languages have been done before).

Verbs in these two groups were matched for frequency usingThe Frequency Dictionary of Modern Russian Language (Lyashevskaya& Sharoff, 2009) and for length (see Appendix). Only unprefixedimperfective verbs were used. We also created two sets of 35nonce verbs. They mimicked the general characteristics of the cor-responding real verb groups (length and phonological properties ofthe stem). However, we tried to avoid close resemblance to partic-ular real words, because this would be an extra factor that is hardto control for.

Two groups of 35 nouns were matched with the verb groups forfrequency and length. All nouns were masculine, belonged to thefirst declension and had the Nom.Pl form ending in �y. In onegroup, the last vowel of the stem is dropped in many forms includ-ing the Nom.Pl form: e.g. koster ‘fire’ – kostry. In the other group,the stem remains the same in all forms, as is standard for Russiannouns: e.g. šofer ‘driver’ – šofery. For the sake of brevity, we willfurther call the first group irregular, although this is a relativelyminor irregularity. As we noted above, Russian nouns have less di-verse inflectional paradigms than Russian verbs, so this was theonly irregular feature we could find in the necessary frequencyrange. Our goal was to analyze whether the effects of regularitywould be comparable in verbs and nouns. We also had two corre-sponding groups of 35 nonce nouns.

Let us note that vowels are dropped only in a subgroup of nounstems ending in particular vowel and consonant clusters (e.g. �er,�or, �el, �ol etc.). We chose stems with such clusters both forirregular and for regular nouns not to make the first group muchmore phonologically homogenous than the second. In the majorityof cases whether the vowel is dropped can be predicted from thecombination of consonants before this vowel and from the positionof the stress. In case of nonce nouns, participants did not knowwhat the intended stress was, so different Nom.Pl forms could belicitly derived from them.

In total, there were eight types of stimuli: regular verbs (RV),irregular verbs (IV), regular nonce verbs (RNV), irregular nonceverbs (INV), regular nouns (RN), irregular nouns (IN), regular noncenouns (RNN) and irregular nonce nouns (INN). We analyze thembelow using Regularity and Lexicality factors (we looked at verbsand nouns separately rather than putting them together and treat-ing word category as the third factor primarily because the type ofirregularity we could find for nouns was very minor compared towhat we had in case of verbs). 21 Subjects took part in our study,so in total we had 735 responses in each category.

2. Results and discussion

Let us start with behavioral results. Participants made fewermistakes with regular verbs than with irregular verbs (22 out of735 and 71 out of 735, or 3.0% and 9.7%, respectively) and fewermistakes with nonce regular verbs than with nonce irregular verbs

N. Slioussar et al. / Brain & Language 130 (2014) 33–41 35

Author's personal copy

(96 out of 735 and 320 out of 735, or 13.1% and 45.7%, respec-tively). A two-way RM ANOVA revealed significant main effectsof Lexicality (F(1, 20) = 83.23; p < 0.001; g2 = 0.81) and Regularity(F(1, 20) = 32.29; p < 0.001; g2 = 0.62), but no significant interac-tion of these factors. No subject had more than 25% mistakes inthe RV, IV and RNV categories, which is often taken as the behav-ioral performance threshold. At the same time, all subjects made alot of mistakes with nonce irregular verbs, so this reflects theobjective difficulty of using unproductive and infrequent conjuga-tional patterns included in this group. Since our stimulus groupswere large, we still had enough imaging data for the analysis, butthe findings from this particular group should be taken with a cer-tain caution.

Before analyzing noun stimuli, we had to take into account thatsignificantly more irregular nonce nouns were inflected as regularones than vice versa (41 out of 735 and 259 out of 735, or 5.6% and32.5%, respectively, p < 0.001 according to the chi-square test).

Since this is licit from the point of view of Russian grammar, wemoved these responses to the appropriate new category whenworking with behavioral and fMRI data. After this was done, wecalculated that participants made fewer mistakes with regularnouns than with irregular nouns (11 out of 735 and 21 out of735, or 1.5% and 2.9%, respectively) and fewer mistakes with regu-lar nonce nouns than with irregular nonce nouns (45 out of 953and 64 out of 517, or 4.7% and 12.4%, respectively). A two-wayRM ANOVA revealed significant main effects of Lexicality(F(1, 20) = 45.17; p < 0.001; g2 = 0.69), Regularity (F(1, 20) = 13.57;p < 0.001; g2 = 0.4) and their interaction (F(1, 20) = 7.47;p < 0.001; g2 = 0.27).

Let us turn to imaging results. A full factorial ANOVA of verbgeneration trials showed that the main effects of Lexicality andRegularity were significant, while their interaction was not. A sig-nificant effect of Regularity was demonstrated for the left inferiorand middle frontal gyrus (IFG/MFG; BA 44/45/46/9/6), the left

Table 1Brain areas sensitive to Lexicality and Regularity.

Brain region Z-value k Peak MNI coordinates

x y z

1. Main effect of Lexicality for verb trialsa

1.1. Real verbs > nonce verbsL. angular gyrus (BA 39/40) 6.83 149 �51 �64 28L. middle frontal gyrus (BA 8) 5.75 18 �30 29 49L. precuneus (BA 31) 5.04 17 �9 �49 37R. angular gyrus/superior temporal gyrus (BA 39) 5.12 16 51 �58 22

1.2. Nonce verbs > real verbsL. inferior parietal lobule (BA 40/7) 5.36 134 �30 �52 46L. inferior frontal/precentral gyrus (BA 9/6) 5.84 80 �45 5 31R. superior parietal lobule/angular gyrus (BA 7) 5.86 61 27 �64 46R. inferior parietal lobule/supramarginal gyrus (BA 40) 5.90 59 42 �37 40R. cerebellum 5.70 39 39 �61 �29L. medial frontal gyrus (BA 6) 5.05 22 �6 11 52

2. Main effect of Regularity for verb trialsb

2.1. Regular verbs > irregular verbsR. angular gyrus (BA 39) 5.42 27 51 �61 25

2.2. Irregular verbs > regular verbsL. precentral gyrus (BA 44) 6.61 542 �45 5 31L. middle frontal/inferior frontal gyrus (BA 44/45/46/9) 6.44 �45 23 34L. middle frontal gyrus (BA 10) 6.23 �45 50 1L. inferior parietal/superior parietal lobule (BA 40/7) 6.46 240 �30 �52 43L. insula 5.6 59 �30 23 �2R. cerebellum 5.91 27 33 �64 �29L. supplementary motor area (BA 6) 5.35 26 �6 11 52

3. Main effect of Lexicality for noun trialsa

3.1. Real nouns > nonce nounsL. precuneus/posterior cingulate (BA 31) 5.55 134 �6 �52 31L. angular gyrus (BA 39) 5.96 110 �48 �64 25R. angular gyrus/ superior temporal gyrus (BA 39) 5.06 20 51 �58 25

3.2. Nonce nouns > real nounsL. precentral gyrus (BA 44) 7.78 986 �45 5 31L. insula 6.54 �33 17 4L. inferior frontal/middle frontal gyrus (44/45/46/9/6) 6.39 �42 29 22L. inferior parietal lobule (BA 7/40) 7.29 471 �39 �49 46L. superior parietal lobule (BA 7) 7.2 �27 �67 46L. supplementary motor area (BA6) 6.6 167 �6 11 52R. superior parietal lobule/angular gyrus (BA7) 5.58 112 27 �67 46L. cuneus/occipital lobe (BA 17/18) 5.33 44 �12 �73 7R. inferior parietal lobule (BA 40) 5.47 26 45 �34 49R. cerebellum 5.33 21 30 �64 �26R. inferior frontal gyrus (BA 45) 5.13 21 48 29 25R. precentral gyrus (BA 44) 5.19 20 48 5 28

BA, approximate Brodmann’s area; L/R, left/right hemisphere; k, cluster size in voxelsa Clusters were identified by applying the mask of positive (‘real stimuli > nonce stimuli’ t-contrast) and negative (‘nonce stimuli > real stimuli’ t-contrast) effects of

Lexicality as it is implemented in the SPM8 full-factorial second level analysis.b Clusters were identified by applying the mask of positive (‘regular stimuli > irregular stimuli’ t-contrast) and negative (‘irregular stimuli > regular stimuli’ t-contrast)

effects of Regularity as it is implemented in the SPM8 full-factorial second level analysis.

36 N. Slioussar et al. / Brain & Language 130 (2014) 33–41

Author's personal copy

insula, the left inferior/superior parietal lobule (IPL/SPL; BA 40/7),the right angular gyrus and the right cerebellum (see Table 1 andFig. 1). Estimation of the beta values within the observed clustersindicated that in all explored regions except for the right angulargyrus, there was greater activity in irregular real and irregularnonce verb conditions. Only one brain area in the right angulargyrus exhibited a smaller decline in activity in the regular real verbcondition compared with the others. A significant effect of Lexical-ity was demonstrated for a similar set of regions in the fronto-pari-etal brain network (see Table 1). Additionally, three brain areasexhibited smaller decline in activity in the real verb conditionscompared to nonce verb conditions. A full factorial ANOVA of noungeneration trials demonstrated only a significant main effect ofLexicality (see Table 1). The set of regions in the fronto-parietalbrain network influenced by this factor is similar to the ones thatare affected by Regularity and Lexicality in the verb conditions.

Let us focus on Regularity first. Our data suggest that irregularverb production leads to greater activation of the fronto-parietalbrain network comprising the inferior frontal gyrus and the infe-rior/superior parietal lobule of the left hemisphere. As we showedin Section 1.4, several authors have recently reported similar re-sults and offered two alternative explanations. Our study confirmsthat this is a robust crosslinguistic phenomenon, which can be ob-served not only in the languages with sharp regularity contrasts,puts it in a broader perspective and provides an important argu-ment for one of the existing explanations.

Choosing between possible explanations, we hypothesized thatif increased activation of this network reflects greater processingload, in particular, greater working memory and response selectiondemands, it should increase not only from regular to irregularstimuli, but also from real to nonce stimuli. The fact that the setof brain regions influenced by Regularity largely overlaps withthe one influenced by Lexicality already points in this direction(see Table 1). To model a linear increase of processing difficultyfrom RV to IV to RNV to INV trials, we used an(RV > B) < (IV > B) < (RNV > B) < (INV > B) parametric t-contrast,where B is an implicitly modeled baseline (null-event). This

contrast revealed clusters of BOLD increase in the left IFG/MFG(BA 44/45/9/6), the right MFG (BA 46), the left insula, the left sup-plementary motor area (SMA; BA 6), the left and right IPL/SPL (BA40/7), the right angular gyrus (BA 7) and the right cerebellum (seeTable 2 and Fig. 2).

Additionally, using conjunction analysis we demonstrated thatthe sets of brain areas sensitive to Regularity and to the increaseof processing difficulty largely overlap. The following common re-gions were identified: the left IFG/MFG (BA 44/9/6), the left IPL/SPL(BA 40/7), the left insula, the right angular gyrus and the right cer-ebellum (see Fig. 3 and Table 3). Our hypothesis is further sup-ported by the fact that the number of errors in subjects’responses was significantly influenced both by Regularity and byLexicality.

Although we did not observe a significant influence of Regular-ity on the BOLD signal changes for nouns, a parametric t-contrastanalogous to the one used for verbs ((RN > B) < (IN > B) < (RNN >B) < (INN > B)) revealed practically the same set of brain areas ofsignificant BOLD changes (see Table 2 and Fig. 2). The number oferrors in subjects’ responses also increased from RN to IN to RNNto INN. The fact that similar results were obtained for two differenttypes of regular and irregular real and nonce stimuli strongly sup-ports our hypothesis. We explain the absence of a significant Reg-ularity effect on the BOLD signal changes for nouns by the fact thata dropped vowel is a relatively minor irregularity for an inflectionalparadigm in Russian.

To conclude, we argue that increased left IFG activation associ-ated with irregular verb stimuli cannot be explained by conflictmonitoring between the regular rule and an irregular form or byinhibition of the regular rule application, as suggested by Sahinet al. (2006). The former option is out because no form is storedfor irregular nonce verbs. The latter option is undermined by thefact that the IFG exhibits greater activity not only when irregularverbs are produced, but also when nonce verbs are generated,and a very similar pattern can be observed for nouns in the para-metric t-contrast. This finding can be explained by the processingload account along the lines proposed by Desai et al. (2006)

Fig. 1. Brain areas sensitive to Regularity (in verb trials). Areas with greater activity in irregular real and irregular nonce verb conditions are shown in the hot scale, the areawith a smaller decline in activity in the regular verb condition is shown in blue. Plots show effect sizes with 95% confidential intervals. All presented clusters are obtainedafter the FWE p < 0.05 correction at the voxel level. L/R, left/right hemisphere; IFG, inferior frontal gyrus; MFG, middle frontal gyrus; SMA, supplementary motor area; IPL,inferior parietal lobule; SPL, superior parietal lobule; RV, regular verbs; IV, irregular verbs; RNV, regular nonce verbs; INV, irregular nonce verbs.

N. Slioussar et al. / Brain & Language 130 (2014) 33–41 37

Author's personal copy

because the increase of processing load is expected in both com-parisons, for verbs and for nouns equally.

Let us clarify this point. Nouns and verbs termed ‘regular’ in ourstudy have inflectional patterns that are more frequent and are ap-plied to a wider variety of words compared to ‘irregular’ stimuli.This makes these patterns more active in the single integrated sys-tem where all forms are processed according to the SS approach, soboth existing and new forms using these patterns pose less atten-tional and response selection demands and are easier to generate.Nonce words of any kind are more difficult to inflect than realwords because no information is stored for them.

The functional role of the left IFG in the context of the presentstudy can be associated with ‘general-purpose’ control processesthat involve attention, working memory and decision-making ratherthan with some specifically linguistic tasks. Similar claims have beenmade by various other authors (e.g. Bornkessel-Schlesewsky &Schlesewsky, 2013; Koechlin & Summerfield, 2007; Novick,Trueswell, & Thompson-Schill, 2005; Novick, Trueswell, & Thomp-son-Schill, 2010). Moreover, these regions were demonstrated toreact to increasing processing load, namely to the difficulty of stim-ulus–response mapping, in non-linguistic experiments (e.g. Brass,Wenke, Spengler, & Waszak, 2009; Schumacher & D’Esposito, 2002).

Table 2Parametric t-contrasts modeling the linear increase of processing difficulty.

Brain region Z-value k Peak MNI coordinates

x y z

Parametric contrast (RV > B) < (IV > B) < (RNV > B) < (INV > B)L. precentral gyrus (BA 44) 7.21 493 �45 5 31L. inferior/middle frontal gyrus (BA 44/45/46/9/6) 6.27 �48 20 31L. inferior parietal lobule (BA 40) 6.73 457 �30 �52 43L. superior parietal lobule (BA 7) 6.11 �24 �61 43L. supplementary motor area (BA 6/32) 6.25 172 �6 11 52R. supplementary motor area /middle cingulum (BA 6) 5.53 9 20 43R. angular gyrus (BA7) 6.56 138 27 �64 46R. superior/inferior parietal lobule (BA 7) 4.74 30 �55 55R. supramarginal gyrus (BA 40) 6.61 126 42 �37 40R. cerebellum 6.71 101 36 �61 �29L. insula 5.75 81 �30 23 �2R. middle frontal gyrus (BA 46) 5.89 34 45 35 22L. superior/middle frontal gyrus (BA 6) 4.92 22 �24 �4 55

Parametric contrast (RN > B) < (IN > B) < (RNN > B) < (INN > B)L. precentral/inferior/middle frontal gyrus (BA 44/45/46/6/9) 7.79 1051 �48 8 31L. insula 6.51 �33 17 4L. inferior parietal lobule (BA 40) 7.5 530 �39 �49 46L. superior parietal lobule (BA 7) 7.33 �27 �67 46L. supplementary motor area 6.59 165 �6 11 52R. superior parietal lobule/superior occipital gyrus 5.56 101 27 �67 46R. angular gyrus (BA 7) 5.50 33 �61 49R. inferior/middle frontal gyrus (BA 45/46) 5.42 38 48 29 25R. inferior parietal lobule (BA 40) 5.49 35 45 �34 49R. inferior frontal/ precentral gyrus (BA 44) 5.18 30 48 5 28L. cuneus/occipital lobe (BA 17/18) 5.09 28 �12 �73 7L. lentiform nucleus/pallidum 5.29 24 �9 �1 �1R. cerebellum 5.43 24 30 �61 �29

BA, approximate Brodmann’s area; L/R, left/right hemisphere; RV, regular verbs; IV, irregular verbs; RNV, regular nonce verbs; INV, irregular nonce verbs; RN, regular nouns;IV, irregular nouns; RNV, regular nonce nouns; INV, irregular nonce nouns; k, cluster size in voxels.

Fig. 2. Brain areas sensitive to the increase of processing difficulty. The increase of processing difficulty was modeled by two parametric t-contrasts:(RV > B) < (IV > B) < (RNV > B) < (INV > B) for verb stimuli (red) and (RN > B) < (IN > B) < (RNN > B) < (INN > B) for noun stimuli (green). Overlapping areas are shown inbrown. All presented clusters are obtained after the FWE p < 0.05 correction at the voxel level. L/R, left/right hemisphere; IFG, inferior frontal gyrus; MFG, middle frontalgyrus; SMA, supplementary motor area; IPL, inferior parietal lobule; SPL, superior parietal lobule.

38 N. Slioussar et al. / Brain & Language 130 (2014) 33–41

Author's personal copy

Thus, while most previously obtained findings were interpretedin favor of DS models, the results of our experiment are more read-ily compatible with the SS approach, where all forms are processedin a single integrated system. However, our results do not bear onthe question of how this system works, in particular, whether itcontains no rules, as in connectionist models, or many rules of dif-ferent status, as in the model developed by Yang (2002) andadapted for Russian in (Gor, 2003).

3. Methods

3.1. Subjects

Subjects were 21 native speakers of Russian (13 women),19–32 years of age, with no history of neurological or psychologi-cal disorders. All participants were right-handed, as assessed bythe Edinburgh Handedness Inventory (Oldfield, 1971). Subjectswere given no information about the specific purpose of the study.All subjects gave their written informed consent prior to the studyand were paid for their participation. All procedures were in accor-dance with the declaration of Helsinki and were approved by theEthics Committee of the N.P. Bechtereva Institute of the HumanBrain, Russian Academy of Sciences.

3.2. Language protocol and experimental fMRI paradigm

Stimuli were presented for 700 ms on a screen mounted insidethe magnet just in front of the subjects’ eyes (Eloquence fMRISystem). We used varied ISI by presenting fixation crosses(‘‘xxxxx’’) between stimuli for 3100, 3200, 3300, 3400 or

3500 ms. Additionally, 280 stimuli were pseudo-randomly inter-mixed with 140 ‘‘null-events’’ (fixation crosses) for attaining base-line level of BOLD signal between events (Friston, Zarahn, Josephs,Henson, & Dale, 1999). We used three experimental runs with2–5 min rest between them. First 10 dummy scans of each runwere discarded. Stimulus administration and synchronization withthe functional image acquisition was controlled by the E-Primesoftware (version 1.1, Psychology Software Tools Inc., Pittsburgh,PA, USA). Before the scanning, each subject performed a practicerun. All oral responses were recorded simultaneously with fMRIdata acquisition by means of the Persaio MRI Noise CancellationSystem (Psychology Software Tools, Inc.).

3.3. Behavioral data analysis

The percentage of errors in different experimental conditionswith verb stimuli was log transformed and then analyzed as anindependent variable using the two-way repeated measure analy-sis of variance (RM ANOVA) with two within-subject factors: Lex-icality (real and nonce verbs) and Regularity (regular and irregularverbs). Partial g2 values were calculated to assess the size of signif-icant main effects and interactions in terms of proportion ofexplained variance. The same analysis was conducted for condi-tions with noun stimuli. The analysis was implemented using theSTATISTICA software (Statsoft, Tulsa, OK, USA).

3.4. MR imaging protocol

Magnetic resonance imaging was performed on a 3 Tesla PhilipsAchieva scanner. In addition to a scout sequence, subjectsunderwent structural and functional imaging. Structural images

Fig. 3. Common brain areas sensitive to Regularity and processing difficulty (in verb trials). Brain areas revealed in the conjunction analysis (blue) of two effects: Regularity(red) and the increase of processing difficulty modeled by the (RV > B) < (IV>B) < (RNV > B) < (INV > B) parametric F-contrast (green). Overlaps between these three types ofclusters are shown in violet. All presented clusters are obtained after the FWE p < 0.05 correction at the voxel level. L/R, left/right hemisphere; IFG, inferior frontal gyrus; MFG,middle frontal gyrus; SMA, supplementary motor area; IPL, inferior parietal lobule; SPL, superior parietal lobule.

Table 3Common brain areas sensitive to Regularity and processing difficulty (in verb trials). Clusters were identified in a conjunction analysis of the main effect of Regularity and the(RV > B) < (IV > B) < (RNV > B) < (INV > B) parametric F-contrast.

Brain region Z-value k Peak MNI coordinates

x y z

L. precentral (BA 44) 6.61 319 �45 5 31L. inferior/middle frontal gyrus (BA 44/9/46) 5.67 �48 8 16L. inferior/superior parietal lobule (BA 40/7) 6.46 185 �30 �52 43L. insula 5.6 51 �30 23 �2L. supplementary motor area (BA 6) 5.36 26 �6 11 52R. cerebellum 5.92 26 33 �64 �29R. angular gyrus (BA 39) 5.42 21 51 �61 25

BA, approximate Brodmann’s area; L/R, left/right hemisphere; k, cluster size in voxels.

N. Slioussar et al. / Brain & Language 130 (2014) 33–41 39

Author's personal copy

were acquired applying a T1-weighted pulse sequence (T1 W-3D-FFE;TR = 2.5 ms; TE = 3.1 ms; 30� flip angle) measuring 130 axial slices(field of view, FOV = 240 mm � 240 mm; 256 � 256 scan matrix) of0.94 mm thickness. Functional images were obtained using anecho planar imaging (EPI) sequence (TE = 35 ms; 90� flip angle;FOV = 208 mm � 208 mm; 128 � 128 scan matrix). 32 Continuous3.5 mm thick axial slices (voxel size = 3 mm � 3 mm �3.5 mm) cov-ering the entire cerebrum and most of the cerebellum were orientedwith reference to the structural image. The images were acquiredwith a repetition time (TR) of 2000 ms. In order to avoid extensivehead motions, we used an MR-compatible soft cervical collar.

3.5. Functional image processing and statistical analysis

Image preprocessing and statistical analysis were performedusing the Statistical Parametric Mapping (SPM8) routines (Well-come Department of Cognitive Neurology, London, UK) runningin the MATLAB platform (Mathworks Inc., Natick, MA, USA). EPIimages of each subject were realigned to the first image in eachtime series and corrected for residual head motions. Subsequently,realigned EPI images were normalized to the MNI standard stereo-tactic space based on the mean functional image generated duringthe realignment. The images were re-sampled to an isometric vox-el size of 3 mm, and 8 mm full-width/half-maximum isotropicGaussian smoothing was applied to all functional images prior toanalysis.

At the first level analysis, all experimental events of a particulartype (onsets of word presentation) were introduced in the generallinear model (GLM) as regressors: RV, IV, RNV, INV, RN, IN, RNN,INN and Errors. Each regressor was convolved with a canonicalHRF, and a temporal high pass filter (cutoff: 128 s) was applied.Translations and rotations in x, y and z directions produced atthe realignment stage were included in the GLM as confoundingcovariates in order to control for residual head movement artifacts.For every subject, t-contrasts between each experimental condi-tion and an implicitly modeled baseline (B) were calculated:RV > B, IV > B, RNV > B, INV > B, RN > B, IN > B, RNN > B, INN > B.

At the second level random effect analysis, obtained beta mapswere analyzed using the full factorial ANOVA with two repeatedmeasure factors: Lexicality with two levels for real and nonce stim-uli and Regularity with two levels for regular and irregular stimuli.For controlling false positives, a family-wise error correction was ap-plied at the voxel level (p < 0.05), and only clusters with more than15 voxels were reported. Additionally beta values from the revealedsignificant BOLD change clusters were extracted by the REX toolbox(http://www.nitrc.org/projects/rex/) and presented in Fig. 1.

Betas obtained in the RV > B, IV > B, RNV > B and INV > B con-trasts were used to model a linear increase of processing difficultyin the (RV > B) < (IV > B) < (RNV > B) < (INV > B) parametric t-con-trast (Fig. 2). Assuming a similar effect for noun stimuli, the(RN > B) < (IN > B) < (RNN > B) < (INN > B) t-contrast was also cal-culated (Fig. 2). To assess whether there are brain areas associatedwith verb production that are sensitive both to Regularity and tothe increase of processing difficulty, a conjunction analysis of themain effect of Regularity and the (RV > B) < (IV > B) < (RNV >B) < (INV > B) parametric F-contrast was used. In these contrastswe applied a corrected p < 0.05 voxel-wise threshold (k > 15). Theanatomical location of revealed significant BOLD changes wasidentified by the Anatomy toolbox (Eickhoff et al., 2005). For illus-tration purposes we used the template developed for MRIcroGL(http://www.cabiatl.com/mricrogl/).

Acknowledgment

The study was partially supported by the Grant #12-06-00706from the Russian Foundation for Humanitarian Sciences (RFHR)

and by the Grant #0.38.518.2013 from St. Petersburg StateUniversity..

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.bandl.2014.01.006.

References

Berent, I., Pinker, S., & Shimron, J. (1999). Default nominal inflection in Hebrew:Evidence for mental variables. Cognition, 72, 1–44.

Beretta, A., Campbell, C., Carr, T. H., Huang, J., Schmitt, L. M., Christianson, K., et al.(2003). An ER-fMRI investigation of morphological inflection in German revealsthat the brain makes a distinction between regular and irregular forms. Brainand Language, 85, 67–92.

Bornkessel-Schlesewsky, I., & Schlesewsky, M. (2013). Reconciling time, space andfunction: A new dorsal-ventral stream model of sentence comprehension. Brainand Language, 125, 60–76.

Brass, M., Wenke, D., Spengler, S., & Waszak, F. (2009). Neural correlates ofovercoming interference from instructed and implemented stimulus–responseassociations. The Journal of Neuroscience, 29, 1766–1772.

Chernigovskaya, T., Tkachenko, E., Dalbi, I., & Svistunova, T. (2007). Osobennostiporozdenija glagol’nyx form pacientami s bolezn’ju Alcgejmera (‘Generation ofverb forms in Alzheimer’s disease’). In Materials of the XXXVI internationalphilological conference (pp. 73–178). St. Petersburg: St. Petersburg UniversityPress.

Clahsen, H. (1999). Lexical entries and rules of language: A multidisciplinary studyof German inflection. Behavioral and Brain Sciences, 22, 991–1060.

Clahsen, H., Aveledo, F., & Roca, I. (2002). The development of regular and irregularverb inflection in Spanish child language. Journal of Child Language, 29, 591–622.

Davidson, D. E., Gor, K. S., & Lekic, M. D. (1996). Russian: Stage one: Live fromMoscow! Dubuque, IO: Kendall Hunt Publishing Company.

de Diego-Balaguer, R., Rodriguez-Fornells, A., Rotte, M., Bahlmann, J., Heinze, H. J., &Münte, T. F. (2006). Neural circuits subserving the retrieval of stems andgrammatical features in regular and irregular verbs. Human Brain Mapping, 27,874–888.

Desai, R., Conant, L. L., Waldron, E., & Binder, J. R. (2006). FMRI of past tenseprocessing: The effects of phonological complexity and task difficulty. Journal ofCognitive Neuroscience, 18, 278–297.

Dhond, R. P., Marinkovic, K., Dale, A. M., Witzel, T., & Halgren, E. (2003).Spatiotemporal maps of past-tense verb inflection. Neuroimage, 19, 91–100.

Eickhoff, S. B., Stephan, K. E., Mohlberg, H., Grefkes, C., Fink, G. R., Amunts, K., et al.(2005). A new SPM toolbox for combining probabilistic cytoarchitectonic mapsand functional imaging data. NeuroImage, 25, 1325–1335.

Friston, K. J., Zarahn, E., Josephs, O., Henson, R. N. A., & Dale, A. M. (1999). Stochasticdesigns in event-related fMRI. NeuroImage, 619, 607–619.

Gor, K. (2003). The rules and probabilities model of native and second languagemorphological processing. St. Petersburg: St. Petersburg University Press.

Gor, K. (2010). Beyond the obvious: Do second language learners processinflectional morphology? Language Learning, 60, 1–20.

Gor, K., & Chernigovskaya, T. (2003). Mental lexicon structure in L1 and L2 631acquisition: Russian evidence. Glossos, 4. <http://slaviccenters.duke.edu/projects/glossos-journal/issues/issue-4>.

Gor, K., & Chernigovskaya, T. (2005). Formal instruction and the acquisition ofverbal morphology. In A. Housen & M. Pierrard (Eds.), Investigation in instructedsecond language acquisition (pp. 103–139). Berlin: Mouton de Gruyter.

Gor, K., & Chernigovskaya, T. (2001). Rules in the processing of Russian verbalmorphology. In G. Zybatow, U. Junghanns, G. Mehlhorn, & L. Szucsich (Eds.),Current issues in formal Slavic Linguistics (pp. 528–536). Frankfurt am Main:Peter Lang.

Gor, K., & Jackson, S. (2013). Morphological decomposition and lexical access in anative and second language: A nesting doll effect. Language and CognitiveProcesses, 28, 1065–1091.

Gor, K., Svistunova, T., Petrova, T., Khrakovskaya, M., & Chernigovskaya, T. (2009).Mental’nyj leksikon pri raspade jazykovoj sistemy u bol’nyx s afaziej:Eksperimental’noe issledovanie glagol’noj morfologii (‘The decay of themental lexicon in aphasia: An experimental study of verb morphology’).Voprosy Jazykoznanija, 5, 3–17.

Hahn, U., & Nakisa, R. C. (2000). German inflection: Single route or dual route?Cognitive Psychology, 41, 313–360.

Indefrey, P., Brown, C., Hagoort, P., Herzog, H., Sach, M., & Seitz, R. (1997). A PETstudy of cerebral activation patterns induced by verb inflection. Neuroimage, 5,S548.

Jaeger, J. J., Lockwood, A. H., Kemmerer, D. L., Van Valin, R. D., Jr., Murphy, B. W., &Khalak, H. G. (1996). A positron emission tomographic study of regular andirregular verb morphology in English. Language, 72, 451–497.

Jakobson, R. O. (1948). Russian conjugation. Word, 4, 155–167.Joanisse, M. F., & Seidenberg, M. S. (2005). Imaging the past: Neural activation in

frontal and temporal regions during regular and irregular past-tense processing.Cognitive, Affective and Behavioral Neuroscience, 5, 282–296.

40 N. Slioussar et al. / Brain & Language 130 (2014) 33–41

Author's personal copy

Koechlin, E., & Summerfield, C. (2007). An information theoretical approach toprefrontal executive function. Trends in Cognitive Sciences, 11, 229–235.

Lavric, A., Pizzagalli, D., Forstmeier, S., & Rippon, G. (2001). A double-dissociation ofEnglish past-tense production revealed by event-related potentials and low-resolution electromagnetic tomography (LORETA). Clinical Neurophysiology, 112,1833–1849.

Lyashevskaya, O. N., & Sharoff, S. A. (2009). Castotnyj Slovar’ Sovremennogo RusskogoJazyka (‘The Frequency Dictionary of Modern Russian Language’). Moscow:Azbukovnik.

MacWhinney, B., & Leinbach, J. (1991). Implementations are not conceptualizations:Revising the verb learning model. Cognition, 40, 121–157.

Marslen-Wilson, W. D., & Tyler, L. K. (1997). Dissociating types of mentalcomputation. Nature, 387, 592–594.

Marslen-Wilson, W. D., & Tyler, L. K. (1998). Rules, representations, and the Englishpast tense. Trends in Cognitive Sciences, 2, 428–435.

McClelland, J. L., & Patterson, K. (2002). Rules or connections in past-tenseinflections: What does the evidence rule out? Trends in Cognitive Sciences, 6,465–472.

Münte, T. F., Say, T., Clahsen, H., Schiltz, K., & Kutas, M. (1999). Decomposition ofmorphologically complex words in English: Evidence from event-related brainpotentials. Cognitive Brain Research, 7, 241–253.

Newman, A. J., Izvorski, R., Davis, L., Neville, H. J., & Ullman, M. T. (1999). Distinctelectrophysiological patterns in the processing of regular and irregular verbs.Journal of Cognitive Neuroscience, 47–48.

Newman, A. J., Ullman, M. T., Pancheva, R., Waligura, D. L., & Neville, H. J. (2007). AnERP study of regular and irregular English past tense inflection. NeuroImage, 34,435–445.

Novick, J. M., Trueswell, J. C., & Thompson-Schill, S. L. (2005). Cognitive control andparsing: Reexamining the role of Broca’s area in sentence comprehension.Cognitive, Affective, & Behavioral Neuroscience, 5, 263–281.

Novick, J. M., Trueswell, J. C., & Thompson-Schill, S. L. (2010). Broca’s area andlanguage processing: Evidence for the cognitive control connection. Languageand Linguistics Compass, 4, 906–924.

Oh, T. M., Tan, K. L., Ng, P., Berne, Y. I., & Graham, S. (2011). The past tense debate: Isphonological complexity the key to the puzzle? Neuroimage, 57,271–280.

Oldfield, R. C. (1971). The assessment and analysis of handedness: The Edinburghinventory. Neuropsychologia, 9, 97–113.

Orsolini, M., Fanari, R., & Bowles, H. (1998). Acquiring regular and irregularinflection in a language with verb classes. Language and Cognitive Processes, 13,425–464.

Orsolini, M., & Marslen-Wilson, W. (1997). Universals in morphologicalrepresentation: Evidence from Italian. Language and Cognitive Processes, 12,1–47.

Pinker, S. (1991). Rules of language. Science, 253, 530–535.Pinker, S. (1999). Words and rules: The ingredients of language. New York: Harper

Collins.

Pinker, S., & Prince, A. (1988). On language and connectionism: Analysis of a paralleldistributed processing model of language acquisition. Cognition, 28, 73–193.

Plunkett, K., & Marchman, V. (1993). From rote learning to system building:Acquiring verb morphology in children and connectionist nets. Cognition, 48,21–69.

Plunkett, K., & Nakisa, R. C. (1997). A connectionist model of the Arabic pluralsystem. Language and Cognitive Processes, 12, 807–836.

Ragnasdóttir, H., Simonsen, H. G., & Plunkett, K. (1999). The acquisition of past tensemorphology in Icelandic and Norwegian children: An experimental study.Journal of Child Language, 26, 577–618.

Rumelhart, D., & McClelland, J. (1986). On learning the past tenses of English verbs.In D. Rumelhart & J. McClelland (Eds.), Parallel distributed processing:Explorations in the microstructure of cognition. Psychological and BiologicalModels (2, pp. 216–271). Cambridge: MIT Press.

Sach, M., Seitz, R. J., & Indefrey, P. (2004). Unified inflectional processing of regularand irregular verbs: A PET study. NeuroReport, 15, 533–537.

Sahin, N., Pinker, S., & Halgren, E. (2006). Abstract grammatical processing of nounsand verbs in Broca’s area: Evidence from fMRI. Cortex, 42, 540–562.

Schumacher, E. H., & D’Esposito, M. (2002). Neural implementation of responseselection in humans as revealed by localized effects of stimulus–responsecompatibility on brain activation. Human Brain Mapping, 17, 193–201.

Svistunova, T. (2008). Organizacija mental’nogo leksikona. Formirovanie i raspadglagol’noj slovoizmenitel’noj morfologii: Eksperimental’noe issledovanie namaterial russkogo jazyka (‘Mental lexicon structure. Development and decay ofthe verbal inflectional morphology: An experimental study of Russian’).Doctoral dissertation, St. Petersburg State University.

Tkachenko, E., & Chernigovskaya, T. (2010). Input frequencies in processing ofverbal morphology in L1 and L2: Evidence from Russian. Oslo Studies inLanguage, 2, 281–318.

Townsend, C. E. (1975). Russian word formation. Cambridge, MA: Slavica.Tyler, L. K., Stamatakis, E. A., Post, B., Randall, B., & Marslen-Wilson, W. (2005).

Temporal and frontal systems in speech comprehension: An fMRI study of pasttense processing. Neuropsychologia, 43, 1963–1974.

Ullman, M. T. (2004). Contributions of memory circuits to language: Thedeclarative/procedural model. Cognition, 92, 231–270.

Ullman, M. T., Bergida, R., & O’Craven, K. (1997a). Distinct fMRI activation patternsfor regular and irregular past tense. NeuroImage, 5, S549.

Ullman, M. T., Corkin, S., Coppola, M., Hickok, G., Growden, J., & Koroshetz, W.(1997b). A neural dissociation within language: Evidence that the mentaldictionary is part of declarative memory and that grammatical rules areprocessed by the procedural system. Journal of Cognitive Neuroscience, 9,289–299.

Yang, C. D. (2002). Knowledge and learning in natural language. Oxford: OxfordUniversity Press.

Zaliznyak, A. (1977). Grammatic�eskij Slovar’ Russkogo Jazyka. Slovoizmenenie (‘TheGrammatical Dictionary of the Russian Language. Inflection’), Moscow: RusskijJazyk.

N. Slioussar et al. / Brain & Language 130 (2014) 33–41 41