Michela Mosca. Research Proposal for PhD Program Sdisu

Michela Mosca Research Proposal for PhD Program Sdisu Working title: Mental lexicon retrieval in multilingual speakers with two typologically near languages Research area for FSE scholarship: Human science Abstract The way bilinguals have access to their mental lexicons is still an uncertain topic in the psycholinguistics field. The latest researches on language production suggest that bilinguals move from using an inhibitory control model to a language-specific selective mechanism during development (Costa, Santesteban, & Ivanova, 2006). The authors state that the level of L2 proficiency affects the way multilingual people retrieve the lexical information of a weaker language (L3), namely through the shift from the inhibitory process of the non-target languages to a language-specific selection of the target language. The goal of the present project is to demonstrate that such a shift could not be always possible. In particular, if we take into account the factor of the typological distance between L2 and L3: a L3 linguistically closer L2 (than L1), becomes the main cause of cross-linguistic interferences in L3 speech production (Hammarberg & Williams, 1998; Cenoz, 2003). An in-depth exploration of the naming performance of bilinguals in a language switching task (L1-L3 and L2-L3) could give a new insight on a relatively underexplored field: the relationship between the role of the typological distance of L2 and L3 and the way bilinguals retrieve the mental lexicons and hence inform us about the validity of the actual switching paradigm. Keywords: mental lexicon, switching costs, typological distance, cross-linguistic influence. INTRODUCTION One of the most important questions in the study of bilingualism is how multilingual people manage to produce pure monolingual language output when the communicative situation requires them to do so. To keep separate the languages they know is one of the most remarkable abilities polyglots have. Indeed even though the speech production of highly proficient bilinguals in a second language (L2) can carry traces of the first

language (L1) (e.g. L1 pronunciation or syntactic structures; Flege, MacKay, & Meador, 1999; Pallier, Colomè, & Sebastiàn-Gallès, 2001; Yeni-Komshian, Flege, & Liu, 2000), it rarely shows L1 lexical intrusions (Poulisse, 1999; Pulisse & Bongaerts, 1994). Models that account for this behavior assume the existence of a language control process that differentially activate or/and inhibit each of the underlying language subsystems. The most famous model of inhibitory control in bilingual speech production is that proposed by Green (1998). According to this model (IC-Model - Inhibitory Control Model) when a bilingual speaks in one language, inhibitory control mechanisms suppress the irrelevant language. Besides, inhibition is proportional to the level of activation of the words that have to be suppressed. This means that the amount of inhibition applied to one language is proportional to the level of proficiency that the speaker has in that language. Hence, speaking in a less dominant L2 requires a strong inhibition of the L1 lexical nodes. In compliance with the IC-Model every inhibition needs a different amount of time to be overcome: the more suppression is applied to a given lexicon, the more time will be needed to retrieve words from that lexicon. The IC-Model assumption has been widely supported empirically by evidence of asymmetrical switching costs during language production task (Costa & Santesteban, 2004; Costa, Santesteban, & Ivanova, 2006; Meuter & Allport, 1999). When unbalanced bilinguals switch from their less proficient language (L2) to their stronger L1 they suffer major magnitude of switching cost then otherwise (from L1 to L2). That is, the switching costs for not highly proficient bilinguals are asymmetrical, as according the IC-Model. The quantity of asymmetry in the switching costs seems to depend on the L2 proficiency: the higher the L2 proficiency, the smaller the asymmetry. Consequently, it was proved that for highly proficient balanced bilinguals the switching costs between L1 and L2 (and vice versa) were symmetrical, as supposed (Meuter & Allport, 1999; Monsell, Yeung, & Azuma, 2000). Nonetheless, there are instances when highly proficient balanced bilinguals exhibit no difference in switching costs between one of their strong language (either L1 or L2) and a weaker L3 (Costa & Santesteban, 2004). In these cases of symmetrical switching costs between two languages (L1 Spanish-L3 English and L2 Catalan-L3 English), where the speaker has a different level of proficiency, researches have argued for an alternative mechanism to the IC-Model: a language specific selection mechanism (Costa, 2005; Costa & Caramazza, 1999; Costa, Miozzo, & Caramazza, 1999; Costa et al., 2006; Roeflos, 1998). The language-specific selection hypothesis assumes that lexical access does not need inhibitory control. According to the authors, the balanced bilingual has developed a lexical selection mechanism that actives only the words belonging to the intended language ( target language ). The lexical nodes belonging to the non-response language ( non-target language ) would not act as competitors

during the lexical selection, and therefore, there is no need for them to be suppressed. Costa et al. argued that the main factor driving the shift from the inhibitory control model to a language-specific selection mechanism was the proficiency achieved in any pair of languages. When this specific selection mechanism is developed, it will be applied to any language, regardless of the achieved speaker s proficiency in that language. Therefore, if the bilingual has not attained balanced performance levels in at least two languages, then his lexical selection mechanism would make use of the inhibitory control process. Further results seemed to support the language-specific selection theory; in particular Costa et al. (2006) showed symmetrical time latencies (switching costs) in highly proficient bilinguals (L1 Spanish, L2 Catalan) between their L2 and a weaker L3 (English). Furthermore the authors investigated on other variables such as: age of L2 acquisition and language similarity. More precisely they found symmetrical switching costs for both highly proficient late bilinguals (L1 Spanish L2 English) and highly proficient bilinguals with different languages (L1 Spanish L2 Basque). Thus, Costa et al. (2006) results have suggested that neither language similarity nor L2 age of acquisition have effect on the control of lexicalization. But do highly proficient bilinguals always show symmetrical switching costs for L2-L3? The aim of the present project is that to answer to this question, taking into account another factor missed out by Costa et al. (2006), that is the role of the typological distance between L2 and L3 in L2-L3 switching tasks. In the last decades there have been interesting findings concerning the role of a L3 typological near L2, during L3 acquisition (Cenoz, 2003; De Angelis & Selinker, 2001; Hammarberg & Williams, 1998; Odlin & Jarvis, 2004; Ringbom, 1986; Stedje, 1977). According to the Typological Distance Theory the language typologically closest to the target L3 is the main cause of the transfer lapses [I]f the two non-target languages are equally close to the target L3, it is the foreign non target-language that dominates crosslinguistic transfer (Cenoz, 2003). So the theory suggests that the acquisition of a L3 can suffer from lexical intrusions by a typologically near language known by the speaker; when the non-target languages are equally close to the target language, it is the foreign language the main source of the cross-linguistic effects. An exhaustive example of it can be offered by the longitudinal single-case study conducted by Hammarberg & Williams (1998). The study showed to which extend the L3 (Swedish) acquisition process of the speaker (L1 English) suffered from L2 (German) cross-linguistic interferences. The subject of the study was Sarah Williams (SW), a British English native speaker. Through a six-year stay in Germany, SW achieved a near-native competence in German that became her prior L2. The study concentrates on her L3 (Swedish) learning process immediately after her arrival in Sweden from Germany.

During the conversations in Swedish the learner made abundant use of language switches (both from English and German), particularly in the early stages of L3 development. The switches appeared differently according to their pragmatic purpose: the speaker appealed to English for pragmatically purposeful switches and to German for automatic, non-intentional switches (Poulisse & Bongaerts, 1994). In this case German (L2) seemed to have for SW an unconscious supplier role for the L3 utterance. The involuntary languages switches during L3 speech production were labeled as Without Indentified Pragmatic Purpose, and shortened to WIPP (Hammarberg, 2001). The authors explained that WIPP do not seem to be caused by any language choice, but rather seem to be lapses in the formulation of L3. In the case study of SW, 92% of all WIPP switches were German (4% from English and 4% from other L2s). Researchers believe that nonadapted languages switches (WIPP) in L3 acquisition from the L2 are due by various factors: status, typological distance, recency and proficiency. Status: by learning a L3, the polyglot tends to use the same acquisition strategies used for the L2, because they are both perceived as foreign languages: there is an unconscious desire to suppress the L1 on the basis of its being not-foreign. Typological distance: linguistic transfers can be caused by the perceived distance or psychotypology (Kellerman, 1983) by the speaker between the languages. The learner is more likely to transfer a structure from one language to another if the two languages are perceived as similar. Recency: learners are more likely to borrow from a language they actively use, because highly frequent lexical items are successful candidates for unintentional lexical transfer. Proficiency: the higher is the proficiency level achieved in a L2, the stronger is its influence during L3 speech production. So if a fluent L2 that is typologically near to the weaker L3 can cause in the learner the majority of non-intentional language switches (e.g. 92% as in the case of SW), it can be supposed, that in such cases, a specific language selection mechanism for the L3 mental lexicon could be unsuccessful. Since the L2 tends not only to remain active during L3 oral production (particularly in the early stages of development) but also to assume the role of default supplier language ( foreign language effect, Meisel, 1983; De Angelis & Selinker, 2001). The goal of this project is that to propose a new point of view, in order to analyze the organization of multilingual mental lexicons. Overview of the experiments

The experiments proposed in this project have two main goals: first to replicate the observations of Costa et al. (2006); second to test whether L2 typological nearness can affect the pattern of performance in L2-L3 switching tasks. Experiment 1 investigates the switching performance of late highly proficient bilinguals (L1 German, L2 English), learners of a L3 (Italian), when switching between their L1-L3 and L2-L3. Experiment 2 explores the switching time latencies of late highly proficient bilinguals (L1 Italian, L2 German) learners of a L3 (English), when the task requires them to switch from the L1 into the L3, from the L2 into the L3 and vice versa. In Experiment 3 early highly proficient bilinguals (L1 German-L2 Italian) learners of a L3 (English) are analyzed, when switching between L1-L2 and L2-L3. Finally, Experiment 4 investigates the performance of late highly proficient bilinguals (L1 German, L2 English) learners of L3 Swedish, during switching tasks in L1-L3 and L2-L3. The experimental procedure follows that of Costa et al. (2006): a picture naming task with language switches. The subjects of the study are instructed to name the pictures in either one of their stronger language (L1-L2) or L3. The target language is determined by a color cue, which indicates the language of production. Participants Twelve participants for each experiment are required. Participants age (18-30 years old) and socio-economic background should be as homogeneous as possible; university students are preferred. Mentally retarded, speech, sight and color defected participants are excluded from the experiments. Language situation of the subjects should be: L1= native speaker; L2= from C.1.1 to C.2.2 level; L3 = from A.1.1 to B.1.1 level. The foreign language proficiency will be valued through CEFR levels tests. Two of the three languages the participants speak have to be typologically close, in order to assess to which extend the typological nearness of a L2 can influence the L3 oral production, causing asymmetrical switching costs for L2-L3 even in highly proficient bilinguals. The two typologically close languages are German and English (both Germanic languages belonging to the Indo-European language family). The one more distant language is Italian (Romanic language of the Indo-European language family). For the last experiment three equally close languages are chosen: German, English and Swedish (all Germanic languages belonging to the Indo-European language family). The aim is to test the possible influence of a typologically near L1 in L3 mental lexicon retrieval.

Materials In accordance with Costa et al. (2006), the experiments of the present project will include a picture list containing 10 black and white line drawings of common objects with nocognate names. The materials used for all the participants will be the same. As in Costa et al. (2006), all 10 pictures will be presented individually on a computer screen. There are two types of trials: non-switch and switch trials. In a non-switch trial the previous trial is named in the same language, whereas in a switch trial the previous trial is named in a different language. Pictures are presented in short sequences ( lists ); a range of 5-14 pictures are randomly placed in 100 lists which contained from 0 to 4 switching trials. The total number of the trials in each experiment is 950 (665 non-switch trials (70%) and 285 switch trials (30%)). Each picture is presented 95 times during the experiment. In half of the non-switch trails (333 trials) participants are asked to name the picture in the stronger language (L1 or L2) and in L3 in the other half (333 trials). The same is applied for the switch trials, so that the participants use their stronger language (L1 or L2) and the weaker language (L3) the same number of time during the experiment (475 stimuli for each language). Procedure All participants are individually tested in a soundproof room. Each of them are given written instructions on a computer screen. They are asked to name the picture so fast as possible, trying to avoid errors. The language to be used is suggested by the color of the cue: blue for the stronger language (L1 or L2) and red for L3. The assignment of color cue to response language is counterbalanced across participants. A training session will precede the standard experiment. The first picture of the list is presented and remains on the screen for 2000 ms or until participant responses. A black pause of 1150 ms follows. A new picture is presented and the circle is repeated until the end of the list. After the presentation of the last picture of the list, an asterisk presented for 1000 ms will signal the end of the list. Reaction time analyses is the best way to find out if any asymmetry in switching performance is to be seen. Differences in time latencies will signal the use of a IC-Model for switching tasks. Data analysis All task responses are tape recorded and coded as correct or incorrect. Only correct responses analyzed. Correct and incorrect responses are both included in the errors analysis. Error rates and naming latencies are submitted to two analyses of variance

(ANOVAs). Two main variables are taken into account in the statistical analyses: Response language (either the stronger or the weaker language), and Type of Trial (Switch versus Non-Switch). Experiment 1 The aim of this experiment is that to replicate the symmetrical switching costs reported by Costa et al. (2006) for late highly proficient bilinguals, switching between their L1 and L3 and between their L2 and L3. The speaker (L1 German) of this group should have been exposed to their L2 (English) relatively late in life, but should have reached a near native level. Furthermore they are considered L3 (Italian) learners. Considering the results of Costa et al. (2006), symmetrical switching costs for L1 and L3, as for L2 and L3 are expected. Experiment 2 In this experiment switching performance of late highly proficient bilinguals with a L2 typologically close to the L3 are tested. The participants are Italian-German late highly bilinguals, learners of English. The goal of the experiment is to assess whether a more linguistic nearness between L2 and L3 can effect the way highly proficient bilinguals have access to their mental lexicon: from a specific-language selection mechanism to the IC-Model. If L2 typological nearness can influences L3 oral speech production (Hammarberg, 2001, Cenoz, 2003), then asymmetrical time latencies for L1-L3 and L2-L3 switching tasks are expected. Experiment 3 Purpose of this experiment is to assess the switching performance of early highly proficient bilinguals, learners of a L2 typologically close L3. The linguistic situation of the speaker is the same as in Experiment 2, except from the age of acquisition of the L2. If the Foreign Language Effect Theory can be applied only for late bilingual learners of a new language, then no switches asymmetry should be registered during this experiment. Switching performance of early highly proficient bilinguals should not suffer from the typological nearness of L2 (German) and L3 (English), because of their different status (first versus foreign language). I expect switching costs between L1-L3 and L2-L3 to be symmetrical, as according to the language-specific selection mechanism formulated by Costa et al. (2006). Experiment 4

Switching performance of late highly proficient bilinguals, learners of a L3 equally close to the L1 as to the L2 are assessed in this experiment. All participants have German as mother tongue, have learned English late in life and can be labeled as Swedish learners. German, English and Swedish are equally close languages. In accordance with the Typological Distance Theory, the status of the L2 makes it to be the main cause of nonintentional language switches (Cenoz, 2003). In this case L1 typological nearness should not affect L2-L3 switching task. The linguistic closeness of L2 and L3 should cause asymmetrical switches costs between L1-L3 and L2-L3, as in Experiment 2. Summarizing scheme of the expected results Experiment 1. L1 German; L2 English; L3 Italian L1-L3 = symmetrical switching costs L2-L3 = symmetrical switching costs Experiment 2. L1 Italian; L2 German; L3 English L1-L3 = asymmetrical switching costs L2-L3 = asymmetrical switching costs Experiment 3. L1s (German-Italian); L2 English L1(a)-L2 = symmetrical switching costs L1(b)-L2 = symmetrical switching costs Experiment 4. L1 German; L2 English; L3 Swedish L1-L3 = asymmetrical switching costs L2-L3 = asymmetrical switching costs Conclusion If the hypothesis that the typological nearness of the L2 to a L3 can affect the way highly proficient bilinguals access to their mental lexicons turns to be true, it means that a review of the overall switching costs theory proposed by Costa et al. (2006) is required. In particular, factors such as typological distance between L2 and L3, and age of acquisition of the L2 could make a shift from the IC-Model (though to be for low proficient bilinguals) to a specific language selection model (generally related to highly proficient bilinguals) no possible. The asymmetrical time latencies registered by some late highly proficient

bilinguals (when L2 and L3 are typologically close) would provide evidence of the use of the IC-Model even in highly proficient bilinguals. On the contrary if no differences in switching performance are registered, it suggests that the typological nearness of L2 and L3 can not affect the way highly proficient bilinguals retrieve the lexical information in a L2-L3 switching task, that is through a specific language selection mechanism, as argued by Costa et al. (2006). References Cenoz, J. (2003). The additive effect of bilingualism on third language acquisition: A review. International Journal of Bilingualism, 7, 71 87. Colomè, A. (2001). Lexical activation in bilinguals_ speech production: Languagespecific or language-independent? Journal of Memory and Language, 45, 721 736. Costa, A., & Caramazza, A. (1999). Is lexical selection in bilingual speech production language-specific?. Further evidence from Spanish English and English Spanish bilinguals. Bilingualism: Language and Cognition, 2, 231 244. Costa, A. (2005). Lexical access in bilingual production. In J. F. Kroll, A. M. B. De Groot (Eds.), Handbook of bilingualism: Psycholinguistic approaches. New York: Oxford University Press. Costa, A., Miozzo, M., & Caramazza, A. (1999). Lexical selection in bilinguals: Do words in the bilingual_s two lexicons compete for selection? Journal of Memory and Language, 41, 365 397. Costa, A., & Santesteban, M. (2004). Lexical access in bilingual speech production: evidence from language switching in highly proficient bilinguals and L2 learners. Journal of Memory and Language, 50, 491-511. Costa, A., Santesteban, M. & Ivanova I. (2006). How do highly proficient bilinguals control their lexicalization process? Inhibitory and language specific selection mechanisms are both functional. Journal of Experimental Psychology: Learning, Memory, and Cognition, 32(5), 1057-1074. De Angelis, G., & Selinker, L. (2001). Interlanguage transfer and competing linguistic systems in the multilingual mind. In J. Cenoz, B. Hufeisen, & U. Jessner (Eds.), Cross-linguistic influence in third language acquisition: Psycholinguistic perspectives (pp. 42-58). Clevedon, UK: Multilingual Matters. Flege, J. E., MacKay, I. R. A., & Meador, D. (1999). Native Italian speakers_ perception and production of English vowels. Journal of the Acoustical Society of America, 106, 2973 2987.Green, D. W. (1986). Control, activation and resource. Brain and Language, 27, 210 223. Green, D. W. (1998). Mental control of the bilingual lexicosemantic system. Bilingualism: Language and Cognition, 1, 67 81. Hammarberg, B. (2009). Processes in third language acquisition. Edinburgh University Press Ltd. Kellerman, E. (1983). Now you see it, now you don t. In S. Gass & L. Selinker (Eds.), Language transfer in language learning (pp. 112-134). Rowley, MA: Newbury House. Meisel, J. (1983). Transfer as a second language strategy. Language and Communication, 3, 11-46.

Meuter, R. F. I., & Allport, A. (1999). Bilingual language switching in naming: Asymmetrical costs of language selection. Journal of Memory and Language, 40, 25 40. Monsell, S., Yeung, N., & Azuma, R. (2000). Reconfiguration of task-set: Is it easier to switch to the weaker task? Psychological Research, 63, 250 264. Odlin, T., & Jarvis, S. (2004). Same source, different outcomes: A study of Swedish influence on the acquisition of English in Finland. The International Journal of Multilingualism (in press). Pallier, C., Colom_e, A., & Sebasti_an-Gall_es, N. (2001). The influence of nativelanguage phonology on lexical access:exemplar-based versus abstract lexical entries. Psychological Science, 12, 445 449. Poulisse, N., & Bongaerts, T. (1994). First language use in second language production. Applied Linguistics, 15, 36 57. Ringbom, H. (1986). Crosslinguistic influence and the foreign language learning process. In M. Sharwood Smith & E. Kellerman (Eds.), Crosslinguistic influence in second language acquisition (pp. 150-162). Oxford, UK: Pergamon Press. Roelofs, A. (1998). Lemma selection without inhibition of languages in bilingual speakers. Bilingualism: Language and Cognition, 1, 94 95. Stedje, A. (1977). Tredjespråksinterferens i fritt tal-en jämförande studie. In R. Palmberg & H. Ringbom (Eds.), Papers from the Conference on Contrastive Linguistics and Error Analysis, 19 (pp. 141-158). Publications of the Research Institute of the Åbo Akademi Foundation. Williams, S., & Hammarberg, B. (1998). Language switches in L3 production: Implications for a polyglot speaking model. Applied Linguistics, 19 (3), 295-333. Yeni-Komshian, G. H., Flege, J. E., & Liu, S. (2000). Pronunciation proficiency in the first and second languages of Korean English bilinguals. Bilingualism: Language and Cognition, 3, 131 141.