University of Connecticut
Three uses for experimental syntax
Thursday, May 19
My goal in this talk is to examine the role that experimental syntax techniques can play in 21st century linguistics (and hopefully inspire some discussion of new experimental projects). I will review three projects that I have been working on recently, each of which suggests a different possible answer. In the first study, I will argue that large corpora of formally collected acceptability judgments can be helpful in re-evaluating some of the foundational assumptions of the field, such as the fundamental nature of the grammar. Specifically, I will use a large judgment corpus that I collected to argue against recent claims in the computational linguistics literature (e.g., Lau, Clark, and Lappin 2015) that generative grammars can be replaced by surface-oriented models like n-grams and recurrent neural networks. In the second study, I will argue that formally collected judgments can reveal interesting patterns of variation, both crosslinguistically, and between languages, especially if we use factorial designs and numerical ratings to eliminate the confound introduced by categorical judgments. I will present the results of three experiments on island effects in Norwegian, which show surprising patterns of variation (at least relative to the patterns that have previously been reported). Finally, in the third study, I will argue that training in experimental syntax can serve as a stepping stone toward linking syntactic theories with neurolinguistic methods. I will present the results of an EEG experiment looking at the neuronal oscillations underlying two A-movement constructions: unaccusatives and passives. Though follow-up studies are required to refine the results, the first experiment suggests a potential oscillatory signature for A-movement in the theta band, which in turn may have implications for the ultimate theoretical analysis of A-movement.
University of Hawai‘i at Mānoa
Learning and Processing Classifiers in L2 Chinese
Thursday, May 19
Classifiers, or measure words, are an integral part of the Chinese nominal system. Recent work in psycholinguistics has shown that Chinese native speakers make use of both grammatical and semantic information encoded by classifiers during language comprehension in real time. I will present two visual-world eye-tracking studies in which we ask to what extent L2 learners of Chinese, for whom classifiers are known to be challenging, make use of this information during real-time processing. Hypotheses and findings will be discussed within the context of previous work on the processing of gender-marked determiners by L2 learners of European languages. Within this context, the particular difficulties associated with acquiring properties of noun class systems in an L2 have been proposed to result from key differences in how child and adult learners process associative relations between nouns and their modifiers. I will present results from a laboratory-based learning study with beginning learners of Chinese that support such an account, and raise questions for L2 instruction and curriculum design.
University of Rochester
Communicative efficiency in language production: Optional case-marking and relative clause production in Japanese
Friday, May 20
Grammatical encoding is one of the earliest stages in linguistic encoding. One broadly accepted view holds that grammatical encoding is primarily or exclusively affected by production ease, rather than communicative considerations (Arnold, 2008; Arnold, Wasow, Asudeh, & Alrenga, 2004; Ferreira, 2008; Ferreira & Dell, 2000; Lam & Watson, 2010; MacDonald, 2013). This contrasts with proposals that speakers’ preferences during grammatical encoding reflect a trade-off between production ease and communicative goals (Jaeger, 2013).
In this talk, I will be presenting production studies in Japanese (Kurumada, 2012; Kurumada & Jaeger, 2015) as well as two artificial language learning studies (Fedzechkina et al. 2012; 2013) to tease apart predictions of the two theories. On the topic of case-marking, the studies show that speakers are more likely to produce case-marking when the properties of the sentence would otherwise bias comprehenders against the intended interpretation. In a relative clause production study, speakers are producing a head noun of a relative clause with longer duration when its syntactic status is less predicable. In both cases, theories that are purely based on production ease won’t predict the patterns in production. I discuss roles of predictability and communicative efficiency in language production and how typologically-rich data sets can advance our understanding of mechanisms underlying real time language processing.
University of Hawai‘i at Mānoa
Korean prosody and sentence processing:
Evidence from speaking, listening, reading, writing, and memory tasks
Friday, May 20
A fundamental – and actively debated – question in psycholinguistic research concerns the connections among language production, comprehension, and learning, such as the extent to which readers and listeners draw on production processes during comprehension, and share representations across the processing system versus employing distinct ones. In this talk I’ll present a set of studies on Korean prosody and sentence processing, and suggest that the findings are best accounted for by a unified processing system. I’ll take the position that the processor builds linguistic structure at multiple levels of representation concurrently and incrementally, and allows each level to continually update connected representations, such as allowing syntactic phrasing to influence prosodic phrasing, and vice versa, in both production and comprehension. Along the way, I’ll describe a range of experimental tasks beneficial to the study of language, including cross-modal naming, self-paced reading, and different types of production tasks.
The use of morphosyntactic cues in online sentence comprehension by young Mandarin-speaking children
Friday, May 20
Language comprehension involves the rapid integration of different types of linguistic and non-linguistic information. Understanding the mechanism underlying this rapid process is a central component of the study of language comprehension. Research on adult sentence processing has demonstrated that when interpreting a sentence, the human sentence processing mechanism (or the parser), incrementally computes the structural representation and possible meanings of the sentence while drawing on different sources of linguistic and non-linguistic information. A key question that has been widely investigated is the role of event knowledge in sentence comprehension.
Knowledge about events includes typical event participants, causal relationships between participants and objects, instruments, time course, and duration. Previous research has shown that event information associated with verbs and verb morphology plays an important role in adult sentence comprehension. Adults can use the information immediately and effectively to construct representations of events. However, not much research has been conducted to investigate children’s use of linguistic information during real-time event comprehension. In this talk, I will present two studies investigating whether 3-year-old Mandarin-speaking children can use morphological/morphosyntactic cues effectively during real-time event comprehension.