Cross-linguistic Representation and Processing of Words, Grammatical features, and Sentences

Abstract

This dissertation investigates how parametric variation of linguistic properties leads to similarities and differences in language processing across the levels of words, grammatical features, and sentences. For a truly generalizable theory of psycholinguistics, the languages surveyed should not be constrained to English (Garnham, 1994; Norcliffe et al., 2015), yet cross-linguistic understanding of language processing at multiple levels is not very clear. For instance, are grammatical features shared across different languages? Do speakers of languages with different canonical word orders go through the same process when producing and comprehending sentences with different argument structures? English and Korean, which are the languages studied in this dissertation, form a good pair to investigate those questions given their typological differences in writing systems, word order, and morphological complexity, among others. Chapter 2 and Chapter 3 examine representations of lexical concepts and grammatical features among Korean-English bilinguals. Chapter 2 uses the priming paradigm to investigate whether lexical concepts as well as grammatical case feature prime between two languages. In Chapter 3, neural decoding is applied to both EEG collected for a single language as well as for two different languages to examine how words and grammatical features (i.e., number and tense) are processed in bilingual brains. Both Chapter 2 and Chapter 3 provide partial evidence for language-independent representation of lexical concepts; robust cross-languages priming effects of lexical concepts are found as well as overlapping time windows for neural decoding in English and Korean, although cross-languages neural decoding does not yield above-chance accuracies. Among the grammatical features, only the number feature showed commonalities in neural decoding. Chapter 4 and Chapter 5 investigate cross-linguistic sentence processing mechanisms during production and comprehension. Chapter 4 reports an eye-tracking study with a visual-world paradigm designed to explore the time-course of sentence production in English and Korean. While both English and Korean speakers show sensitivity to argument structure during sentence planning, group differences are also observed, such that English speakers rely mostly on the verb images to encode argument structure while Korean speakers use both verb and object information. Chapter 5 focuses on the timing of verb pre-activation in the two languages during sentence comprehension. Language differences are again observed; the timing of verb activation reflected the argument structure of the sentences in English, but not in Korean. Taken together, this dissertation highlights that unlike lexical concepts, typological differences in English and Korean lead to separate representations of grammatical features and distinct procedures during sentence production and comprehension. Therefore, although the identical linguistic labels given to grammatical features in different languages suggest language-independent representations of these features, they may in fact be represented based on language-specific details. Also at the sentence level, the distinct canonical word orders in the two languages result in variations in how different types of information is utilized both for production and comprehension, by interacting with other processing-related factors such as memory and morphological systems.