The role of morphological structure in the processing of aspectual verb forms in a Bulgarian-speaking aphasic patient

Journal of Neurolinguistics 17 (2004) 349–369 www.elsevier.com/locate/jneuroling

The role of morphological structure in the processing of aspectual verb forms in a Bulgarian-speaking aphasic patient Rossitza Nikolova*, Gonia Jarema Centre de recherche, Institut universitaire de ge´riatrie de Montre´al, Universite´ de Montre´al, 4565 Queen Mary Road, Montreal, Que., Canada H3W 1W5 Received 25 February 2002; accepted 2 October 2003

Abstract The present study investigates the on-line performance of a non-fluent Bulgarian-speaking aphasic patient and nine matched control participants in the recognition of prefixed verbs. A visual primed lexical decision experiment was employed to probe the role of aspect, semantic transparency, and root status (free vs. bound) in the recognition patterns obtained. Controls’ results showed aspect and transparency effects. The patient demonstrated facilitation for verb forms featuring bound roots, while showing inhibition on transparent forms with free-standing roots, pointing to difficulties with decomposition, but not with accessing stored forms. These results suggest that the patient had a problem at the level of morphological decomposition-based mechanisms. q 2003 Elsevier Ltd. All rights reserved. Keywords: Aphasia; On-line word recognition; Prefixed aspectual verbs; Semantic transparency; Root status; Bulgarian

1. Introduction Psycholinguistic theories differ significantly on the issue of how polymorphemic words are represented and processed. Holistic theories assume that morphologically complex words are stored in the lexicon and retrieved as whole words, similarly to morphologically simple words (Butterworth, 1983; Lukatela, Gligorievic´, Kostic´, & Turvey, 1980; Segui & Zubizaretta, 1985). In contrast, in a full-parsing model morphemes are the basic unit of * Corresponding author. Fax: þ 1-514-349-3548. E-mail address: [email protected] (R. Nikolova). 0911-6044/$ - see front matter q 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.jneuroling.2003.10.002

350

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

lexical representation and the components of complex forms are represented and processed separately, i.e. in decomposed form (Kempley & Morton, 1982; Taft, 1981; Taft & Forster, 1975). An intermediate position is that regular and phonologically or semantically transparent words are always processed through their constituent morphemes, whereas irregular idiosyncratic forms are stored as a whole (Pinker & Prince, 1994). Other authors argue that the whole-word and morphemic routes are processed in parallel (Caramazza, Laudanna, & Romani, 1988; Caramazza, Miceli, Silveri, & Laudanna, 1985). Furthermore, word frequency, frequency of the morphological components of the lexical item, productivity and transparency can play a role in determining the way in which polymorphemic words are processed (Baayen, Dijkstra, & Schreuder, 1997; Frauenfelder & Schreuder, 1991; Marslen-Wilson, Tyler, Waksler, & Older, 1994; Schreuder & Baayen, 1995; Zwitserlood, 1994). Unlike the modular models described above, connectionist models (Plaut & Gonnerman, 2000; Rumelhart & McClelland, 1986) specify that morphology is not explicitly represented, but appears from the correspondence between form and meaning. Accordingly, a general processing system treats all forms (regular/irregular, transparent/opaque) in the same manner and priming effects are due to the orthographic or semantic similarity of the forms, rather than to their morphological relatedness. Bybee’s (1995) network of associations model hypothesizes that high-frequency regular words could be stored in the lexicon and accessed directly, whereas low-frequency regular words require some access to the regular affixation schema. In other words, it is the frequency of use that determines the nature of storage and access, not structure. In the present study, we investigated prefixed aspectual verbal forms in Bulgarian. Several factors were manipulated to examine the processing of these forms. First, we addressed the role of semantic transparency of prefixed aspectual verbs while keeping inflectional suffixes constant. A polymorphemic word is semantically transparent if its meaning is composed of the meaning of its constituents. In contrast, the meaning of opaque complex forms is not compositional (Aronoff, 1976). Semantic transparency is important in determining how a morphologically complex word can be represented at the level of the lexical entry and can influence processing (Jarema, Busson, Nikolova, Tsapkini, & Libben, 1999; Marslen-Wilson et al., 1994; Sandra, 1990; Schreuder & Baayen, 1995; Zwitserlood, 1994). In their meta-model of morphological processing, Schreuder and Baayen (1995) have proposed the addition of concept nodes to the lexicon when the meaning of a complex word is not fully transparent semantically and is not obtained as a union of representational sets of the constituent elements. In different studies, facilitation was obtained for transparent forms and not at all, or a smaller facilitation, for opaque forms (Marslen-Wilson et al., 1994; Nikolova & Jarema, 2002; Rastle, Davis, Marslen-Wilson, & Tyler, 2000). In a visual simple lexical decision (LD) experiment, Slabakova (2001) investigated prefixed perfective verbs in Bulgarian and found slower response latencies for opaque forms when compared to transparent forms, and proposed a morphological account for both transparent and opaque forms. In her model she postulates two levels, a lexical level and a grammatical level. Prefixes of transparent forms have only the function of perfectivization (accomplishment of action), they do not have a lexical meaning. The recognition of the transparent perfective forms is successful after accessing only the grammatical level of the model. In contrast, prefixes in the opaque forms, apart

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

351

from the function of perfectivization, also have a lexical meaning because they change the meaning of the verbs. For opaque forms, both levels (grammatical and lexical) must be accessed, which accounts for their slower response latencies. It should be noted that our study differs from that of Slabakova (2001) in one important respect, notably the degree of opacity of the verb forms. Slabakova tested patients with opaque perfectives that contain a prefix denoting both the completion of the event and specification of the mode of action, or Aktionsart (e.g. obruli (transparent) ‘he shook down fruit from the tree’ vs. nabruli (opaque) ‘he shook down a lot of fruit from the tree’). Whereas the meaning is constant in both verbs (i.e. shaking down fruit from the tree) the second verb contains the concept of the intensity of the action (an Aktionsart). Moreover, we investigate a broader spectrum of linguistic phenomena, namely aspect (comparison between perfectifs and secondary imperfectifs) and root status in addition to semantic transparency. Neurolinguistic studies have investigated the issue of semantic transparency in compounds (Blanken, 2000; Libben, 1998). The patient of Libben (1998) did not recognize the majority of the compounds tested and tended to treat them as novel forms, i.e. as semantically transparent. However, some opaque compounds were produced correctly. Libben explained the patient’s performance with the notion of an interplay of activation and inhibition, as proposed in the automatic progressive parsing and lexical excitation (APPLE) model (Libben, 1994). In a study probing compounds in German, Blanken (2000) also concluded that transparent and opaque compounds are accessed by morpheme-based processes, despite the fact that the proportion of componential substitutions and simplifications decreased as the opacity in the patients’ performance increased. The second goal of this study was to examine the role of aspect in the processing of prefixed verbal forms. Bulgarian features a contrast between perfective (P) and imperfective (IMP) verbs. Prefixation is the common mechanism to derive a perfective form from the imperfective one. We opposed prefixed perfective and secondary imperfective verbs to keep prefixation constant. Secondary imperfectivization is a very productive phenomenon in Bulgarian. In a long lag priming experiment on SerboCroatian, Feldman (1994) found greater facilitation for verbs following perfective than for those following imperfective primes. In contrast, Jetchev and Bertinetto (submitted) did not find a difference between perfectives and imperfectives in Bulgarian. However, in two experiments using a visual simple lexical decision and a masked priming task, respectively, Nikolova and Jarema (2002) showed an effect of aspect, where prefixed perfective verbs were recognized faster and exhibited a greater magnitude of priming compared with prefixed secondary imperfective verbs. In a neurolinguistic experiment on Polish, Perlak and Jarema (2000) found in a simple lexical decision task that control subjects recognized perfectives faster than imperfectives for verbs with a bound root. Their patient, on the other hand, did not show a difference between the two forms. The authors concluded that the patient had a weakening of the links between the derivationally related forms. Aspect was also investigated in a longitudinal off-line study in Polish by Ulatowska, Sadowska, and Ka¸dzielawa (2001); the authors described a patient that substituted prefixes, particularly in the perfective aspect. In Czech, Lehec˘kova` (2001) showed that perfective forms were systematically replaced by imperfective forms. The author did not explain this observation. Is it because patients were unable to use the past tense (perfective verbs do not have present tense) and exchanged the perfective verbs for

352

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

imperfectives, or were imperfective verbs more frequent than perfectives ones? This is a surprising result, because perfective verbs are considered marked forms (Comrie, 1976) compared with imperfective verbs and, furthermore, imperfectives have been posited as the base on which perfectives are formed. The third aim of the study is to investigate the effect of the status of the root (free vs. bound) in the recognition of prefixed aspectual verbs. Bulgarian has not only bound grammatical morphemes (derivations, inflections) but also bound lexical morphemes, i.e. bound roots. To surface to the level of the word they need an affix. The goal of this study is to determine whether free roots have a particular status as units of recognition and facilitate the processing of verbs to a greater extent than bound roots. Very few psychoand neurolinguistic studies have probed the status of the root. In a cross-linguistic study investigating the on-line performance of Greek- and Polish-speaking non-brain-damaged subjects on a primed lexical decision (LD) task, Kehayia and Jarema (1994) observed morphological priming in Polish with prime – target pairs featuring free roots but not with bound roots. Accordingly, they argued that only roots that are free (i.e. that are words) are lexical units in the mental lexicon. Perlak and Jarema (2001) report similar results from an LD experiment on Polish verb forms. Probing non-brain-damaged and aphasic speakers of Polish on a visual LD task, Perlak and Jarema (2000) found a root status effect for verbs with free-standing roots and proposed a dual-route access for aspectual verb forms. For Bulgarian, another Slavic language, Nikolova and Jarema (2002) did not find a root status effect in the recognition of aspectual verbs. However, both, free and bound verb roots facilitated verb recognition. It was therefore suggested that both free and bound roots are represented in the mental lexicon and are lexical units of access. As reviewed above, to date few neurolinguistic studies have investigated aspect, semantic transparency and root status. However, the neurolinguistic literature provides the following general findings for verb processing. To summarize, the studies showed that: (a) patients have greater difficulty with tasks involving purely linguistic stimuli than with tasks involving picture stimuli (Berndt, Mitchum, Haendings, & Sandson, 1997; Kemmerer, Trandel, & Barrash, 2001), (b) non-fluent patients have more difficulty retrieving verbs than nouns (Miceli, Silveri, Villa, & Caramazza, 1984; Zingeser & Berndt, 1990), (c) a dissociation is observed between different semantic classes of verbs, such as concrete and abstract verbs (Kemmerer & Tranel, 2000; Breedin, Saffran, & Coslett, 1994) and semantically simple and complex verbs (Breedin, Saffran, & Schwartz, 1998; Kemmerer & Tranel, 2000), and (d) verb retrieval is influenced by the transitivity of the verb, both at the word and the sentence levels (Jonkers & Bastiaanse, 2000). Another important finding is that dissociations can occur between knowledge of the semantic properties of verbs, such as motion features and participant roles, and knowledge of the syntactic properties of verbs, such as their subcategorization frames (Breedin & Martin, 1996). These dissociations provide strong support for the representational modularity of semantic and syntactic information (Jackendoff, 1997). The present investigation is a case study of a patient suffering from transcortical motor aphasia (TCMA), which is considered as non-fluent aphasia. TCMA was first reported by Lichtheim (1885). The language profiles and CT anatomy of TCMA suggest that

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

353

the essential lesion causes disruption of connections at sites between the supplementary motor area and the frontal perisylvian speech zone (Freedman, Alexander, & Naeser, 1984). The non-fluent speech in patients with TCMA is characterized by problems initiating verbal output (Albert, Goodglass, Helms, Rubens, & Alexander, 1981) and includes poor list generation, impoverished story completion and simplified grammar (Lhermitte, 1986). Botez, Lecours, and Berube (1983) consider TCMA as an aphasia limited to language. In the neurolinguistic literature, to our knowledge, no studies have investigated the performance of patients suffering from TCMA on lexical recognition of isolated words in the visual modality. Nonetheless, Shuren and Heilman (1995) described a patient with TCMA with a reading comprehension deficit. He was unable to follow written commands, although he accurately performed these same commands when they were given verbally. Ullman et al. (1997) showed that patients with anterior frontal lesions have deficits in retrieving regular forms from the lexicon, because these forms could be retrieved only by applying a morphological rule. We thus hypothesize difficulties for our patient in accessing transparent verbal forms if they must be retrieved by following morphological rules, as opposed to opaque forms, which are purportedly accessed whole (Ullman, Corkin, Coppola, Mc Kok, Growdon, Koroshetz, & Pinker, 1997). With respect to aspect, secondary imperfective forms where we assume decomposition of the secondary imperfective suffix, should be more difficult to access than perfective forms. Regarding the status of the root, we postulate that there will be no difference between verbs featuring free roots and verbs featuring bound roots, following the results of a previous psycholinguistic study on Bulgarian (Nikolova & Jarema, 2002).

2. Description of the Bulgarian verbal system Aspect plays a central role in the Bulgarian verbal system. In an attempt to differentiate between tense and aspect, Comrie (1976:2,3) maintains that ‘tense relates the time of the situation referred to some other time, usually to the moment of speaking’ and ‘aspects are different ways of viewing the internal temporal consistency of a situation’. Bulgarian exhibits a systematic contrast between perfective (P) and imperfective (IMP) verbs. Through prefixation, suffixation or phonological change, one verb of an aspectual pair can be derived from the other. Most commonly, the base form of an aspectual pair is the imperfective, e.g. pi-ja (IMP), ‘I am drinking’ þ iz ¼ iz-pi-x (P), ‘I finished drinking’. Using the most frequent mechanism of derivation, prefixation, a perfective is derived from an imperfective form by attaching a prefix. Prefixes in aspectual pairs of verbs can be divided into two types: one kind of aspectual prefix solely serves to encode completion (telicity) of the action, e.g. iz-pi-x (P), ‘I finished drinking’, and does not fundamentally alter the lexical (semantically transparent) meaning of the base; another type of prefix, in addition to encoding telicity, contributes to the idiosyncratic (semantically opaque) meaning of the base, e.g. po-pi-x (P), ‘I wiped’. For example, both forms izpix and popix are derived from the verb pi-ja (IMP), ‘I am drinking’,1 the concept of drinking being 1

The citation form of verbs in modern Bulgarian is the first person singular present tense form; Bulgarian has lost the infinitive form.

354

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

transparent (TR) in the form izpix and opaque (OP) in the form popix. Despite the fact that the root pi is common to both forms, its transparent meaning is lost due to the combination with a particular prefix. Therefore, the meaning of opaque forms is not compositionally derivable from the meaning of their constituents (Aronoff, 1976). Bulgarian aspectual prefixes are polysemic, but their default meaning is the meaning of completion of the activity. Below we exploit the opportunity presented by Bulgarian to investigate semantic transparency in forms that share the same base (root) but vary with respect to the prefix and meaning. Verb roots can be either free (e.g. pi, ‘you, she/he drank’, nested in iz-pi-x (P, TR), ‘I finished drinking’ and in po-pi-x (P, OP), ‘I wiped’) or bound (e.g. gub, nested in iz-gub-i-x (P, TR), ‘I lost’ and po-gub-i-x (P, OP), ‘I ruined (someone)’). To maintain prefixation constant between perfective and imperfective forms we chose prefixed perfective/imperfective pairs by exploiting the tendency in Bulgarian to derive secondary imperfectives from prefixed perfectives. A secondary imperfective is thus created from a prefixed perfective through imperfective suffixation /-va-/, yielding a new complementary form within an aspectual pair (e.g. iz-pi-x– iz-pi-va-x). In our experiment, all the experimental stimuli are thus prefixed.

3. Methodology In this study we used an on-line primed lexical decision task, because there is evidence in the psycholinguistic literature that this experimental paradigm taps morphological structure during word recognition. Generally, decision latencies to visual targets are facilitated by the prior presentation of another word that shares the same base morpheme (Feldman, 1994; Fowler, Napps, & Feldman 1985; Grainger, Cole´, & Segui, 1991). Base morphemes as well as affixes play a role in how prime – target pairs formed from the same base morpheme influence each other in a word recognition task. Moreover, Feldman (2000) suggested that full nesting of the prime within the target benefited target recognition.

4. Participants 4.1. Patient The patient (TM) was a right-handed 53-year-old man with a university education. He was a military officer. He sustained an aneurysm rupture on the left anterior artery (area A1 – A2, as shown by the angiogram) 1 year prior to the present investigation. The CT scan before the surgery showed a hyperdensity in the sagittal line before the third ventricle. The ventricular system had a normal disposition but was larger than normal size. The patient underwent surgery, during which the aneurysm was clipped. The postoperative CT-scan showed a hypodensity in the frontal left area around the clip. The neurological diagnosis was subarachnoid haemorrhage with right hemiparesis and transcortical motor aphasia. Patient TM was also tested on the Bulgarian adaptation

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

355

(Alexandrova, Terzieva, Tarnev, & Mavlov, 1994) of the BDAE (Goodglass & Kaplan, 1983) for an assessment of his off-line deficits 1 year post-onset. This examination showed a mild dysarthria, a small number of phonemic and semantic paraphasias and a slight phoneme reduction in pronunciation. His speech output was reduced (in some cases telegraphic, and with verbal stereotypes and word-finding pauses), but largely grammatical. However, he produced mostly short sentences and had slight problems in reading comprehension. His repetition, naming, writing and auditory comprehension were normal. Language abnormalities were magnified when the patient was asked to tell a story (Little Red Riding Hood) as a narrative speech task. This condition was manifested by a word-finding difficulty, a problem with initiating speech, and frequent blocking. TM produced only simple sentences (no embedded sentences were recorded) with perseverations, few morphological paraphasias (substitutions of pronouns and inflected auxiliary forms, e.g. 3p.sg of to be for the 2p.sg.), and few morphological omissions (in verbs). He told the tale in the present rather than the past tense, which is the appropriate tense for tales. He did not show any signs of a visual or auditory deficit, or anterograde or retrograde amnesia. 4.2. Control participants We also tested nine control participants matched with the patient. All were righthanded males, 53– 57 years-old, with an university education. All participants were native speakers of Bulgarian and tested in Sofia, Bulgaria. All had normal or correctedto-normal vision. None of the control participants reported any past or present neurological disease. 4.3. Stimuli A total of 64 verbs (32 perfectives and 32 imperfectives) the critical stimuli. All forms were in the 1p.sg. simple past tense, half of the verbs featured free roots (FR) that correspond to the 2/3 p.sg. past tense form, and the other half bound roots (BR). One hundred and ninety two prime –target pairs were obtained by priming each of the 64 critical target forms with the simple root, the prefixed root, and an unrelated control item. Unrelated primes were morphologically and semantically unrelated to targets and did not share any letter in the same position as the targets (e.g. slizam ‘I go down’ primed izmix ‘I finished washing’). Non-words (n ¼ 128) were created by changing the first letter of either the prefix or the root of the critical stimuli. Nouns and adjectives were used as fillers in equal number to the verbs ðn ¼ 64Þ: Non-word fillers were derived by changing the first letter of each filler. The critical stimuli comprised two groups of verbs with a free root (FR): semantically transparent (e.g. iz-pi-x(P), ‘I finished drinking’) and semantically opaque (e.g. po-pi-x (P), ‘I wiped’), distinguished by the aspectual prefix. As is evident from these examples, opaque and transparent verbs with FRs have the same root (e.g. pi, ‘you, she/he drank’) and differ in the aspectual prefixes and overall meaning. However, the meaning of the opaque forms in not compositional, the root having lost its original transparent meaning. Prefixes used for transparent FR verbs included iz-, na- and u- (for two items)

356

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

Table 1 Example of aspectual derivation in Bulgarian for perfective and secondary imperfective verbs Imperfective þ prefix ¼ perfective þ imp. suffix ¼ second imperfective Free root PIX þ IZ ¼ IZPIX þ /-VA-/ ¼ IZPIVAX Bound root GUBIX þ IZ ¼ IZGUBIX þ /-VA-/ ¼ IZGUBVAX

and those used for opaque FR verbs po, pro- and ob- (for one item). Semantically transparent prefixes encode only the grammatical information of completion of the event and are thus basically telicity morphemes. Semantically opaque prefixes have, in addition to telicity, an idiosyncratic lexical meaning. Verbs with a bound root (BR) were also divided into transparent and opaque forms. For example iz-gub-ix (P), ‘I lost” is transparent, and po-gub-ix (P), ‘I ruined (someone)’ is opaque and both verbs derive from the verb gub-ja, ‘I am losing’. Again, the same root (here the bound root gub-) appears in the opaque and transparent verbs. Prefixes used for transparent BR verbs included iz- and na- and those for opaque BR verbs po- and pro-. Both transparent and opaque verbs with FRs and BRs featured perfective/imperfective pairs. For example, iz-pi-x, ‘I finished drinking’ is a transparent verb in the perfective (P) form, with the FR pi-, ‘he/she drank‘. The secondary imperfective form (IMP) of this verb is iz-pi-va-x, ‘I was finishing drinking’, which contains the imperfective suffix -va-. Together, they form the aspectual pair iz-pi-x –iz-pi-v-a-x. An example of a transparent perfective verb with a BR is iz-gub-i-x, ‘I lost‘ and its secondary imperfective form is iz-gub-va-x, ‘I was finishing losing‘; this results in the aspectual pair iz-gub-i-x –iz-gubv-a-x. The imperfective marker -va- is the suffix with which all transparent and opaque secondary imperfectives are formed. An example of aspectual derivation is shown in Table 1.

Table 2 Average stimulus frequency per 500.000 (Frequency Dictionary of Written Bulgarian, Todorova and Panc˘ovska (1995)) Category of verb

Frequency

FRopImp FRopPerf BRopImp BRopPerf FRtrImp FRtrPerf BRtrImp BRtrPerf

10 10 3 19 9 7 5 12

op ¼ opaque, tr ¼ transparent, P ¼ perfective, IMP ¼ imperfective, FR ¼ free-standing root, BR ¼ bound root.

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

357

Table 3 Frequency of occurrence of individual prefixes Prefixes

Number of verb roots with which the prefix occurs

Cumulative frequency

PO (op) PRO (op) OB (op) IZ (tr) U (tr) NA (tr)

117 40 26 117 21 67

1516 511 322 1511 255 688

op ¼ opaque, tr ¼ transparent.

All stimuli were phonologically transparent and did not contain phonological changes. They were matched for frequency2 and length (three syllables), except for the perfective verbs containing a free root, which had one less syllable. Stimuli were all of low frequency, ranging from 3 to 19 per 500.000 (Table 2). The cumulative frequency of prefixes was also matched, i.e. the frequency of opaque prefixes was identical to that of transparent prefixes. The frequency of a particular prefix was checked by counting its number of occurrences in the frequency dictionary of written Bulgarian as summarized in Table 3.

5. Procedure Participants performed a masked visual lexical decision task using the PsyScope 1.2.1. Program. The advantage of this approach is that, because the prime is presented briefly and is masked by a combination of forward and backward masking (the latter coming from the target), the prime itself is usually unavailable for report and participants have very little or no direct awareness of the prime. This helps to reduce the possibility of any priming effect due to conscious recognition that the prime and the target share a common morpheme, and priming effects are therefore more likely to be produced by automatic, unconscious processes. In each trial, a mask (######) appeared for 500 ms, followed by a prime for 70 ms, presented in smaller letters (uppercase 18-point Tolstoy font) than the target. Immediately following the prime (ISI ¼ 0) the target appeared in uppercase 24-point Tolstoy font, and stayed on the screen of a portable Macintosh Powerbook 1400 computer until the participant responded. Because primes were shorter than targets, there was no alignment to the first letter. The inter-trial interval was 1000 ms. All participants performed lexical decision judgments using the index and middle fingers of their left hand (for the patient only, because of the weakness of his right hand) and right hand (for controls). The stimulus set was divided into three lists and in each, a target appeared only once. All participants saw the full set of stimuli in all priming conditions in three testing sessions separated minimally by 1 week. There was a training session of 27 stimuli, and seven pauses were introduced, one between the training session and 2

The frequency dictionary for written Bulgarian used in this study (Todorova & Panc˘ovska, 1995) is based on the cumulative frequency of words.

358

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

the test set, and six during the experiment. All subjects were remunerated for their participation in the experiment.

6. Results 6.1. Analysis of response times and magnitude of priming Errors and extreme response times (over 2000 ms for controls and over 4000 ms for the patient) were removed from the analysis, other times (more than two standard deviations) were also taken out. Control subjects produced a 2% error rate. Patient TM had an error rate of 8% for words and 10% for non-words. For this patient, we removed two items (one perfective and the corresponding imperfective of the same verb) because of errors in all priming conditions. 6.2. Control participants For controls, the analyses of reaction times (RTs) revealed significant priming effects for all categories of verbs for both by-subject ðF1 Þ and by-item analyses ðF2 Þ [F1 ð2; 16Þ ¼ 20; 74; p , 0:01; F2 ð2; 112Þ ¼ 4; 28; p , 0:15], (Table 4). We also performed an analysis of the magnitude of facilitation, i.e. the difference in scores between the control prime and the simple root prime conditions, and between the control and the prefixed root prime conditions. For the control subjects the results were analyzed using ANOVA repeated measures. The independent variables were transparency, aspect, root status and prime type. The dependent variable was the difference score between the unrelated condition and each related one. Decision latencies to target after morphologically related primes were reduced relative to their respective control primes, i.e. all related forms produced facilitation. We found a main effect of prime type only in Table 4 Reaction times (RTs) and standard deviations (SDs) in paranthesis over subjects and patient by prime type (unrelated, simple roots and prefixed roots primes) Control participants

opFRperf opFRimp trFRperf trFRimp opBRperf opBRimp trBRperf trBRimp

Patient TM

Unrelated

Simple root

Prefixed root

Unrelated

Simple root

Prefixed root

922(169) 1007(147) 864(120) 998(155) 1044(210) 1065(246) 1071(220) 1040(163)

942(189) 953(159) 765(110) 990(162) 950(210) 1063(260) 942(171) 1038(207)

857(119) 948(221) 808(71) 934(183) 934(252) 1003(246) 906(122) 1058(233)

1277(125) 1364(101) 1255(168) 1229(101) 1335(244) 1336(211) 1294(165) 1348(200)

1141(95) 1314(32) 1254(136) 1362(236) 1306(202) 1242(225) 1211(93) 1239(151)

1264(176) 1277(167) 1296(202) 1263(232) 1210(110) 1158(162) 1214(163) 1211(148)

op ¼ opaque, tr ¼ transparent, perf ¼ perfective, imp ¼ imperfective, FR ¼ free-standing root, BR ¼ bound root.

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

359

the subject analysis, such that the prefixed root facilitated recognition more than the simple root prime [F1 ð1; 8Þ ¼ 4; 97; p , 0:05; F2 ð1; 56Þ ¼ 0; 007; p ¼ 0:93]. A main effect of aspect was significant only in the subject analysis [F1 ð1; 8Þ ¼ 6; 72; p , 0:05]; F2 ð1; 56Þ ¼ 1; 75; p ¼ 0:19], with more facilitation for perfective than for imperfective forms. Most importantly, the interaction between transparency, aspect, root and prime type was significant in the subject analysis [F1 ð1; 8Þ ¼ 6; 38; p , 0:05], and marginally significant in the item analysis F2 ð1; 56Þ ¼ 2; 98; p ¼ 0:08]. There was a significant triple interaction between transparency, root and prime for perfective verbs [F1 ð1; 8Þ ¼ 4; 71; p ¼ 0:06] and a double interaction between transparency and prime for perfective verbs with a FR [F1 ð1; 8Þ ¼ 13; 85; p , 0:01]. Simple effects revealed that primed with a simple root prime, transparent perfective forms with a FR showed more facilitation than opaque perfective forms with a FR [F1 ð1; 8Þ ¼ 7; 94; p , 0:05], and opaque perfective forms with a FR primed by prefixed roots were more facilitated than the same forms primed with a simple root prime [F1 ð1; 8Þ ¼ 6; 57; p , 0:05]. We also found a marginally significant interaction between root, aspect and prime for transparent verbs [F1 ð1; 8Þ ¼ 4; 00; p ¼ 0:08], as well as an interaction between root and aspect for transparent verbs primed with a prefixed root prime [F1 ð1; 8Þ ¼ 4; 28; p ¼ 0:07]. Furthermore, simple effects revealed that primed with the prefixed root prime, transparent perfective forms with a BR show more facilitation than transparent imperfective forms with a BR [F1 ð1; 8Þ ¼ 6; 89; p , 0:05]. 6.3. Patient For the patient, only item analyses (factorial) were performed. Data analysis of patient TM’s RTs showed an interaction between root and prime type only in the prefixed root prime condition [F1 ð1; 100Þ ¼ 3:88; p ¼ 0:05]. The simple effects showed facilitation only for the verbs with a BR [F1 ð1; 100Þ ¼ 8; 13; p , 0:01]. The verb category with a FR did not show any priming effects, irrespective of transparency [F1 ð1; 100Þ ¼ 0:00; p ¼ 0:97] (Table 4). The non-parametric Fisher exact test for the error analysis did not show any significant effect of verb category, p ¼ 0.63. In an additional analysis we measured the magnitude of priming between the verb categories. Patient TM showed more priming for verbs with a BR than verbs with a FR [F1 ð1; 95Þ ¼ 4:18; p , 0:05]. We performed a supplementary analysis only for the forms featuring free roots. This analysis showed an effect of transparency in the simple root prime condition [F1 ð1; 23Þ ¼ 3; 9; p ¼ 0:06], where we found inhibition for the transparent forms there was an (Table 4).

7. General discussion The linguistic literature contains much debate about the nature of aspectual verbs in the Slavic languages. Specifically, are they derivational or inflectional forms? In general (e.g. Comrie, 1976; Mel’c˘uk, 1993), perfective verbs are considered derivational forms and secondary imperfective verbs are considered inflectional forms. In the psycholinguistic literature, the most traditional conceptualization of mental storage and computation of

360

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

polymorphemic words is that inflectional forms with a regular morphology are decomposed and derivational forms are stored. According to that conception, we can expect that all transparent and opaque perfective verbs (considered as derivative forms) will be accessed via the global route and that secondary imperfectives (considered as inflexional forms) will be accessed via a morphological rule. However, studies using different languages report findings that differ markedly from the traditional understanding of the storage and computation of complex words. For example, Baayen et al. (1997) reported evidence for extensive storage of regular plural nouns in Dutch, and also Sereno and Jongman (1997) showed similar results in English. The Dutch (Baayen et al., 1997) and Italian (Burani, Dovetto, Thornton, & Laudanna, 1997) studies show that distributional and linguistic properties of individual affixes affect the way in which words with these affixes are processed. Taking into account cross-linguistic differences between morphological systems may contribute to the understanding of the balance between storage and computation in lexical processing. This is precisely what we attempted to do in the present study. Employing a primed visual lexical decision task, we investigated the performance of an aphasic patient and nine control participants in the processing of prefixed aspectual verbs in Bulgarian, involving root status, aspect and transparency. First, we will discuss the control participants’ performance on verbs with FRs, followed by a discussion of verbs with BRs and, lastly, we will interpret the findings for patient TM in light of the results obtained for control participants. In the controls’ results, we found a transparency effect for verbs with a FR. In the simple root prime condition, transparent perfective forms showed greater priming than opaque perfective forms, and opaque perfective forms showed more facilitation in the prefixed root prime condition than in the simple root prime condition. In this condition opaque perfective verbs with a FR (e.g. pi-popix) did not show facilitation compared with transparent perfective verbs with a FR (e.g. pi-izpix), because the meaning of the simple root pi is lost in the opaque form popix, unlike in the transparent form izpix. Opaque forms do not have compositional meaning, as the simple root (e.g. pi) does not contribute to their meaning (e.g. popix). Moreover, opaque perfective forms with a FR showed more priming when primed with a prefixed root prime (e.g. popi-popix) than with a simple root prime (e.g. pi-popix), because the meaning of the prefixed root popi is present in the meaning of the verb popix. These findings seem to point to a whole-word access route to opaque perfective forms with FRs and decomposition for transparent perfective forms with FRs. With respect to the comparison between transparent perfective and transparent imperfective forms with FRs, we did not find a statistically significant simple effect. However, we obtained a main effect of aspect where perfective forms showed greater facilitation than imperfective ones. Secondary imperfective forms are created by adding the very common and productive aspectual suffix -va- to their corresponding perfective forms. We expected secondary imperfectives to involve a processing cost because of the additional morphological computation (þ suffix -va-), which may slow response latencies and attenuate facilitation. According to Burani, Thornton, Iacobini, and Laudanna (1994) numerosity (affixal productivity), i.e. the number of times that an affix occurs in a language, is one of the most important factors for the processing of polymorphemic words. Aronoff (1976) defines affixal productivity as ‘the ease with which you can make a new

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

361

word formation with a certain affix’. The more numerous an affix is, the more likely it will be used as a processing unit. This is precisely what we concluded in a previous study (Nikolova & Jarema, 2002), where we found a difference in the magnitude of priming between prefixed transparent perfectives (with FRs and BRs) and prefixed transparent secondary imperfectives (with FRs and BRs), where the perfectives showed more priming than the imperfectives in the simple root prime condition. An effect of aspect was obtained in the present study for transparent verbs with BRs, however in the prefixed prime condition, not in the simple prime condition as in Nikolova and Jarema (2002). Transparent perfectives (e.g. izgubix) showed a greater magnitude of priming than the transparent imperfectives (e.g. izgubvax). However, verbs with BRs did not show a transparency effect. Thus, opaque forms did not elicit differential priming compared with transparent forms featuring BRs. This result in the prefixed prime condition is not surprising, because for opaque forms the semantic relation between the prefixed prime and the target is transparent (e.g. pogub-pogubix, where the prime pogub- is the prefixed root of the verb pogubix). Our finding suggests that both transparent and opaque forms with BRs are accessed via the direct route, i.e. via the prefixed root. For example, to access the transparent verb izgubix the lexical entry could be the prefixed root izgub-, even though the simple root gub- exists. This suggestion is in line with the lack of difference between transparent and opaque verbs with BRs in the simple root prime condition. If opaque and transparent forms are accessed in a direct way, the simple root should facilitate neither opaque nor transparent forms. However, the control participants’ results revealed differential processing between opaque popix and transparent izpix forms with a FR; transparent verbs obtained more facilitation when primed with a simple root prime. If transparent perfectives with FRs (e.g. izpix) and transparent perfectives with BRs (e.g. izgubix) were accessed differentially, why did the results obtained not show a significant difference in the magnitude of priming? We suggested access through decomposition for the transparent perfective verbs with FRs (e.g. iz-pi-x) and whole route access via the prefixed root for the transparent perfective verbs with BRs (e.g. izgub-ix). We interpret the lack of differences between transparent forms with FRs and transparent forms with BRs as reflecting a ‘no cost’ in accessing the prefix, as suggested by Forster and Azuma (2000), who also did not find differential facilitation between prefixed prime –target pairs and pairs with unprefixed primes and prefixed target. Similar results were reported by Grainger, Cole´, and Segui (1991) in masked priming in French and Feldman and Larabee (2001) in a priming experiment in the visual modality. Our previous study did not reveal differential access between verbs with FRs and verbs with BRs (Nikolova & Jarema, 2002). This finding supports the view that the status of the lexical root does not play a role in the recognition of aspectual verbs in Bulgarian and suggests that both types of verbal root (free and bound) are lexical units in the mental lexicon and are equally represented. How can we explain the differences between the result of this study and that of Nikolova and Jarema (2002)? One hypothesis may pertain to the difference in age of the participants tested: those in the present study are older. The literature (Tainturier, Tremblay, & Lecours, 1992; Tsapkini, 2001) reveals that age and education can affect lexical processing. Alternately, changes in the lexical representations may occur after age

362

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

50. We may then postulate that the way in which lexical entries are represented in the lexicon itself changes with age. The lexicon can thus be perceived as a dynamic component, in which the force of links between the entries is subject to change and new representations may be formed. This assumption is consistent with the interactive lexical access models whereby parameters such as frequency of usage or semantic idiosyncrasy can contribute to such changes (Levelt, Roelofs, & Mayers, 1999; Schreuder & Baayen, 1995). In light of our results, it may be the case that patients change access strategies as they age, and patients at an advanced age opt for the global access route for transparent forms with bound roots (BRs). 7.1. Patient Patient TM showed priming effects (in the response latencies analysis) for all the verbs with a bound root, primed with a prefixed root, irrespective of aspect and transparency, but failed to show priming effects for the transparent verbal forms with a FR. Moreover, transparent forms with a FR were inhibited, compared with opaque forms with FRs, for which priming was obtained. Patient TM thus showed a difficulty in retrieving transparent verbs featuring FRs. In his performance on the Boston test, he also showed a slight deficit in reading comprehension. Shuren and Heilman (1995) described a patient with TCMA who exhibited a failure to follow written commands. They assume that this deficit is often transient, lasting hours to days after onset, and is a secondary deficit to an attentional rather than linguistic deficit. However, 1 year post-onset our patient still showed slight reading comprehension difficulties which, combined with his specific pattern of performance characterized by an absence of priming for the transparent forms with FRs suggests a linguistic deficit. In the literature, a claim is that non-fluent aphasics have an access impairment or a slowed activation (Milberg, Blumstein, Katz, Gershberg, & Brown, 1995; Prather, Zurif, Love, & Brownell, 1997; Prather, Zurif, Stern, & Rosen, 1992). Our patient achieved priming for verb forms featuring BRs only in the prefixed prime condition in a short SOA (70 ms). However, patient TM did not show priming for all categories. Our findings are in line with the claim that non-fluent patients activate lexical information in a slower-thannormal fashion. There remains, however, some strong opposition to this position. Patients of Hagoort (1993, 1997) and Ostrin and Tyler (1993) showed priming for all kind of stimuli leading the authors to conclude that the immediate lexical activation (the automatic access) is normally rapid for Broca’s aphasics, and that slower activation is not the explanation of Broca’s patients deficits. The authors cite evidence that patients show priming at a very short SOA, which has been claimed to mainly tap automatic spleading of activation within the semantic lexicon (Neely, 1991). In view of the fact that patient TM demonstrated priming for verbs with BRs only in the prefixed root prime condition, but not in the simple root prime condition, suggests that he accessed these forms whole. Moreover, he did not access opaque and transparent verbs with BRs differentially. It seems that patient TM could not compute transparent forms with FRs that are presumed to be decomposed, suggesting that the decomposition-based mechanism was impaired. Thus, the problem appears to be manifested only when the computational demands exceed his capacities. This is a common finding for many

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

363

non-fluent aphasics who exhibit difficulties in inflectional morphology. In an on-line experiment Kehayia (1997) showed that morphology, i.e. rules of word-formation and underlying representations of morphemes, are not lost in aphasia. Rather, what is lost is access to morphemes and to the mechanisms that combine them and the processing of morphologically complex words is consequently impaired. Our patient presents the profile common to patients with anterior lesions (Ullman et al., 1997), namely problematic treatment of transparent forms with free roots (FRs), which appear to be accessed via the decomposition route. A possible interpretation of the results could be in the framework of the model proposed by Tsapkini (2001). The access procedure in this model features several stages. Because Bulgarian has an extensive inflectional system and all verbal forms are affixed with inflectional suffixes (for person, number and tense), decomposition can be assumed for all forms, at least in the initial stages of lexical access. Furthermore, decomposition is conceivable for the inflectional suffixes that are common to both opaque popi-x and transparent izpi-x forms. The decompositional process in the initial stage does not concern the additional suffix -va- (Tsapkini, 2001), i.e. the rule-based aspectual marker for secondary imperfectives (e.g. popiva-x, OP; izpiva-x; TR). This does not exclude the possible application of the whole-word route for the remaining forms (which have their own lexical entry) after the stripping of the inflectional suffix (e.g. in our case -x-, which is the 1p.sg. for the past). Both routes may apply in the initial stages of lexical access. What differentiates morphological processing of the opaque and transparent forms is what happens in later stages, after the stripping of the common inflectional suffix, i.e. what happens at the stem level (simple and prefixed). Tsapkini’s proposition of an additional intermediate level of representation of the form with an aspectual suffix is consistent with Booij (1994), who suggests that inherent (aspect, in our case) and contextual inflection operate in the lexicon and that inherent inflection can trigger word formation (e.g. napiva þ ne ¼ napivane, whose base form contains the imperfectivizing suffix). Results obtained in recent models highlight the possibility that inherent inflection is part of the lexicon. For example, transparent forms, although regular, may have a global access representation if the forms are frequent and storage is more economical for the processing system (Schreuder & Baayen, 1997; Bertram, Schreuder, & Baayen, 2000). Patient TM seems to access transparent forms with a BR without distinguishing between aspects (e.g. izgubi-x and izgubva-x), because after he decomposes the inflection, he accesses both roots (perfective and imperfective) via the direct route. Control participants showed global access recognition for these verbal forms. The only forms for which TM showed inhibition were transparent verbs with FRs. Following the results obtained by control participants, we suggested that these verbs were recognized via a decomposition route. Thus, to access transparent forms with FRs (iz-pi-va-x et iz-pi-x), the patient must access the constituent morphemes of these forms, i.e. he must decompose them. His deficit therefore seems to involve the decomposition mechanism and the processing of additional affixes (such as the suffix -va-, a marker of secondary imperfectivization, along with the prefixes), in other words, affixes that are additional to the inflectional affix of tense, person and number, obligatory for all finite verb in Bulgarian. To determine whether his deficit affects affixes in general, additional experiments would be required with derived forms (nominal and adjectival), suffixed and prefixed, which would involve much more complex

364

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

processing of suffixes. Note that inflectional verbal suffixes are more regular than nominal and adjectival derivational suffixes. For example, in the noun napivane, derived from the verb napivam, a secondary imperfective verb to which the derivational suffix -ne is added, would the patient also experience difficulty decomposing the derivational suffix -ne? The patient’s results cannot be interpreted within the framework of connectionist models based on visual recognition (Plaut & Gonnerman, 2000; Rueckl, Mikolinski, Raveh, Miner, & Mars, 1997; Seidenberg & McClelland, 1989). In contrast with the dualroute models, the connectionist models propose a single recognition route for all words, including complex words. Morphological information is not explicitly represented in the mental lexicon, independent of the representations of formal and semantic information. Morphological links create correlation links between formal and semantic characteristics, which is why the degree of the effect of morphological priming is assumed to be determined jointly by the form and meaning similarity. Our patient exhibits priming effects in the prefixed-root priming condition. One might assume that the prefixed root provides more formal similarity than the simple root. Nonetheless, our patient exhibited this priming effect solely for the group of verbs with bound roots (e.g. izgub-izgubix), not for verbs with free roots (e.g. izpi-izpix). In view of the significant morphological effect obtained, our patient’s results cannot be explained within the framework of connectionist models. Moreover, Feldman (2000) and Rastle et al. (2000) have demonstrated that morphological effects are greater than the sum of the orthographic and semantic effects combined. These results lead us to the question of the locus of the patient’s impairment within the lexicon. If the impairment affects access to the semantic characteristics of the verbs, it should affect all categories of verbs equally. This is, however, not the case. The literature reveals cases where patients are unable to access either spoken or written forms of verbs (Caramazza & Hillis, 1991; Hillis & Caramazza, 1995; Rapp & Caramazza, 1998, 2002) which suggests that it is a modality-specific rather than a semantic deficit, since patients can produce verbs in the unimpaired modality. Rather, in these cases it is likely that modality-specific representations of verbs are inaccessible, either because the representations are themselves damaged or because they have been disconnected functionally from the semantic system (Rapp & Caramazza, 1998). Hagoort (1993) claims that the deficit of patients can be in either the morphological parsing of complex words into stems and affixes, or the on-line retrieval of the syntactic features of the inflectional suffixes (see also, Badecker, 1997). Our patient was tested only on isolated words and it would be difficult to locate his deficit at the morphosyntactic level. In the literature, inflectional impairment has often been explained by a failure at the morphosyntactic level (Badecker, 1997; Luzzatti & De Bleser, 1996). However, the performance of patient SJD described by Badeker and Caramazza (1991) who made errors in regular affixation of verbs while his production of irregularly inflected forms was intact points to a deficit at the morphophonological level. Tsapkini, Kehayia, and Jarema’s (2002) patient exhibited difficulties with verbal inflections that also appeared to be located at the morphophonological level of verbal processing. Patient HG of Shapiro and Caramazza’s (2003) study had little difficulties with irregular verbs which may point to an impairment in retrieving idiosyncratic lexical forms that need to be memorized as lexical units. Moreover, Luzzatti, Mondini, and Semenza (2001) report a patient who could access the syntactic structure of complex words, but had

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

365

a problem with the phonological form of these words. The authors conclude that the deficit of their patient was not at the lemma level, where morphosyntactic features are represented (Levelt et al., 1999), but at the lexeme level, where the morphophonological information is represented. The BDAE results revealed that patient TM did not have a comprehension deficit, nor a deficit when reading aloud. Our patient seems to have a deficit in affix stripping in the morphological decomposition process of transparent verbs with a free root in the visual modality. More specifically, his difficulty was in performing the necessary computations that involve the processing of prefixes at the root level. His impairment thus appears to be in the decomposition mechanism, at least at the orthographic lexical level. A complementary study in production could reveal whether patient TM has difficulties in decomposition not only at the orthographic input level, but also at the phonological input level. In conclusion, in this study we have discussed manifestations of difficulty in the on-line recognition of complex words by a Bulgarian-speaking non-fluent aphasic patient. We investigated prefixed aspectual verb forms and argued for an impairment in the decomposition-based mechanism of the patient’s performance. Further research is required to shed more light on word recognition patterns particular to aphasia in speakers of Bulgarian.

Acknowledgements The research reported here was supported by an MCRI grant from the Social Sciences and Humanities Research Council of Canada (#412-95-0006) awarded to Gonia Jarema (Universite´ de Montre´al), director and co-principal investigator; Eva Kehayia (McGill University), co-principal investigator, and Gary Libben (University of Alberta), coprincipal investigator. We thank Kyrana Tsapkini for helpful discussions on earlier versions of this paper. We are grateful to patient TM for his participation and his patience throughout the testing. We are indebted to Katia Stoianova, Elizabeth Alexandrova and Margarita Terzieva for their help and assistance in recruiting patients.

References Albert, M. L., Goodglass, H., Helm, N. A., Rubens, A. B., & Alexander, M. P. (1981). Clinical aspects of dysphasia. New York: Springer. Alexandrova, B., Terzieva, M., Tarnev, I., & Mavlov, L. (1994). Adaptation in Bulgarian of the Boston diagnostic aphasia examination booklet (1983). Aronoff, M. (1976). Word formation of generative grammar. Cambridge, MA: MIT Press. Baayen, R. H., Dijskstra, T., & Schreuder, R. (1997). Singular and plural in Dutch: evidence for a parallel dual route model. Journal of Memory and Language, 37, 94–117. Badecker, W. (1997). Level of morphological deficit: indications from inflectional regularity. Brain and Language, 60, 360 –380. Badeker, W., & Caramazza, A. (1991). Morphological composition in the lexical output system. Cognitive Neuropsychology, 8, 335– 367. Berndt, R. S., Mitchum, C. C., Haendinges, A. N., & Sandson, J. (1997). Verb retrieval in aphasia. 1. Characterizing single word impairments. Brain and Language, 56, 68–106.

366

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

Bertram, R., Schreuder, R., & Baayen, R. H. (2000). The balance of storage and computation in morphological processing: the role of word formation type, affixal homonymy, and productivity. Journal of Experimental Psychology: Learning, Memory and Cognition, 26(20), 489–511. Blanken, G. (2000). The production of nominal compounds in aphasia. Brain and Language, 74, 84–102. Booij, G. E. (1994). Against split morphology. In G. E. Booji, & J. v. Marle (Eds.), Yearbook of morphology 1993 (pp. 27–49). Dordrecht: Kluwer Academic. Botez, M. I., Lecours, A. R., & Berube, L. (1983). Speech and language in the frontal syndrome. In A. R. Lecours, F. Lhermitte, & B. Bryan (Eds.), Aphasiology (pp. 124 –140). London: Balliere-Tindall. Breedin, S. D., & Martin, R. C. (1996). Patterns in verb impairment of aphasia. An analysis of four cases. Cognitive Neuropsychology, 13, 51–91. Breedin, S. D., Saffran, E. M., & Coslett, H. B. (1994). Reversal of the concretness effect in a patient with semantic dementia. Cognitive Nueropsychology, 11, 617– 660. Breedin, S. D., Saffran, E. M., & Schwartz, M. F. (1998). Semantic factors in verb retrieval: an effect of complexity. Brain and Language, 63, 1–31. Burani, C., Thornton, A. M., Iacobini, C., & Laudanna, A. (1994). Investigating morphological non-words. In W. U. Dressler, & C. Burani (Eds.), Cross-disciplinary approaches to morphology. Wien: Verlag der Osterreichische Akademie der Wissenschaften. Burani, C., Dovetto, M., Thornton, A. M., & Laudanna, A. (1997). Accessing and naming suffixed pseudo-words. In G. E. Booij, & J. van Marle (Eds.), Yearbook of morphology 1996 (pp. 55–73). Norwell, MA: Kluwer Academic. Butterworth, B. (1983). Lexical representation. In B. Butterworth (Ed.), Language production, 2 (pp. 257–294). London: Academic press. Bybee, J. (1995). Regular morphology and the lexicon. Language and Cognitive Processes, 10(5), 425 –455. Caramazza, A., & Hillis, A. (1991). Lexical organization of nouns and verbs in the brain. Nature, 349, 788 –790. Caramazza, A., Laudanna, A., & Romani, C. (1988). Lexical access and inflectional morphology. Cognition, 28, 297–332. Caramazza, A., Miceli, G., Silveri, M. C., & Laudanna, A. (1985). Reading mechanisms and the organization of the lexicon: evidence from acquired dyslexia. Cognitive Neuropsychology, 2, 81–114. Comrie, B. (1976). Aspect. Cambridge, UK: Cambridge University Press. Feldman, L. B. (1994). Beyond orthography and phonology: differences between inflections and derivations. Journal of Memory and Language, 33, 442–470. Feldman, L. B. (2000). Are morphological effects distinguishable from the effects of meaning and shared form? Journal of Experimental Psychology: Learning, Memory and Cognition, 26(6), 1431– 1444. Feldman, L. B., & Larabee, J. (2001). Morphological facilitation following prefixed but not suffixed primes: lexical architecture or modality-specific processes? Journal of Experimental Psychology: Human Perception and Performance, 27, 33–39. Forster, K. I., & Azuma, Y. (2000). Masked priming for prefixed words with bound stems: does submit prime permit ? Language and cognitive processes, 15, 539–561. Fowler, C. A., Napps, S. E., & Feldman, L. B. (1985). Relations among regular and irregular morphologically related words in the lexicon as revealed by repetition priming. Memory andCognition, 13(3), 241– 255. Frauenfelder, U. H., & Schreuder, R. (1991). Constraining psycholinguistic models of morphological processing and representation: the role of productivity. In G. Booij, & J. van Marle (Eds.), Yearbook of morphology (pp. 165–183). Dordrecht, The Netherlands: Kluwer. Freedman, M., Alexander, M. P., & Naeser, M. A. (1984). Anatomic basis of transcortical motor aphasia. Neurology, 34, 409 –417. Goodglass, H., & Kaplan, E. (1983). The assessment of aphasia and related disorders, Boston diagnostic aphasia examination booklet. Philadelphia, USA: Lea and Febiger. Grainger, J., Cole´, P., & Segui, J. (1991). Masked morphological priming in visual word recognition. Journal of Memory and Language, 30, 370– 384. Hagoort, P. (1993). Impairments of lexical-semantic processing in aphasia: evidence from the processing of lexical ambiguities. Brain and Language, 45, 189–232. Hagoort, P. (1997). Semantic priming in Broca’s aphasics at a short SOA: no support for an automatic access deficit. Brain and Language, 56, 287–300.

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

367

Hillis, A., & Caramazza, A. (1995). Representation of grammatical categories of words in the brain. Journal of Cognitive Neuroscience, 7, 396 –407. Jackendoff, R. (1997). The architecture of the language faculty. Cambridge: MIT Press. Jarema, G., Busson, C., Nikolova, R., Tsapkini, K., & Libben, G. (1999). Processing compounds: acrosslinguistic study. Brain and Language, 68, 362–369. Jetchev, G., & Bertinetto, P. M. (submitted). Lexical access in Bulgarian perfective vs imperfective verbs. Proceedings of 9th International Morphology Meeting, Vienna, 2000. Jonkers, R., & Bastiaanse, R. (2000). Verb retrieval in isolation and sentence context in Broca’s aphasics: the effect of transitivity. Brain and Language, 74(3), 33–36. Kehayia, E. (1997). Morphological representation of inflected and derived words in Greak speaking individuals. Proceedings from the Second International Conference on Greek Linguistics, October, 1997. Kehayia, E., & Jarema, G. (1994). Morphological priming (or prim#ing) of inflected verb forms: a comparative study. Journal of Neurolinguistics, 8(2), 83–94. Kemmerer, D., & Tranel, D. (2000). Verb retrieval for action naming in brain-damaged subjects: 1. Analysis of stimulus, lexical, and conceptual factors. Brain and Language, 73, 347 –392. Kemmerer, D., Tranel, D., & Barrash, J. (2001). Patterns of dissociation in the processing of verb meaning in brain-damaged subjects. Language and Cognitive processes, 16(1), 1–34. Kempley, S. T., & Morton, J. (1982). The effects of priming with regularly and irregularly related words in auditory word recognition. British Journal of Psychology, 73, 441– 454. Lehec˘kova`, H. (2001). Manifestation of aphasic symptoms in Czech. Journal of Neurolinguistics, 14, 179–208. Levelt, W. J. M., Roelofs, A., & Meyer, A. S. (1999). A theory of lexical access in speech production. Behavioral and Brain Sciences, 22, 1–75. Lhermitte, F. (1986). Human autonomy and the frontal lobes. II. Patient behaviour in complex and social situations: the environmental dependency syndrome. Annals of Neurology, 19, 335–343. Libben, G. (1998). Semantic transparency in the processing of compounds: consequences for the representation, processing and impairment. Brain and Language, 61, 30–44. Libben, G. (1994). How is morphological decomposition achieved? Language and CognitiveProcesses, 9(3), 369–391. Lichtheim, L. (1885). On aphasia. Brain, 4, 463–484. Lukatela, G., Gligorievic´, B., Kostic´, A., & Turvey, M. T. (1980). Representation of inflected nouns in the internal lexicon. Memory and Cognition, 8(5), 415–423. Luzzatti, C., & De Bleser, R. (1996). Morphological processing in Italian agrammatic speakers: eight experiments in lexical morphology. Brain and Language, 54, 26–74. Luzzatti, C., Mondini, S., & Semenza, C. (2001). Lexical representation and processing of morphologically complex words: evidence from the reading performance of an Italian agrammatic patient. Brain and Language, 79, 345 –359. Marslen-Wilson, W., Tyler, L. K., Waksler, R., & Older, L. (1994). Morphology and meaning in the English mental lexicon. Psychological Review, 101(1), 3–33. Mel’c˘uk, I (1993). Cours de morphologie ge´ne´rale, vol. 2. Significations morphologiques. Les Presses de l’Universite´ de Montre´al, CNRS E´ditions. Miceli, G. A., Silveric, C., Villa, G., & Caramazza, A. (1984). The basis for the agrammatic’s difficulty in production main verbs. Cortex, 20, 207– 220. Milberg, W., Blumstein, S. E., Katz, D., Gershberg, F., & Brown, T. (1995). Semantic facilitation in aphasia: effects of time and expectancy. Journal of Cognitive Neuroscience, 7, 33 –50. Neely, J. H. (1991). Semantic priming effects in visual word recognition: a selective review of current findings and theories. In D. Besner, & G. Humphreys (Eds.), Basic processes in reading: Visual word recognition. Hillsdale, NJ: Erlbaum. Nikolova, R., & Jarema, G. (2002). The interaction of morphological structure and prefix transparency in the processing of Bulgarian aspectual verb forms. Brain and Language, 81, 649–665. Ostrin, R., & Tyler, L. (1993). Automatic access to lexical semantics in aphasia: evidence from semantic and associative priming. Brain and Language, 45, 147– 159.

368

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

Perlak, D., & Jarema, G. (2000). Accessing aspectual verbs in Polish: a case study. Brain and Language, 74(3), 423–425. Perlak, D., & Jarema, G. (2001). Computation and storage of aspectual verb forms in the mental lexicon: evidence from Polish. In G. Zybatow, U. Junghanns, G. Mehlhorn, & L. Szucsich (Eds.), Current issues in formal Slavic linguistics (pp. 521 –527). Lang: Frankfurt am Main. Pinker, S., & Prince, A. (1994). Regular and irregular morphology and the psychological status of rules of grammar. In S. D. Lima, R. L. Corrigan, & G. K. Iverson (Eds.), The reality of linguistic rules. Amsterdam/ Philadelphia: John Benjamins. Plaut, D. C., & Gonnerman, L. M. (2000). Are non-semantiic morphological effects incompatible with a distributed connectionist approach to lexical processing? Language and CognitiveProcesses, 15, 445–486. Prather, P., Zurif, E., Love, T., & Brownell, H. (1997). Speed of lexical activation in nonfluent Broca’s aphasia and fluent Wernicke’s aphasia. Brain and Language, 59, 391 –411. Prather, P., Zurif, E., Stern, C., & Rosen, T. J. (1992). Slowed lexical access in nonfluent aphasia: a case study. Brain and Language, 43, 336–348. Rapp, B., & Caramazza, A. (1998). A case of selective difficulty in writing verbs. Neurocase, 4, 127–140. Rapp, B., & Caramazza, A. (2002). Selective difficulty with spoken nouns and written verbs: a single case study. Journal of Neurolinguistics, 15, 373–402. Rastle, K., Davis, M. H., Marslen-Willson, W. D., & Tyler, L. K. (2000). Morphological and semantic effects in visual word recognition: a time–course study. Language and Cognitive Processes, 15(4/5), 507–537. Rueckl, J. G., Mikolinski, M., Raveh, M., Miner, C. S., & Mars, F. (1997). Morphological priming, fragment completion, and connectionist networks. Journal of Memory and Language, 36, 382–405. Rumelhart, D. E., & McClelland, J. L. (1986). On learning the past tenses of English verbs. In J. L. McClelland, & D. E. Rumelhard (Eds.), Parallel distributed processing: Explorations in the microstructure of cognition (Vol. 2) (pp. 216–271). Psychological and biological models, Cambridge, MA: MIT Press. Sandra, D. (1990). On the representation and processing of compound words: automatic access to constituent morphemes does not occur. The Quarterly Journal of Experimental Psychology, 42a, 529–567. Schreuder, R., & Baayen, R. H. (1995). Modeling morphological processing. In L. B. Feldman (Ed.), Morphological aspects of language processing (pp. 131 –154). Hillsdale, NJ: Lawrence Erlbaum Associates. Schzeuder, R., Baayen, R. M. (1997). How complex simplex words can be. Journal of Memory and Language, 37, 118–139. Segui, J., & Zubizarreta, M.-L. (1985). Mental representation of morphologically complex words and lexical access. Linguistics, 23, 759– 774. Seidenberg, M. S., & McClelland, J. L. (1989). A distributed developmental model of word recognition and naming. Psychological Review, 96, 523–568. Sereno, J., & Jongman, A. (1997). Processing English inflectional morphology. Memory and Cognition, 25, 425–437. Shapiro, K., & Caramazza, A. (2003). Looming a loom: evidence for independent access to grammatical and phonological properties in verb retrieval. Journal of Neurolinguistics, 15(2–3), 85 –111. Slabakova, R. (2001). Aspectual constraints in the mental lexicon. Journal of Psycholinguistic Research, 30(2), 197–217. Shuren, J. E., & Heilman, K. M. (1995). Reading comprehension in a transcortical motor aphasic. Neurology, 45, 1418. Tainturier, M.-J., Tremblay, M., & Lecours, A. R. (1992). Educational level and the word frequency effect: a lexical decision investigation. Brain and Language, 43, 460–474. Taft, M. (1981). Prefixed stripping revisited. Journal of Verb Learning and Verb Behavior, 20, 289 –297. Taft, M., & Forster, K. I. (1975). Lexical storage and retrieval of prefixed words. Journal of Verb Learning and Verb Behaviour, 14, 638 –647. Todorova, E., Panc˘ovska, R (1995). Frequency dictionary of written Bulgarian (1944–1989) (c˘estoten rec˘nik na bulgarskata publitzistika (1944–1989)). Kedar, Sofia. Tsapkini, K (2001). A psycholinguistic and neurolinguistic investigation of morphological processing and representation of verbs and nouns: evidence from Greek. PhD dissertation. Universite´ de Montre´al.

R. Nikolova, G. Jarema / Journal of Neurolinguistics 17 (2004) 349–369

369

Tsapkini, K., Jarema, G., & Kehayia, E. (2002). Manifestations of morphological impairments in Greek aphasia: a case study. Journal of Neurolinguistics, 14, 281 –296. Ulatowska, H. K., Sadowska, M., & Ka¸dzielawa, D. (2001). A longitudinal study of agrammatism in Polish: a case study. Journal of Neurolinguistics, 14, 321 –336. Ullman, M. T., Corkin, S., Coppola, M., Hickok, G., Growdon, J. H., Koroshetz, W. J., & Pinker, S. (1997). A neural dissociation within language: evidence that the mental dictionary is part of declarative memory, and that grammatical rules are processed by the procedural system. Journal of Cognitive Neuroscience, 9, 266–276. Zingeser, L. B., & Berndt, R. S. (1990). Retrieval of nouns and verbs in agrammatism and anomia. Brain and Language, 39, 14– 32. Zwitserlood, P. (1994). The role of semantic transparency in the processing and representation of Dutch compounds. Language and Cognitive Processes, 9, 341 –368.