Casaponsa, Aina and Duñabeitia, Jon Andoni and Carreiras, Manuel (2016) Bilingualism and Visual Word Recognition : Sub-lexical modulation of language detection and lexical access in bilinguals. PhD thesis, University of the Basque Country (UPV/EHU).
Full text not available from this repository.Abstract
Individuals raised in bilingual societies such as the Basque Country are constantly exposed to both their languages (e.g., Basque and Spanish). On a daily basis, such bilinguals have to rely on two language systems to interpret and represent their environment, to communicate, and to comprehend the world around them. How do they handle representations corresponding to different language codes? How do they know the language in which they are operating? This PhD Thesis focuses on such questions in the context of reading, particularly in visual word recognition. Then, how and when do bilinguals know the language of the words they read, and how does language identification affect the access to representations of words in two different codes?Although several studies in the domain of psycholinguistics have already shown that bilingual populations access representations of words (word forms) in a language non-selective fashion, little is known about the potential role that intrinsic properties of languages play in word recognition. For instance, it is more difficult for a Catalan-Spanish bilingual to differentiate the language in which a given word is written than for a Basque-Spanish bilingual. One of the reasons for this is that the structure of words is more similar between Catalan and Spanish than between Basque and Spanish. However, both Catalan-Spanish and Basque-Spanish bilinguals can figure out that the word “mártir” (martyr in both English and Basque, and “màrtir” in Catalan) is imperatively a word of Spanish. Neither Basque nor Catalan has the grapheme “á” in its repertoire. Thus, language attribution in this case can be established based on a single grapheme. Even if they do not know its meaning, or do not have a lexical representation of the word “mártir”, just based on this single grapheme they can rule out that at least this word not belongs to one of their languages (Catalan/Basque). Although this ability to detect the intrinsic characteristics of words in each language seems rather intuitive, a vast majority of literature in the domain of bilingual word recognition has supported the idea that same-script bilinguals (e.g., English-Welsh, Catalan-Spanish) faced with ambiguous language contexts, do not access language membership information until the lexical representations are “activated” in the bilingual mind. That is, once the bilingual mind find a match for the word form “mártir” in the lexicon (e.g., see Bilingual Interactive Activation plus model, Van Heuven & Dijkstra, 2002). This PhD Thesis seeks to demonstrate that these intrinsic characteristics of languages (i.e., orthographic regularities) can also be used by bilinguals to disentangle the language of words, prior to the activation of lexical entries in the lexicon. Moreover, it also seeks to characterise the extent to which the existence of different language detection mechanisms (sub-lexical and lexical) affect visual word recognition processes and cross-language interactions in different contexts. Recent studies have suggested that sub-lexical characteristics of words (e.g., specific graphemes and/or specific letter combinations) might help to speed up conscious language attribution (Oganian et al., 2015; Vaid & Frenck-Mestre, 2002; Van Kesteren et al., 2012), and even word identification (Lemhöfer et al., 2008). However, little is known about the underlying mechanisms guiding language detection at sub-lexical levels of processing and the specificity in which they occur. Are they automatic processes independent of context? Do they affect other stages of visual word recognition (i.e., lexical competition)? Are they specific for bilinguals?This thesis comprises ten experiments that vary in the demands of the task at hand (i.e., language detection, letter and word identification, lexical decision, and semantic categorisation), the language of operation (L1, L2 or both), the level of language proficiency of the participants (Spanish-Basque balanced bilinguals, Spanish-Basque highly proficient bilinguals, Spanish-English highly proficient bilinguals, and Spanish monolinguals), and the technical approach used (behavioural and electrophysiological methods). Experiments are designed and divided around three thematic axes: (A) Orthotactics and language identification, (B) Time-course of language detection; and (C) Orthotactics and cross-language activation. The work presented in stream (A) demonstrates that indeed bilinguals make use of the statistical orthographic regularities of languages (e.g., probabilistic combinations of letters) to disambiguate the language of the words, speeding up language attribution processes (Experiment 1 and 3: language decision tasks) and word identification (Experiment 2 and 4: word identification tasks). Furthermore, this sensitivity to orthographic regularities is not exclusive of bilingual minds (Experiments 3 and 4), but rather an emergent characteristic of language acquisition. Although this orthotactic sensitivity is not specific for bilinguals, the mechanisms involved and their influence to the different stages of word processing are highly mediated by the presence of one or two languages systems in participants’ mind. Hence, results from this first thematic axis (Chapter 2) corroborate the presence of two routes to access language membership information: sub-lexical and lexical. Furthermore, the activation of the sub-lexical language route via the presence of language specific orthographic information (i.e., letter combinations that clearly point out to a specific language) mediates lexical activation and ultimately visual word recognition. The series of experiments presented in axis (B) further demonstrate the automaticity in which this sub-lexical language detection mechanism occurs. In a series of event-related potentials (ERPs) experiments in combination with the masked language switching priming paradigm, results consistently demonstrate that when operating in a single language context the bilingual mind detects unconsciously perceived language changes as early as 200 ms. This unconscious language detection only occurred when sub-lexical language information was available (language specific orthography), showing a consistent lack of any signature related with lexical language detection mechanisms taking place when prime words followed the orthographic regularities of the target language. Hence, axis (B) demonstrates that language detection mechanisms at the sub-lexical level are automatic in essence, whilst language detection mechanisms at the lexical level are mediated by language context, not taking place when language detection is not relevant for the task. Furthermore, results were consistent independently of the combination of the languages known by the bilinguals (L2 being immersed in the society: Spanish-Basque, Experiment 5 and 7; L2 being a foreign language: Spanish-English, Experiment 6), and the methodology used to account for sub-lexical characteristics of languages (Experiment 5 versus Experiment 7). Moreover, a control group of monolinguals (Experiment 8) further support our interpretations by showing language effects for unconsciously perceived pseudowords with both language specific and language common orthography. To finalise, the work presented in axis (C) shows that this automatic language detection mechanism at the sub-lexical level clearly constrains cross-language lexical activation even when language specificity points out to the same language in which the bilingual is operating and when word processing is not even required during the task (Letter identification task: Experiment 9). Furthermore, the last Experiment of this thesis (Experiment 10) shows that the presence of language specific orthography not only constrain cross-language lexical activation but enables language selective lexical access. This is shown by masked translation priming effects emerging only when the orthographic characteristics of unconsciously perceived translations follow the orthotactics of the target language. In sum the work presented in this thesis suggests that despite the demands of the task at hand, the strategies used by the participants, and the language of operation, sub-lexical language information plays a critical role in bilingual word processing and in conscious and unconscious language detection. When bilinguals read words in both their languages they use a sub-lexical and a lexical route to access language membership information. Importantly, the sub-lexical route is automatically activated even when bilinguals are immersed in a single language context, whilst this does not seem to be the case for the lexical route. Moreover, language identity resolution via sub-lexical orthographic characteristics enables language selective mechanisms to emerge reducing unnecessary cross-language activation at the lexical level. Hence, the result of this thesis provides a comprehensive approach that helps us to better understand and characterise the processes underpinning bilingual visual word recognition. A modification of the most recent model of bilingual language processing (i.e., Bilingual Interactive Activation + extended model, Van Kesteren et al., 2012) is thus proposed at the end of this thesis in an attempt to account for all the work presented here and the results in the literature of visual word recognition in bilingual populations.