Linguistics Theses and Dissertations
Permanent URI for this collection
Browse
Recent Submissions
Now showing 1 - 5 of 114
- ItemGENERATING AND MEASURING PREDICTIONS IN LANGUAGE PROCESSING(2023) Nakamura, Masato; Philips, Colin; Linguistics; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Humans can comprehend utterances quickly, efficiently, and often robustly against noise in the inputs. Researchers have argued that such a remarkable ability is supported by prediction of upcoming inputs. If people use the context to infer what they would hear/see and prepare for likely inputs, they should be able to efficiently process the predicted inputs.This thesis investigates how contexts can predictively activate lexical representations (lexical pre-activation). I address two different aspects of prediction: (i) how pre-activation is generated using contextual information and stored knowledge, and (ii) how pre-activation is reflected in different measures. I first assess the linking hypothesis of the speeded cloze task, a measure of pre-activation, through computational simulations. I demonstrate that an earlier model accounts for qualitative patterns of human data but fails to predict quantitative patterns. I argue that a model with an additional but reasonable assumption of lateral inhibition successfully explains these patterns. Building on the first study, I demonstrate that pre-activation measures fail to align with each other in cases called argument role reversals, even if the time courses and stimuli are carefully matched. The speeded cloze task shows that “role-appropriate” serve in ... which customer the waitress had served is more strongly pre-activated compared to the “role- inappropriate” serve in ... which waitress the customer had served. On the other hand, the N400 amplitude, which is another pre-activation measure, does not show contrasts be- tween the role-appropriate and inappropriate serve. Accounting for such a mismatch between measures in argument role reversals provides insights into whether and how argument roles constrain pre-activation as well as how different measures reflect pre-activation. Subsequent studies addressed whether pre-activation is sensitive to argument roles or not. Analyses of context-wise variability of role-inappropriate candidates suggest that there are some role-inappropriate pre-activations even in the speeded cloze task. The next study at- tempts to directly contrast pre-activations of role-appropriate and inappropriate candidates, eliminating the effect of later confounding processes by distributional analyses of reaction times. While one task suggests that role-appropriate candidates are more strongly pre- activated compared to the role-inappropriate candidates, the other task suggests that they have matched pre-activation. Finally, I examine the influence of role-appropriate competitors on role-inappropriate competitors. The analyses of speeded cloze data suggest that N400 amplitudes can be sensitive to argument roles when there are strong role-appropriate competitors. This finding can be explained by general role-insensitivity and partial role-sensitivity in pre-activation processes. Combined together, these studies suggest that pre-activation processes are generally insensitive to argument roles, but some role-sensitive mechanisms can cause role-sensitivity in pre-activation measures under some circumstances.
- ItemThe Learning and Usage of Second Language Speech Sounds: A Computational and Neural Approach(2023) Thorburn, Craig Adam; Feldman, Naomi H; Linguistics; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Language learners need to map a continuous, multidimensional acoustic signal to discrete abstract speech categories. The complexity of this mapping poses a difficult learning problem, particularly for second language learners who struggle to acquire the speech sounds of a non-native language, and almost never reach native-like ability. A common example used to illustrate this phenomenon is the distinction between /r/ and /l/ (Goto, 1971). While these sounds are distinct in English and native English speakers easily distinguish the two sounds, native Japanese speakers find this difficult, as the sounds are not contrastive in their language. Even with much explicit training, Japanese speakers do not seem to be able to reach native-like ability (Logan, Lively, & Pisoni, 1991; Lively, Logan & Pisoni, 1993) In this dissertation, I closely explore the mechanisms and computations that underlie effective second-language speech sound learning. I study a case of particularly effective learning--- a video game paradigm where non-native speech sounds have functional significance (Lim & Holt, 2011). I discuss the relationship with a Dual Systems Model of auditory category learning and extend this model, bringing it together with the idea of perceptual space learning from infant phonetic learning. In doing this, I describe why different category types are better learned in different experimental paradigms and when different neural circuits are engaged. I propose a novel split where different learning systems are able to update different stages of the acoustic-phonetic mapping from speech to abstract categories. To do this I formalize the video game paradigm computationally and implement a deep reinforcement learning network to map between environmental input and actions. In addition, I study how these categories could be used during online processing through an MEG study where second-language learners of English listen to continuous naturalistic speech. I show that despite the challenges of speech sound learning, second language listeners are able to predict upcoming material integrating different levels of contextual information and show similar responses to native English speakers. I discuss the implications of these findings and how the could be integrated with literature on the nature of speech representation in a second language.
- ItemON SERIAL VERB CONSTRUCTIONS(1992) Wu, Daoping; Hornstein, Norbert; Linguistics; University of Maryland (College Park, Md); Digital Repository at the University of MarylandThe term serial verb construction (SVC) refers to a construction in which more than one verb are not connected by any lexical device such as conjunction and punctuation, etc.. This construction is quite popular in Chinese, Caribbean creoles, West African languages, and Dravidian languages. Structurally, the SVCs may be compounds, clauses or phrases. The clausal SVCs have been attested in all the serialized languages. The compound SVcs are reported in Chinese and Edo. Only a few instances of phrasal SVCs have been found in Dravidian languages. The compound SVCs in Chinese can consist of two verbs or a verb plus an adjective. The productive compound can only have the following structures: V trans. +V intrans., V trans + A and V intrans. + A. Among the three compounds, the head of the compound must link to both the external and the internal arguments when it is transitive. The nonhead adjective can link to either the external argument or the internal argument, while the nonhead verb can only link to the internal argument, and in most cases, the intransitive verb must be unaccusative. Feature Percolation Convention and the Case requirements generate all the grammatical compound SVCs and rule out all the ungrammatical ones. These rules also can well account for the differences between the VV compound and the VA compound. This conforms to what the Feature Percolation Convention and the Case requirement predict. It is suggested that the syntactic SVCs under discussion are clausal because of facts concerning their binding properties. This analysis is first proposed by Bickerton & Iatridou(1989) in their study of the Caribbean creoles. With this analysis, if the first verb in an SVC is in the matrix clause, the second verb is in an adjunct clause attached to the V' position of the matrix clause. This analysis obtains support from Chinese. In this respect, the only difference between Chinese and the Caribbean creoles is that the adjunct clause has two adjoining positions. Furthermore it is proposed in this thesis that the adjunct clause is a CP rather than an IP.
- ItemIsland Constraints: What is there for children to learn?(2022) Hirzel, Mina; Lidz, Jeffrey; Lau, Ellen; Linguistics; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)This dissertation presents behavioral studies that target the early syntactic representations of wh-movement during infancy and early childhood. Previous studies show that by 20 months-old, infants represent wh-movement and use this knowledge to respond to wh-questions during language comprehension tasks (Gagliardi 2012; Gagliardi et al., 2016; Seidl et al., 2003). Studies probing the nature of early representations of wh-movement show that by around 4 years-old, children represent island constraints (e.g., de Villiers et al., 1990; de Villiers & Roeper, 1995a, 1995b; Fetters & Lidz, 2016; Goodluck et al., 1992). It remains unclear how knowledge of wh-movement develops. What is the source of this ‘empirical gap’ between the onset of knowledge of wh- movement, and the observation that children respect island constraints? One possibility is that knowledge of island constraints is a component of Universal Grammar (e.g., Chomsky 1965, 1986; Hornstein & Lightfoot 1981). In this case, the ‘empirical gap’ in the linguistic abilities of infants compared to young children isn’t indicative of their linguistic knowledge, but rather the difficulties with testing infants and toddlers on complex syntax. Another possibility is that knowledge of island constraints is acquired via experience (e.g., Pearl & Sprouse, 2013). In this case, the ‘empirical gap’ reflects a knowledge gap, and there’s no evidence for knowledge of island constraints during infancy because it has yet to be acquired. Experiment 1 shows that by 19 months-old, infants have knowledge of wh-movement, and use this knowledge during language comprehension. Results are consistent with recent work which shows that 18 month-olds, but not 17 month-olds, know that wh-phrases co-occur with gap positions in wh-object questions (Perkins & Lidz, 2021). Experiment 2 shows that 3 year-olds respect locality constraints on wh-movement in wh- questions, and Experiment 3 shows that adults behave as expected on this task. Experiments 4 and 5 test children and adults on locality constraints on wh- movement in relative clauses, but these results are inconclusive (likely due to difficulties with moving the task online). The results of Experiment 3 raises challenges for learning hypotheses of island constraints which emphasize the role of linguistic experience. Learning models which propose that linguistic experience is the key factor in the acquisition of island constraints must consider these behavioral results when estimating the amount of data that the learner needs to solve the acquisition problem. These behavioral results are consistent with the hypothesis that knowledge of island constraints is innate, but further work is needed to close the ‘empirical gap’ between the onset of knowledge of wh- movement and the onset of knowledge of island constraints.
- ItemWhat could go wrong? Linguistic illusions and incremental interpretation(2022) Muller, Hanna; Phillips, Colin; Linguistics; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)The systems underlying incremental sentence comprehension are, in general, highly successful - comprehenders typically understand sentences of their native language quickly and accurately. The occasional failure of the system to deliver an appropriate representation of a sentence is therefore potentially illuminating. There are many ways the comprehender's general success could in principle be accomplished; the systematic pattern of failures places some constraints on the possible algorithms. This dissertation explores two cases of systematic failure, negative polarity illusions and substitution illusions (sometimes called "Moses illusions") with the goal of identifying the specific circumstances under which the illusion arises, and, as a consequence, the specific constraints placed on possible implementations of linguistic knowledge. In the first part of this dissertation, I explore the profile of the negative polarity illusion, a case in which a sentence containing an unlicensed negative polarity item and a preceding, but not structurally-relevant licensor is perceived as if it is acceptable, at least in early stages of processing. I consider various proposals for the grammatical knowledge that determines the restricted distribution of negative polarity items, and possible algorithms for using that grammatical knowledge in real time to process a sentence containing a negative polarity item. I also discuss possible parallels between negative polarity illusions and superficially-similar illusory phenomena in other domains, such as subject-verb agreement. Across sixteen experiments, I show that the profile of the illusion is more restricted than previously thought. Illusions do not always arise when an unlicensed negative polarity item is preceded by a structurally-irrelevant licensor, and the circumstances under which they do arise are quite specific. These findings suggest that the negative polarity illusion may be meaningfully distinct from other illusory phenomena, though this conclusion does not necessarily require stipulating a separate mechanism for every illusion. I discuss the implications of these findings for possible real-time implementations of grammatical knowledge. In the second part of this dissertation, I turn to the substitution illusion, a case in which a word in a trivia fact is swapped out for another word, making the sentence a world knowledge violation, but comprehenders do not consciously detect the anomalous nature of the sentence. Here I attempt to develop specific and testable hypotheses about the source of the illusion, paying particular attention to how the same mechanism that "fails" in illusion sentences (in that it does not allow the comprehender to detect the anomaly) serves the comprehender well in other circumstances. I demonstrate that the substitution illusion, like the negative polarity illusion, is more restricted than previously thought - some stimuli yield very high illusion rates while others yield very low illusion rates, and this variability appears to be non-random. In seven experiments, I pursue both a correlational approach and an experimental manipulation of illusion rates, in order to narrow the space of possible explanations for the illusion. These investigations collectively demonstrate that occasional errors in comprehension do not necessarily reflect the use of "shortcuts" in sentence processing, and can be explained by the interaction of the linguistic system with non-linguistic components of the cognitive architecture, such as memory and attention. While neither illusion phenomenon is ultimately fully explained, the research presented here constitutes an important step forward in our understanding of both domains and their broader implications.