Research master thesis | Linguistics (research) (MA)
under embargo until 2025-07-31
2025-07-31T00:00:00Z
While multiple models of language production exist, it remains difficult to find conclusive evidence for one or the other model. The present study is one of the first to incorporate both behavioral...Show moreWhile multiple models of language production exist, it remains difficult to find conclusive evidence for one or the other model. The present study is one of the first to incorporate both behavioral data and EEG data in a native Dutch noun phrase production task to investigate the processing of grammatical gender. We used a picture-word interference paradigm in which speakers had to name a picture while being distracted by a superimposed word. The distractors were manipulated for grammatical gender congruency with and semantic relatedness to the picture names. In separate blocks, participants had to produce either bare nouns, or determiner noun constructions. We found a robust effect of the semantic relatedness effect in the behavioral data, in which targets belonging to the same semantic category as the distractor were named significantly later than unrelated targets. The EEG data on the semantic relatedness effect were less clear, possibly due to opposing ERP components in (partly) overlapping time windows. More importantly, we replicated the gender congruency effect in reaction times, showing that naming latencies were longer for gender-incongruent trials than for congruent trials. This behavioral effect disappeared in bare noun naming. The EEG data showed similar results, in which a more positive peak was found between 385-585 milliseconds for incongruent than for congruent trials, again only in the determiner noun block. The result seems to indicate a P300 effect rather than the expected enhanced N400, suggesting higher conflict resolution processes for incongruent trials. The absence of the gender effect in bare noun naming in both data types suggests that grammatical gender is not automatically accessed or selected in Dutch noun (phrase) production. This in turn has implications for speech production models.Show less
This paper investigates how well adult Spanish-Dutch bilinguals have acquired grammatical gender agreement in Dutch and whether individual speakers apply specific strategies to resolve difficulties...Show moreThis paper investigates how well adult Spanish-Dutch bilinguals have acquired grammatical gender agreement in Dutch and whether individual speakers apply specific strategies to resolve difficulties associated with gender in Dutch. Both Spanish and Dutch have two-fold gender systems: Spanish differentiates between a feminine and a masculine gender; and Dutch distinguishes between a common and a neuter gender. In Spanish each gender is marked on the determiner and the adjective with a distinct morphological marker (e.g. the indefinite articles un – amasc and una - afem). In Dutch the indefinite article converges into one form for both genders (i.e. een) and the rules of adjectival agreement are less straightforward than the rules in Spanish. To examine how adult bilinguals handle the opaque Dutch gender system, data from four late Spanish-Dutch bilinguals (L1 Spanish, L2 Dutch), seven early Spanish-Dutch bilinguals and six monolingual Dutch speakers were collected by means of elicited production and analyzed for grammatical gender agreement on definite determiners and adjectival inflection. It was found that the four highly proficient late speakers of Dutch show non-target-like performance in their data involving agreement on both definite determiners and adjectives, regularly overgeneralize to the common gender, and tend to produce fewer indefinite than definite phrases in comparison to the monolinguals and early bilinguals. The findings in this study are discussed in light of previous research regarding the different mechanisms of child and adult (second) language acquisition of grammatical gender in Dutch (Blom et al., 2006, 2008; Prevost and White, 2000; White, 2003). The evidence from the current experiment suggests that: a. adults use two different routes for acquiring gender agreement in Dutch - a lexical route for agreement at the determiner, and a rule-based route for agreement at the adjective (Blom et al., 2006, 2008); b. adult speakers may have the knowledge of the rules of gender agreement in Dutch but are not successful in applying the rules consistently in their production (Prevost and White, 2000; White, 2003).Show less
The current thesis aims to investigate the set of factors operative in gender assignment process in Greek-Turkish nominal constructions, that is in code-switching where Greek determiners (gendered)...Show moreThe current thesis aims to investigate the set of factors operative in gender assignment process in Greek-Turkish nominal constructions, that is in code-switching where Greek determiners (gendered) and Turkish nouns (non-gendered) are combined together. MacSwan (2005) using the Minimalist Program (Chomsky, 1995) suggests that in determiner-noun code-switches determiners can only stem from the gendered language, since it encodes the largest amount of uninterpretable features compared to the non-gendered language it interacts with, while nouns as non-function words might as well originate in the non-gendered language. Likewise, Matrix Language Frame was proposed by Myers-Scotton (1993b) in an attempt to set up a theoretical framework within which insertional CS production occurs and is framed by constraints. The latter theory posits that in bilingual speech only on language is dominant, the Matrix Language, while other-language constituents stem from the Embedded Language(s). Bilingual NP Hypothesis complementing MLF model suggests that system morphemes can only stem from the Matrix Language. As opposed to system morphemes, content morphemes may be lexical items from the Embedded Language. In either way, Matrix language accounts for the morpho-syntactic frame in bilingual speech. This results in DPs where Matrix Language, that is Greek, is responsible for determiner. Drawing on the abovementioned frameworks, a 20-person sample of Greek Turkish bilinguals were called to participate in the Director-Matcher task. Through this controlled elicitation and artificial technique Greek-Turkish bilinguals were instructed in such a way that they were prompted to use Turkish nouns within Greek phrases and sentences. Except for the Director-Matcher task, the same subjects were asked to reply to an online linguistic questionnaire where they had to answer crucial for this research questions as well as to evaluate their language skills. The results from the Director-Matcher task indicate that Greek-Turkish bilinguals in Western Thrace use neuter as default, while phonological and analogical factors play no role in gender assignment to Turkish nouns. Significantly enough, the persistent use of neuter in the data points to that, despite the perpetual affiliation of Greek and Turkish among the bilinguals in Western Thrace, they treat Turkish nouns as foreign. This study shows that this is true even for the bilingual system of individuals who are members of a community that has been bilingual for a century. Furthermore, the prevalence of neuter as default gender in Greek-Turkish nominal constructions proves the claim that the criterion determining the factors based on which grammatical gender is assigned to the nouns depends on the language pair under study. Another essential issue which drew my attention while transcribing the data I recorded is the absence of articles (definite or indefinite) in several data points (n=124). It is necessary to highlight that these occurrences do not adhere to the rules of Greek syntax and as, a result, violate the grammaticality of the Greek language system. Last but not least, this study confirms that neuter is the default gender in Greek.Show less