Below is an interactive version of the conference program. Presentations highlighted in purple
are part of the Special Session on Linguistic Theory Electrified. The full program can be found here: http://cuny2018.ucdavis.edu/program-pdf.html
are part of the Special Session on Linguistic Theory Electrified. The full program can be found here: http://cuny2018.ucdavis.edu/program-pdf.html
CUNY 2018 Conference Schedule | |||||
Thursday | Friday | Saturday | |||
8:00-9:00 AM | Coffee and Registration | 8:00-9:00 AM | Coffee and Registration | 8:00-9:00 AM | Coffee and Registration |
9:00-9:10 AM | Opening Remarks | ||||
Session 1 | Session 3 | Session 5 | |||
Time | Event | Time | Event | Time | Event |
9:10-9:40 AM | 110 | 9:00-9:30 AM | 82 | 9:00-9:30 AM | 108 |
9:40-10:10 AM | 187 | 9:30-10:00 AM | 107 | 9:30-10:00 AM | 191 |
10:10-10:40 AM | 209 | 10:00-10:30 AM | 287 | 10:00-10:30 AM | 205 |
10:40-11:00 AM | Break | 10:30-10:50 AM | Break | 10:30-10:50 AM | Break |
11:00-11:40 AM | Invited Speaker | 10:50-11:30 AM | Invited Speaker | 10:50-11:30 AM | Invited Speaker |
11:40-12:10 PM | 259 | 11:30-12:00 PM | 368 | 11:30-12:00 PM | 220 |
12:10-2:00 PM | Lunch (not catered) | 12:00-2:00 PM | Poster Session B | 12:00-2:00 PM | Poster Session C |
Session 2 | Session 4 | Session 6 | |||
Time | Event | Time | Event | Time | Event |
2:00-2:30 PM | 125 | 2:00-2:30 PM | 58 | 2:00-2:40 PM | Invited Speaker |
2:30-3:00 PM | 147 | 2:30-3:00 PM | 163 | 2:40-3:10 PM | 177 |
3:00-3:30 PM | 213 | 3:00-3:30 PM | 387 | 3:10-3:30 PM | Coffee |
3:30-3:50 PM | Coffee | 3:30-3:50 PM | Coffee | 3:30-4:00 PM | 26 |
3:50-4:30 PM | Invited Speaker | 3:50-4:30 PM | Invited Speaker | 4:00-4:30 PM | 96 |
4:30-5:00 PM | 138 | 4:30-5:00 PM | 249 | 4:30-4:45 PM | Break |
5:00-5:15 PM | Break | 5:00-5:15 PM | Break | 4:45-5:15 PM | 298 |
5:15-5:45 PM | 307 | 5:15-5:45 PM | 261 | 5:15-5:45 PM | 161 |
5:45-6:15 PM | 380 | 5:45-6:15 PM | 75 | ||
6:00-8:00 PM | Poster Session A | 7:30 PM | Party! |
Thursday March 15, 2018 | |||
Coffee | 8:00 AM - 9:00 AM | Opening Remarks: Fernanda Ferreira | 9:00 AM - 9:10 AM | Session 1 | 9:10 AM - 12:10 PM |
Submission | Title | Authors | |
Chair: | Liz Schotter | ||
110(abstract) | Cross-modal Bilingual Activation in English & American Sign Language Bilinguals: The role of language experience We investigate cross-modal lexical priming in bimodal bilinguals (American Sign Language & English). Specifically, we ask whether phonological priming of English words occurs during an ASL-only lexical decision task when the underlying English translation equivalents are phonologically related. We also examine the role of language experience (i.e., age of ASL exposure). Using mixed-effects models, we find a significant effect of prime type (English-related, Unrelated) and a further qualified prime type by group interaction (p=.007). The interaction reveals that for late signers (both deaf and hearing), English-related priming inhibited responses and, conversely, for early signers it facilitated responses. | Robin Thompson and Clifton Langdon | 9:10 AM |
187(abstract) | Verb learning in English and Japanese: Statistical and non-statistical effects Acquiring argument structure involves distributional evidence, but less is known about effects of event and verb knowledge. To examine this, English and Japanese speakers learned novel verbs based on varying distributional cues. In the English-like familiarization, events with internal agents (die) occurred with intransitives only, and generated a stronger preference in English compared to Japanese speakers. Events with external agents (rock) always occurred with transitives and intransitives, but a spurious preference for transitives emerged in the Japanese-like familiarization. Together, this suggests that distributional evidence is assessed based on existing verb patterns, and semantic differences motive a bias for syntactic distinctions. | Yi Ting Huang, Mary Bounds and Yuichi Suzuki | 9:40 AM |
209(abstract) | Zooming in on zooming out: Dynamic tuning of bilingual language control during reading The relative contributions of bottom-up input and top-down control during bilingual language processing are underspecified. We conducted two eye-tracking experiments to assess the time course and degree of accessibility of each language in different global language contexts. When critical words were presented overtly in Experiment 1, code switches disrupted reading early during lexical processing, but not as much as pseudowords did. In Experiment 2, critical words were presented covertly as parafoveal previews, and code-switched previews were treated like pseudowords. These data provide evidence for the flexible adjustment of the degree of activation of each language during bilingual reading. | Liv Hoversten and Matt Traxler | 10:10 AM |
Break | 10:40 AM | ||
Chair: | Victor Ferreira | ||
Invited Speaker | Electrifying Psycholinguistics: A Historical Perspective on Cognitive Electrophysiology, Expectancy, and Language | Marta Kutas | 11:00 AM |
259(abstract) | Ergativity, transitivity, and subject-object processing asymmetries in Niuean This study investigates the processing of wh questions in the ergative-absolutive Polynesian language Niuean. We use visual-world eye-tracking to examine how subject vs. object processing asymmetries are affected by (i) the case marking of the subject (ergative vs. absolutive) and object (absolutive vs. oblique), and (ii) verb transitivity (obligatory oblique object vs. optional oblique object). Results suggest that Niuean does not exhibit an overall subject or object advantage. Instead, we find that the subject vs. object asymmetry is affected by both case (e.g., marked ergative vs. unmarked absolutive) and transitivity (whether the object is obligatory). | Rebecca Tollan and Daphna Heller | 11:40 AM | Session 2 | 2:00 PM - 5:45 PM |
Submission | Title | Authors | |
Chair: | Albert Kim | ||
125(abstract) | Do SL tasks capture stable individual differences in children? An investigation of task reliability across modalities Are statistical learning measures reliable in children? A growing literature examines individual differences in SL and their relation to language learning outcomes. However, these studies use tasks whose suitability for assessing individual differences has been questioned (Siegelman-et-al-2017ab). We examine the reliability of three SL tasks (two auditory and one visual) in children (mean age 8;2) two months apart. While children showed learning at a group-level, all tasks had low reliability with test-retest and split-half reliability well below psychometric standards. These findings raise significant concerns about the use of current SL tasks to predict and explain individual differences in development. | Inbal Arnon | 2:00 PM |
147(abstract) | The Effects of Aging on Verb Bias Knowledge and Learning Listeners rely on co-occurrence frequencies between verbs and syntax-verb biases-to disambiguate sentences with ambiguous structure (e.g., Feel the frog with the feather). Listeners learn new verb biases from exposure to verb-structure pairings, indicating syntactic representations are malleable. However, given age-related declines in relational memory and syntactic priming in production, we ask: Does aging impair syntactic co-occurrence learning? We find that (1) adults use verb bias to guide interpretation across ages, however (2) biases appear to converge towards Equi-bias with aging. Finally, (3) when exposed to new pairings of verbs and structures, adults of all ages update verb biases. | Rachel Ryskin, Zhenghan Qi, Melissa Duff and Sarah Brown-Schmidt | 2:30 PM |
213(abstract) | Do speakers adapt object descriptions to listeners under load? We investigated whether speakers use more redundant descriptions for listeners under increased cognitive load. We conducted a referential communication experiment in a driving simulator, predicting that speakers would be more redundant when listeners perform a difficult as compared to an easy driving task. The results confirmed this prediction, but only when the speaker had previously taken the role of driver, and only in the first block of the experiment. These findings suggest that speakers only take the listener’s needs into account when there are strong cues that adaptation is necessary, and only adapt to their first assessment of listeners' needs. | Jorrig Vogels, David Howcroft, Elli Tourtouri and Vera Demberg | 3:00 PM |
Coffee | 3:30 PM | ||
Chair: | Fernanda Ferreira | ||
Invited Speaker | Towards a model of linguistic structure building: what do ERPs contribute? | Ellen Lau | 3:50 PM |
138(abstract) | Syntactic adaptation effects do not transfer across tasks Studies on syntactic adaptation have demonstrated that processing biases can be altered by manipulating the probability of competing structures in the input (e.g., Fine et al., 2013), but these demonstrations alter the distribution within a single task. Two eye tracking during reading experiments utilizing the distribution of gap positions provide evidence that syntactic adaptation effects do not transfer from one task to another. While the active gap filling bias is suppressed when exposure to unexpected gap positions (i.e., prepositional object gaps) is presented within-task, this bias remains when the exposure phase is masked as a separate sentence recognition study. | Emily Atkinson, Ian Rigby, Naomi Shapiro, Brent Woo and Akira Omaki | 4:30 PM |
Break | 5:00 PM | ||
307(abstract) | Individual differences in language knowledge and the syntactic processing P600 ERP In this work we explored systematic individual differences in cognitive ability correlated with classic ERP responses to semantic and syntactic anomalies. We constructed three latent variables (domain general and verbal working memory, and language knowledge) drawn from a large battery of behavioral tasks and simultaneously regress these variables on latent factor ERP representations. Overall, we find that language knowledge predicts the syntactic anomaly P600 response and no reliable correlation between N400 and cognitive ability factors. | Shannon McKnight, Akira Miyake, Donald Bell-Souder and Albert Kim | 5:15 PM | Poster Session A | 6:00 PM - 8:00 PM |
Submission | Title | Authors | |
2(abstract) | Is the LAN effect in morphosyntactic processing an ERP artifact? The present project examined whether the LAN effect is an artifact due to the overlap between two different ERP effects (N400 and P600). We investigated local agreement in a Romance language (Spanish) in a large sample of subjects (n=80) and items (n=120). | Sendy Caffarra, Martha Mendoza and Doug Davidson | Lobby |
3(abstract) | Synchronization of Neural Oscillations with Speech Facilitates Syntactic Information Processing Synchronization of neural oscillations with syntactic phrases has been described previously, but its purpose for information transmission is unclear. We hypothesized that synchronization implies the alignment of electrophysiological excitability and syntactic information, facilitating sentence processing. Our study distributed morpho-syntactic violations across phrases; violations occurred at points differing in syntactic surprisal. Violation detection accelerated with decreasing surprisal—in correlation with oscillatory phase, which had synchronized to the stimuli. Surprisal and reaction times were more strongly correlated with phase than with each other, suggesting that phase was intervening. Synchronization of neural oscillations with phrases may thus serve to optimize information transmission. | Lars Meyer and Matthias Gumbert | Lobby |
9(abstract) | Representing inverse semantic scope in L2-Spanish In this study, we have investigated the processing of inverse semantic scopes by Chinese-speaking L2-Spanish learners with a picture-sentence verification task. The preliminary results suggest that (i) inverse scope readings in existential-universal configurations are difficult, albeit possible, even for native speakers, and (ii) L2-proficiency/experience may play a role. | Jun Lyu and Lijun Zhang | Lobby |
12(abstract) | Passive priming requires function word overlap The clearest piece of evidence to date for purely syntactic priming comes from Bock and Loebell’s (1990) seminal finding that intransitive locatives with prepositional phrases primed passives as much as passive primes did. However, their intransitive and passive sentences all included the preposition by, raising the possibility that by was responsible for, or at least essential to, priming across the different constructions. Here we demonstrate that abstract syntactic phrase structure is neither sufficient for passives to be primed nor is it necessary. We conclude that passive priming is largely or completely attributable to the by-phrase adjunct. | Jayden Ziegler, Adele Goldberg and Jesse Snedeker | Lobby |
15(abstract) | When is the verb a potential gap site? How filler maintenance affects active gap-filling Throughout an open filler-gap dependency, some features of the filler are being actively maintained in working-memory, while others decay. The current study asks whether the maintained features guide active gap-filling, determining which verbs can act as potential gap sites. In Experiment 1, we show that the filler’s animacy is maintained. In Experiment 2 we test under which circumstances an attempt to resolve the dependency is made despite implausibility, showing that violation of the verb's animacy requirement prevents an attempt to resolve the dependency (unlike violation of other selectional restrictions), suggesting that active gap-filling is indeed guided by the maintained features. | Tal Ness and Aya Meltzer-Asscher | Lobby |
17(abstract) | Memory Retrieval During the Processing of Adjuncts and Coordination Grammatical illusions, where ungrammatical sentences are perceived as grammatical, are key evidence of interference (Jäger et al., 2017). Recently, illusions of plausibility have also been reported (Cunnings & Sturt, under review). We conducted two experiments further investigating illusions of plausibility in temporal adjuncts and coordination (The detective/criminal stood by the cop/car {after arresting/and then arrested} the robber). Although the results were not as robust as our previous study, the data suggested interference in both constructions. Interference may have been attenuated here because the distractor (cop/car) did not match the [+SUBJECT] retrieval cue of the critical verb (VanDyke & McElree, 2011). | Ian Cunnings and Patrick Sturt | Lobby |
20(abstract) | Noisy-channel agreement (mis-)comprehension Errors in subject-verb agreement are often misperceived: erroneous utterances like ‘’The key to the cabinets *were’’ elicit the illusory acceptability of ungrammatical verbs (‘attraction’; e.g. Wagers et al, 2009) and an increase in non-literal responses to comprehension questions asking about the number of the head (accepting ‘’keys’’ as the head; Patson & Husband, 2016). In a visual-world eye-tracking experiment, we demonstrated that both phenomena have the same source: individuals fixate a plural version of the head noun in response to hearing plural cues from other nouns in the sentence as well as in response to an ungrammatical plural verb. | Laurel Brehm, Carrie Jackson and Karen Miller | Lobby |
22(abstract) | Indexing the N400 and the P600 in a Computational Model of Sentence Comprehension The N400 and the P600 component are the most salient ERP components for sentence processing, but their functional interpretation is subject to ongoing debate. Further advances require greater emphasis on explicit computational models, and an important first step in this direction is a recent neurocomputational model instantiating the Retrieval-Integration account of the N400 and the P600, which has been shown to account for semantically-induced ERP-effects. Here, we show that this model can be scaled up to (a) achieve broader linguistic coverage, (b) account for a broader range of signature N400 and P600 phenomena, and (c) be reconciled with surprisal-based models. | Harm Brouwer, Noortje Venhuizen and Matthew Crocker | Lobby |
23(abstract) | The overt pronoun in European Portuguese prefers animate antecedents We ran two experiments, a self-paced reading and an off-line questionnaire, to verify whether the null and the overt subject pronouns in European Portuguese are sensitive to the animacy features of antecedents. The results show the overt pronoun is preferentially interpreted as referring back to antecedents that are both in non-subject position and animate. The findings do not fit completely neither the Position of Antecedent Hypothesis (Carminati, 2002), because overt pronoun interpretation does not solely depend on syntactic features, nor the Accessibility Theory (Ariel, 1996), because there is not a straightforward relation between explicitness of anaphoric forms and semantic prominence. | Sara Morgado, Maria Lobo and Paula Luegi | Lobby |
28(abstract) | Impact of interlocutor identity on grammatical processing: the case of Basque allocutive Pragmatic information concerning the social context of an utterance is an essential part of the linguistic message. Contrasting evidence is found concerning the time course of pragmatic information analysis during online sentence processing. In an ERP study we capitalized on Basque allocutive agreement to investigate the impact of interlocutor’s gender in the processing of morphosyntactic information contained in verbs. The allocutive agreement violation effects that emerged in the P600 time window speak against the idea of a quick integration of any relevant social context information. Rather, information concerning interlocutor’s identity may be taken into account during late reanalysis processes. | Simona Mancini, Max Wolpert, Dana Scarinci and Sendy Caffarra | Lobby |
29(abstract) | Singular they: online and offline interpretation effects among L1 and L2 speakers Adopting a cue-based retrieval account of language processing, we investigate how L1 and L2 English speakers use competing number cues when interpreting English singular they. In a self-paced reading task, participants read sentences containing referential (e.g., that jogger) or nonreferential (e.g., a jogger) subjects, which were referred to using singular (he/she) or plural (they) pronouns. When interpreting the number of the subject noun phrase, L1 speakers exhibited an interaction between pronoun type and referential status of the antecedent. L2 speakers showed no such interaction, suggesting that they may be less able to integrate multiple, potentially conflicting cues than L1 speakers. | Neil Shook, Laurel Brehm, Carrie Jackson and Holger Hopp | Lobby |
34(abstract) | Older adults show comprehension costs in dual tasking for high (but not very high) cloze targets Prior research has shown that limitations in working memory contribute significantly to age-related declines in predictive language processing. Here, we investigated the relevance of working memory capacity by means of a dual task. Groups of younger and older adults judged the meaningfulness of highly predictable and less predictable sentence continuations, while also performing a secondary tracking task. Older adults demonstrated effortful, yet intact, comprehension of highly expected sentence continuations, but a comprehension cost for less expected sentence continuations. Thus, increased demands on working memory limit the extent to which older adults generate expectations during language processing. | Katja Häuser, Vera Demberg and Jutta Kray | Lobby |
35(abstract) | Multimodal Predictability in the N400 and the Index of Cognitive Activity A word's predictability or surprisal, as determined by Cloze probabilities or language models, has been related to processing effort, often indexed by the ERP component N400. A word’s predictability can also depend on the non-linguistic context it appears in. In the VWP, anticipatory eye movements show that comprehenders use the visual context to predict upcoming words. So far, however, it is unclear how visual context affects predictability and processing effort. We present evidence that visually-determined expectations for a target word predict graded processing effort for that word, in a pupilometric (ICA) and in an ERP (N400) measure. | Christine Ankener and Maria Staudte | Lobby |
39(abstract) | Processing bound singular 'they' While a number of studies have investigated the processing of "they" with singular referential antecedents (as in “I stopped someone and asked them for the time”, Foertsch & Gerns-bacher 1997, Sanford & Filik 2007, Doherty 2013, Doherty & Conklin 2017), the experimental record is mixed as to whether this poses a processing cost. We report two studies on bound variable "they" with a quantified singular antecedent (“Everyone said that they were the tallest”). Our results show that singular "they" is readily interpreted as a bound variable and, most importantly, it is sensitive to structural constraints (c-command, Reinhart 1976). | Chung-hye Han, Keir Moulton, Trevor Block, Queenie Chan, Holly Gendron, Kyeong-min Kim, Sara Williamson and Kazuki Yabe | Lobby |
40(abstract) | L2 Spanish Verbal Morphology Mastery Sequence Departing from Skill Acquisition Theory and Connectionism, students of 3 different levels were tested on present and preterit tense conjugation for regular, irregular, and cognate verbs. Response accuracy was measured and the second year level students underwent an experiment. The training method was Focus on Form instruction. Results show that present is acquired before preterit, and regular and cognate forms before irregular forms. Training did not yield significant improvement. Results point to a possible mastery sequence accounted by input nature (frequency, cue cost), although it is not clear whether the sequence can be altered anyhow. | Laura Fernandez Arroyo | Lobby |
43(abstract) | Korean L2 learners’ sensitivity to prosodic boundaries in syntactic ambiguity resolution Prosodic boundaries influence syntactic parsing and resolve syntactic ambiguity in spoken sentences. This study examined whether L2 learners can recognize and use prosodic boundaries in syntactic disambiguation, and moreover, whether L2 proficiency affects the aspect of the use of prosody. 60 Korean learners of English participated in an end-of-sentence comprehension task which consisted of temporarily ambiguous early/late closure sentences and were tested on their L2 proficiency. The results showed that L2 learners in this study were sensitive, but unable to utilize prosodic information in syntactic processing. In terms of L2 proficiency, learners in high proficiency group generally comprehended sentences faster. | Seung-Eun Kim | Lobby |
47(abstract) | The P600 for singular 'they': How the brain reacts when John decides to treat themselves to sushi We ran an ERP study looking at the processing effort for singular ‘they’ for four kinds of singular antecedents: specific unambiguous gender (John vs John and Mary), specific ambiguous gender (the participant vs the participants), generic unambiguous gender (some man vs some men) and generic ambiguous gender (someone vs some people). We found a P600 for specific antecedents with unambiguous gener (John), but failed to find a P600 for antecedents with ambiguous gender (the participant/someone) suggesting that there is no processing effort associated with using singular ‘they’ for antecedents that are construed as being unmarked for gender. | Grusha Prasad, Joanna Morris and Mark Feinstein | Lobby |
48(abstract) | The mess reveals the system: People use top-down cues to resolve errors in contexts with highly random noise, but not with highly structured noise Listeners seem to use their grammatical knowledge to correct for noise in speech, but what roles does contextual and statistical information play in determining if and how to correct? To investigate this, we first replicate Mack et al. (2012), a study which investigated usage preferences for null-subject constructions using a speech restoration paradigm. By manipulating the placement of the noise in the fillers, we construct two additional experiments with either random or highly-structured noise distribution. While still ongoing, preliminary results from the study show listeners seem to place less importance on grammaticality-based correction strategies when faced with highly structured noise. | Suhas Arehalli and Eva Wittenberg | Lobby |
49(abstract) | The role of L1 and event type in processing aspectual mismatches in L2 English: an ERP study This study explores what ERPs can tell us about how English native speakers and Russian/German learners of English process matching/mismatching aspectual cues during sentence comprehension online and offline. Offline data were similar across groups but online processing differed. Only English native speakers and Russian learners differed online between match-mismatch conditions. German learners did not, despite patterning indistinguishably from the other groups offline. These results suggests that crosslinguistic differences arise in how rapidly learners activate knowledge about aspectual information, and they bring novel support for the view that online processing in the L2 varies as a function of crosslinguistic structural similarity. | Norbert Vanek and Leah Roberts | Lobby |
50(abstract) | Does the result justify the means? Verbal and non-verbal memory of resultative events in Mandarin, Dutch and Spanish speakers. Spanish differs from Mandarin and Dutch in the way results of events are verbally encoded. While Mandarin and Dutch use resultative constructions (verb + resultative marker) to encode both manner and result of actions, Spanish uses only main verbs to encode either manner or result. We compared how native speakers memorized event results when they explicitly described videos showing resultant changes of state (verbal experiment) and when not (non-verbal experiment). While Mandarin and Dutch speakers showed a good performance in the verbal experiment only (‘thinking for speaking’ effects), Spanish participants showed a good performance in both experiments (linguistic relativity effects). | Miguel Santín, Ciara Hobbelink, Angeliek van Hout and Monique Flecken | Lobby |
59(abstract) | Contrasting lexical pre-activation, prediction-based and post-lexical integration accounts of the N400 ERP effect in priming studies Participants were exposed to several lists of prime-target word pairs, containing 80% related word pairs reflecting a specific semantic relationship, 10% reflecting other semantic relationships and 10% of unrelated pairs. They had to judge if words within each pair were related or not. Two types of reductions of the N400 ERP effect were found: relatedness priming started after 300ms, while relational priming (majority vs. minority semantic relationship) was not observed until 400ms. The latter effect is best explained by post-lexical integration processes. A second study using a longer SOA attempts to distinguish it from a possible delayed prediction-based priming. | Alexandre Herbay, Phaedra Royle, John Drury, Lauren Fromont and Karsten Steinhauer | Lobby |
60(abstract) | Can second language learners generate predictions at the level of the discourse? An event-related potentials study We used ERP to investigate L2 prediction at the level of the discourse. Our results show that both L1-English speakers and L1-Spanish L2-English learners can use an it-cleft construction to anticipate the Noun Phrase that can be assigned Focus in a discourse context, although they do so differently, as signaled by their qualitatively different brain responses to unexpected relative to expected articles (L1 speakers: N400; L2 speakers: Anterior Positivity). In addition, prediction in both L1 and L2 speakers was impacted by processing speed but not by working memory. | José Alemán Bañón, Clara Martin and Elena Fano | Lobby |
66(abstract) | Bilinguals’ referential choice in common and privileged ground. The present research tests the ability of the Interface Hypothesis to explain bilingual referential choice by examining the production of pronouns and noun phrases in unbalanced Spanish-English bilinguals in common and privileged ground. The results reveal that both monolinguals and bilinguals are sensitive to the demands of privileged ground, and tend to be more explicit in their choice of referring expressions when the information is not shared with the addressee. However, the difference in referential use between bilinguals and monolinguals was similar in common and privileged ground, inconsistent with the Interface Hypothesis. | Carla Contemori and Iva Ivanova | Lobby |
68(abstract) | Incremental processing relies on underspecified morphosyntactic information German exhibits many instances of syncretism that allow for grammatical function ambiguities. One way to account for syncretisms is by means of underspecification. To investigate whether these theoretical assumptions have consequences for incremental sentence processing, we conducted an ERP study that examined the buildup of grammatical function ambiguities. The results at the determiner and the noun suggest that the processing of a noun phrase depends on underspecified morphosyntactic features. We argue that the graded ERP effects are—for feature integration (reflected in a negativity) and re-evaluation of the syntactic context (reflected in a positivity)—only compatible with an underspecification account. | Florian Bogner and Petra Schumacher | Lobby |
69(abstract) | ERP evidence revisits lexical differentiation in production and comprehension Speakers elaborate their referring expressions with modifiers if a different exemplar from the same category has previously been mentioned in the discourse, a phenomenon called lexical differentiation. We examine lexical differentiation in both production and comprehension by testing (in)appropriately differentiated expressions while recording participants’ electrical brain activity. Speakers’ brain responses revealed that planning appropriate differentiation requires attention and recollection from memory – an increase in the later LPC. In two comprehension studies, enhanced LPCs were seen for over-differentiation. Thus, both speakers and listeners are sensitive to discourse history and clearly track the use of referring expressions across a discourse. | Si On Yoon and Kara Federmeier | Lobby |
70(abstract) | Production predictions: Word duration under two theories A variety of factors influence speakers' decisions to lengthen and shorten utterances. Under one theory, the Facilitation Reduction Hypothesis, ease in production at any level of the production system (i.e., message, lexical, and phonological representations), results in reduction (Kahn & Arnold, 2012; 2015). Another explanation, the Two Domain Theory (TDT), suggests that speakers slow down depending on where they are in the production process. In this study, we examined how production difficulty is related to word duration to compare these accounts. Results suggest that production takes place over at least two domains (lexical and phonological), providing evidence for the TDT. | Nicholas Tippenhauer, Cassandra L. Jacobs and Duane Watson | Lobby |
71(abstract) | Verb learning mediated by sentence processing––understanding both the learning process and the learning outcome Children use a verb’s linguistic context to learn its meaning, but linguistic context can be hard to process. We examine preschoolers’ processing of an upstream subject and their learning of a downstream verb. The subject is either light (e.g., “the boy is pilking”) or heavy (e.g., “the tall boy is pilking”); when the subject is heavy, it is either necessary (e.g., given a tall boy and a short boy) or unnecessary (e.g., given a boy and a girl). A combination of gaze measure during learning and pointing measure after learning revealed interesting relations between language processing and word learning. | Angela Xiaoxue He and Sudha Arunachalam | Lobby |
72(abstract) | Does planning explain predictability effects on word duration? Production is facilitated when talking about something predictable, allowing for earlier linguistic formulation and more fluent delivery. In sentences like "The Duchess handed a painting to the Duke", speakers tend to refer to goal referents because they're more predictable. In this study, we examine the duration of definite descriptions (the Duke) and test whether these references are acoustically reduced because they are pre-planned, as measured by eyetracking. We found that pre-planning doesn't have a direct effect on duration, but is mediated by latency to begin speaking. This supports the fluency account of reduction; overall durations are faster when subjects pre-plan. | Valerie Langlois, Sandra Zerkle and Jennifer Arnold | Lobby |
73(abstract) | Minding the Gap?: (Some) Conditions on Resumption in English I present three experiments that investigate the conditions under which a pronoun will be treated as a resumptive. I argue that resumption is more acceptable when (1) the pronoun and the filler are coreferential (Erteschik-Shir 1992), and (2) when working memory resources are limited. I claim that resumption is licensed when resources are overloaded, and comprehenders cannot maintain a representation of a predicted gap. | Dustin Chacón | Lobby |
74(abstract) | Processing non-argumental dependencies: tense and aspect in Spanish In an EEG violation study, we explored the processing of two time-related non-argumental grammatical features: tense and aspect. The study was carried out in Spanish, a language that expresses these features in the same verbal morpheme. Despite their similarities, the tense violation elicited a biphasic negative-positive pattern, whereas the aspectual violation only showed a late positive deflection. We conclude that, even when morphosyntactic information is housed in the same morpheme, the parser seems to use different strategies to integrate different types of grammatical features. | Dana Scarinci, Manuel Carreiras and Simona Mancini | Lobby |
77(abstract) | Adapting to the unreliable - speech adaptation in real-time interpretation of intonation Intonation plays an integral role in comprehending spoken language. It is also remarkably variable, Nevertheless, listeners rapidly integrate intonational information to predictively map a given pitch accent onto respective speaker intentions. We use mouse-tracking to investigate two questions: (i) whether listeners draw predictive inferences from the presence and absence of an intonational marking and (ii) how listeners adapt their online interpretation of intonational cues when these are stochastically unreliable. Our results suggest that comprehenders rapidly and rationally integrate available intonational information, that they expect reliable intonational information initially, and that they adapt these initial expectations during exposition to unreliable input. | Timo Roettger and Michael Franke | Lobby |
79(abstract) | Prosody and information structure: what is universal and what is not? We study prosodic phenomena related to Information Structure in a language that uses specialized morphemes to convey IS — in Adyghe (Northwest Caucasian family). One production and two comprehension experiments were conducted. When speakers read texts aloud, focused words were more prosodically prominent. But when other speakers listened to these sentences cut out of context, they could not reconstruct their IS. Thus, the tendency to make focused information prosodically prominent may be universal (if the phonology of the language allows for this), but not all languages grammaticalized this tendency and use it to encode IS. | Natalia Slioussar and Natalia Tyshkevich | Lobby |
81(abstract) | Distance matters during adverb-verb tense processing: evidence from ERPs This ERP study (N=23, Italian native speakers) aims at investigating the effect of linear distance during the processing of the temporal concord between a deictic temporal adverb and the verb (e.g. Yesterday I went/*will go to the concert). Data show that distance affects the processing of adverb-verb tense violations during later stages of the verb processing. Distal tense mismatches trigger a larger lateP600 (in frontal sites) compared to local tense mismatches. We propose that in the distal condition the re-interpretation of the mismatch is more demanding because a richer set of (discourse-related) information needs to be examined, thanks to distance. | Nicoletta Biondo, Emma Bergamini and Francesco Vespignani | Lobby |
84(abstract) | The role of categorical and probabilistic thematic information in sentence processing In a self-paced reading experiment, we asked what guides processing: probabilistic thematic information or categorical thematic information coupled with the need to satisfy the Theta Criterion. Optionally transitive verbs varying in their transitivity bias were incorporated into Garden Path sentences of two structures, one in which rapid satisfaction of the Theta Criterion would inevitably lead to processing difficulty, and one in which it would not. Participants were not sensitive to the verb’s transitivity bias in the first type of structure, suggesting that the Theta Criterion, a global grammatical constraint, takes precedence over probabilistic information during processing. | Lola Karsenti and Aya Meltzer-Asscher | Lobby |
88(abstract) | Withholding speech: does the EEG signal reflect planning for production or attention? In conversation, turns between speakers happen in rapid succession. It has been suggested that planning for production occurs simultaneously with listening, and a fully planned response is withheld till the turn end. EEG could be a useful measure to test whether this is true, but first it must be clear how these separate processes are reflected in the neural signal. Here we tested if planning for production and withholding a response is visible in the EEG signal or whether neural changes reflect attention shifts by using a simple withhold paradigm both in the visual and auditory modality. | Suzanne Jongman, Vitoria Piai and Antje Meyer | Lobby |
89(abstract) | Good Enough Processing of Filler-gap Dependencies in Native and Non-Native Speakers Language processing is sometimes ‘good enough’ (Ferreira et al., 2001), but how this influences filler-gap dependencies is unknown. We investigated temporary ambiguity in filler-gap dependencies (John saw the car which Ben stopped the bicycle near vs. John saw the car near which Ben stopped the bicycle). In Experiment 1 native/non-native participants answered comprehension questions (Did Ben stop the bicycle?). In Experiment 2 they read texts testing how the temporary ambiguity was resolved. Comprehension accuracy was lower in ambiguous sentences, and reading times indicated initial misinterpretations were not fully erased from memory, suggesting good-enough parsing is a general property of comprehension. | Hiroki Fujita and Ian Cunnings | Lobby |
91(abstract) | Focus particles strongly draw attachment Recent work shows that pitch accents can affect syntactic attachment ambiguities (Schafer et al., 1996; Lee & Watson, 2011; Carlson & Tyler, 2017). In the sentence "Tom reported that Bill was bribed # last May," contrastive accents on the first or second verb increased interpretations with the final phrase modifying the accented verb. On the hypothesis that this is because the accented word indicates focus position, two experiments explored the effect of focus particles ("only") and accents on attachment. The position of "only" had very strong effects on attachment, suggesting it is a stronger indicator of focus status than accents. | Katy Carlson | Lobby |
92(abstract) | Predicting discourse status: N400 effects of determiner expectation Comprehenders deploy top-down mechanisms to make use of existing information to predict upcoming input. However, it is debated at which level(s) of linguistic representation predictions are generated. We report data from an ERP study asking whether comprehenders predict not just upcoming words or concepts, but the discourse status of upcoming referents. We find that N400 amplitude tracks the predictability of definiteness on the determiner, even before the subsequent noun has been presented. This suggests comprehenders generate predictions about discourse status, and indicates that the predictive facilitation indexed by the N400 extends to processes evoked by functional elements like determiners. | Zoe Schlueter, Anna Namyst and Ellen Lau | Lobby |
94(abstract) | Modeling psycholinguistic effect timecourses with deconvolutional time series regression This work proposes a new approach to modeling psycholinguistic time series – deconvolutional time series regression (DTSR) – and applies it to existing datasets in order to show that temporal diffusion of effects may play a major role in subject behavior. DTSR learns best-fit impulse response functions that mediate the temporal relationship between independent and dependent variables, providing fine-grained estimates of the timecourse of an independent variable’s influence on the response. Our results show that, on datasets with sufficiently long time series, DTSR models make better predictions and learn substantially different effect estimates than either LME or GAM. | Cory Shain and William Schuler | Lobby |
97(abstract) | Processing right-extraposed relative clause in Hindi: Evidence for locality effect In head-final languages, the evidence for locality effect is weak while anti-locality effect has been observed robustly. In these languages, locality effect has been attested only in high memory-load configurations (e.g., Vasishth & Drenhaus, 2011; Levy & Keller, 2013). We describe three self-paced reading experiments to investigate whether locality effect can be seen in Hindi. We have found that locality effect can be observed in certain constructions in a head-final language without using high memory-load configurations. With the increased expectation, locality effect gets attenuated but it was not completely canceled (cf. Husain et al., 2014; Levy et al., 2012). | Samar Husain and Himanshu Yadav | Conference Room A |
99(abstract) | Natural reading facilitates forming fine-grained orthographic expectations: ERP Evidence We used ERPs to test how control over stimulus presentation affects sensory perceptions. Participants read sentences containing: 1) an expected target; 2) the target with two characters transposed; 3) a pseudoword neighbor; or 4) a string of consonants. In Experiment 1 (rapid-serial visual presentation), results showed early neural sensitivity to egregious mismatches only. In Experiment 2 (self-paced reading), results showed graded N170 effects to all targets with anomalous visual information. P600s were elicited by all conditions containing unexpected orthography in both experiments. Results support a nuanced view of visual processing, where processing is more fine-grained with more control during reading. | Nyssa Bulkes, Kiel Christianson and Darren Tanner | Conference Room A |
103(abstract) | ‘Good enough’ processing of English subject-verb agreement violations differs when listening to native vs. foreign-accented speech This study used ERPs to investigate if and how speaker identity (native vs. foreign (Mandarin) accent) affects the processing of two types of English subject-verb agreement violations: errors of commission and omission. ERP results show an effect of speaker identity, but only for errors of commission, with an N400 effect for the native accent, but no effect for the foreign accent. The absence of ERP effects typically associated with morphosyntactic violations is discussed in light of ‘good enough’ processing. The findings highlight the broad importance of speaker identity and listener expectations regarding different types of ungrammaticality when processing speech. | Carmen Kung, Rebecca Holt, Elaine Schmidt and Katherine Demuth | Conference Room A |
104(abstract) | Towards a psychological evaluation metric for semantic representations When people read, they update their mental model of the world. What exactly is the semantic/conceptual representation that is being updated? Given that theories of semantics/concepts are motivated by different empirical phenomena, how do we even compare semantic representations? We propose an information-theoretic framework for evaluating semantic representations independent of their origin. Fundamentally, representation schemes are encodings of semantic information. As a result, channel limited systems (humans) must have a utility function specifying which information to privilege when encoding and thus, which information is most likely lost/forgotten. We evaluate the sensitivity of our metric using a story re-telling paradigm. | Francis Mollica and Edward Gibson | Conference Room A |
105(abstract) | Parafoveal influences on oculomotor control and word recognition: Insights from oral reading Linguistic processing directly influences eye movements during reading, potentially independently of word recognition. We compared silent to oral reading, for which we measured eye movements and utterances concurrently, in a boundary study that dissociated high- and low-frequency preview and target words. Oral reading reduced the degree to which parafoveal information was used to trigger saccades and to complete word recognition. However, readers occasionally (at least partially) uttered the preview rather than the target and this was more likely when it was high frequency, suggesting parafoveal processing was more efficient. More errors were corrected mid-utterance when the preview was low frequency. | Elizabeth Schotter and Alyssa Thompson | Conference Room A |
109(abstract) | Early, but not very early, effects of implicit causality and consequentiality in the visual world paradigm Three visual world experiments investigate the use of implicit causality and consequentiality information in spoken language. The first two experiments replicate findings that implicit causality information is available early (just after "because" in "John blamed Bill because he was careless". In the third experiment we replaced "because" with "and so" and followed it with an explicit consequence. Consequentiality biases are usually the opposite of causality biases, and we found that participants looked early (just after "and so") to the implicit consequence, not the implicit cause. The studies illuminate the time course of the use of these two types of information. | Alan Garnham, Sam Hutton and Bojana Ivic | Conference Room A |
111(abstract) | Comparing retrieval of antecedents for referential and ‘donkey’ pronouns (in Norwegian) We investigated whether the processing of coreferential and so-called 'donkey' pronominal dependencies employs different mechanisms for antecedent retrieval. In two SPR experiments we varied the gender match between a critical pronoun and an RC-internal indefinite NP. We also varied whether the head of the RC was either referential or quantificational, which determined whether the pronoun could establish a coreferential or ‘donkey’ relation with the indefinite NP. In both experiments we found immediate gender-mismatch effects at the pronoun, irrespective of dependency type, suggesting that antecedents for donkey and referential pronouns are retrieved in the same manner and at the same time. | Dave Kush and Ragnhild Eik | Conference Room A |
112(abstract) | Listeners rapidly use unexpected information to update their predictions: Evidence from eye-movements Comprehenders can incorporate rich contextual information to predict upcoming input on the fly and they can detect cues that conflict with their predictions very quickly. However, to date little is known about whether and how listeners use unexpected information to revise their existing predictions. Here we took advantage of the rich classifier system in Mandarin Chinese to examine whether and how comprehenders update their noun prediction upon encountering an unexpected classifier. We present evidence from a visual world eye-tracking experiment which suggests that listeners can quickly use prediction-mismatching classifiers to revise their predictions. | Wing-Yee Chow and Di Chen | Conference Room A |
115(abstract) | ERP priming effects in the processing of coreference: a Bayesian analysis This work addressed the time course of antecedent reactivation during coreference. Using five ERP experiments and Bayesian analyses we asked: how quickly can semantic antecedent information be used during reading comprehension? | Sol Lago, Lena Jäger, Anna Namyst and Ellen Lau | Conference Room A |
116(abstract) | Disambiguation: Source or consequence of word learning? To break down the complexity of word learning children might use certain language/domain-specific inferential learning constraints, such as ‘disambiguation’. This describes the assumption that new words tend to refer to new referents precluding the application of a novel label to a familiar object. Prior studies showed that multilingual children make less use of disambiguation. The research questions we addressed were (a) whether children disambiguate and retain a trained word-object mapping, and (b) how age and language background modulate disambiguation and retention. Eye-tracking data from 18, 24, and 30-month-old monolingual and multilingual infants were collected. | Kate Repnik, Antonella Sorace and Vicky Chondrogianni | Conference Room A |
117(abstract) | She knows/ She thinks/ She doesn’t know that X: Presuppositonal Effects trigger Context Sensitivity of Language-Induced Motor Activity: A Grip-Force Study The recent neuroscientific literature suggests that language-induced motor activation is task- and context-dependent. We investigate the effect of information-layering on grip force activation by comparing assertive information with information embedded under a presuppositional factive verb construction (Study 1) and a non-factive verb construction (Study 2). Furthermore, we examine whether the projection behavior of a factive verb construction modulates grip force activation under negation (Study 3). The data show that the crosstalk between language and motor systems is sensitive to presuppositional contexts: The presupposed factive complement triggers an increase in grip force, even under negation, whereas the non-factive complement does not. | Rob Reinecke, Tatjana Nazir and Jacques Jayez | Conference Room A |
120(abstract) | Event-referent activation in the visual world: Persistent activation is guided by both lexical and event representations Gradient-competition models predict that previously-activated information may persist despite incompatible bottom-up information. A visual-world-study examined the persistence of activation of referents favored by verb-argument structure versus event-related typicality. Prior to encountering bottom-up role-filler information, participants fixated both typical and atypical role-fillers (favored only by verb-argument structure) more than unrelated distractors. After encountering role-filler information, persistent activation was found for both typical and atypical role-fillers, even when the opposite entity was explicitly mentioned. Argument-structure modulated these effects, such that argument role-fillers(recipients/instruments) elicited more fixations than adjunct role-fillers(locations). These results suggest that verb-argument representations can have independent effects, dissociated from event-related typicality. | Haley Dresang, Michael Walsh Dickey and Tessa Warren | Conference Room A |
122(abstract) | The interaction of conceptual number and morphophonology in subject-verb agreement: A role for working memory We investigated the relative impact of working memory, notional number, and morphophonological ambiguity on agreement with complex noun phrases (NPs) among 54 Dutch native speakers (items based on Antón-Méndez & Hartsuiker, 2010). Mixed-logit models revealed that participants made more agreement errors with NPs that were notionally plural and contained homophonous determiners. Participants with higher OSpan scores made fewer agreement errors when the subject NP contained homophonous determiners, and this effect was greater when the subject NP was notionally singular. These findings suggest that the ability to establish correct subject-verb agreement—especially with morphophonological ambiguity—is modulated by working memory capacity. | Heidi Lorimor, Carrie Jackson and Janet van Hell | Conference Room A |
130(abstract) | Center-embedded sentences: An on-line problem or deeper? Doubly center-embedded sentences are a long-standing problem for psycholinguistics. In the past, this has been attributed to a variety of causes, such as syntactic structure density. We present evidence that part of the problem is the mismatch between syntactic and prosodic structure. We employ a new methodology: participants read target sentences both without preview (with a timing click to confirm compliance) and after taking whatever time they need to think about the sentence structure. Our results indicate that a significant number of naïve English speakers do have the linguistic ability to parse center-embedded syntax if the prosodic phrasing cooperates. | Janet Dean Fodor, Benjamin Macaulay, Danielle Ronkos, Tally Callahan and Tyler Peckenpaugh | Conference Room A |
131(abstract) | Communicative efficiency and implicature in multi-party conversation In the current study, we present a new interactive paradigm to test the semantic and pragmatic meaning of utterances. We provide the first experimental demonstration that embedded implicatures are communicated reliably and efficiently. Our paradigm makes implicatures relevant through an explicit decision problem, avoiding meta-linguistic judgments. We have connected the paradigm to a game-theoretic model (Benz, 2012) which allows us to reason backwards from a given action-choice to the underlying interpretation as well as participants’ production strategies. Overall, our interactive paradigm opens up the possibility to do controlled experiments on a variety of contextual and speaker-related factors in multi-party conversation. | Nicole Gotzner, Lisa Raithel and Anton Benz | Conference Room A |
132(abstract) | The incremental effect of conceptual specificity in subjects, verbs and objects: MEG evidence Research has shown that activity in the left anterior temporal lobe (LATL) increases during semantic composition, and that this effect is modulated by the conceptual specificity of the input items. As this effect has only been demonstrated for noun phrases, we tested if the same effect is found in the verb phrases, by varying specificity in the subject, verb and object positions of sentences. Results showed a main effect of verb specificity in the LATL and a marginally significant effect of subject specificity such that specific subjects elicited more activity than general subjects during the verb presentation in the LATL. | Songhee Kim and Liina Pylkkanen | Conference Room A |
135(abstract) | The role of visual salience on the production of plural nouns The current study used a sentence completion task to investigate the role of visual salience in plural noun production. Participants viewed photographs with either a single person, multiple people (plural condition), or multiple people with one visually salient individual (VIP condition) depicted. The photo was presented with a photo caption that was missing a word to describe the person/people in the photograph. Participants were asked to provide the missing word. They were less likely to use a plural noun in the VIP condition compared to the plural condition. The results suggest that visual salience can determine referential equivalence. | Nikole Patson and Alasdair Clarke | Conference Room A |
140(abstract) | Syntactic category prediction during reading: evidence from adjectival modification in Standard Arabic Using MEG, we asked whether syntactic processing during reading comprehension shows evidence of predictive processing. We employed adjectival and complementizer constructions in Standard Arabic that allowed us to vary syntactic surprisal and entropy. We found main effects correlating with entropy in a cluster in the right inferior and middle temporal gyri, and with surprisal in a right temporal cluster (comprising the inferior, middle, and superior temporal gyri). These results are consistent with previous fMRI literature showing sensitivity to predictive processing in these areas. | Suhail Matar and Alec Marantz | Conference Room A |
142(abstract) | tDCS to premotor cortex changes action verb understanding: Complementary effects of inhibitory and excitatory stimulation When people process action language, premotor cortex (PMC) circuits for planning actions show increased activity. However, experimental disruption of PMC activity paradoxically made people respond faster to action verbs. Here, we use transcranial Direct Current Stimulation (tDCS) to either inhibit or excite left PMC hand area. Inhibitory stimulation increased how accurately people responded to unimanual action verbs, whereas excitatory stimulation decreased accuracy. tDCS did not affect responses to abstract verbs. These complementary effects of inhibitory and excitatory tDCS demonstrate that PMC activity affects, not only how fast, but also how well people process action language. | Tom Gijssels, Richard Ivry and Daniel Casasanto | Conference Room A |
146(abstract) | EEG correlates of covert dependency formation in Mandarin wh-questions Event-related potential (ERP) components are sensitive to the processes underlying how questions are understood. This study shows that Mandarin wh-questions do not elicit anterior negativities associated with memory maintenance, in contrast to prior work with both covert and overt wh-dependencies. N=37 native speakers of Mandarin Chinese read Chinese questions and declarative sentences word-by-word during EEG recording. Consistent with prior work with Japanese covert dependencies, we find no P600 at the wh-word. However, in contrast to the prior Japanese results, no anterior negativity was found at the wh-word nor between the cue "wonder" and the wh-word in indirect questions. | Chia-Wen Lo and Jonathan Brennan | Conference Room A |
149(abstract) | Long-before-short vs. time-before-location: Word order preference in the production of Korean sentences with multiple adjuncts We investigated the word order preference in Korean sentence production using the sentences with a subject, a temporal adjunct, a locative adjunct, an accusative object, and a verb. The results indicated that Korean speakers produce the temporal adjunct prior to the locative one irrespective of the manipulated length difference between the adjuncts. This ‘time-before-place’ preference is argued to reflect the production strategy to put the sentence constituents describing the event structure as closely as possible, thereby minimizing the cost for communicating the propositional contents of the predicate. | Yunju Nam, Upyong Hong and Jewook Yoo | Conference Room A |
150(abstract) | A shift in gap manifestation incurs processing cost: Evidence from Hebrew This paper explores the processing effects of a reanalysis that is due to a realization of an optionally overt element. Two self-paced reading experiments demonstrate that, in Hebrew, direct-object resumptive pronouns are read faster when they appear before the verb and slower when they appear in the direct-object's canonical position after it. We take this effect to reflect a processing cost we attribute to a shift in dependency resolution, triggered by the need to integrate a resumptive into the parse after the dependency has already been resolved by positing a gap in object position. | Julie Fadlon, Maayan Keshev and Aya Meltzer-Asscher | Conference Room A |
153(abstract) | About not losing the head in RC processing Subject relative clauses (SRCs) are easier to process than object relative clauses (ORCs) in most languages, such as French and English. Previous research focused mainly on syntactic factors like structural or linear distance (Hsiao & Gibson, 2003) or on semantics factors like animate/inanimate heads (Frauenfelder et al., 1980 ; Mak et al., 2006 ; Gennari et al., 2012). We looked at the effect of implicit causality on RC processing in French and in English. With acceptability judgment tasks in both languages as well as an eye-tracking experiment in English, we conclude that verb bias can explain some of the ORC disadvantage. | Céline Pozniak and Barbara Hemforth | Conference Room A |
156(abstract) | Is it easier to think positive? An ERP study on quantifier processing in German Using picture-sentence verification, the present ERP study investigated whether positive and negative quantifiers are processed incrementally when compositional-semantic processing is disentangled from extra-semantic factors like world knowledge. An immediate N400 for false vs. true sentences was restricted to positive quantifiers. Processing difficulties already occured while processing negative quantifiers, but both quantifiers alike were evaluated against picture contexts before the end of the sentence, as indexed by a positivity for false sentences. We interpret the N400 as an index of predictions based on the semantic properties of the quantifier, and the positivity as being related to updating the current mental model. | Petra Augurzky, Fabian Schlotterbeck and Rolf Ulrich | Conference Room A |
160(abstract) | Subject island violations involve increased semantic processing, but not increased verbal working memory resources: evidence from fMRI Island effects are central to syntactic theory, and many distinct underlying sources have been suggested, including the hypothesis that islands result from an overload of working memory resources. To provide additional insight, we used functional magnetic resonance imaging to identify the brain networks activated by subject islands and phrase structure violations. Subject islands activated networks involved in conceptual-semantic processing, while phrase structural violations activated a distinct network involved in verbal working memory. This dissociation suggests: that subject islands do not result an overload of verbal working memory resources, but rather from a semantic processing problem or trigger semantics-intensive reanalysis. | William Matchin, Diogo Almeida, Jon Sprouse and Gregory Hickok | Conference Room A |
166(abstract) | Tomorrow is not always a day away English tomorrow has long been considered a pure indexical, even as other expressions once considered strictly utterance-time-oriented, have been shown to be context-sensitive. This paper presents data showing that tomorrow is not always interpreted relative to utterance time (UT) and thus is not a strict indexical; and suggests two alternative analyses of tomorrow. We evaluated American English speakers' acceptance of non-UT tomorrow under speech verbs through a comic-captioning task on Mechanical Turk. Our results show that participants accept non-UT readings of tomorrow, but to a lesser degree than the next day, suggesting that tomorrow is more context-sensitive than previously thought. | Carolyn Anderson | Conference Room A |
167(abstract) | Rates of scalar inferences beyond ‘some’ – A corpus study Previous research suggests that different scalar expressions give rise to scalar inferences (SIs) at different rates. This phenomenon has become known as scalar diversity. In current study, we argue that measuring SI rates in a corpus-based approach could provide a better estimate of variability in SI rates. The scalar diversity pattern was established in our corpus-based paraphrase study, but the observed variation differs from previous investigations. | Chao Sun, Ye Tian and Richard Breheny | Conference Room A |
169(abstract) | Agreement attraction does not depend on time pressure Participants in typical agreement attraction studies are placed under time pressure. We test whether the pattern of agreement attraction errors changes when participants are given an unlimited amount of time to choose whether a verb should be singular or plural. We used the materials of Bock and Cutting (1992), who contrasted attraction errors in prepositional phrases and relative clauses. We replicated their results, but did not find evidence of a qualitatively different pattern of results between speeded and unspeeded tasks: errors persisted even when subjects could reread the sentence as many times as they wished. | Tal Linzen and Brian Leonard | Conference Room A |
170(abstract) | Processing role-ambiguous sentences in German: An ERP study in 7-8 year old children German uses case-markers to distinguish different argument roles regardless of word-order. Feminine arguments are case-ambiguous and thus role-ambiguous in German. Since resolving such ambiguities is necessary to interpret everyday transitive utterances, this ERP study investigated whether 7 year old children have learnt this aspect of the language & are able to resolve this ambiguity at NP2. Whilst the children successfully resolve role ambiguities involving feminine arguments, their brain responses show crucial differences with those of adult speakers. Our findings suggest that neural mechanisms underlying language processing are still in development, and have not yet attained adult-like status at age 7. | R. Muralikrishnan, Alessandro Tavano, Angela D. Friederici and Jens Brauer | Conference Room A |
173(abstract) | Extra time to get results? ERP data on complex predicates in Mandarin We used Mandarin resultative compound verb construction (Mom washed-ruined the clothes, meaning that Mom washed the clothes such that the clothes was ruined) to study the predictive mechanism of sentence processing. We examined whether the ruining-by-washing relation is computed quickly enough to inform predictions about the subsequent object. Our ERP results suggest complex predicate computation is too slow to impact predictions for the subsequent noun, as indexed by the absence of N400 predictability effects. We observed a late positivity at the complex verbs; we tentatively associate it with complex event relation computation, which could delay the updating of predictions. | Chia-Hsuan Liao and Ellen Lau | Conference Room A |
174(abstract) | Segmenting sentences: Robustness of the statistical word segmentation literature We report the first set of results in a multi-year project to replicate every adult statistical word segmentation study. The purpose is both to assess the strength of the findings but also to provide more accurate effect size estimates. We replicated successful learning in all six replications. However, many theoretically important failed to replicate. Moreover, learning success was generally much lower than in the original studies. These results may be due to differences in subject populations, low power in the original studies, or some other factor. Regardless, these initial results suggest taking caution in relying on the originally reported findings. | Lauren Skorb and Joshua K. Hartshorne | Conference Room A |
180(abstract) | The effect of lexical accessibility on Spanish-English codeswitching Bilinguals face lexical competition across their two languages, and switching languages (codeswitching) may be a consequence of this competition. This study investigates the hypothesis that at points of high lexical competition, bilinguals are more likely to codeswitch. Pictures that were more often labeled in English on a lexical accessibility assessment were more likely to be switched into English in a Spanish sentence production task, and had longer speaking durations whether or not they were codeswitched. However, codeswitched utterances had the longest speaking durations. The present work suggests that relative lexical accessibility of the competing labels may influence producers’ codeswitching choices. | Justin Sarkis, Jessica Montag and Judith Kroll | Conference Room A |
181(abstract) | Processing of regular and irregular past tense verbs in Korean sentence processing: ERP evidence for the full parsing model The present study addresses the question of whether regular and irregular past tense verbs of Korean are processed rule-based or memory-based. We conducted ERP experiment using violation paradigm that manipulated past tense morphemes. As a result, the P600 were found in both regular and irregular conditions, whereas only the early N400 was elicited by the incorrect regular forms compared to their correct counterparts. We argue that the P600 indicates that the processing of regular and irregular past tense forms takes advantage of a single identical mechanism of rule-based parsing in Korean. Therefore, our results support the full parsing model. | Dongsu Lee, Upyong Hong, HyeNyeong Chung and Kihyo Park | Conference Room B |
183(abstract) | The direct relation between prediction and adaptation The current study evaluated the Error-based Implicit Learning accounts by investigating the role of prediction in adaptation. Consistent with previous findings, in the prediction task using visual world paradigm, native English speakers made predictions using semantic cues from the given context. They also showed adaptation toward the current linguistic input as measured by change in interpretation of relative clause (RC) attachment at the pre and post tests after exposure to sentences forcing low attachment interpretation (block 1) and high attachment interpretation (block 2). Importantly, however, individuals' predictive ability did not seem to influence their adaptation. | Eunjin Chun, Joshua Daniels, Hope Klingensmith and Edith Kaan | Conference Room B |
185(abstract) | Syntactic Priming Influences Comprehension Patterns in Spanish-English Codeswitchers In this work we explore structural priming by bilinguals while reading code-switched sentences, with the aim of exploring how previous exposure to syntactic structures shapes comprehension and ease of processing. We recorded EEGs while Spanish-English bilinguals read two types of constructions that vary in frequency and structure, but not in meaning. (habían cleaned… and had cleaned …). Results showed that target trials were facilitated in bothconditions with different neural responses depending on frequency (P600 - less frequent / anterior negativity - more frequent). These results that bilingual speakers are sensitive to the statistical regularities found in code-switched language corpora. | Rhonda Mudry, Laura Rodrigo and Paola E. Dussias | Conference Room B |
192(abstract) | Different determiners, different algorithms: two majority quantifiers in Cantonese bias distinct verification strategies Lidz et al. (2011) propose that meanings provide canonical strategies for evaluation and argue for a particular way of specifying the English determiner most. We consider two Cantonese determiners: daai-do-sou, which has a proportional meaning like most, and zeoi-do, which has a "largest subset" meaning. When asked to evaluate statements about sets of dots, both determiners led participants to rely on their Approximate Number System. But whereas daai-do-sou biased the same strategy as most, zeoi-do biased a different procedure, even in cases where either strategy was viable. This adds cross-linguistic support to the claim that meanings provide default verification procedures. | Tyler Knowlton, Athena Wong, Justin Halberda, Paul Pietroski and Jeffrey Lidz | Conference Room B |
193(abstract) | Preschoolers and adults adapt to verb bias in online sentence processing We asked whether previously observed verb-bias adaptation in comprehension reflected distributional learning about the syntactic-semantic combinatorial privileges of each verb or learning about word events each verb describes. We trained four- and five-year-olds and adults using the dative alternation because the DO and PD structures describe the same world events. After brief training, children and adults fixated to the potential theme (e.g., lemon) more for PD-trained verbs than for DO-trained verbs; suggesting that they had linked to each verb new information about its syntactic likelihood. This adaptation likely reflected learning about verb argument structure rather than about events. | Yi Lin and Fisher Cynthia | Conference Room B |
197(abstract) | Brain potentials of expressive content: New evidence for semantic theory How are expressives processed and represented? We explore expressives via swear words within a lexical decision task and EEG. Results show swear words are more effortful than other words similar in their negative affect, meaning there is more to the expressive dimension than merely a heightened emotional state. EEG shows negative words and swear words eliciting larger late positivity effects than neutral words, but both with distinct scalp distributions. This is the first study we know of exploring expressives neurophysiologically from a linguistic perspective. We thus probe our understanding of the semantic-pragmatic boundary and its intersection with social meaning. | Stanley Donahoo and Vicky Lai | Conference Room B |
200(abstract) | Early detection of conflicting constraints? ERP evidence from Georgian Much work suggests comprehenders rapidly generate expectations about upcoming input. The current ERP study uses Georgian, a split ergative-absolutive language, to ask whether conflicts between expectations themselves have the same status as conflicts with the bottom-up input. Because ergative-marked nominals predict aorist tense and future adverbials predict non-aorist tense, their combination creates a conflict not between the cues themselves but between their predictions. If these cues drove predictive updates to the syntactic representation, we might expect an immediate response to the conflict. However, our ERP results demonstrate no evidence of processing difficulty until the downstream verb. | Ellen Lau, Nancy Clarke, Michaela Socolof, Rusudan Asatiani and maria polinsky | Conference Room B |
202(abstract) | How interference guides the missing VP effect in German Multiple center-embeddings lacking one of the verbs are occasionally perceived as grammatical. Various models account for this, blaming memory overload, language statistics or interference. We present an eye-tracking experiment in German with sentences containing three verb-final clauses, which were grammatical or lacked VP2 or VP3. The number of the highest subject and corresponding VP1 were manipulated. The ungrammatical sentences had longer reading times. In addition, VP3 conditions were read longer than VP2 conditions. Participants (25 of 48) who detected the ungrammaticality showed susceptibility to the number manipulation. These findings are in agreement with the interference account. | Katja Suckow and Jana Häussler | Conference Room B |
203(abstract) | Disconfirmed lexical predictions linger in highly constraining contexts Previous studies on lexical prediction typically address cases where a lexical prediction is disconfirmed, and focus on the effects of another, unexpected word found in its place (Delong et al., 2014; Frisson et al., 2017). In contrast, we address the status of the originally predicted word once the prediction is disconfirmed in a memory probe task and self-paced reading task. We find evidence that words predicted from locally and globally constraining contexts are not suppressed, as disconfirmed predictions continue to influence memory and sentence comprehension, suggesting that words gradually decay or are actively maintained in memory. | Stephanie Rich and Jesse Harris | Conference Room B |
206(abstract) | Memory retrieval in comprehension is sensitive to production alternatives Most current memory retrieval models for sentence comprehension assume that retrieval candidates are previously processed linguistic input. In three experiments that examine antecedent retrieval for ellipsis resolution, we show that memory retrieval not only targets the actual previous linguistic input, but also salient production alternatives that could have been uttered. We suggest that the mechanism to introduce these alternatives into the candidate space could naturally fall out from a general memory retrieval architecture if we view retrieval as structure reproduction based on previously obtained semantic representations [Potter and Lombardi, 1990]. | Ming Xiang and Josef Klafka | Conference Room B |
208(abstract) | Syntax and recall serial: The effect of positional statistics on serial ordering Much sentence processing research assumes that the ordering of verbal material in WM is separate from linguistic LTM. We demonstrate that linguistic LTM in the form of grammatical sequences impacts WM performance in a serial recall task while controlling for other, informative cues. Participants were better at recalling grammatical sequences consistent with real-world language patterns compared to reversed sequences. These results have implications for theories of language production and comprehension, and suggest a closer relationship between language production/comprehension experience and verbal working memory ability. | Steven Schwering and Maryellen MacDonald | Conference Room B |
210(abstract) | Priming the Metonymic Expressions in Japanese Sentence Production The present study reports on a recall-based sentence production experiment investigating how people construct metonymic expression (“read Dickens” instead of “read Dickens’s book”) in Japanese. Results showed that recalling the target with metonymic expressions was influenced by primes with the metonymic expressions (“read Dickens”), but not primes with the non-metonymic (“read Dickens’s book”) and literal expressions (“met Dickens”). Also, the difference in the proportions of the non-metonymic and literal expressions was not statistically significant. These results suggest that speakers can access the metonymic expression directly. I discuss these findings in the context of theories of production. | Mikihiro Tanaka | Conference Room B |
214(abstract) | A priority list of factors for syntactic encoding (in German) In two event description experiments with eye tracking, visual cueing of event participants was combined with the manipulation of the event participant’s discourse status to investigate which factors are given more weight in syntactic encoding. Results suggest the following priority order of factors influencing syntactic choices: 1. discourse status of the referents, 2. event structure (thematic roles), 3. attentional starting point. | Xiaogang WU and Johannes Gerwien | Conference Room B |
218(abstract) | The role of animacy and structural information in relative clause attachment: Evidence from Chinese In the current study, we investigated the interplay of animacy and relative clause type in Chinese relative clause attachment. The results showed that while relative clause attachment in Chinese is modulated by animacy of a potential head NP (Exp1: production study), the animacy effect is constrained by a relative clause type. However, the results are only partially compatible with the previous findings that SRs and ORs tend to occur with an animate and inanimate head NP respectively in Chinese (Pu, 2007; Wu, 2009), as the effect of animacy is mainly limited to SRs (Exp2: comprehension task). | Deborah Ong and Nayoung Kwon | Conference Room B |
223(abstract) | Cost due to re-ranking reflected in the same ERP component during the processing of Japanese ditransitives Human parser makes a prediction about upcoming information and they are updated at every word of a sentence. According to Surprisal Theory, the input which narrows down the upcoming information causes a large re-ranking, which results in large cost. In Japanese head-final ditransitive sentences, pre-verbal arguments are predicted to incur re-ranking cost, and this leads to lower processing cost for the final verb. An important question is whether the processing cost observed at two positions are caused by the same re-ranking processing, as Surprisal Theory maintains. Our ERP experiment confirmed N400 components at both positions, providing support for Surprisal Theory. | Shodai Uchida, Manabu Arai, Yuki Hirose and Takane Ito | Conference Room B |
226(abstract) | Cue reliability and re-weighting in speech perception There has been much interest in how listeners weight cues during speech perception. This work has generally assumed that cue weights are static. However, cue reliability is not constant across situations and could potentially lead to re-weighting. Here, we exposed two between-subjects groups to differing levels of cue conflict between acoustic and semantic cues in a discrimination task. Listeners who encountered high levels of conflict gradually down-weighted semantic cues, relying primarily on acoustic cues. These findings highlight the adaptivity of speech perception, pointing to a strategy that takes advantage of the current statistics of the environment to support optimal processing. | Wednesday Bushong and T. Florian Jaeger | Conference Room B |
232(abstract) | Subject/Object Relative Clause Asymmetry is shaped by Role Tracking Difficulties We examined whether the relative clause (RC) asymmetry originates in non-linguistic visual processing using an object tracking task where the identical circles interacted in ways that would best be described using a subject RC or an object RC. The participants’ accuracy in producing RC structures was significantly higher for the SRC-target trials than the ORC-target trials. This effect persisted in replications using passive ORCs and simple active transitive sentences. This cannot be explained by linguistic properties or experience, but suggests that difficulties in tracking role switches could weaken the link between meaning and ORC forms. | Andrew Jessop and Franklin Chang | Conference Room B |
234(abstract) | The role of coherence and finiteness in extraction from adjunct islands: An acceptability study of Swedish and English We present results from two acceptability judgment experiments investigating the effect of semantic coherence and finiteness on extraction from adjunct islands in Swedish and English. A coherence relation (e.g., a causal, as opposed to a purely temporal relation) between the events described in the matrix and the adjunct clause was found to significantly improve extraction in both languages. Furthermore, finiteness was seen to significantly degrade extraction from coherent adjuncts, but only in English. Swedish showed no effect of finiteness or finiteness by coherence interaction. Our results have implications regarding filler-gap association in island domains and cross-linguistic variation. | Christiane Müller, Damon Tutunjian and Anna-Lena Wiklund | Conference Room B |
244(abstract) | Agreement attraction in grammatical sentences We propose an interference-based explanation for the failure to find an illusion-of-ungrammaticality effect in agreement attraction and manipulate experimental materials in order to diminish the interference. Unfortunately, a 3-regions-long spillover effect renders the reading data uninterpretable. However, in the question response data, we find strong support for the feature percolation account and evidence for semantic misinterpretation accompanying agreement attraction. | Anna Laurinavichyute and Titus von der Malsburg | Conference Room B |
297(abstract) | Adjective interpretation across the senses: On the subjective nature of sight, smell and taste When processing subjective adjectives (e.g. fun, tasty), one needs to know whose opinion is being expressed. Although these adjectives can involve different sensory modalities (e.g. sight, taste, smell), current linguistic theories do not make clear distinctions based on sensory modality. We tested whether interpretation of who is the attitude holder/experiencer of an adjective depends on whether the situation involves seeing, smelling or tasting (e.g. 'it looked/smelled/tasted disgusting'). The results suggests that yes, sensory modality impacts identification of the attitude-holder. There are more first-person/narrator interpretations (vs. third-person/character-in-narrative interpretations) in the visual than the gustatory or olfactory domains. | Elsi Kaiser, Catherine Wang and Gwenyth Portillo-Wightman | Conference Room B |
302(abstract) | The facilitatory role of regressions in recovery from processing difficulty: an application of the reverse boundary-change technique We investigated the role of regressions in recovery from processing difficulty during reading, using a reverse boundary change technique. In the change condition, a word several words left of fixation changed to a lexical neighbour (e.g. "house" -> "horse" in 1) when the reader's gaze crossed an invisible boundary (marked "|" in 1). (1) There was an old house->horse that John had| ridden when he was a boy. Processing was facilitated in (1), relative to a condition that remained implausible throughout the trial (i.e. "house" did not change). This indicates that lexico-semantic information is used to aid recovery during regressions. | Patrick Sturt and Nayoung Kwon | Conference Room B |
321(abstract) | Features of Verb Complexity for People with Aphasia in a Verb-Final Language The current study examined verb complexity effects in people with aphasia using a verb-final language. We developed a coding system that operationalized the predicate complexity as the number of verbs in the predicate. Eighteen individuals with aphasia (9 anomic and 9 Broca) and 10 neurotypical controls participated in the study. Individuals with Broca’s aphasia produced less complex predicate compounds than persons with anomic aphasia or controls. However, the metric for capturing these differences is somewhat different than that used for English speakers. The results highlight the importance of developing theories of the linguistic complexity in the context of crosslinguistic studies. | Jee Eun Sung and Gayle DeDe | Conference Room B |
335(abstract) | ERP effects distinguishing Negative polarity versus Free choice: A look at English ANY and Mandarin RENHE English ANY (and Mandarin RENHE) lives a double-life. It can serve as a Negative Polarity Item (NPI) or a Free Choice Item (FCI). To examine the relationship between the NPI/FCI-variants of ANY, we first examined ERP responses to EVERY/ANY in four embedded contexts in English, varying the presence/absence of a main clause subject licensor and whether the Target occurred in the embedded subject position or the pivot position of an existential construction. A second study conducted in Mandarin compares NPI RENHE (‘any’) to its free choice variant, which is derived via combination with the subsequent adverbial DOU (‘all’). | Jun Lyu, Hongchen Wu, Aydogan Yanilmaz and John Drury | Conference Room B |
379(abstract) | Lateral spatial biases in scene comprehension Transitive action scenes are comprehended faster when agents are on the left and patients on the right (Maas & Russo, 2003). Additionally, response hand is a crucial factor in these spatial-linguistic interactions (Boiteau & Almor, 2016). In the current study (E1), we manipulated sentence voice and response hand, testing whether sentence processing leads to discourse representations in which agents and patients are associated with lateralized spatial-motor processes. In E2 we examined how response hand interacts with the orthographic direction of participants’ language. Our results suggest that multiple spatial frames (e.g., motoric, perceptual) are involved in processing action scenes. | Tim Boiteau, Randy Lowell and Amit Almor | Conference Room B |
392(abstract) | P600 vs. N400 indicators of comprehension: Dynamic effects of cognitive-control engagement on real-time parsing We investigated the role of cognitive control in sentence processing by interleaving trials of the Stroop task with semantically anomalous sentences in an ERP study. Following Congruent Stroop trials, sentence anomalies elicited a P600 effect. Following Incongruent Stroop trials, the same category of anomalies elicited an N400 effect. We suggest that dynamic cognitive-control engagement modulated readers’ parsing commitments. When cognitive control was relatively un-engaged (Congruent Stroop), readers attempted to structurally repair anomalies according to a preferred analysis (P600). Engagement of cognitive control (Incongruent Stroop) allowed readers to override the preferred interpretation and pursue a syntactically dictated but implausible interpretation. | Jared Novick, Nina Hsu, Sara Milligan and Al Kim | Conference Room B |
402(abstract) | Counting exceptions: Exception Phrases and quantification Exception Phrases (EPs) have been claimed to be degraded when they associate with existential quantifiers. We test this claim and find that the decline in acceptability is significantly more pronounced for negated universals compared to bear existentials. Next, we test the acceptability of EPs associated with multidimensional adjectives (e.g., healthy/sick)---which have been argued to encode universal and existential quantification over adjectival dimensions---, and find parallel patterns of acceptability to those displayed by quantifiers. We suggest that these results can be explained as a trade-off between the semantic and the pragmatic interpretation of EPs associated with existential quantifiers. | Helena Aparicio and Roger Levy | Conference Room B |
Friday March 16, 2018 | |||
Coffee | 8:00 AM - 9:00 AM | Session 3 | 9:00 AM - 12:00 PM |
Submission | Title | Authors | |
Chair: | Adrian Staub | ||
82(abstract) | Hear no evil or speak no evil: Lengthening of phonological competitors Speakers tend to lengthen words that sound similar to previously named words (e.g. beaker and beetle). We test two accounts of this effect: speakers lengthen because producing two similar-sounding words leads to interference; or, speakers lengthen strategically to reduce confusability. In an event description task, speakers produced sentences where words phonologically overlapped with another referent that was either unnamed, named by another speaker, or named by the participant. Speakers only lengthened in the latter two conditions, counter to established theories. Instead, our results parallel theories of reduction that have proposed speakers remember what they have said when deciding to lengthen. | Andrés Buxó-Lugo, Cassandra L. Jacobs and Duane Watson | 9:00 AM |
107(abstract) | Context, conflict, and the time course of interpreting irony Comprehending irony requires listeners to infer that speakers mean the opposite of what they say, but it remains unclear how context combines with literal meanings. Using visual-world tasks, participants heard sentences about characters spoken by Literal Lucy and Ironic Ike (“What a fabulous/terrible chef s/he is”). Pronouns identify Targets, but adjectives provide earlier cues when combined with speaker gender. Experiment 1 and 2 demonstrate that unlike opposites, interpreting irony depend on separable semantic and pragmatic analysis. Experiment 3 reveals that these computations result in conflicting representations. This suggests that ironic interpretations arise from real-time procedures and are not pre-compiled meanings. | Rachel Adler, Jared Novick and Yi Ting Huang | 9:30 AM |
287(abstract) | Cue reliability affects anticipatory use of prosody in processing globally ambiguous sentences In the two visual-world eye-tracking experiments, we examined the anticipatory effect of prosody in processing globally ambiguous sentences, and whether listeners place less weight on prosodic information in structural decision when filler items have ‘unreliable’ prosody. The results showed participants responded differently depending on prosodic cues prior to the critical word, reflecting their anticipatory use of prosody. It was also shown that participants relied less on prosody when filler items had unreliable prosody in Experiment 2, suggesting that listeners track the reliability of prosody and adjust the extent to which they use prosody in making anticipatory judgments during structural analysis. | Chie Nakamura, Jesse Harris and Sun-Ah Jun | 10:00 AM |
Break | 10:30 AM | ||
Chair: | Lars Meyer | ||
Invited Speaker | Rhythmic computation of linguistic structure | Andrea E. Martin | 10:50 AM |
368(abstract) | Prosody, Poetry and Processing: ERP Evidence for Hierarchical Metric Structure in Silent Reading We recorded ERPs while participants read metrically predictable rhyming couplets with and without metrically unpredictable words. We manipulated the lexical stress pattern (i.e., beat position: STRONG-Weak, Weak-STRONG) and metrical consistency (consistent, inconsistent) of the target word in a 2x2 design. We observed interactive effects of metric consistency and stress pattern on ERP amplitudes between 325-400ms over left and medial-frontocentral scalp regions. Metrically inconsistent STRONG-Weak targets elicited negativities relative to consistent STRONG-Weak targets. Consistent and inconsistent Weak-STRONG targets did not differ. This finding provides further evidence of implicit metric representations during silent reading, supporting the Implicit Prosody Hypothesis. | Michelle Oraa Ali, Ahren B. Fitzroy and Mara Breen | 11:30 AM | Poster Session B | 12:00 PM - 2:00 PM |
Submission | Title | Authors | |
1(abstract) | Language control and lexical access in diglossic speech production: evidence from variety switching in speakers of Swiss German Two switching experiments, including Swiss German and Standard German (Experiment 1) or Swiss German and Tamil (Experiment 2), investigated the mechanisms involved in diglossic language control and lexical access. Switch costs in diglossic switching were comparable to switch costs in bilingual switching. Switch costs were asymmetric, favouring the sociolinguistically preferred variety or the less dominant language. Cognate gains were higher for Swiss German and occurred even for Tamil responses. These results suggest that diglossic switching underlies the same mechanism of language control as bilingual switching, that inhibition is involved and that cognates have integrated lexical representations in diglossic speakers. | Constanze Vorwerg, Sumanghalyah Suntharam and Marie-Anne Morand | Lobby |
8(abstract) | Look at that coordinated dialogue: Deixis and prediction in sentence processing The present study evaluated whether listeners can use spatial deixis (e.g., this, that) to predict a speaker’s likely referent. Adults and 5-year-olds viewed scenes while listening to deictic sentences (e.g., Look at that beautiful baby) and neutral sentences (e.g., Look at the beautiful baby). We found that both adults and children used deixis to predict the plurality of the referent, but only adults used deixis to predict the proximity of the referent (e.g., using this to anticipate a referents proximal to the speaker). Thus, the present findings reveal specific developmental changes in how prediction occurs during language processing. | Tracy Reuter and Casey Lew-Williams | Lobby |
16(abstract) | Predictive pre-updating: ERP correlates and the influence of working memory capacity Participants with high and low working-memory (WM) capacity read strongly and weakly constraining sentences. ERPs were measured on the predictable noun and on its preceding verb, where the prediction is generated. For participants with high WM capacity, increased P600 amplitude was observed at the verb in the high-constraint sentences, reflecting “pre-updating”, i.e. integration of predicted content into the context’s representation built in WM. Easier integration of a pre-updated word was observer at the noun, reflected as decreased P600 amplitude. These effects were not observed in the low-span group, indicating that the tendency to engage in pre-updating depends on WM capacity. | Tal Ness and Aya Meltzer-Asscher | Lobby |
24(abstract) | Communicating, interpreting, and learning from uncertainty: Could be a promising approach Source reliability and degree of commitment to a proposition is implicated in many domains within pragmatics. Communicating reliability requires speakers to signal confidence in an assertion, and listeners to interpret that signal. This is complicated by individual and group differences (e.g., politeness constraints, "mansplaining", etc). The evidence from this series of studies is a first-step in developing methodologies to evaluate how speakers convey their uncertainty and how listeners titrate their expectations on the inferred uncertainty of a speaker. In future studies we investigate how listeners use this information to determine the reliability of their interlocutor, and how that affects learning. | Amanda Pogue and Mike Tanenhaus | Lobby |
27(abstract) | Japanese 3rd person overt pronouns carry anti-logophoricity This study investigated the possibility that Japanese 3rd person overt pronouns carry anti-logophoricity, i.e., they tend not to refer to antecedents when the antecedents’ speech or thoughts are reported in the sentence. An off-line questionnaire-based experiment was conducted with 43 native Japanese speakers as participants. Two sentence discourses are given, which included a 3rd person overt pronoun and two competing antecedents. The results indicated that Japanese 3rd person overt pronouns show a significant dis-preference for the antecedents when they are agents of speech or thoughts. Overall, this study found that Japanese 3rd person overt pronouns are anti-logophoric pronouns. | Shinichi Shoji | Lobby |
41(abstract) | Telicity across languages Accomplishment predicates (like ‘peel an apple’) describe events that have a duration and a culminating point (Vendler 1967). When combined with perfective aspect they have a completion entailment that the telos has been reached (the apple is completely peeled). Using behavioural and reading methods, previous studies show that participants do not immediately commit to the telicity of event descriptions (Pickering et al. 2006). Our study employs the visual world eye-tracking paradigm in a cross-linguistic investigation of Dutch, Russian and Spanish, that differ as to how perfectivity is realized. | Miguel Santin, Francesca Foppolo and Angeliek van Hout | Lobby |
51(abstract) | The Processing Of Object-First Declarative Sentences And Wh-Questions Including Weak Masculine Nouns In Comparison In the present poster we present on-line (visual world eye-tracking paradigm) and off-line data (picture-matching task) of German 7-year-old L1c and German-Turkish L2c as well as German L1a. We focus on the processing and comprehension of a) German unambiguous declarative OVS sentences introduced by a masculine accusative first NP, marked on the determiner (den Opa, the grandpa) b) German unambiguous wh-questions including an accusative question pronoun (Welchen TigerØ, which tiger) and c) German unambiguous wh-questions including an accusative question pronoun and an accusatively marked noun (Welchen Affen, which monkey; Affe, nom.; Affen, acc.). | Valentina Cristante, Anja Binanzer and Andreas Bittner | Lobby |
53(abstract) | The resolution of anaphoric relations in discourse Several studies showed that antecedents ruled out by syntactic binding constraints still influence online processing of anaphora in some stages of processing. This undermines the general view that binding constraints make antecedents strictly inaccessible throughout processing. In this experiment, we generalize these findings by looking at inter-sentential anaphoric relations. As in the syntactic literature, in semantic accounts like Discourse Representation Theory (DRT), formal constraints on such relations are formulated in terms of the accessibility of the antecedent by an anaphor. Using a self-paced reading experiment, we examined the role of discourse factors like non-referentiality in online sentence processing. | Morwenna Hoeks and Jakub Dotlacil | Lobby |
76(abstract) | Forms and features in agreement processing We report a self-paced reading experiment studying how the inflectional class (declension) a noun belongs to and its gender influence the processing of gender agreement in Russian. Russian has three genders and three declensions, according to the most widespread classifications. Most consonant-final Nom.Sg forms are masculine (2nd declension), most feminine Nom.Sg forms end in -a/ya (1st declension), with 3rd declension consonant-final feminine nouns being less usual. Agreement errors were noticed significantly later with 3D-F nouns than with 1D-F or 2D-M ones. However, error-related delay on subsequent words was more pronounced for M subjects than for F ones (1D and 3D). | Natalia Slioussar and Pavel Shilin | Lobby |
80(abstract) | D-linking and syntactic islands in Mandarin Chinese There are two main theories on D-linking effect in wh-islands. The featural Relativized Minimality ascribed the effect to partial feature overlap between the wh-operator and the wh-intervener, whereas the argument of the working memory account focuses on easier retrieval. The present study contributes to this debate by investigating the D-linking effect in Chinese subject and RC islands, which are "selective" islands without wh-interveners. The results are congruent with the working memory account. | Jun Lyu | Lobby |
83(abstract) | Encoding semantically rich words: The relative effects of reactivation and distinctiveness We investigated the relative effects of "head-reactivation" and "distinctiveness" on activation levels of associated noun phrases (NPs) by employing pre- vs. post-modifiers. In 4 experiments, we showed that post-modified NPs (e.g., the actor who was frustrated) exhibit greater activation levels than not-modified NPs (e.g., the actor; as manifested in more pronominal reference to them during production and faster reading times on pronouns referring to them during comprehension). However, pre-modified NPs (e.g., the frustrated actor) exhibited weaker boosts in activation. Thus, our results suggest that although both head-reactivation and distinctiveness increase activation, head-reactivation leads to relatively greater activation than distinctiveness does. | Hossein Karimi and Fernanda Ferreira | Lobby |
90(abstract) | Referent availability in the comprehension of weak definites Weak definites (Frank went to the psychologist) highlight a concept and background the referent of the noun. Results from a visual world and a mouse tracking experiment suggest that weak definites come with lower referent accessibility than indefinites. They also show that comprehenders can quickly accommodate a referential reading of weak definites when they make sense of larger text. | Klaus von Heusinger and Andreas Brocher | Lobby |
98(abstract) | Processing crossing dependencies in Hindi Given that the verb can be strongly expected in a head-final language, we investigate the role of prediction and working-memory constraints in processing crossing dependencies in Hindi. Using a self-paced reading experiment, we have shown that processing crossing dependency even in a head-final configuration is difficult and strong expectation attenuates this difficulty. However, this processing cost reflects different aspects of the prediction processes - prediction maintenance in certain dependencies will be contingent on whether the prediction gets revised due to intervening material or not. | Himanshu Yadav and Samar Husain | Lobby |
121(abstract) | Influences on the agreement in German hybrid nouns – distance and syntactic domain Hybrid nouns are nouns like German Mädchen (‘girl’), which can occur with either syntactic agreement (i.e. with the grammatical gender feature) or semantic agreement (i.e. with the grammatical gender feature). Two acceptability studies and a production experiment should determine which factors influence the agreement pattern, focusing mainly on the linear distance between noun and pronoun and the effect of the syntactic domain, i.e. if the pronoun is in a clause subordinated to that of the antecedent or in a new main clause. The results show an effect of linear distance but no effect of syntactic domain. | Astrid Gößwein | Lobby |
123(abstract) | Discourse constraints on referential predictability: is the subject predictable? Are grammatical subjects referentially predictable? Using sentences like “Panda Bear is having lunch with Puppy,” we tested whether there’s an expectation for subjects to be mentioned. Exp1 tested metacognitive judgments about who’s likely to be mentioned next, and found an overall preference for the non-subject – contrary to the subject bias found in natural texts. Exp2 and Exp3 tested which character is more frequently mentioned first in story continuations, and found that referring to the subject is more likely in longer narratives (3- vs. 1-sentence). This raises questions about the use of simple experimental tasks to estimate probability of reference. | Sandra Zerkle and Jennifer Arnold | Lobby |
126(abstract) | What Information is Facilitated by a Repeated Verb? Previous syntactic priming studies in comprehension have observed facilitated processing of a reduced-relative (RR) target sentence following a RR prime sentence with the same initial verb. The present study uses eye-tracking while reading to compare RR target processing when the prime was a main clause, locative with a by-phrase (and same verb), RR with the same verb or RR with a different verb. Decreases in reading times for targets following RR but not locative primes suggests that the repeated verb boosted structural processing by facilitation in predicting/parsing the relative clause (inherent to this structure) rather a lexically-based by-phrase. | Tooley Kristen M. | Lobby |
128(abstract) | Reading or hearing it: How modality affects the flexibility of novel word representations Although the acquisition of a vocabulary is a task often associated with the early stages of language development, learning novel words is not restricted to child populations. In fact, the vast majority of the words adults know are learned after toddlerhood. Some of these words are acquired auditorily, while others are acquired visually. Here, we examine to what extent novel word representations are generalizable across the two domains, and how variation in word tokens during word learning might affect subsequent word recognition. Results show that modality-specific effects on novel word representations can be mitigated by task demands and variability. | Gregory Byrne and Marieke van Heugten | Lobby |
134(abstract) | Balloons, banks, and other blown-up things: listeners' use of associative and combinatorial information in generating on-line predictions Comprehenders must identify relationships between constituents in accordance with the sentence’s syntactic structure. However, low-level associative effects have also been suggested to be at play. In two experiments, we examine the influence of associative cues alongside combinatorial information in the earliest moments of processing by using verb-driven predictions. Results suggest that both factors influence incremental interpretation, yet associative effects are modulated by a cue’s proximity to the verb. In addition, predictive processing is initially mildly affected by affective information and the ability to engage in predictive processing declines to some (small) extent over the adult lifespan. | Adriana Baltaretu and Craig Chambers | Lobby |
139(abstract) | Processing adjunct control: Rapid use of structural information in anaphora resolution Does it matter to the timing of anaphora resolution whether the surface cue to anaphora is a pronoun (‘he’ in (1a)), or a non-finite verb (‘eating’ in (1b))? These forms of anaphora differ in two ways: only in (1a) is the cue dedicated to reference. Second, in (1b) the eater must be Mickey, while in (1a) it may in principle be any salient male. We show in two visual-world eyetracking experiments that listeners can use relevant information to rapidly resolve both forms of anaphora, despite differences in the cue. (1) Mickey talked to Minnie before… (a) he ate. (b) eating. | Jeffrey Green, Michael McCourt, Alexander Williams and Ellen Lau | Lobby |
145(abstract) | To adapt or not to adapt: No evidence that readers adjust their expectation for a disjunction in the either...or structure There is growing evidence that comprehenders adapt their structural expectations according to local probabilities. In three self-paced reading studies, we investigated whether comprehenders show this adaptation by manipulating cues in the environment that change the strength of predictions motivated by the presence of the word either. Across studies, comprehenders read or and its following disjunction quicker when either was present vs. absent. However, while we observed evidence for the prediction of a disjunction, we didn’t find any adaptation of this expectation, suggesting that comprehenders may not be as predictive as possible during processing and adapt to every local syntactic change. | Michelle Colvin, Tessa Warren and Michael Walsh Dickey | Lobby |
152(abstract) | Expectation management in online discourse comprehension: an ERP investigation of Dutch inderdaad ‘indeed’ and eigenlijk ‘actually’ The Dutch discourse markers (DMs) inderdaad ‘indeed’ and eigenlijk ‘actually’ encode a (mis)match between what is said and what is expected on the basis of the discourse. In two ERP reading experiments we asked to what extent such high-level, expectation-managing cues affect online discourse comprehension. We expected inderdaad to strengthen and eigenlijk to weaken predictions about incoming words, as measured by N400 and late positivity modulations. Results show that DMs do not significantly modulate predictive pre-activation of incoming words compared to control adverbs, but inderdaad and eigenlijk differentially affect the online integration of subsequent input with the wider discourse model. | Geertje van Bergen, Marlou Rasenberg and Joost Rommers | Lobby |
165(abstract) | Objects and self-portraits: when logophoricity varies across languages and structures Is the processing of reflexive pronouns sensitive? Slogget & Dillon (2017) argue that agreement attraction effects found in English reflect a “logophoric” reading (i.e. coreference with the point-of-view holder). Our study suggests that a logophoric antecedent for an object position reflexive is not available in Hebrew online processing (in line with a more restricted distribution). However, once the reflexive was further embedded under a picture noun, attraction effects emerged. The fact that such antecedents are accessible in this environment might reflect either a differential availability of the logophoricity across structures, or another function of the pronoun (e.g. intensive reflexives). | Maayan Keshev, Noa Bassel and Aya Meltzer-Asscher | Lobby |
189(abstract) | Being themself: processing and resolution of singular (im)personal they What knowledge is accessed and checked during gendered coreference dependency formation? English encodes information about gender in pronouns and names, and coreference dependency formation relies on antecedent gender of matching that of the anaphor. However, human gender is not binary, and nonbinary genders are increasing in visibility. We investigate whether nonstandard coreference dependencies are processed differentially across the population. We find higher acceptability and less processing cost among people with regular contact with transgender/nonbinary communities, particularly younger speakers. We suggest experience with gender variation influences speakers’ mental representations of gender, which are what is accessed during gendered coreference dependency formation. | Lauren Ackerman | Lobby |
190(abstract) | The neurocognitive mechanism of gender agreement processing by Chinese EFL learners Three ERP experiments were conducted to explore the neurocognitive mechanism of L2 gender agreement processing by Chinese EFL learners, as well as the potential influence of L2 proficiency on such mechanism. Experiment 1 and 2 respectively tested learners’ real-time access to gender information when processing a noun in their L1 and L2. Experiment 3 tested their online matching of gender features when processing a reflexive. The results in the 3 experiments revealed that for learners differing in L2 proficiency, qualitative neural differences may exist, leading to lower sensitivity to gender disagreement among less proficient learners. | Zhibin Yu and Hongming Zhao | Lobby |
199(abstract) | How long can you hold the wh-filler? We investigate how wh-fillers are maintained in memory in two wh-filler-gap dependency (WhFGD) formations: "reactivated" WhFGD formation (the filler is linked to the verb once and the wh-filler is reactivated later) and "active" WhFGD formation. We ask whether Wh-fillers are maintained in a distinguished memory state until the dependency is resolved. No agreement attraction is observed for the coordination sentences, suggesting that only partial information (e.g., the head or the syntactic category of the NP) is retrieved. The active-filler showed agreement attraction, where the wh-dependency is kept until the licensor appears, suggesting fine-grained information is maintained for the active filler. | Nayoun Kim, Laurel Brehm, Patrick Sturt and Masaya Yoshida | Lobby |
204(abstract) | The role of contextual-pragmatic information in speech perception: an eye-tracking study Recent studies have shown that listeners integrate information from domains as disparate as phonetic perception and pragmatic inferences regarding upcoming coreference (e.g., Rohde & Ettlinger 2012). Pragmatic information, however, comes in many forms. Using a Visual World Paradigm, we investigate whether and how a different kind of pragmatic information, the contrastive function of prenominal adjectives (Sedivy et al. 1999) can affect listeners’ perception of voicing in initial plosives. Our results suggest that, pragmatic contrast inference did not affect the behavioral judgments on phonetic categorization, but it did have (albeit limited) influence during the online processing of VOT perception. | Yenan Sun, Eszter Ronai, Alan Yu and Ming Xiang | Lobby |
216(abstract) | Processing of ambiguous and unambiguous morphological cues in object-verb-subject sentences: Is there an SVO-bias? Comprehension of OVS sentences is associated with increased processing demands. The goal of the present study, which comprises three experiments, is to explore processing of locally case-ambiguous as well as case-unambiguous German SVO and OVS sentences using a modified version of the visual-world paradigm: eye-tracking during sentence-picture matching. The results reveal an SVO-bias for locally case-ambiguous sentences, which is rapidly revised towards an OVS structure after processing of an unambiguous inflectional cue. Furthermore, the SVO bias is persistent even in case-unambiguous sentences, although processing strategies are different when passive sentences are included as fillers in the experiment. | Sandra Hanne, Frank Burchert and Shravan Vasishth | Lobby |
231(abstract) | Do Different Feature Encodings in Spoken vs. Written Pronouns Influence Pronoun Use? Are production processes shared between written and spoken production when they differ in their encoding of information? The present study aims to address the question by investigating how the different feature specification in written and spoken Mandarin influences reference production. We found that Mandarin speakers showed different sensitivity to gender information in spoken and written production. Mandarin speakers’ pronoun use was significantly affected by gender information in written production but not in spoken production. Our results suggest that written and spoken language involve different referential processes when their pronouns differ in their feature specification. | Heeju Hwang | Lobby |
235(abstract) | Interaction of language and magnitude perception: gradable adjectives rely on extraction of ratio-level information In this study, we look at the relationship between the meaning of gradable adjectives such as tall, short, long, loud etc. and the human perceptual system for extracting and representing magnitude information. Specifically, we asked whether form of a gradable adjective (comparative vs. positive) can alter the perceptual process in magnitude extraction and whether the meaning of gradable adjectives would be sensitive to ratio-level information. | Arnold Kochari and Robert van Rooij | Lobby |
237(abstract) | Reconsidering asymmetries in voice-mismatched verb phrase ellipsis Whereas it is well-established that Verb Phrase Ellipsis (VPE) with is degraded when antecedent and ellipsis clauses mismatch in voice, studies have also found differences within mismatches: active VPE with passive antecedents tends to be more acceptable than passive VPE with active antecedents. Here we are argue against recent memory-based explanations of this mismatch asymmetry by examining mismatches in cataphoric contexts. Contrary to memory-based accounts, we find no evidence that the order of antecedent and ellipsis clauses matters. Rather, the mismatch asymmetry appears to be driven by a mismatch-independent penalty for passive ellipsis clauses. | Till Poppels and Andrew Kehler | Lobby |
239(abstract) | Incremental contrasts: Why spotting a blue triangle is different in English than in Spanish In referential communication, adjectives are normally used to preempt an ambiguity between entities of the same kind, but they are often used redundantly. In a reference-production study, we show that English speakers produce more redundant color adjectives than Spanish speakers (replicating Rubio-Fernández, 2016). In three eye-tracking experiments, we tested the hypothesis that color contrast is established across-categories in English (pre-nominal modification: blue vs. non-blue figures) while it is established within-categories in Spanish (post-nominal modification: triangle blue vs. triangle red). Our results support the view that color contrast is established incrementally, which affects both the production and processing of color adjectives. | Paula Rubio-Fernández and Francis Mollica | Lobby |
240(abstract) | Response bias modulates the grammaticality asymmetry: Evidence for a continuous valuation model of agreement attraction In agreement attraction, the grammaticality asymmetry—the fact that errors in comprehension lead to illusions of grammaticality, but not illusions of ungrammaticality—has been used to favor retrieval accounts (e.g. ACT-R) over feature percolation accounts (e.g. Marking & Morphing). In three speeded 2AFC experiments, we show the grammaticality asymmetry is due to a bias towards grammatical sentences, rather than retrieval dynamics. When response bias is neutralized, attraction occurs in both grammatical and ungrammatical sentences. The results are used to support a drift diffusion account of attraction, where illusions arise due to equivocal evidence about number marking on the subject. | Christopher Hammerly, Adrian Staub and Brian Dillon | Lobby |
242(abstract) | The fate of the unexpected: consequences of misprediction on ERP repetition patterns This study investigated the potential downstream benefits or costs of prediction-violating words. Participants read unexpected but plausible words in strongly and weakly constraining sentences. Three sentences later the critical word was repeated at the end of a weakly constraining sentence. At the N400 and LPC, words initially encountered as prediction violations seemed to be processed similarly to words that were simply repeated. These findings suggest that while prediction violations are not difficult to encode, neither do they provide downstream memory benefits. Thus, prediction violations may be neither as costly – nor as critical – for comprehension as sometimes assumed. | Melinh K. Lai and Kara Federmeier | Lobby |
243(abstract) | Semantic attraction in sentence processing We demonstrated a new type of attraction effect -- a semantic attraction effect in sentences in which the morphosyntactic agreement between the subject and the verb is intact. The semantic attraction effect does not significantly differ from the morphosyntactic attraction effect in size, the effects are additive. Based on these results, we tentatively conclude that semantic and grammatical attraction effects are the results of similar or even the same underlying mechanisms. | Anna Laurinavichyute and Titus von der Malsburg | Lobby |
245(abstract) | Unaccusative and unergative predicates under ergativity: N400 signaling agent/ergative interpretation difficulties. This study explores the electrophysiology of unergative vs. unaccusative predicates in an ergative language. Unergatives have agent subjects, unaccusatives theme subjects (Perlmutter 1978). In ergative languages like Basque, agents carry ergative case, themes do not. We studied the ERP patterns generated by subject-agreement violations in both predicate types and found different signatures for each, with unergatives aligning with previously reported findings for transitive sentences, where an N400 preceded the generally found P600 component. We argue that this N400 found in both unergative and transitive predicates signals difficulty in agent/ergative identification, crucially missing in the case of unaccusatives. | Gillen Martinez de la Hidalga, Itziar Laka and Adam Zawiszewski | Lobby |
247(abstract) | The Effect of Context on Noisy-Channel Sentence Comprehension Participants from Mechanical Turk were presented with a questionnaire containing 20 critical items consisting of two context sentences followed by an implausible test sentence, which was formed by swapping a pair of words from a plausible version (materials edited from Gibson, Bergen & Piantadosi, 2013). There was also a comprehension question with each item, whose answer determined whether the reader understood the item literally, or inferred the more plausible meaning. As predicted by the noisy-channel framework, participants were more likely to make plausible inferences when sentences were preceded by a supportive context than by a non-supportive context. | Sarah Nathaniel, Rachel Ryskin and Edward Gibson | Lobby |
248(abstract) | Forgetting effects due to Local Coherence in Hindi Speakers of head-final languages are assumed to be particularly good at making and maintaining predictions about upcoming material (Levy and Keller, 2013). We report a study in the head final language Hindi where speakers are unable to sustain the prediction of the matrix verb that was to be integrated with the head noun in the presence of a center-embedded relative clause. This forgetting effect is concomitant with local coherence effects (Tabor et al., 2004) – in an SPR+acceptability task, RTs are lower at the matrix verb in ungrammatical sentences with a locally coherent parse compared to those without local coherence. | Sakshi Bhatia and Samar Husain | Lobby |
250(abstract) | Referential Entropy influences Overspecification: Evidence from Production In a visually-situated language production experiment, we investigated the possibility that referential overspecifications are driven by information-theoretic factors like entropy reduction. That is, speakers may include “redundant” information in order to help restrict the search space, thus reducing cognitive effort for themselves and/or the comprehender. We manipulated whether the redundant or the necessary adjective reduced entropy more. Preliminary results (1) replicate a well-established speaker bias to overspecify for colour, and (2) indicate that speakers used overspecification as a rational strategy, when the redundant adjective reduced entropy more than the necessary adjective. | Elli Tourtouri, Les Sikos and Matthew Crocker | Lobby |
251(abstract) | If I were you: Inducing Non-default Interpretations of Perspective-sensitive Items in the Visual World Some expressions of natural languages depend on a perspective for their interpretation (e.g., relative locative expressions like left). The default is that the speaker constitutes the perspectival center, but this default is defeasible. In a visual world eye-tracking study, we investigated two factors likely to affect a shift of this perspectival center: mode of presentation of a sentence (speech vs. thought), and form of utterance. Here, we pitted imperatives against counteridenticals (‘If I were you/he...’) which arguably constitute an extreme case of perspective-shifter. We provide evidence for the shifting effect of both, the counteridentical and speech conditions. | Johanna Klages, Elsi Kaiser, Anke Holler and Thomas Weskott | Lobby |
253(abstract) | Referential production in aging: A multiple-measures study The current study explored potential differences in referential production in older and younger adults. We used a wide variety of measures to provide a comprehensive perspective on adults' performance, and referents were everyday objects presented in either contrastive or non-contrastive contexts. The findings demonstrated clear age-related differences in referential descriptions, with the most notable effects involving older adults' greater range in modifier choice and increased tendency to provide redundant information. We suggest the observed patterns may be best explained in terms of differences in inhibitory control during lexical selection—a hypothesis that can be explored more directly in future research. | Raheleh Saryazdi, Julie Bannon and Craig Chambers | Conference Room A |
254(abstract) | Syntax predicts, semantics revises: filler-gap dependencies into adjunct clauses We explore the effect of semantics on the processing of filler-gap dependencies into adjunct clauses. In Experiment 1, we show filler-gap dependencies may cross into non-finite adjunct clauses. In Experiment 2, we find a reverse plausibility mismatch effect (i.e., plausibility increases processing difficulty), if the FGD plausibly resolves with a gap in a type-compatible adjunct clause. We argue that this demonstrates no gap is initially predicted in the adjunct clause. However, if the semantics of the sentence permits, the comprehender will reanalyze the sentence to resolve the FGD into the adjunct clause, increasing processing time. | Annika Kohrt, Trey Sorensen and Dustin Chacón | Conference Room A |
257(abstract) | Tracking quantifier scope: EEG evidence for active representational updating Keeping track of QP scope is important during incremental sentence comprehension because a QP’s scope determines where and when it can exert its semantic influence. However, the exact mechanisms by which the incremental parser manages QP scope information are unknown. We sought to determine whether there was any evidence for active update procedures hypothesized to underlie scope tracking online. We found a P3a component coincides with words that signal the end of a QP’s scope domain. | Dave Kush and Ashley Lewis | Conference Room A |
258(abstract) | Selective effects of case marker deletion on sentence processing in Korean The purpose of this paper is to examine the effects of case marking deletion positions (subject vs. non-subject position), word-order canonicity, and the syntactic structure (passive vs. active comparisons) on sentence processing in a verb-final language. A total of 35 Korean-monolingual individuals participated in this study. Korean speakers with a free word order language demonstrated strong preference toward the canonical word order presenting the first-NP-as-an-agent-strategy. The non-subject-CM deletion elicited worse performance than subject-CM deletion in passive sentences, indicating that case marking deletion for the agent-NP has greater impacts on sentences processing, especially when the syntactic structure becomes more complex. | Jee Eun Sung and Seunghun Lee | Conference Room A |
264(abstract) | An ERP study on the interaction of humanness and adjective agreement in Arabic In Modern Standard Arabic (MSA), plural nouns can be of two types, namely sound plurals and broken plurals. Broken plural nouns referring to human entities trigger full agreement in number and gender, whereas those referring to non-human entities, even if they are grammatically masculine, trigger singular feminine agreement, called deflected agreement. In this ERP study, we employed noun-adjective agreement in simple intransitive sentences to investigate how the processing system resolves this. Results suggest that the spoken variety of our participants influences how they process MSA online, and humanness interacts with morphosyntax both at the early and late stages of processing. | Ali Idrissi, R. Muralikrishnan, Hagar Hussain, Aya Mohamed, Tariq Khwaileh and Eiman Mustafawi | Conference Room A |
265(abstract) | An investigation of cue-based anticipatory processing in low constraint sentences using ERPs This study uses ERPs to examine the extent to which phonological information and grammatical number information are used to anticipate nouns in English. We further examine the role that task difficulty plays in anticipatory language processing. Results find that phonological information conveyed by the determiners a/an is reliably used to anticipate impending nouns, but only when predicting is relatively easy. Number marking on demonstrative (this/these) and on copular 'be' (is/are) was not used to anticipate, regardless of task difficulty. Findings indicate that the extent to which comprehenders anticipate may depend on the ease of anticipating and on cue salience. | Kailen Shantz and Darren Tanner | Conference Room A |
266(abstract) | Combined ERP-pupillometry measures differentially reveal effects of facilitation and effort in the comprehension of meaning shift Shifts in meaning, such as in the rosé has paid, typically involve processing costs due to the transfer of reference, but supportive context may reduce these costs. ERP research showed that the N400 reflects facilitation through contextual pre-activation/expectation, while the late positivity reflects the integration of an unexpected meaning shift. We investigated reference shifts by using combined ERP-pupillometry measures. Our results suggest that the N400 reflects facilitation due to pre-activation of conceptual features. In the absence of clear integration effects in the ERP signal, pupil size is associated with effort in the integration of the pre-activated features in sentence context. | Franziska Kretzschmar and Andreas Brocher | Conference Room A |
268(abstract) | Attentional cueing in non-canonical OVS sentences in German The present EEG study investigated whether changes in word order in German (SVO vs. OVS) function as attention cueing and modulate activity in the alpha frequency band. Participants read short stories with i) inferred and given/repeated information and which ii) varied in word order. The results show an alpha power decrease for inferred vs. given target words in sentence initial position in OVS sentences; no such effect was found in SVO sentences. We interpret this finding as evidence for an increased attention to the contrast between inferred and given noun phrases in OVS structures in German. | Diana Dimitrova and Petra B. Schumacher | Conference Room A |
270(abstract) | Where *is the jeans? Adapting to L2 grammatical errors Listeners adapt to regularities in their interlocutor’s grammar (Fine et al., 2013; Fraundorf & Jaeger, 2016). However, when such information is incorrect and unreliable, listeners no longer use it to facilitate processing (van Heugten & Christophe, 2013). In this study, we investigated whether listeners could adapt to grammatical errors that were consistent and reliable. This was done in a visual world study looking at predictive looks to pluralia tantum nouns with singular verb agreement (e.g., Where *is the pants?). Results show that listeners learned item specific regularities in verb agreement, and could even generalize to previously unmentioned pluralia tantum nouns. | Thomas St. Pierre and Jean-Pierre Koenig | Conference Room A |
272(abstract) | Integrating Listener and Speaker Characteristics into the Coordinated Interplay Account We argue for the integration of socially interpreted context (including speaker information) and comprehender characteristics into real-time language processing accounts. Extant real-time processing accounts are underspecified regarding the integration of social (speaker) information and comprehender characteristics. We extend the Coordinated Interplay Account (CIA, Knoeferle & Crocker, 2006; Knoeferle et al., 2014) and demonstrate how the extended social Coordinated Interplay Account (sCIA, Münster & Knoeferle, 2018) accommodates recent findings (i.e., Carminati and Knoeferle, 2013; Van Berkum et al., 2008). Integrating social information into the CIA is the first step toward acknowledging the impact of “the social” on language comprehension. | Katja Münster and Pia Knoeferle | Conference Room A |
274(abstract) | Loss aversion bias is affected by L2 proficiency, not by more rational behavior in the L2 L2 reasoning has been argued to be more rational than L1 reasoning because of a reduction in emotionality associated with the L2. While we do not replicate the finding that L2 users in general are “more rational” in their decision making than native speakers, we find that greater proficiency is associated with a stronger loss aversion bias, suggesting that any difference between L1 and L2 speakers is not due to less dominant heuristics in L2 reasoning, but rather a result of subtle differences between native and non-native interpretations: a decision that is irrational under one interpretation is rational under another. | Zoe Schlueter, Chris Cummins and Antonella Sorace | Conference Room A |
276(abstract) | Complexity effects in A- and A'-dependencies We examine the effect of complex fillers on acceptability of filler-gap dependencies. Filler complexity is known to improve acceptability in island environments, but working memory accounts of this effect (Kluender 1998; Hofmeister & Vasishth 2014) predict that the amelioration should occur very broadly: with gaps in all locations (not just in islands), with many types of complex fillers (not just “D-linked” ones), and with many dependency types (not just filler-gap (A’) dependencies). We present new evidence that all three predictions are correct, suggesting that the filler complexity effect is more pervasive than previously thought, in accord with working memory accounts. | Dayoung Kim and Grant Goodall | Conference Room A |
279(abstract) | This is the structure that we wonder why anyone produces it: Resumptive pronouns in English hinder comprehension English resumptive pronouns (RPs; e.g.,"...the senator that we knew why HE resigned") are enigmatic: despite being unacceptable, speakers regularly produce them. Recent evidence shows that, relative to gaps ("...the senator that we knew why _ resigned"), RPs are rated more comprehensible and lead to decreased reading times. These findings were taken as evidence that RPs facilitate comprehension. In three experiments we show that RPs are interpreted less correctly than gaps. Thus, contrary to the facilitation hypothesis, RPs hinder English sentence comprehension. This underscores the need for caution in studies which assess processing difficulty by measuring reading times or collecting ratings. | Adam Morgan, Titus von der Malsburg, Victor Ferreira and Eva Wittenberg | Conference Room A |
281(abstract) | Children fail to revise their syntactic analysis, even when their initial interpretation is driven by prosodic cues Three experiments investigated how children and adults interpret an ambiguous noun-verb homophone (e.g., flies) when sentences began with prosodic information supporting either a noun or verb interpretation, but ended with lexical content supporting the opposite interpretation. The results showed that all participants rely heavily on the initial information provided by prosody to constrain their parsing. However, contrary to adults, children do not recover from their initial commitment: the later-arriving lexical content did not override children’s initial interpretation. This suggests that prosody is not treated as an inferior source of information, to be ignored in case of conflict. | Alex de Carvalho, John Trueswell and Anne Christophe | Conference Room A |
283(abstract) | The breadth of phonological interference in word choice Interactive models of language production predict that phonological properties can influence word choice, but evidence is largely limited to error data and sentence production. In a picture naming task, we presented pictures with clear dominant and secondary names (coat / jacket) and manipulated the phonological overlap of the preceding word (code / jackal / walrus). Proportion of dominant and secondary word choice was significantly affected by phonological interference (Study 1), and the effect held across a wide range of dominant name agreement (Study 2), providing novel evidence that the interactive nature of lexical access bears directly on word choice. | Mark Koranda and Maryellen MacDonald | Conference Room A |
284(abstract) | Learning absolute adjective meanings through non-absolute exemplars In our experiment, we ask whether learners infer meanings of novel adjectives by attributing observed visual variability to contextual explanations (e.g., 90% full cup is sufficiently full or it will spill). Adults exposed to contextual explanations were more likely to infer a novel adjective with a maximum standard, suggesting they extrapolated a maximum standard by explaining away contextually-conditioned visual variability. We also examined the influence of prosody and absolute exemplars in Exposure, finding that absolute exemplars, not prosody manipulations, were necessary when inferring novel absolute meaning. This demonstrates the intricate relationships between inferential and associative-based mechanisms underlying word learning. | Crystal Lee and Chigusa Kurumada | Conference Room A |
285(abstract) | Exploring heterogeneous P600 satiation in an attempt to reveal N400s in semantic P600 sentences Semantically anomalous yet syntactically well-formed sentences such as "The hearty meal was devouring the kids." have been shown to elicit P600, rather than N400, ERP effects. Our goal is to test whether these sentences elicit an N400 effect in addition to a P600 (which might be obscuring these N400s). We use an ERP-based satiation methodology to try to specifically satiate only the P600 effect in these sentences. We found that distinct syntactic violations do not satiate each others' P600s and that the P600 effects to several syntactic violations are distinct from each other in latency and scalp distribution. | Emma Nguyen and Jon Sprouse | Conference Room A |
286(abstract) | Effects of production on predictive processing in young and older adults The present study used a self-paced reading task, either without a production task, or with interleaved naming trials in which the target word was low- or high-frequency to investigate the effects of production task on predictive processing with young and older adults. We expected that the addition of a production task and successful naming in the high-frequency trials would make readers anticipate upcoming words to a larger extent. Consistent with previous research, young adults anticipated upcoming words more actively with a concurrent production task. Older adults in our study predicted during reading regardless of the engagement of the production system. | Aleuna Lee and Edith Kaan | Conference Room A |
288(abstract) | Predictive processing and reliability in the influence of prosody in L2 structural analysis This study investigated L2ers’ anticipatory use of prosody in processing globally ambiguous sentences in English, and whether structural analysis in L2 is affected by the reliability of prosodic cues. The results from two visual-world eye-tracking experiments, in which filler items had either reliable or unreliable prosody, showed L2ers use prosody in structural analysis but not as early as native speakers. The results also showed that the reliability effect was weak with L2ers, suggesting the degree to which listeners weigh prosodic information according to the reliability is related to the listeners’ statistical knowledge about the mapping between prosody and incoming input. | Chie Nakamura, Jesse Harris, Sun-Ah Jun and Yuki Hirose | Conference Room A |
289(abstract) | Gender Effects on Gender Ambiguous Anaphor Resolution Anaphor resolution depends on coordinating different types of constraints. These constraints often co-occur making it difficult to tease apart their effects. Chinese provides a good test bed in that pronouns/reflexives are homophones with gender represented only in the written form. Exp1&2 examined people's sensitivities to structural and gender cues during Chinese anaphor resolution when both cues were presented. Exp3 examines how constraints are combined when the gender cue is ambiguous in spoken form. When the critical gender cue was unavailable, instead of using the structural cue only, people showed an increased uncertainty in resolving anaphors. | Yuhang Xu, Yedan Tian, Jeffrey Runner and Mike Tanenhaus | Conference Room A |
291(abstract) | Case and misinterpretation in number attraction effects in Eastern Armenian We report three experiments on number attraction effects in Eastern Armenian. Using self-paced reading paradigms combined with comprehension questions targeting the interpretation of the subject-verb dependency, we show: (1) that attraction occurs in Armenian; (2) that attraction causes misinterpretation of the subject-verb dependency, either by misleading participants to interpret the attractor as the subject of the verb or by overwriting the original number of the subject’s head. Furthermore, using a case manipulation in a forced-choice paradigm we demonstrate that an overt (accusative) case marker on the attractor ameliorates attraction rates. | Serine Avetisyan, Sol Lago and Shravan Vasishth | Conference Room A |
294(abstract) | Gender agreement and semantic processing in French children: ERP effects of age and proficiency We present event-related potential (ERP) responses to gender and semantic anomalies in children and adults, and correlated them with grammatical abilities. Paricipants listened to sentences describing visual scenes. Gender agreement errors (‘a shoe.M brown.F’) and visual-semantic errors were created (e.g., a brown hat). Linear regressions were run on peak-to-peak amplitudes with effects condition, and variables age and grammatical abilities. Adults elicited a N400+P600 for semantic mismatches and a LAN-P600 for gender errors. Children showed N400+P600s for both, suggesting that they may rely on lexical-semantic cues to process agreement. Response magnitude and distribution was positively correlated with grammatical abilities. | Emilie Courteau, Lauren Fromont, Phaedra Royle and Karsten Steinhauer | Conference Room A |
295(abstract) | Be timely: how turn-taking gaps constrain the integration of alternatives in implicatures A novel overhearing paradigm tested listeners' integration of Questions Under Discussion (QUDs) into implicature processing. The position of scalar terms varied (initial and final) as well as the length of turn gap to test the incremental integration alternatives. Implicature rates were at chance for initial and final positions at early (250ms) and only at medium gaps (750ms) did rates part from chance. At later intervals implicature rates dropped below chance, suggesting longer gaps are understood as hesitations. Mouse-path data showed similar implicature patterns as found in Tomlinson et al (2013) at all intervals, albeit only for some-initial positions. | John Michael Tomlinson, Jr. | Conference Room A |
296(abstract) | Verb class is early used during the processing of subject-verb agreement (in Italian) We studied ERPs correlates of subject-verb number agreement mismatch in Italian by manipulating the class of the verb: unergative vs transitive. Our hypothesis was that transitive verbs could elicit no LAN since Italian allows for preverbal objects and thus the mismatch in number between a verb and the preceding DP does not necessary imply an outright violation. Results were at odd with this prediction, in that a larger LAN/N400 (and smaller P600) was found for transitive verb than for unergative ones, leaving open the question about the large variability in the negativities elicited by subject-verb agreement violations in Italian. | Francesco Vespignani, Emma Bergamini and Nicoletta Biondo | Conference Room A |
300(abstract) | How individual difference measures inform event processing in monolingual and early and late bilingual speakers We assessed how executive function ability predicted event comprehension performance in monolinguals and bilinguals. Participants listened to sentences (The bride will unwrap/accept the gift.) while viewing images depicting the initial and end state (unwrapped/wrapped gift) and completed AX-CPT and pro-saccade tasks. Greater inhibitory control in individuals with native proficiency (monolinguals/early bilinguals) predicted better anticipation. In contrast, inhibitory control in late learners of English did not influence anticipation; rather, greater attentional skills predicted better anticipation. This suggests that, during sentence processing, late learners and native speakers rely on differing strategies to comprehend events. | Kyra Krass, Gitte Joergensen, Cloe Zeiden, Megan Zirnstein, Gerry Altmann and Eleonora Rossi | Conference Room A |
301(abstract) | Planning scope in British Sign Language (BSL) sentence production We investigate the timing of conceptual and grammatical encoding processes for production in British Sign Language. For spoken languages, planning scope prior to sentence onset appears to be at the phrasal level, not lexical or clausal levels (Smith&Wheeldon, 1999). Production latencies were measured for matched-length sentences beginning with either a simple subject (e.g., boy) or coordinate subject (e.g., boy and girl). Similar to spoken language, BSL signers were significantly slower to begin producing the initial clause for sentences beginning with a coordinated NP. Results suggest planning is amodal and not influenced by differences in motor output speed (slower for signs). | Robin Thompson, Katrien Segaert and Linda Wheeldon | Conference Room A |
303(abstract) | What is dropped in Topic Drop? – A rating study on the relationship between Topicality and Topic Drop in German We present the results of an acceptability rating study investigating the eponymous effect of topicality on Topic Drop, i.e. the omission of a preverbal constituent, in German. 48 participants rated the acceptability of items presented as text message dialogues on a 7-point Likert scale. On the one hand, the data reveal no significant interaction between topicality and Topic Drop questioning the impact of topicality for the omission. On the other hand, a significant interaction between Topic Drop and person shows that Topic Drop is less acceptable with 3rd person compared to 1st person confirming previous corpus findings. | Lisa Schäfer, Robin Lemke and Ingo Reich | Conference Room A |
304(abstract) | The role of context predictiveness in younger and older adults: EEG coherence in alpha and theta frequency bands We investigated differences in predictive preactivation of lexical representations and their subsequent verification/violation in older and younger participants using a picture-word matching task. We examined alpha and theta coherence before the presentation of the target word (predictive vs non-predictive), and N400 ERP post word onset (match vs mismatch). No coherence differences were found between participant groups. However, older participants showed a N400 effect in both Predictive and Non-predictive items. This may indicate that both groups share similar prediction processes, but that older adults may be more affected by context effects resulting to prediction errors even when context is less predictive. | Spyridoula Cheimariou | Conference Room A |
305(abstract) | Systematic variability in language related ERPs: Distribution, effect size, and latency As part of a large-scale effort to examine individual differences in language-related event-related potentials (ERPs), we examined variability in ERP patterns in response to semantically and syntactically anomalous words during sentence reading. From the initial sample of 159 participants, we randomly sub-sampled groups of 20, 30, and 40 participants (1000 per each sub-sample size) in order to estimate effect size of classic N400 and P600 effects. P600 effects were smaller and more variable in onset and scalp distribution. In addition, we qualitatively assess important response characteristics and suggest important avenues for future descriptive and explanatory work in this space. | Shannon McKnight, Akira Miyake, Donald Bell-Souder and Albert Kim | Conference Room A |
308(abstract) | Don’t be uninformative! – Experimental studies on information theory and fragment usage Fragments (“Your ticket, please!”) can communicate the same meaning as a full sentences (“Show me your ticket, please!”) do. But why and when do we use fragments? We hypothesize that, following Levy & Jaeger's (2007) Uniform Information Density hypothesis, specifically those parts of the utterance which are predictable are omitted. We addressed this with a rating study that shows that fragments are relatively preferred if they refer to an event which is likely given the previous ones in a script-based context. The rating study is complemented with a production study that confirms our utterance likelihood estimates. | Robin Lemke, Lisa Schäfer and Ingo Reich | Conference Room A |
309(abstract) | Shared mechanism underlying non-embedded and embedded enrichments: evidence from enrichment priming Recent studies have shown that the enrichment of ‘some’ to ‘some but not all’ can occur in the scope of non-upward monotone quantifiers. By adopting enrichment-priming paradigm, we investigated whether there is evidence for a shared mechanism underlying non-embedded and embedded enrichments. | Chao Sun and Richard Breheny | Conference Room A |
310(abstract) | Coreference and discourse focus in broad-coverage stimuli Previous work argues that discourse prominence facilitates coreference resolution. However, these studies use constructed stimuli with specific syntax (e.g. cleft constructions) which could have frequency confounds. This work explores the generalizability of a discourse prominence effect on coreference resolution in a broad-coverage naturalistic analysis. The current work proposes and evaluates new estimators of prominence as predictors of reading behavior in a corpus of self-paced reading times. While distance-based measures are not predictive, MentionCount (a measure of thematization) is a significant predictor in this naturalistic setting, providing evidence for the generalizability of discourse prominence effects to broad-coverage stimuli. | Evan Jaffe, Cory Shain and William Schuler | Conference Room A |
312(abstract) | U-shaped development in children’s discourse bootstrapping How does children’s developing ability to process discourse affect their word learning? Two experiments with children (aged 2 to 8) compared use of the connectives so and but for so-called discourse bootstrapping (e.g., “Katy was cold, so she wore a dax” versus “Katy was cold, but she wore a dax”). We found a striking pattern of U-shaped development: Accuracy at using inferential “so” improved monotonically, but accuracy using contrastive “but” declined from 2 to 6 before improving. We suggest that this U-shaped development results from children’s improving skills at prediction, which surprisingly cause a developmental disadvantage at word learning. | Hugh Rabagliati, Nora Wolf, Barbora Skarabela and Hannah Rohde | Conference Room B |
315(abstract) | Two mechanisms underlying agreement errors The study explores the causes of attraction errors by combining a verb selection task and a comprehension task. Results show that although more verb selection errors occur with number mismatch between the agreement head and the attractor, more comprehension errors occur with number match. Finer analyses suggest that agreement errors arise both because of incorrect structure building (sometimes, the wrong NP controls agreement) and because of feature contamination (the distractor’s feature is transferred, though the correct structure is built), a view which is compatible with two principles of self-organized sentence processing: similarity-based interference at encoding and feature passing. | Samuel Schmid, Sandra Villata, Whitney Tabor and julie franck | Conference Room B |
316(abstract) | What do eye movements in the visual world reflect? A case study from adjectives A common dependent measure used in visual-world eye-tracking experiments is the proportion of looks to a visually depicted object in a certain time window after the onset of the critical stimulus. We supplement an eye-tracking dataset with an offline incremental decision task to measure participants’ beliefs about the intended referent at various points in the unfolding sentence and and assess to what extent these beliefs predict the eye-tracking data. The results suggest that belief about the referent is only one factor that affects eye movements in referential tasks and we discuss the methodological implications. | Ciyang Qing, Daniel Lassiter and Judith Degen | Conference Room B |
217(abstract) | Irony processing in adolescents: an ERP study Irony comprehension requires complex inferences about a speaker’s mind. Behavioural data indicates that while children between ages 6-10 begin to show some appreciation of the communicative intent in an ironic remark, full appreciation of speaker attitude might be developing beyond middle childhood and into adolescence. Additionally, recent findings suggest that the online use of Theory of Mind (ToM) shows a prolonged development through late childhood and adolescence. We provide ERP evidence that suggests that the two age-groups process irony differently and engage different cognitive mechanisms. This is the first study to investigate figurative language comprehension in adolescents through ERPs. | Irene Symeonidou, Iroise Dumontheil, Wing Yee Chow and Richard Breheny | Conference Room B |
317(abstract) | Evidence for multiple memory representations from discourse anaphora and ellipsis Sentence processing models which are based on ACT-R represent memory as a unified store from which linguistic material is retrieved. However, it may be that different types of retrieval use distinct representations: discourse anaphors find antecedents in the discourse representation, while verb phrase ellipsis finds antecedents in the syntactic representation. We present evidence of different distance effects for discourse anaphora and VP ellipsis in two experiments. We take these results as a challenge to models which propose a single unified memory store. In contrast, a proposal where linguistic memory storage consists of multiple representations seems to make the correct predictions. | John Duff and Lyn Frazier | Conference Room B |
323(abstract) | Pronouns at the right frontier: discourse structure affects accessibility of final appositives This study tested the predictions of two approaches to the findings from Syrett & Koev (2015) on the increased accessibility of sentence-final (compared to sentence-medial) appositive relative clauses (ARCs): the speech act account by Frazier et al. (2017) and the discourse structure account by Jasinskaja (2016) and Hunter & Asher (2016). Both accounts were supported in that both RC-type and the discourse relation between matrix clause and RC affected the accessibility of final ARCs for propositional and individual anaphors. Moreover, the two anaphor types behaved differently in a way, which is surprising for both accounts and raises further questions. | Alexander Göbel | Conference Room B |
325(abstract) | Filling the gap in gap-filling: long-distance dependency processing in sentence production In long-distance dependency processing, it is widely known that comprehenders fill the gap before the gap becomes apparent in the bottom-up input (Crain & Fodor, 1985; Stowe, 1986) and even before encountering gap-hosting verbs (Omaki et al., 2016). In contrast to the gap- filling process in sentence comprehension, the gap-filling process in sentence production is poorly understood. Here we show that speakers plan the gap before the production onset of wh-questions, and that this gap-planning precedes the planning of gap-hosting verbs. We suggest that long-distance dependency processing in comprehension and production share the same active, top-down structure building process. | Shota Momma and Victor Ferreira | Conference Room B |
326(abstract) | Structural Priming and ERP Language processing strongly relies on predictive mechanisms which have been shown to adapt to reliable cues. We demonstrate syntactic adaptation to a complex, but highly reliable, cue (combination of a specific verb class with a specific tense marking). Since adaptation is methodologically relevant for methods that require large number of trials (e.g. ERP) we present a follow up study designed to attenuate reliability of the cue. | Bruno Fernandes, Artemis Alexiadou, Wing-Yee Chow, Andrea Santi and Nino Grillo | Conference Room B |
329(abstract) | Animacy and possession type guide interpretation of ambiguous VP ellipsis "Bill washed his car, and John did, too." has two possible interpretations if the overt "his" represents Bill: coreferential (John washed Bill's car) or bound variable (John washed his own car). What guides comprehenders' selection of one over the other? We conducted an experiment testing how four possession relations (inalienable, ownership, animate relational, and kinship) modulate interpretational preference. Our results indicate that the overt possession's animacy is an important factor in the resolution of the elided possessive pronoun. We claim that animate possessions are more likely than inanimates to receive independent discourse representations and consequently be available for coreferential interpretation. | Jesse Storbeck and Elsi Kaiser | Conference Room B |
330(abstract) | Exploring? Adjective? Ordering Preferences via Artificial Language Learning Adjectives are subject to cross-linguistically robust ordering preferences. Recent research showed that adjective ordering is predicted by subjectivity: the more subjective an adjective, the farther away from the noun it occurs. To test whether speakers productively apply a semantic generalization, we conducted an artificial language learning study. Participants learned two Alien adjectives whose meanings differed in subjectivity. They then indicated their preference among the two possible Alien adjective orderings. Results show that higher ratings were given when the subjective adjective preceded the objective one. This suggests that artificial language learning is a promising avenue for studying adjective ordering preferences. | Michael Hahn, Richard Futrell and Judith Degen | Conference Room B |
331(abstract) | A wandering mind is not a good comprehending mind- Evidence from brain oscillations We developed a method to 1) detect the mind wandering moments during reading by performing the Spectral Similarity Analysis (SSA) between 5-minutes resting state qEEG and 5-minutes qEEG during reading, and 2) investigated how the mind wandering ratio during reading relates to comprehension. The SSA examined the spectral similarity between a given segment of EEG data and a spectrogram segment of resting state qEEG. We found that the mind wandering ratio at O1 significantly predicted reading comprehension, with a higher ratio relating to worse comprehension. This novel method allows researchers to detect mind wandering moments without interrupting natural reading process. | Peiyun Zhou, Chantel Prat, Brianna Yamasaki and Andrea Stocco | Conference Room B |
333(abstract) | Neural correlates for individual differences in processing ‘some’ scalar implicatures The current study tries to reconcile the seemingly contradictory outcomes of three different studies trying to account for the individual differences in judgement of underinformative scalar implicatures (USI) (Noveck & Posada, 2003; Nieuwland et al., 2010; Antoniou et al., 2016), by trying to correlate USI behavior with autistic-like traits. Preliminary results show that correlations with AQ score are lower than previously found, but participants with high and low AQ scores differ in effect size. An unsupervised learning algorithm found that the averaged waveforms do not represent significant individual differences, which are not account for by AQ and require further research. | Maria Goldshtein and Darren Tanner | Conference Room B |
334(abstract) | Complex sentence planning in L1 & L2 Turkish The current elicited production study set out to explore to what extent a complex sentence with an embedded clause is planned in advance and whether this planning is verb-centered during first (L1) and second language (L2) production of a strongly head-final language like Turkish. Our data suggest that both speaker groups may have a tendency to initiate their utterance with a simple structure. Yet, our results further suggest that the degree of advance sentence planning might be moderated by the case-assignment properties of the matrix verb, and interestingly, that this modulation has an opposite direction in L1 and L2 speakers. | Nur Basak Karatas, Kira Gor, Robert Slevc and Ellen Lau | Conference Room B |
336(abstract) | Relational versus Plural Concepts: The Role of the Left Angular Gyrus The left Angular Gyrus (lAG) plays a role in relational (Williams et al.2017;Thompson et al.2007) and quantity processing (Domahs et al.2014); relational concepts (‘mother’) and plural concepts ('ladies') activate the lAG more than singular non-relational ones ('lady'), respectively. Does a single function unify both findings? We present singular and plural relational (sister/sisters) and non-relational (lady/ladies) nouns in isolation. Analyzing MEG data with a 2(relational/non-relational) x 2(bare/plural-marked) x 3(ambig/nouny/verby) ANOVA isolates an interaction of relationality with category, 190-320ms after target onset. No effect of plural marking was discovered; the hypothesis that relationality and plurality share a single function was not supported. | Adina Williams and Liina Pylkkanen | Conference Room B |
337(abstract) | Children’s N400 is sensitive to both predictability and frequency: evidence from natural listening The adult N400 sensitively indexes ease of lexical activation and semantic integration. Using a natural listening paradigm, we explore whether children’s N400s are similarly sensitive to frequency and sentence/discourse level constraints. Adults and children (5-10) listen to a story as ERPs time-locked to the onset of every word are recorded. Content words are coded for frequency and four different measures of predictability. Our findings suggest that both adults and children use top-down constraints from semantic/syntactic/pragmatic levels during lexical processing. In adults, these top-down constraints dominate over frequency. In children, lower level information like frequency continues to play a critical role. | Tanya Levari and Jesse Snedeker | Conference Room B |
338(abstract) | Better to be reliable than early: Cognitive-control effects on developmental parsing Children parse sentences word-by-word, which can lead to interpretation errors when initial misanalyses conflict with later-arriving evidence. Such challenges have been attributed to immature cognitive-control, though it remains unclear how exactly they are linked. We manipulated cognitive-control engagement via interleaved Stroop trials (congruent/incongruent) before Sentence trials (Exp1: early-arriving verbs, Exp2: later arriving verbs). We found that Incongruent Stroops caused children to more readily interpret sentences according to information gleaned from the verbs, regardless of whether verbs were prediction or revision cues. This suggests that cognitive-control fulfills the very general function of increasing reliance on reliable parsing cues like verb-specific biases. | Zoe Ovans, Jared Novick and Yi Ting Huang | Conference Room B |
339(abstract) | How accessible in the discourse model are implicit arguments? An investigation using sluicing In this work, we take a closer look at the discourse status of implicit arguments (e.g., the implicit object in “Lisa was reading Ø last night”) and investigate (i) IAs’ accessibility as antecedents and (ii) context effects on IA interpretation by looking at d-linked sluicing constructions with overt and implicit antecedents (e.g. “Lisa was reading {a book} last night, but I’m not sure {which one/which book}). In an acceptability judgement task (n=45, 5.scale), results reveal that implicit objects are as acceptable antecedents as overt objects, but only when the retrieval cue is informative (‘which book’). | Ana Besserman and Elsi Kaiser | Conference Room B |
349(abstract) | Now You See it, Now You Don’t: Visual Cues Modulate Short vs Long Passive Production The syntactic representation of (i) long (‘The cat was brushed by the girl.’) versus (ii) short (‘The cat was brushed.’) passives is an open question. Recent work by Messenger et al. (2010) argues in favor of shared structure. However, their results could also be linked to a task-based bias to mention all elements in the images used to induce passives. We present two priming studies testing this possibility. In line with the mention-all bias, we find that the presence/absence of the to-be-described agent image plays an important role in speakers’ choice of short versus long passives. | Monica Do and Elsi Kaiser | Conference Room B |
351(abstract) | Does attraction lead to systematic misinterpretation of NP number? Probably not. Three experiments (two self-paced reading, one RSVP followed by a speeded comprehension question) investigated whether agreement attraction within complex noun phrases influences comprehenders' eventual interpretation of the head noun of the noun phrase (e.g., after reading "…the key to the cabinets…" do readers conceptualize more than one key?). Results showed no evidence for misrepresentation of number during discourse reading (more naturalistic comprehension), whereas direct probe questions (metalinguistic task) about noun number showed some evidence for misplaced plural features. We argue that these results are consistent with mis-binding of plural features to lexical hosts, but not systematic number misrepresentations. | Darren Tanner, Jack Dempsey and Kiel Christianson | Conference Room B |
352(abstract) | Effects of discourse givenness on children’s parsing of heavy-NP-shifted sentences Discourse structure influences syntactic choices in production and expectations about upcoming constituents in comprehension. Prior studies have shown that children can use discourse structure to interpret ambiguous phrases such as pronouns. We asked whether children also expect discourse givenness to affect the ordering of post-verbal arguments. Using a visual-world-paradigm, we found that five- and six-year-olds, like adults in Arnold & Lao (2008), expected discourse-given referents early in un-shifted sentences and discourse-new referents later in heavy-NP-shifted sentences. Our results suggest that children can link information structure to syntactic ordering. | Yi Lin and Cynthia Fisher | Conference Room B |
355(abstract) | Preschoolers’ partner-specific online processing of disfluency We examine whether children flexibly interpret disfluency to suit a current partner’s knowledge, testing 4-year-olds in a task involving live conversation with two different partners, each of whom shared different knowledge with the child. We measured children’s eye-gaze when they heard speakers’ fluent or disfluent instructions. For disfluent expressions, children looked more at the unfamiliar than the familiar object (from the speaker’s view) during the initial processing of the disfluency. Our findings show that children can represent two different partners’ knowledge states and use this information accordingly when interpreting a partner’s disfluency in a live conversation. | Si On Yoon, Kyong-sun Jin, Sarah Brown-Schmidt and Fisher Cynthia | Conference Room B |
356(abstract) | How broad is the effect of reading exposure on language processing? Perhaps readers are more practiced at maintaining a representation of information structure, and thus show greater sensitivity to accenting (i.e. variation in pitch, and word duration) during spoken language comprehension. Participants heard sentence pairs, one sentence established object given-ness and a second was gated at either an accented or de-accented word. Participants were asked to select the object to which the gated sentence referred. Participants were then evaluated on print exposure. Listeners chose the given object more when they heard the de-accented word (p < .0001). However, print exposure did not account for variation in accenting biases (p = .7146). | Elyce Williams and Jennifer Arnold | Conference Room B |
373(abstract) | Sentence-by-sentence N400 variability during multimodal cue integration reflects prediction violation: Evidence from pitch accent and beat gesture This study investigated how trial-level variability in the N400 event-related potential reflects integration of beat gesture and pitch accent, which convey importance in the visual and auditory modalities, during online sentence processing. The results demonstrate increased N400 variability when pitch accent is present but beat gesture is absent, as well as when beat gesture is temporally misaligned with pitch accent. These findings suggest that N400 variability reflects violation of predictions based on syntactic structure for occurrence of cues to importance within individual sentences, indicating that pitch accent and beat gesture convey focus and that this is reflected in listeners' predictions. | Laura Morett, Nicole Landi, Julia Irwin and James McPartland | Conference Room B |
375(abstract) | Implicit gender biases in the production and comprehension of pronominal references Production and comprehension of language draw on a broad range of knowledge and beliefs, including general world knowledge, contextually variable information, and social pragmatics. All of these factors are exemplified in gender processing for role nouns. Von der Malsburg et al. (2017) found a bias in both production and comprehension against female pronouns in descriptions of the next president during the 2016 election cycle. Here, by studying the relationship between world knowledge and pronoun-use expectations for a wide variety of role nouns, we show that this bias generalizes quite broadly, at least in production. | Veronica Boyce, Titus von der Malsburg, Till Poppels and Roger Levy | Conference Room B |
393(abstract) | Real-time pragmatic processing with a novel lexicon When we expect to hear a scalar-modified noun phase (e.g., a large cup), do we pre-encode the entire phrase or process information as it unfolds? Using a novel combination of artificial word learning and a visual-world eye-tracking paradigms, we examine the effects of familiarity of head nouns on prenominal adjectives. Our results suggest that listeners can inferentially derive pragmatic interpretations of pre-nominal adjectives even when their head nouns are less readily available. It supports the ideas that 1) lexical items are processed as they unfold in the input and 2) pre-encoding of entire referential expressions may not be strictly necessary. | Bethany Gardner, Madeline Clark, Amanda Pogue and Chigusa Kurumada | Conference Room B |
399(abstract) | Contextual Search in Presupposition Satisfaction: Integrating Self-Paced Reading and ERP Data We investigate the online processing of copula variation in Spanish (ser/estar) across three dialects. Estar presupposes that the embedded predication holds contingently at utterance time (ser is neutral). Estar-licensing contexts are thus those that allow the predication to be construed as contingent by making accessible alternative situations at which the predication may be false. Findings: 1) Contextual search for presupposition satisfaction is reflected in longer reading times for estar with neutral contexts. 2) Presupposition satisfaction is reflected in negative deflection peaking at 400 ms after adjective. 3) Contextual search is reflected in late positivity between 600-800 ms after the adjective. | Sara Sanchez-Alonso, Maria Pinango and Ashwini Deo | Conference Room B |
403(abstract) | What is “good enough” for L2 predictive processing and incorporation? While predictive processing has gained recent interest in language processing work on second language (L2) predictive processing is relatively scarce. Previous work focuses largely on either integration or prediction in L2 processing, and so little is known about the relationship between the two in the L2 acquisition process. This study begins to connect the two by reporting results of the same group of (L1 English) L2 Spanish learners in both an “integration” and a “predictive” self-paced reading task. Learners show delayed effects of integration even though they show timely effects of prediction. | Gabrielle Klassen | Conference Room B | Session 4 | 2:00 PM - 6:15 PM |
Submission | Title | Authors | |
Chair: | Matthew Husband | ||
58(abstract) | The role of alternatives in the processing of affirmative and negative sentences The role of probabilistic prediction in language processing and its relation to the N400 has been a matter of debate in the ERP literature. We present results of ERP experiments where cloze probabilities of sentence completions are a function of the visual scenario that provides unique or multiple potential sentence continuations. We test the hypothesis that the N400 is larger if the context model allows for multiple true alternatives relative to the case when it only allows for a unique true sentence completion. The effect of the number of alternatives is compared for affirmative and negative sentences. | Maria Spychalska, Viviana Haase, Jarmo Kontinen and Markus Werning | 2:00 PM |
163(abstract) | The temporal dynamics of structure and content in the language network We performed a parallel fMRI and MEG experiment in the same subjects in order to identify the temporal dynamics of sentence processing in the brain. While the anterior temporal lobe and inferior frontal gyrus showed minimal effects of structure, the posterior temporal lobe showed increased activity for sentences at the end of the subject noun phrase, consistent with left-corner parsing models, and the temporo-parietal junction showed effects consistent with the incremental interpretation of event semantics. Our results help elucidate the dynamics of sentence processing in the brain, including novel support for the posterior temporal lobe in predictive lexical-syntactic processing. | William Matchin, Christian Brodbeck, Christopher Hammerly and Ellen Lau | 2:30 PM |
387(abstract) | Comprehending events on the fly: inhibition and selection during sentence processing Language comprehension entails keeping track of entities introduced into the discourse and of any changes they undergo as a consequence of events described by that language. In the present study, we used time-frequency analysis of EEG to examine processing of representations of the same object in different states (e.g., an onion before and after being chopped) as unfolding language selects one of the states. Increase in alpha power was observed when language referred back to substantially (compared to minimally) changed objects. Given the role of alpha oscillations in inhibition, we interpret this effect as indicating suppression of the irrelevant state. | Yanina Prystauka, Zachary Ekves and Gerry Altmann | 3:00 PM |
Coffee | 3:30 PM | ||
Chair: | Matthew Traxler | ||
Invited Speaker | Prediction, Production, and Memory | Colin Phillips | 3:50 PM |
249(abstract) | Prediction-inconsistent information leads to prediction updating – an ERP study on sentence comprehension. Many ERP studies show effects of predictions on words preceding the predictable word, e.g. on an adjective whose gender marking is inconsistent with the gender of the predictable noun. The present study tests a novel hypothesis whereby this prediction-inconsistency effect reflects prediction updating that has causal consequences for the N400 on the upcoming noun. By manipulating informativity of the adjective, we show there are two mechanisms of prediction updating: 1) reconfiguring activations of already predicted nouns, 2) activating previously unpredicted words. Both mechanisms have causal consequences for the N400 on the noun and are partially under strategic control of participants. | Jakub Szewczyk | 4:30 PM |
Break | 5:00 PM | ||
261(abstract) | N400 evidence for parallel lexical predictions We used ERPs to ask whether multiple probable candidate words are predicted in parallel in medium-constraint contexts. Participants read short story contexts that generated predictions for at least two probable words. These contexts were then continued with either the Best (i.e. most probable) or the Second Best (i.e. second most probable) word. We find a three- way dissociation of N400 amplitudes: Best < Second Best < Unpredicted, consistent with a parallel gradient account of prediction. | Emily Morgan, Nate Delaney-Busch, Minjae Kim, Lena Warnke, Eddie Wlotko and Gina Kuperberg | 5:15 PM |
380(abstract) | Competing predictions drive N400 sensitivity to argument role reversals Argument role reversals create strong cloze probability contrasts, but do not elicit N400 contrasts, possibly because the parser is unable to use argument role information without a verb. If so, providing a verb early in the context should allow role information to impact the N400. Our EEG experiment contrasted noun-noun-verb reversals with noun-verb-noun reversals. Surprisingly, we found an N400 contrast in both NNV and NVN contexts. Our stimuli were designed to elicit symmetrically strong, distinct verb predictions in canonical and reversed orders. Our data suggest that argument role information only results in N400 contrasts when arguments yield competing verb predictions. | Lara Ehrenhofer, Neomi Rao, Julia Buffinton, Ellen Lau and Colin Phillips | 5:45 PM |
Saturday March 17, 2018 | |||
Coffee | 8:00 AM - 9:00 AM | Session 5 | 9:00 AM - 12:00 PM |
Submission | Title | Authors | |
Chair: | Hugh Ragabliati | ||
108(abstract) | Neural correlates of attentional (re-)orientation: The case of German demonstratives Attentional orienting is a central aspect of communication. The focus of attention can be shifted from one referent to another and certain devices for attentional orienting are available to the speaker. We investigate two types of German demonstrative pronouns and hypothesize that these referential forms take priority in the brain and that attentional reorientation should be computationally demanding. Data from two ERP studies indicate that the German demonstrative pronoun ‘dieser’ is an important attention orienting device and that the N400 and late positivity reflect the neural correlates associated with its attention orienting functions. | Melanie Fuchs and Petra Schumacher | 9:00 AM |
191(abstract) | Person blocking in reflexive processing: when "I" matter more than "them" In two experiments, we show that English reflexives which mismatch their local antecedents in person are rated worse and read more slowly than reflexives which mismatch in number. However, we observed that locally person-mismatched reflexives were actually less likely to attend to feature-matched non-local referents than their number-mismatched counterparts, a finding apparently at odds with current models of antecedent resolution. We suggest instead that this behavior demonstrates a "person blocking" effect, consistent with cross-linguistic constraints which disallow non-local reflexive reference in the presence of a first/second person pronoun (e.g. in Mandarin). | Shayne Sloggett and Brian Dillon | 9:30 AM |
205(abstract) | Production and comprehension of referential expressions show divergent behavior Rational models of language communication commit to a critical premise that language comprehension is closely connected to production, at least at the computational level, since the speaker and the listener are assumed to be engaged in recursive pragmatic reasoning about each other’s communicative intentions [Goodman&Frank 2016]. It remains an open question, however, how comprehension and production can be linked mechanistically, with some behavioral data even challenging a listener-centered speaker model [Ferreira,2008]. The current study examines the comprehension and production of referential expressions, with novel findings showing divergences between the two systems that raise important questions for the radical rational view. | Ming Xiang, Allison Kramer, Timothy Leffel and Chris Kennedy | 10:00 AM |
Break | 10:30 AM | ||
Chair: | Zenzi Griffin | ||
Invited Speaker | Intra-sentential codeswitching in bilingual speakers: Electrophysiological approaches | Janet van Hell | 10:50 AM |
220(abstract) | Dissociable effects of prediction and integration on the N400: Evidence from a large-scale replication study What makes predictable words easier to process than unpredictable words? Are predictable words genuinely predicted, or more plausible and therefore easier to integrate with sentence context? Using data from a large-scale replication study, we investigated effects of word predictability and plausibility on the N400. Our analysis revealed overlapping effects of predictability and plausibility on the N400, albeit with distinct spatiotemporal profiles. Our results challenge the view that semantic facilitation of predictable words reflects effects of only prediction or integration, and suggest cascading processes that access and integrate word meaning with context into a sentence-level meaning. | Mante S. Nieuwland, Dale Barr, Federica Bartolozzi, Simon Busch-Moreno, Emily Darley, David Donaldson, Heather Ferguson, Xiao Fu, Evelien Heyselaar, Falk Huettig, Matt Husband, Aine Ito, Nina Kazanina, Vita Kogan, Zdenko Kohút, Eugenia Kulakova, Diane Mézière, Stephen Politzer-Ahles, Guillaume Rousselet, Shirley-Ann Rueschemeyer, Katrien Segaert, Jyrki Tuomainen and Sarah Von Grebmer Zu Wolfsthurn | 11:30 AM | Poster Session C | 12:00 PM - 2:00 PM |
Submission | Title | Authors | |
5(abstract) | Children’s use of word order and morphosyntactic markers in Tagalog comprehension: Evidence from eye-tracking We used a verb-initial language to investigate children’s difficulty in interpreting sentences with non-canonical word order. Tagalog is interesting because patient voice sentences (patient is the subject) are highly frequent, and thematic roles are unambiguous throughout the sentence. Our accuracy and eye-tracking data (sentence-picture matching) showed that five- and seven-year-old children correctly interpreted more patient-initial sentences in the patient voice than in the agent voice. It seems that children relied on a word order strategy in the agent voice, but used the morphosyntactic markers in the patient voice. This patient voice advantage is best explained by the frequency account. | Rowena Garcia, Jens Roeser and Barbara Höhle | Lobby |
6(abstract) | Delayed but on-line effects of metric factors: Silent reading in Turkish The Implicit Prosody Hypothesis (Fodor, 2002) was tested with a late/early closure ambiguity in Turkish. Experimental sentences had four versions manipulating length (balanced, unbalanced) and syntactic structure (LC, EC). Participants read the sentences silently and their eye-movements were tracked. The analyses revealed an LC advantage in the disambiguating region; whereas there was a preference for balanced lengths in the spillover region. Although used online, length effects seem to be delayed compared to syntactic biases, suggesting (i) syntactic information is used earlier than metric information/ (ii) the parser uses metric information once the ultimate length distribution is known, at spillover region. | Nazik Dinçtopal Deniz | Lobby |
18(abstract) | Antecedent Retrieval During the Resolution of Reciprocal Anaphors Cue-based parsing (Lewis et al., 2016) predicts anaphor resolution should be susceptible to interference. However, evidence of interference during the resolution of reflexive anaphors is mixed (Jäger et al., 2017). We monitored participants’ eye-movements as they read sentences which factorally manipulated the number (mis)match between a reciprocal anaphor and two antecedents (The girl/s said the boy/s hurt each other). Reading times were significantly affected by the number of the local antecedent (boy/s) but not the nonlocal antecedent (girl/s). These results suggest reciprocals resist interference and suggest that syntactic constraints constitute more highly weighted cues to antecedent retrieval than number agreement. | Ian Cunnings and Patrick Sturt | Lobby |
21(abstract) | Interaction of Language Use Experience and Attraction Effects in Processing SubjectVerb Honourific Agreement in Korean We explored attraction effects involving Korean subject-verb honourific agreement in a non-clausal boundary (possessive constructions) manifested by native speakers of Korean with different language use experience: university students (STD), and flight attendants and ground staff in an airline company (AIR). Results showed that, despite their stable knowledge of the agreement relation, the AIR group suffered less from attraction effects than the STD group. Our results suggest that attraction effects involving honourification in Korean do happen in a non-clausal boundary, and that experience of language in use modulate the degree to which language users experience attraction effects in real-time sentence processing. | Gyu-Ho Shin and Hyunwoo Kim | Lobby |
36(abstract) | Idioms show effects of meaning relatedness and dominance similar to those seen for ambiguous words The tension between word-level and phrase-level processing has been a central focus of idiom research (e.g. Gibbs, 1980). Additionally, the relatedness of the meanings of an ambiguous unit, whether a single word or an idiom, influences processing, suggesting flexible processing across grain size and literality status. We found longer fixation times when idioms had related literal and figurative meanings but a highly dominant figurative meaning, similar to previous research on ambiguous word processing. These results demonstrate that lexical-level constructs can generalize to processing at the phrase level, and suggests that ambiguous figurative multiword phrases and single words are processed similarly. | Evelyn Milburn and Tessa Warren | Lobby |
42(abstract) | Modeling lexical access in ACT-R One robust parameter affecting latencies and accuracies in lexical decision tasks is frequency. Since Howes and Solomon (1951), it is accepted that lexical access can be approximated as a log-function of frequency. In their Exp. 1, Murray and Forster (2004) (M&F) collected responses and response times in a lexical decision task using words from 16 frequency bands, and showed that log-frequency provides an imperfect fit to the data. We provide an ACT-R model of the M&F data and embed it in a Bayesian model to estimate its parameters. The results cast doubt on some common assumptions in ACT-R psycholinguistic models. | Adrian Brasoveanu and Jakub Dotlacil | Lobby |
57(abstract) | Is object dispreference absolute or featurally modulated across languages? Comparing Cantonese and French relative clauses Languages typically display a subject-object asymmetry in processing relative clauses. Whether the asymmetry is to be related to structural distance, to linear distance or to intervention facts is in part still controversial. SVO languages with prenominal RCs (ex. Cantonese) are crucial, since they display a configuration in which object gaps are linearly closer to their filler, but structurally farther from it. Our previous studies found a controversial result in production experiment and in spontaneous data, which points to a structural intervention hypothesis. We shall present the results of a production experiment aiming at testing this hypothesis on Cantonese and French. | Jiaying Huang and Caterina Donati | Lobby |
85(abstract) | Readers utilise proper noun capitalisation to determine syntactic class prior to direct fixation We examined eye movements during the reading of sentences containing object relative (ORCs) or subject relative clauses (SRCs) in which the relevant noun phrase consisted of a capitalized proper noun (e.g. The crafty defence lawyer who [Shannon advised/advised Shannon] was dishonest). Sentences were presented in either normal sentence casing or entirely in upper case. Participants fixated prior to the relative clause (e.g. girl who) for longer given an ORC than SRC, but only in sentences written in title case. We propose that readers infer the syntactic class of a capitalized parafoveal word, and feed this into the sentence parser early. | Michael G. Cutter, Andrea E. Martin and Patrick Sturt | Lobby |
87(abstract) | Reanalysis Processes in Native and Non-native Sentence Processing Non-native speakers often misinterpret garden-path sentences (Jacob & Felser, 2016). However, why they fail in reanalysis is debated. We conducted two experiments investigating garden-path sentences, where native/non-natives answered comprehension questions and read sentences manipulating gender (mis)match between a temporarily-ambiguous NP and a reflexive (After the neighbour visited(,) Ken’s dad/mum prepared himself…; see Slattery et al., 2013). Comprehension accuracy was lower for ambiguous sentences and reading times were longer following gender mismatches, irrespective of ambiguity. Neither effect was modulated by group. These results suggest misinterpretation in native/non-native speakers results from persistence of the initially-assigned interpretation, rather than failure to complete reanalysis. | Hiroki Fujita and Ian Cunnings | Lobby |
93(abstract) | Evidence of semantic processing difficulty in naturalistic reading We present evidence from three broad-coverage reading time corpora of a positive relationship between semantic distance and processing effort, such that words are read longer if they are further in the semantic space from preceding words in the sentence. We compute semantic distance by taking mean vector distance between embeddings of a word and the words in its preceding context. Using likelihood ratio testing, we find a significant positive effect of semantic distance on reading time, over and above a strong baseline model including multiple measures of incremental surprisal. | Cory Shain, Richard Futrell, Marten van Schijndel, Edward Gibson, William Schuler and Evelina Fedorenko | Lobby |
100(abstract) | Linearity effect in weak crossover bleeding in Mandarin Chinese In two offline experiments, we showed that both overt and covert movement of wh-operators can bleed weak crossover (WCO) in Chinese, as long as the operator linearly precedes the bound pronoun. We discuss the results with regard to the current theories on WCO. | Jun Lyu and Dawei Lu | Lobby |
106(abstract) | Attraction is Subject-ive: Dissociating Illusions and Intrusions in Agreement Attraction I examine the processing of agreement errors for sentences with clausal subjects. Clausal subjects are unspecified for number, triggering default singular agreement, unlike NP subjects. In Experiment 1, I show that sentences in which an NP subject contains an attractor are improved, indicating agreement attraction. However, this is not the case for sentences with clausal subjects containing attractors. This is because the comprehender expected to form an agreement relation between an NP and the verb in the NP-subject condition, but not the clausal subject condition. Thus, a plural NP can only induce a grammatical illusion in the first case. | Dustin Chacón | Lobby |
114(abstract) | Animacy-driven expectations in Norwegian relative clause processing We present evidence from an SPR study that indicates animate heads license strong expectations for subject RC continuations during incremental processing in Norwegian. Inanimate heads, on the other hand, appear not to license consistent expectations for either subject or object RCs. Results from a corpus study and a sentence completion task are consistent with the hypothesis that these expectations are partially driven by distributional patterns in language-users’ experience. | Dave Kush and Ragnhild Eik | Lobby |
118(abstract) | Presupposition Processing: Discourse Attachment and Answerhood to Questions – The Case of Factive Verbs: An EEG Study Presuppositions are considered distinct from the main content in terms of (i) discourse continuation, and (ii) answerhood to questions. We investigated the processing of factive verb constructions measuring event-related potentials in order to test whether continuations or answers to the less relevant content – the presupposition – are cognitively more demanding than continuations to the more relevant one – the main content. In conclusion, the main finding of our three studies can be presented as such that there may, indeed, be a salience difference between the main content versus the presupposition of factive verb constructions; however, such a difference is discourse-dependent. | Jacques Jayez and Rob Reinecke | Lobby |
124(abstract) | Grammatical weight and relative clause processing in German and Cantonese This paper investigates the word orderings of head noun and its modifying relative clause (RC) in German and Cantonese from the processing perspective. The principles of ‘Early Immediate Constituents’ (EIC, Hawkins, 1994, 2004) was tested by an experiment using acceptability judgement tasks. EIC predicts that rearranging RC in both languages from its canonical position similarly facilitates parsing due to the minimised distance between the constituents (i.e. head noun and the main verb), especially when the RCs are heavy (in length and complexity). In spite of the similar predictions in parsing advantage, the experiment found different results for the two languages. | Yat Han Lai | Lobby |
129(abstract) | Speakers Learning to Read in Cyrillic Follow Different Developmental Trajectories In an eye-tracking reading project, we established basic descriptive corpus statistics of eye movements in 3 different populations within one language, namely, Russian. Participants (monolingual adults and children, and heritage Russian adults) read sentences from the Russian Sentence Corpus or its child version. Following the cross-linguistic protocol for collecting eye movements in reading, the corpora included sentences designed around target words selected in a 3 x 2 design (length and frequency). The observed differences in eye movement characteristics across all groups are discussed with respect to developmental trajectories of reading fluency in Russian. | Olga Parshina, Anna K. Laurinavichyute, Aleksei A. Korneev, Svetlana Alexeeva, Ekaterina Y. Matveeva, Tatiana V. Akhutina and Irina A. Sekerina | Lobby |
133(abstract) | A crosslinguistic corpus study of crossing dependencies using psycholinguistically motivated measures Considerable research has pointed to processing adaptability based on typological properties of languages. In this work we carry out a corpus study(12 typologically diverse languages) to investigate the effect of head directionality on crossing dependencies which are known to be costly. Our work shows that both linear and hierarchical distance which are processing constraints on a crossing dependency interact with head directionality of the language. Under the assumption that the corpus data reflects processing constraints, our work points to processing adaptability with respect to crossing dependencies based on certain typological properties of a language. | Vishakha Shukla, Himanshu Yadav, Ashwini Vaidya and Samar Husain | Lobby |
137(abstract) | The role of informativity on the disfluency effect Does the informativity of disfluency diminish for speakers who are frequently disfluent or disfluent in an unusual way? In this behavioral study we test this by examining the expectation for new information following disfluency, and ask whether it disappears for highly disfluent speakers. Exp.1 manipulated the frequency of disfluency, though disfluencies still occurred in expected locations. Exp.2 changed the distribution of disfluency, signaling an unnatural correlation between disfluency and given information. In both experiments, the disfluency-new bias remained a robust effect, suggesting that there may be limits to adaptation effects. | Valerie Langlois and Jennifer Arnold | Lobby |
155(abstract) | Crossed hands and the syntax-space effect Previous research has shown that action scenes are comprehended faster when the agent is located to the left of the patient (Chatterjee, Southwood & Basilico, 1999). We have extended this effect to reading transitive sentences, showing that this effect is mainly present with left hand responses (Boiteau & Almor, 2016). In this experiment we test whether hand positioning (crossed vs. normal) has any bearing on this syntax-space effect and find that indeed it is not only the response hand that matters but also the side of space towards which one responds. | Tim Boiteau, Cameron Smith and Amit Almor | Lobby |
157(abstract) | Time-course of Orthographic Effects in Chinese Spoken Word Recognition This study uses Chinese single-character words to evaluate whether spoken words activate orthographic representations via sound-spelling mappings or via schematic visual/perceptual representations. Exp.1 demonstrates the separation of phonological and orthographic representations during Chinese spoken word recognition. Exp.2 showed that orthographic effects emerged as soon as there is sufficient information in the speech signal for participants to know the orthographic information of the target word (Exp. gating). The results are consistent with the claim that mapping of spoken words onto printed words at short preview is mediated by visually-based orthographic information that becomes available as a word is processed. | Yuhang Xu, Mengya Xie, Qingrong Chen and Mike Tanenhaus | Lobby |
158(abstract) | Modelling Memory Retrieval Processes with Drift Diffusion We show that Drift Diffusion Model can be used as an alternative to Multiple Response SAT to model speed-accuracy tradeoffs during memory retrieval in sentence processing. We report a replication of Martin & McElree’s (2011) finding for direct access of antecedents in sluicing, and then extend the DDM technique to examine memory retrieval in picture-NP reflexives and anaphoric presuppositions, finding evidence for serial search and direct access for antecedents respectively. Modeling memory retrieval processes with DDM provides convergent evidence to SAT, and has advantages in requiring fewer response time measurements to recover meaningful parameters. | Sherry Yong Chen and E. Matthew Husband | Lobby |
159(abstract) | Teasing apart structure-building and structure-completion during naturalistic story listening using EEG Recent work combining computational psycholinguistics and neurolinguistics have shown success in developing more rigorous accounts of the neural bases of sentence comprehension. These efforts offer quantitatively precise accounts of candidate algorithms implemented by neural language circuits. We extend those efforts with a focus on refining theories of predictiveness and also aim for generalizability across the range of sentence-types that appear in natural language. We compare incremental parsing models with different degrees of predictiveness against electroencephalography signals recorded while participants listen passively to an audiobook story. Results show correlates of predictive structure-building processes that are distinct from downstream correlates of structure-completion. | Jonathan Brennan and John Hale | Lobby |
162(abstract) | Complement coercion induces greater garden path cost: Two eye tracking studies Two eye tracking while reading studies investigate the effect of complement coercion on recovering from misanalysis in Subject/Object garden path ambiguities ("After John started / read the book was sold"). Experiment 1 finds that coercion of the noun ("start the book") induces greater processing difficulty over controls ("read the book") upon revision, and that the coercion penalty increases the longer the erroneous interpretation remains valid. Experiment 2 finds that coercion of an object noun reduces processing penalties when the matrix verb also requires coercion ("… was completed"), suggesting that the coerced interpretation remains accessible to the processor during repair processes. | Jesse Harris | Lobby |
168(abstract) | Approaching scalar diversity through (RSA with) Lexical Uncertainty Previous research suggests that different scalar expressions give rise to scalar inferences at different rates. This phenomenon has become known as scalar diversity (SD). To explore the source of SD, we use insights from Rational Speech Act approaches that allow for local lexical enrichments. These approaches predict scalar diversity is possible if the priors for lexical enrichments vary across different scalar terms. Here we provide evidence that liability for lexical enrichments differ across different scalar terms and we relate SD to liability of local enrichments. | Chao Sun and Richard Breheny | Lobby |
172(abstract) | Effects of linguistic focus and lexical frequency on reading times In two eyetracking-while-reading experiments, we examined whether manipulations of linguistic focus modulate the magnitude of the word-frequency effect. Experiment 1 manipulated focus using cleft constructions. Results showed overall longer processing times when words were focused versus defocused; in addition, the magnitude of the word frequency effect was reduced when target words were focused. Experiment 2 manipulated focus using a contrast between simple sentences and relative clauses. Although effects of frequency emerged, there were no effects of the structure manipulation. The results replicate and extend previous use on the use of syntactic structure as a linguistic focusing device. | Matthew Lowder and Fernanda Ferreira | Lobby |
175(abstract) | Power transforms distort interaction patterns in mixed-effects analyses of chronometric data Power transforms (log- and inverse-transforms) have been routinely applied to chronometric data to avoid violating the assumption of residual normality in mixed-effects models. However, researchers have revived the idea that power transforms are not innocuous. Specifically, power transforms distort the scale in which cognitive processes are known to occur, thereby potentially introducing spurious effects or obscuring real effects. This project investigates the costs of power transforms in mixed-effects models by running realistic simulations that were based on analyses of three word-recognition megastudies. Consistent with previous reanalyses of small datasets, power transforms obscured interaction effects present in the raw time scale. | Van Rynald Liceralde and Peter Gordon | Lobby |
178(abstract) | The flexibility of multiword sequences: Evidence from eye tracking Two eye tracking experiments investigated the flexibility of multiword sequences by testing whether altering the form of binomial phrases (salt and pepper) affects prediction of the constituents in these phrases. Both experiments showed that readers’ prediction of the predictable conjunct (pepper) was robust, even when the phrase contained variant conjunctions (and also in Experiment 1; as well as in Experiment 2). However, the cloze study in Experiment 1 showed that lexical prediction was modulated by the variant conjunction, highlighting differences between offline vs. online tasks. Overall, our findings suggest that multiword sequences are processed as flexible units during online reading. | Suphasiree Chantavarin and Fernanda Ferreira | Lobby |
179(abstract) | Resolving cross-method conflicts in the timing of cohort competition using jTRACE Different methodologies diverge on the question of whether syntactic context can constrain the generation of lexical candidates during auditory word recognition. We show in simulations with jTRACE that the latency of phonological competition effects is systematically related to the composition of the response candidate set allowed by the task. We observe that neural and behavioral cohort measures appear to reflect distinct distributional properties, and that visual world experiments not concerned with syntactic category may still impose category restrictions on the cohort. This raises the possibility that the existing literature is in fact fully consistent with an immediate syntactic category constraint. | Phoebe Gaston, Ellen Lau and Colin Phillips | Lobby |
182(abstract) | The direct relation between prediction and adaptation The current study evaluated the Error-based Implicit Learning accounts by investigating the role of prediction in adaptation. Consistent with previous findings, in the prediction task using visual world paradigm, native English speakers made predictions using semantic cues from the given context. They also showed adaptation toward the current linguistic input as measured by change in interpretation of relative clause (RC) attachment at the pre and post tests after exposure to sentences forcing low attachment interpretation (block 1) and high attachment interpretation (block 2). Importantly, however, we could not find the effect of individuals' predictive ability on their adaptation. | Eunjin Chun, Joshua Daniels and Hope Klingensmith | Lobby |
184(abstract) | Generalizing dependency length minimization: Crosslinguistic corpus evidence for information locality The theory of dependency length minimization holds that, for reasons of processing efficiency, linguistic heads and dependents will usually be close in linear order. We present evidence for a generalization of the theory, called information locality, arising from a recent unified model of surprisal and locality effects in language processing (Futrell & Levy, 2017). Information locality predicts that, beyond dependency length minimization, words with high mutual information (which predict each other strongly) should be close in linear order. We provide evidence for this claim in corpus studies of over 50 languages using the Universal Dependencies corpora. | Richard Futrell, Edward Gibson and Roger Levy | Lobby |
186(abstract) | Epistemic inferences in passing conversation: Pragmatics as a test of Theory of Mind accounts According to the 2-system account of Theory of Mind proposed by Apperly and Butterfill (2009), reasoning about other people's beliefs is slow and effortful (as opposed to reasoning about others' perceptual registrations, which is automatic). While this account might explain the results of false-belief tasks with infants and children, we argue that it does not account for the epistemic inferences that we derive in everyday conversation. We report two self-paced reading experiments showing that adults reason about others' beliefs fast and efficiently in passing conversation with strangers and acquaintances, challenging Apperly and Butterfill's 2-system account of Theory of Mind. | Paula Rubio-Fernández, Michelle Oraa Ali and Edward Gibson | Lobby |
195(abstract) | Non-native English (L2) speakers generalize on the basis of meaning Native English speakers judge sentences with novel combinations of verbs and constructions (e.g., Amber explained Zach the answer) to be less acceptable when there is a competing alternative (CA)(i.e., Amber explained the answer to Zach). We show that non-native English (L2) speakers judge novel sentences with CAs to be more acceptable than native speakers. Recent exposure to CAs does not change their judgments. However, L2 speakers were well-above chance at remembering what they had heard verbatim. Thus, L2 speakers know what they’ve heard but do not take CAs into account. | Karina Tachihara and Adele Goldberg | Lobby |
198(abstract) | Antecedent retrieval in ellipsis and pronominals This study investigates the antecedent retrieval process of three anaphoric constructions, NP-ellipsis (NPE), Anaphoric One and Pronoun (Pro). Two eye-tracking while reading experiments and a self-paced reading experiment showed that while NPE exhibited an illusion of grammaticality (the agreement attraction effect in ungrammatical sentences), One and Pro did not, and thus, the antecedent is retrieved from memory differently in the processing of NPE, and One/Pro. We suggest that the structure is built at the NPE-site, where the reanalysis is prompted when the mismatch between the retrieved NP and the verb arises. | Nayoun Kim, Laurel Brehm and Masaya Yoshida | Lobby |
201(abstract) | Local constraint is inhibited by global context in concessive structures Facilitation from lexical prediction in highly constraining contexts is found across methodologies (Rayner & Well, 1996; Federmeier & Kutas, 1999), and from either globally or locally constraining context (Fitzsimmons & Drieghe, 2013). We examine how globally and locally constraining contexts interact within a single sentence when the local context suggests an upcoming word w, but the larger context biases against w. Our results suggest that early lexical processing is facilitated by supporting constraint from local context, and is not overruled by conflicting constraint from global context. Conflicting global context, in turn, resulted in processing penalties in later measures of reading. | Stephanie Rich and Jesse Harris | Lobby |
211(abstract) | Early vs late cues in the processing of Italian relative clauses In Italian, relative clauses are syntactically ambiguous between a subject and an object reading. In two self-paced reading studies, we investigate the effectiveness of morpho-syntactic and syntactic cues in triggering reanalysis by comparing early and late cues, with the aim to adjudicate between the cue-based memory retrieval model and the self- organized sentence processing model. The results show that reanalysis is attained more effortless for early cues, while reanalysis fails when the structure has stabilized, which happens when a great portion of the structure has already been built (digging-in effect), in line with predictions from the self-organized parsing model. | Sandra Villata and Paolo Lorusso | Lobby |
212(abstract) | SES Differences in the Structure of Child-directed Speech In this study, we ask if SES impacts the structural organization of children’s input. Specifically, we examine the effect of SES on the use of variation sets (VSs, successive utterances with partial self-repetitions) in child-directed speech (CDS). VSs have been found to facilitate language learning, but have been studied only in high SES. Here, we examine their use in naturalistic speech in two languages (Hebrew and English) for both low and high SES. We find that VSs are more frequent in high SES compared to low SES in both languages, suggesting that SES also impacts the organization of the input. | Shira Tal and Inbal Arnon | Lobby |
219(abstract) | Proximity and same case marking do not increase attraction effect in comprehension: Evidence from eye-tracking experiments in Korean In the present study, we investigated whether attraction effects would be modulated by memory representation of a distractor, manipulating proximity of a distractor to a licit antecedent and similarity of case marking of a distractor to that of a licit antecedent. However, the current study did not find evidence in support of the hypothesis that a greater number of matching cues of a distractor would trigger more mis-retrieval, in contrast to a previous finding that a greater number of (mis)matching cues of a licit antecedent did so (Parker, 2014). | Nayoung Kwon and Patrick Sturt | Lobby |
397(abstract) | Low-level language statistics affect reading times independently of surprisal Surprisal theory has provided a unifying framework for understanding many phenomena in sentence processing (Hale, 2001; Levy, 2008), positing that a word’s probability in context determines processing difficulty. Problematically for this claim, one low-level statistic, word frequency, has been shown to affect processing independently of surprisal. We present the first clear evidence that a more complex low-level statistic, word bigram probability, also affects processing independently of surprisal. These findings suggest a broad, independent role of low-level statistics in processing and motivate research into new generalizations of surprisal that can also explain why local statistical information should have an outsized effect. | Adam Goodkind and Klinton Bicknell | Lobby |
95(abstract) | What do readers adapt to in Syntactic Adaptation? We explored whether the syntactic adaptation effects found for reduced relative clauses sentences (RRC) in Fine (2013) Exp't. 3 could be replicated and whether violations of a transitivity bias in the ambiguous region also elicited syntactic adaptation and whether this additional violation influenced adaptation in the disambiguating region of sentences with reduced relative clauses. Fine-grained analyses revealed adaptation to violations of both D.O. and main clause expectations. While adaptation to D.O. expectations did not predict adaptation to RRCs, processing of and adapting to prior violations may nonetheless interfere with RRC adaptation and delay its emergence to later words. | Avery Malone and Gail Mauner | Lobby |
221(abstract) | How free is the processing of free word order? ERP evidence on the positional preference of temporal adverbs in Korean sentence comprehension The current study investigated whether there is a grammatically invisible word order preference in Korean, especially focusing on the adverb position in sentences which consist of the multiple verbal arguments or adjuncts. 30 Korean speakers’ ERPs were recorded at the verb position which are different in terms of (i) the position of involved temporal adverb, (ii) syntactic/semantic property of other co-occurring constituents (verbal arguments or adjuncts). The ERP result was characterized by a large positivity observed in the condition where the temporal adverb was placed right before the verb in 200-700ms, irrespective of the semantic property of the involved constituents. | Yunju Nam and Upyong Hong | Conference Room A |
224(abstract) | Resumptive pronouns can ameliorate island violations in real-time comprehension In contrast to intuition, much psycholinguistic work of the past decade has claimed that resumptive pronouns (RPs) in English provide no benefit to the comprehension of islands. This claim has been made on the basis of offline measures. In the present study, using self-paced reading, we find evidence of real-time benefits for RPs in island violating structures: compared to gaps, in non-island conditions, RPs are associated with longer reading times, while RPs in island-violating structures facilitate reading. The results provide support for an account where RPs give bottom-up evidence of the location of gaps, making filler-gap dependencies easier to form. | Christopher Hammerly | Conference Room A |
227(abstract) | A construction-conflict explanation of the subject-island constraint We present acceptability-rating studies in English and French, showing that extraction out of nominal subject in relative clauses (RCs) is acceptable in both languages, contrary to the subject-island constraint. We show that similar extractions in wh-questions are not possible in either language. We propose: the discourse status of the extracted element should match the discourse function of the construction (wh-question or RC); and it should match the discourse status of its head (here of the subject or object noun). In a wh-question, the extracted element should be a focus (foreground information), in an RC, it should be a topic (backgrounded). | Anne Abeille, Barbara Hemforth, Elodie Winckel and Edward Gibson | Conference Room A |
236(abstract) | Relative weight and givenness in constituent ordering of typologically different languages: Evidence from French and Persian Grammatical weight and givenness of the referent are two factors known to affect constituent ordering across languages. Constituents tend to appear in order of increasing weight in VO and of decreasing weight in OV languages. Given referents tend to be produced before new referents cross-linguistically. We provide comparable experimental data from French and Persian to contrast weight and givenness effects in VO and OV languages. Our study clearly supports the role of weight in OV and VO languages production. Our French data show a given-first effect, while using the same protocol, we previously failed to find this effect in Persian. | Pegah Faghiri and Juliette Thuilier | Conference Room A |
238(abstract) | Expectations and prediction in sentence comprehension: German particle verbs as a test case Does strong expectation entail specific, lexical prediction? We tested this question using German particle verbs. In the sentence "Das Fest ging früh los", the verb "ging" strongly predicts the particle "los". However, other verbs are compatible with many particles. We predicted that a small set of compatible particles would encourage prediction of a specific particle and speed up reading, but that a large set of compatible particles would increase uncertainty, discouraging prediction and slowing reading. Instead, SPR/ET reading times were slower when particle set size was small. We speculate that making and maintaining a specific prediction may incur some cost. | Kate Stone, Shravan Vasishth and Titus von der Malsburg | Conference Room A |
246(abstract) | Effects of cross-linguistic differences in structural flexibility on production planning In this study, we assess the effects of structural flexibility on production planning by comparing causative motion expressions in Mandarin and English. We hypothesize that compared to English, Mandarin should show more structural options that speakers typically use to talk about causative motion. Our elicitation task confirms this difference. The syntactic priming task tests whether the observed differences affect speakers' sensitivity to the recently activated syntactic frame. We predict that Mandarin speakers should have more syntactic frames available for retrieval and therefore be more susceptible to syntactic priming than English speakers. Our findings confirm this prediction. | Chun Zheng and Elaine Francis | Conference Room A |
252(abstract) | Retrieval interference in the resolution of long-distance dependencies: An eye-tracking study comparing L1 & L2 processing Based on the cue-based parsing theory, the present study tests syntactic and semantic retrieval interference in L1 and L2 processing to understand the source of comprehension difficulty and L1-L2 differences during sentence reading. The observed similarity-based interference effects in L1 and L2 data provide supportive evidence for the cue-based retrieval parsing. However, the lack of reliable two-way interaction in L2 data may suggest that different from skilled L1 readers, L2 readers may have difficulty in employing multiple cues simultaneously during parsing, possibly due to limited computational resources, taxed by the activation and inhibition of L1 or reduced automaticity in L2. | Nayoung Kim and Kiel Christianson | Conference Room A |
263(abstract) | Gender assignment variability modulates the anticipatory use of grammatical gender in an L2 This study uses ERPs with a cued lateralized picture monitoring task to test the claim of the Lexical Gender Learning Hypothesis (LGLH) that variability in gender assignment underlies the reduced use of gender as an anticipatory cue by adult L2 learners. Using the N2pc as an index of anticipation, we find evidence that advanced L1-English L2 learners of German use gender to predictively direct their covert visual attention toward a target noun prior to its acoustic onset in a short utterance, but only for nouns for which participants have stable gender assignments. Data are thus consistent with the LGLH. | Kailen Shantz and Darren Tanner | Conference Room A |
269(abstract) | Before or after? Category marker position and category feature transparency determine what gets learned Categories in language such as noun gender can be marked by preceding category markers (premarkers, e.g., gendered article in German: Der Fisch) but also by succeeding markers (postmarkers) or by both (e.g., gendered article and relative pronoun in German: Der Fisch, der …). In a computational simulation (naïve discrimination learning, Baayen et al., 2011) and an artificial language learning study, we showed that postmarkers outperform premarkers in linguistic category discrimination but premarkers outperform postmarkers in within-category item discrimination. This effect can be explained with error-driven learning (Rescorla and Wagner, 1972) and depends on whether categories are overlapping or distinct. | Dorothée B. Hoppe, Jacolien van Rij and Michael Ramscar | Conference Room A |
271(abstract) | Broccoli or *cauliflower? Adaptation to L2 lexical errors When referring to objects, listeners typically expect speakers to follow precedent and continue using previously established labels (Barr & Keysar, 2002; Metzing & Brennan, 2003). A question remains whether following precedence extends to cases where previously used labels are incorrect. A visual world experiment addressed this question by examining whether looks to previously mislabeled objects (e.g., a picture of broccoli labelled*cauliflower) competed with a phonologically overlapping target (columns) relative to when the object was labelled correctly (broccoli). Participants indeed looked more at the broccoli in the incorrect condition, suggesting that listeners had learned to expect incorrect labels for specific objects. | Thomas St. Pierre and Jean-Pierre Koenig | Conference Room A |
273(abstract) | How idiomaticity and verb voice affect sentence reading: an eye-tracking study on Italian In this eye-tracking experiment we compared the reading times of Italian idiomatic and literal phrases in active form, passive form with preverbal subject and, as allowed by Italian syntax, passive form with postverbal subject, to test whether passive idioms are read more slowly than passive literals and whether preserving the canonical verb-noun order in the second passive form facilitates processing. While active idioms were read faster than active literals, passive forms took longer to read regardless of idiomaticity and postponing the subject disrupted reading even more. Familiarity and semantic transparency also appeared to make the reading of passive idioms easier. | Marco Silvio Giuseppe Senaldi, Paolo Canal, Alessandro Lenci and Pier Marco Bertinetto | Conference Room A |
277(abstract) | Why are English and Hebrew resumptive pronouns different? Resumptive pronouns (RPs; e.g., "the computer that I know why IT broke") are claimed to be grammatical in Hebrew but not English. What might lead to this difference? One obstacle to addressing this question derives from the inherent difficulties of comparing across languages. To better characterize the facts, we collected production, comprehension, and acceptability data on RPs in two experiments: one in English, and a second which was identical except stimuli were translated into Hebrew. Results: Hebrew speakers rate RPs higher and produce more RPs than English speakers. We investigate the relationships between these metrics within individuals. | Adam Morgan, Julie Fadlon, Aya Meltzer-Asscher and Victor Ferreira | Conference Room A |
282(abstract) | Things that should affect agreement attraction in comprehension, but don't Although agreement attraction occurs in production and comprehension, it is debated whether the underlying mechanism is the same across both. We investigate this by exploring whether the same factors that impact attraction in production also affect it in comprehension. In production, attraction rates are affected by non-syntactic factors like notional number and the plausibility match between the attractor and the verb. We present data from two self-paced reading studies that investigate the effect of notional number and plausibility on agreement attraction in comprehension but find no evidence that either has an effect. This suggests divergent mechanisms in comprehension and production. | Zoe Schlueter, Dan Parker and Ellen Lau | Conference Room A |
292(abstract) | Gender Agreement and Predictive Lexical Processing in Czech 23-month-olds: Emerging Sensitivity to Bound Gender Inflections The study examined the early knowledge of noun gender and gender agreement in Czech. Monolingual children aged 21 to 24 months (N=34) listened to noun phrases such as “takový hezký banán/*kniha” (such a nice banana/*book) in which the initial two words, demonstrative and adjective, carried adjectival ending encoding the gender of the upcoming noun that either matched or mismatched the gender of the subsequent head noun. Results show that some children before 2 use gender information from agreement morphemes to anticipate the upcoming noun. However, this was only true about children with above-average performance on offline language tasks. | filip smolik and Veronika Bláhová | Conference Room A |
293(abstract) | Developmental shifts from bottom-up lexical biases to top-down plausibility in syntactic parsing In a visual-world study, we investigate adults’ and children’s use of top-down plausibility and bottom-up lexical biases to resolve PP-attachment ambiguities. Participants heard Modifier-biased, Equi-biased, or Instrument-biased verbs in sentences that had either low or high global plausibility (Find the zebra with the sponge | magnifying glass). Our data show adults rely more on plausibility cues, whereas children rely on lexical information. This suggests that bottom-up information may be prioritized in child comprehension, while strong top-down constraints dominate adult processing. These findings are consistent with prior work on children’s use of prosody and lexical information and disuse of contextual cues. | Anthony Yacovone, Carissa Shafto, Amanda Worek and Jesse Snedeker | Conference Room A |
311(abstract) | When NPI illusions fail: the case of strict NPIs and neg-words in Romanian Psycholinguistic research shows that structurally inaccessible negation can create an ‘illusion of grammaticality’ for NPIs. Previous studies on NPI illusions have shown that weak NPIs (‘any’, ‘ever’) are subject to illusory licensing. It is presently unknown whether strong NPIs (‘in years’) - necessarily syntactically licensed, are susceptible to illusory licensing effects. The results of our SPR experiment on Romanian (strict negative concord language) show comprehenders were immediately sensitive to whether strict NPIs and neg-words were licensed, but we failed to find evidence of illusory licensing, thus favoring semantic-pragmatic approaches to illusory NPI licensing. | Rodica Ivan and Brian Dillon | Conference Room A |
313(abstract) | Top-down language learning in children: Insights from degraded speech What mechanisms allow perceptual learning? An important debate for theories of language concerns the role that high-level (i.e., top-down) information plays in how we learn to perceive the environment (e.g., in learning to perceive distorted speech). But perhaps surprisingly, this debate has rarely focused on the paradigmatic human learner: children. Here, we test whether and how preschool children integrate high-level information in order to interpret distorted sine wave speech. We find that the well-documented “insight effects” seen with adults are interestingly absent in children, and that top-down influence is overall much smaller, suggesting important implications for theories of learning. | Hugh Rabagliati, Hanna Jarvinen and Matthew Davis | Conference Room A |
318(abstract) | Frequency Effects of Proper Names as Measured by Fixation-Related Potentials. Lexical frequency affects the ease of word recognition during sentence reading in both eye-movement measures- where participants have control of the reading rate and access to parafoveal information- and ERPs, where they typically do not. The current work examines the neural effects of lexical frequency in a more natural environment in which eye-movements and the EEGs are simultaneously recorded as participants read sentences by moving their eyes freely. Additionally, it isolates lexical frequency from the closely associated construct of plausibility by studying proper names, which vary in their frequency of occurrence but are equally plausible within a sentence. | Giulia Christine Pancani and Peter Gordon | Conference Room A |
322(abstract) | Says who?: The influence of speech and mental predicates on epithet interpretation Epithets express the attitude of a judge toward their referent. This judge is typically the speaker of the utterance. However, shifted interpretations with a non-speaker judge have been observed in certain contexts, including inside and outside of embedding. I present two interpretation experiments demonstrating the importance of speech and mental predicates in a preceding sentence on the availability of non-embedded shifted readings. The evidence indicates that mental predicates most strongly promote shifts, contradicting the speech-centered predictions of accounts of perspective-shifting which rely on context-shifting operators in the grammar. Instead, I consider a new pragmatic explanation from language-independent theory-of-mind cognition. | John Duff | Conference Room A |
324(abstract) | Effects of phonetic cues on processing Mandarin-English code switches in sentence comprehension Though bilinguals code-switch frequently and with apparent ease, code-switches have been argued to incur a processing (“switch”) cost. However, recent research on Spanish-English bilinguals suggests listener switch cost is mitigated by phonetic shifts signaling an impending code switch. We hypothesized that phonetic cues might be language-dependent: shifts in pitch might serve as cues to Mandarin-English code-switches. Using concept monitoring and eye tracking experiments, we investigated how bilingual listeners process English sentences where cues to an upcoming Mandarin code-switch were removed by splicing. Results indicate that while not completely offsetting cost, phonetic cues enable listeners to anticipate code-switches in Mandarin-English speech. | Alice Shen, Susanne Gahl and Keith Johnson | Conference Room A |
327(abstract) | Mutual Information Impacts Adjective Ordering Across Languages Across languages, adjectives show preferences for some orderings over others. No consensus exists on the origin of these preferences. We suggest that, due to the effect of memory limitations on prediction, adjectives that have higher mutual information with the noun should occur closer to it. We verify this prediction across six languages. We then show that powerful statistical language models subject to memory limitations, instantiated by recurrent neural networks, are affected by these differences. Scontras et al. (2017) showed that adjective ordering is predicted by subjectivity. We show that mutual information and subjectivity independently impact adjective ordering. | Michael Hahn, Richard Futrell, Judith Degen, Dan Jurafsky and Noah Goodman | Conference Room A |
332(abstract) | Availability-based production, not Uniform Information Density, predicts Mandarin classifier choice What general principles govern speaker choice in the face of optionality when (near)-semantically invariant alternation exists? Studies have shown that optional reduction in language is sensitive to contextual predictability. Yet it is unclear whether speaker choice is geared toward an audience design, or toward facilitating production. Here we argue that for a different optionality phenomenon, namely classifier choice in Mandarin Chinese, Uniform Information Density and availability-based production make opposite predictions regarding speaker choice in relation to the predictability of upcoming materials. In a corpus analysis, we show the distribution of speaker choices supports the availability-based production account, and not UID. | Meilin Zhan and Roger Levy | Conference Room A |
341(abstract) | A morphological cue beats a semantic constraint in Turkish agreement attraction In this study we investigate the animacy constraint on subject-verb number agreement in Turkish: plural animates can take plural verb morphology, but plural inanimates cannot. Using plural inanimates as attractors thus constitutes an ideal test of two conceptions of repair: if repair is satisfied simply by successful retrieval of a noun+plural chunk from memory, inanimates should drive illusions of grammaticality as well as animates, but if repair involves syntactically licensing the plural feature on the verb with the attractor, they should not. Results from our speeded acceptability judgment experiment support the first, ‘shallow repair’, hypothesis. | Nur Basak Karatas and Ellen Lau | Conference Room A |
342(abstract) | L1 effects on the production of non-adjacent dependencies Non-adjacent dependencies (NADs) occur cross-linguistically in multiple domains. Previous research using artificial grammars shows that NADs are difficult to learn unless a similar one is present in a speaker’s L1 (LaCross 2015). Languages using NADs adapt borrowings to fit their grammatical system, including morphologically (Bat-El 1994), and may change their grammar over time to incorporate more concatenative morphology (e.g., Maltese; Borg & Azzopardi-Alexander 1997; Twist 2006; Drake 2018). But why does this pattern not work in reverse? I explore this question and provide reasons for this imbalance in morphological typology using a production task. | Shiloh Drake | Conference Room A |
343(abstract) | Determining Obligatory Inversion in Spanish WH-Extraction Unlike declaratives in Spanish, sentences with filler-gap dependencies are judged unacceptable with Subject-Verb order. The effect is weaker for adjunct wh-phrases. This is traditionally described as a grammatical asymmetry (Torrego 1984; Gallego 2006), yet Goodall (2008) offers an explanation based on working memory demands. Our two acceptability judgment studies tested the respective contributions of grammatical and processing factors. Our findings suggest that grammatical argument or adjunct status, alone, cannot account for the asymmetry in required vs. optional inversion. | Mary Christensen and Dustin Chacón | Conference Room A |
344(abstract) | Cortical phase-locking to predictable temporal sequences in the absence of periodicity When people comprehend speech, neural oscillations lock onto pulses in volume. Does cortical phase-locking rely on oscillatory resonance with periodic stimuli, or is temporal predictability sufficient to drive synchronization? We recorded EEG while participants watched sequences of images that varied in their periodicity and temporal predictability. We computed mutual information (MI) between stimulus onsets and EEG activity as a function of lag between signals. We found stronger prestimulus MI to predictable than to unpredictable sequences, even when neither sequence was periodic. Cortical phase-locking does not rely exclusively on oscillatory resonance, but may occur whenever sequences are temporally predictable. | Geoff Brookshire and Daniel Casasanto | Conference Room A |
346(abstract) | Visual Context Effects during Bilingual Language Processing: Evidence from Eye-tracking The current visual world eye-tracking study (N = 32) examined English-French bilinguals’ reliance on recently seen events, focusing on their ability to predict a plausible future action during spoken sentence comprehension. | Dato Abashidze, Pavel Trofimovich, Kim McDonough and Julien Mercier | Conference Room A |
347(abstract) | Subjecthood is Privileged in Message Formulation: Real-time Production of wh-Questions in English and Mandarin We present two visual-world eye-tracking experiments investigating which element of a message speakers formulate and linguistically encode first. Exp1 de-coupled effects of linear word order from subjecthood using English declaratives and object wh-questions. If these processes are subjecthood-driven, speakers should initially fixate subjects in both sentence types. Otherwise, they should first fixate subjects in declaratives (SVO) but objects in questions (OSV). Exp2 used Mandarin declaratives and questions, to test if Exp1’s findings were due to information focus associated with questions. We find linguistic encoding is simultaneously influenced by subjecthood and linear word order, but not sensitive to informational focus. | Monica Do, Elsi Kaiser and Pengchen Zhao | Conference Room A |
354(abstract) | Processing imprecision: the interpretation of round numerals in context Aparicio et al. (2017) find that the processing of imprecise interpretations of round numbers is costlier than their precise counterparts when (im)precision is signaled via a slack regulator (e.g. exactly 100 or about 100). The current experiment tests whether similar processing patterns emerge when non-compositional cues, e.g. pragmatic reasoning about conversational goals, are used to bias interpreters towards (im)precise interpretations. In line with previous findings, our results suggest that imprecision incurs a higher processing cost than precision when (im)precision is signaled pragmatically, albeit the difference is only detectable when participants are confident about the precision level intended by the speaker. | Helena Aparicio, Christopher Kennedy and Ming Xiang | Conference Room A |
358(abstract) | Ironic emojis are processed like ironic words: evidence from three ERP experiments Across three experiments we address the question of whether ERP brain responses to an ironic emoji will be the same as those to ironic words, namely a P200 + P600 effect complex. Indeed, in all three experiments, the same P2 + P6 effect complex surfaces when the wink emoji is interpreted as a conveyor of irony. This is taken as neural evidence that emojis are capable of conveying linguistic information in sentence contexts. | Benjamin Weissman and Darren Tanner | Conference Room A |
360(abstract) | Inferred boundaries from sentence processing set the domain of statistical learning If there are structural expectations of the incoming language stream, how does statistical learning operate? We present experiments showing that when words from an artificial language are bracketed given structural expectations to belong to the same structure, participants succeed in learning dependencies between these words. However, when the same dependency is presented such that the words making up the dependency belong to different structures, dependency learning would fail. These experiments demonstrate that statistical learning operates within a domain set from linguistic structures. Linguistic structures from sentence processing determines whether learners acquire even the most robust statistical structure (i.e., 100% co-occurrence). | Felix Wang, John Trueswell, Jason Zevin and Toby Mintz | Conference Room A |
361(abstract) | Pragmatic predictions from the visual world modulate neural semantic activity This ERP study characterized neural effects of visually-derived pragmatic inferences. Participants heard “The target is the (adjective) (noun)” (e.g., tall pitcher). We varied four-picture visual displays to manipulate pragmatic expectedness (expected: tall/short pitcher; unexpected: tall/short ladder; intermediate: both sets). Pragmatically-unexpected nouns, relative to pragmatically-expected nouns, evoked larger N400s (indicating semantic facilitation) and late P600 effects (suggesting structural reanalysis). Further, an anterior negativity between 400-600ms evoked by pragmatically-unexpected nouns perhaps reflected demands of suppressing strong pragmatic predictions and enhancing lower-probability event representations. Thus, visually-derived pragmatic information modulates neural activity related to semantic prediction, and can prolong neural processing for disconfirmed predictions. | Meredith Brown and Gina Kuperberg | Conference Room A |
362(abstract) | Aging Defies Domain-General Explanations of Domain-General Cognitive Control in Syntactic Ambiguity Resolution Cognitive control abilities contribute to individual differences in resolution of syntactic ambiguity. This study investigates the relationship between the processing of temporarily ambiguous sentences and cognitive control in older adults. Although there were moderate correlations between Age and WM and Inhibition in that older adults performed worse, this diminished cognitive capacity did not affect syntactic ambiguity resolution in either their actions or eye movements. We conclude that age-related decline in cognitive control does not necessarily hinder syntactic ambiguity resolution and argue that our results support a domain-specific account for language processing. | Iris Strangmann, Jungmee Yoon, Youngmi Park, Irina Sekerina and Loraine K. Obler | Conference Room B |
365(abstract) | Learning to recognize familiar words across accents in infancy The pronunciation of words can vary greatly between accents. Such between-accent differences have been claimed to impede infants’ word recognition. In fact, children do only appear to start recognizing familiar words in unfamiliar accents in the months before their second birthday. It is currently unclear, however, what drives this growing ability to contend with accented speakers. This is at least partially due to individual studies testing relatively homogeneous groups of children within narrow age ranges, thereby limiting the variability on important predictor variables. By using a more variable subject population, we here assess what factors best explain this developmental pattern. | Marieke van Heugten and Michelle Tulloch | Conference Room B |
366(abstract) | Interpreting an interpretive illusion: What the Dative Illusion tells us about processing I provide evidence of a new interpretive illusion with dative verbs. In double object datives, goals must be recipients; locational goals are unlicensed (John sent { Bill / #Davis, CA } a package). However, with extracted goals, participants allowed unlicensed locational goals with send-type but not give-type dative verbs. Shallow processing models of interpretive illusions have difficulty accounting for this; a second study suggests that correction-based accounts face issues, too, since participants were sensitive to instruction manipulations in ill-formed controls but not experimental items. I discuss how these results bear on different theories of argument structure, favoring a lexicalist approach. | Michael Wilson | Conference Room B |
367(abstract) | Mapping Memory Retrieval and Structure Building in the Brain Language comprehension is viewed as being subserved by a left-lateralized perisylvian network of brain areas such as frontal regions and temporal regions. Retrieval of memorized elements has been linked to the posterior temporal regions while structural composition has been linked to the anterior temporal regions. This study aims to differentiate between these two cognitive processes based on an analysis of fMRI timecourses collected during naturalistic language comprehension. We find that these two operations localize to spatially distinct regions of the brain. Our results corroborate previous neuroimaging studies and highlights a functional characterization of the different components of the language network. | Shohini Bhattasali and John Hale | Conference Room B |
370(abstract) | Multimodal cue integration affects referent prediction in real-time sentence comprehension: Evidence from pitch accent and beat gesture This study used a visual world eye-tracking task to investigate how beat gesture and pitch accent, which convey importance via the visual and auditory modalities, respectively, affect reference resolution during online sentence processing. Evidence from gaze fixations indicated that, similar to contrastive pitch accent, beat gesture encouraged a contrastive interpretation of referents. Additionally, beat gesture influenced interpretation of contrastive pitch accent, such that reference interpretation was contrastive when these two cues occurred conjointly but non-contrastive when contrastive pitch accent was present and beat gesture was absent. | Laura Morett and Scott Fraundorf | Conference Room B |
372(abstract) | Interference type and distractor prominence in processing reflexive-antecedent dependencies In two self-paced reading experiments, this study investigates the mechanism of retrieving the antecedent from the memory for English reflexives. It provides additional evidence for the cue-based models (e.g. Lewis and Vasishth, 2005) which predict an inhibitory interference effect when the distractor's gender and number match the retrieval cue. Reflexives and the following spillover regions were read slower and the delay was more significant when the interference was retroactive, i.e. the distractor was located between the reflexive and its antecedent. The distractor's prominence, which is related to its syntactic position, was not found to be a determinant in this process. | Zhong Chen | Conference Room B |
374(abstract) | “Some” tasks are not optimal: Concerns about truth value judgment tasks for assessing scalar implicatures It is commonly held that generating scalar implicatures is effortful. Evidence supporting this position mainly comes from verification tasks of sentences like "Some elephants are mammals." Comprehenders with fewer cognitive resources, accept such sentences more often. This study provides evidence that cognitive difficulty in this task arises from the awkwardness of trying to make a binary acceptability judgment about these awkward sentences and not from the implicature process itself. There is thus little reason to think that implicature generation is itself effortful. | Les Sikos, Minjae Kim, Jacqueline Lane and Daniel Grodner | Conference Room B |
376(abstract) | The processing of Turkish and Korean NPI licensing and intrusion: ERP evidence Our study investigates NPI intrusion effect in Turkish and Korean where word order properties yield as the typical case NPI/Licensor dependencies that are predictive/prospective (i.e., where NPIs linearly precede licensors). | So Young Lee, AYDOGAN YANILMAZ, Jiwon Yun and John Drury | Conference Room B |
377(abstract) | Verbs, not subjects, drive subject-as-agent misinterpretation in children’s comprehension of passives Child learners of many languages experience persistent difficulty in comprehending passives, possibly because of difficulty revising a subject-as-agent misinterpretation. Exp. 1 exploited the V2 properties of German matrix clauses to investigate whether this misinterpretation takes place at the subject, but found no impact on children’s passive comprehension. Exp. 2 investigated whether the misinterpretation takes place at the main verb, using VFinal properties of German embedded clauses. Children’s passive comprehension dropped significantly. Our results suggest that subject-as-agent misinterpretation is responsible for children’s poor passive comprehension, but that this takes place at the verb, rather than at the subject as previously assumed. | Lara Ehrenhofer, Kazuko Yatsushiro, Tom Fritzsche, Barbara Höhle, Jeffrey Lidz, Colin Phillips and Yi Ting Huang | Conference Room B |
378(abstract) | Semantic effects of adjective classes: evidence from memory How does the production of adjectives affect the representation of unmentioned objects that are indirectly relevant? We consider relative (tall/short), absolute (full/empty) and non-gradable adjectives (striped/square). A surprise memory task revealed that the unmentioned-but-relevant object had a stronger representation after relative adjectives, compared to both absolute and non-gradable adjectives. This is surprising given that both relative and absolute adjectives evoke a scale, which should make the unmentioned object salient. Instead, this reflects the fact that the encoding of the unmentioned-but-relevant object is stronger in the relative case because this object affects whether the adjective is true of the described object. | Kelly-Ann Blake and Daphna Heller | Conference Room B |
382(abstract) | Beating around the evergreen bush: Conventionality violations elicit P300 We used event-related potentials (ERP) to test the flexibility of pattern recognition in conventional strings using familiar idioms containing unexpected, inserted adjectives semantically congruent with the context (The group broke the freezing ice by introducing themselves). Participants saw either idioms, idioms containing inserted adjectives, or literal controls with or without the adjectives. Idioms with inserted adjectives showed an enhanced positivity (P3b) relative to literal controls with adjectives. Literal phrases elicited a late positivity relative to idioms. Findings support a view of comprehension where nonliteral, fixed language processing is categorically distinct. | Nyssa Bulkes and Vicky Lai | Conference Room B |
383(abstract) | Audience Design and Referential Success: A Test of the Simultaneity Approach This work presents evidence for the "simultaneity approach" to audience design (Mozuraitis, Stevenson and Heller, 2018), where, in tailoring referring expressions, speakers “mix” their own perspective with the perspective of their addressee. The current paper uses this approach to model three well-known results in the audience design literature, showing that the relative weighing of perspective is affected by factors such as the goals of speakers, their experience with the task and their level of uncertainty. Most importantly, these results call into question the idea that reference is governed globally by maximizing communicate success. | Daphna Heller and Suzanne Stevenson | Conference Room B |
384(abstract) | Beat gestures encode spatial semantics Theories of gesture agree that beat gestures are meaningless. Here we show that beats are pervasive and meaningful. Participants told stories suggesting literal or metaphorical motion in one of four directions: up, down, left, or right. They produced beat gestures in the direction implied by the story, much more frequently than would be expected by chance. Beat gestures were congruent with the story direction not only during literal spatial language, but also when participants used spatial metaphors for abstract motion, and when they expressed the same ideas without using any spatial language, revealing the spatial foundations of abstract thoughts. | Defu Yap, Geoff Brookshire and Daniel Casasanto | Conference Room B |
391(abstract) | Theta-band oscillatory power in free reading of stories using EEG and EM coregistration Using EEG and eye movement coregistration, we examined the effect of context during free reading of stories on theta-band power. Theta power increases to experimental language manipulations have been suggested to reflect difficulty in language processing, such as increased working memory load during sentence reanalysis, and increased demands in lexico-semantic retrieval. We found increases in theta-band power to fixated words with low compared to high contextual fit, suggesting that cognitive mechanisms to account for language processing difficulty are employed even under conditions with rich predictive affordances such as naturalistic story reading. | Max Cantor, John Trueswell and Albert Kim | Conference Room B |
394(abstract) | Intervening and non-intervening Interference Processing of the non-adjacent dependencies is subject to memory failures when interrupted by interference. One type of interference called "intrusion effect" has been observed during the online processing of Negative polarity items when parsing mechanisms appear to be "tricked" by the input, resulting in a situation where structurally ineligible licensor/NPI relationships are temporarily processed as if they were licit. Existing findings have been mainly elicited in limited contexts in terms of NPI, licensor and intrusive context types (ever, no, object relative clauses respectively). Here, we present evidence of intrusion driven by different NPI and intrusive context types. | AYDOGAN YANILMAZ and John Drury | Conference Room B |
396(abstract) | Self-organized parsing predicts encoding interference slowdowns in agreement attraction contexts A simple extension (RACE/A; van Maanen et al., 2009) of the successful ACT-R cue-based retrieval approach to sentence processing (Lewis & Vasishth, 2005) predicts that reading a verb after “the canoe by the kayaks” should be faster than after “the canoe by the cabins” due to mutual activation between “canoe” and “kayaks” stemming from their semantic similarity. By contrast, a self-organized sentence processing (SOSP) model produced slower processing at the verb with similar NPs because they compete to be the subject. Testing with self-paced reading, we found slower processing at the verb, consistent with self-organization and contra similarity-based activation enhancement. | Garrett Smith, julie franck and Whitney Tabor | Conference Room B |
398(abstract) | Nouns and verbs behave differently as fillers: expectation and interference in constructing long-distance dependencies The present study investigates how Mandarin filler-gap dependencies are comprehended when the main verb has ambiguous complement structure. We conducted two self-paced reading tasks with Mandarin relative clauses where the filler has two potential gap sites but only the second gap is correct. Our results show that the animacy feature of N-like fillers, but not V-like fillers, can be used to locate the correct gap, suggesting that verbs are processed differently from nouns as filler complements. The animacy of NP1 interferes with the retrieval of an animate N-like filler, lending support for the cue-based interference effect. | Yiwen Zhang, Hai Hu and Charles Lin | Conference Room B |
405(abstract) | Auditory perceptual simulation (APS) of voices by L2 English readers: Effects on eye movements and comprehension Auditory Perceptual Simulation (APS) refers to when readers mentally simulated the characteristics of the characters depicted in the text (e.g. Harry Potter) or other speakers’ voices (e.g. Morgan Freeman). This study examined how APS of different speech modulated L2 English readers' eye movements and comprehension during silently reading. Results showed that participants read faster when imagining the faster native English speaker’s voice than when imagining the non-native English speaker’s voice. Individual differences in L2 English speakers’ native languages (the same as the non-native speaker or not) and working memory modulated their sentence processing patterns, reading speeds, and comprehension accuracy. | Peiyun Zhou and Kiel Christianson | Conference Room B |
407(abstract) | Semantic activation and integration across parafoveal and foveal vision during sentence reading: Evidence from event-related brain potentials Parafoveal perception plays a critical role in reading. Yet our understanding of the neural mechanisms of parafoveal processing are only in their infancy. The aim of the current experiment was to use ERPs to isolate the effects of semantic activation, reflected by the N400, and semantic integration, reflected by the Late Positive Component, across parafoveal and foveal vision. We found that semantic activation was initiated in parafoveal vision, but central foveal attention was necessary to enact higher-level and task-dependent integrative processing. Findings are consistent with accounts posing serial stages of semantic activation and integration. | Brennan Payne and Kara Federmeier | Conference Room B |
409(abstract) | Not all filler-gap dependencies are perceived alike: Evidence from Tagalog We asked whether speakers could use verbal agreement in Tagalog—also known as voice morphology in grammatical descriptions—to predict the identity of the gap, and whether the validity of the agreement cue affects how aggressively they do so. We tested three types of FGD—wh-questions, relative clauses, and ay-topicalizations—and we found that agreement was indeed used to predict the gap. However, it did so in such a way that varied across construction, and that was not directly tied to cue validity. | Jed Pizarro-Guevara and Matt Wagers | Conference Room B |
411(abstract) | Does predictability effect pronominalization rate in the passage completion paradigm? The uniform information density hypothesis predicts that speakers prefer shorter to longer expressions for predictable meanings. The effects of predictability on referring expression (RE) choice are, however, contested. We predict that next-mention biases (relative probabilities of mention) following transfer-of-possession verbs are stronger or more robust than those following implicit causality verbs; and that increasing the length of antecedents increases pronominalization both overall, and in cases where the referent is more predictable. We find no evidence for an effect of predictability on RE choice following either verb type, however, although longer expressions are more likely to be pronominalized. | Ekaterina Kravtchenko and Vera Demberg | Conference Room B |
415(abstract) | Structural predictions by agreement errors Our study investigates whether the parser starts building a particular type of bi-clausal structure, namely, a relative clause (RC), upon encountering a local NPI mismatch without any bottom-up evidence (i.e., the RC head noun). We used Visual World Paradigm, and the participants' looks to the RC-head entity-to-be increased upon the detection of the NPI mismatch before the RC head was heard. The results together show that local NPI–NEG mismatch triggers the parser’s prediction of the RC analysis prior to the availability of syntactic evidence for the RC. | Takeshi Kishiyama and Yuki Hirose | Conference Room B |
4(abstract) | Use it or lose it: Delayed syntactic integration in L2 English speakers Native speakers process sentences and integrate morphosyntax incrementally, but it is unclear whether L2 speakers can do the same. L1 English speakers and L2 English (L1 Mandarin) speakers heard passive and progressive active sentences in an eye-tracking study. L2 speakers looked at the correct image during the second noun phrase or the agentive by-phrase, while L1 speakers looked at the correct image sooner (after hearing the entire verb). However, L2 speakers were not slower to process passive sentences than active sentences, unlike the L1 speakers. This suggests that L2 speakers did not integrate morphosyntax incrementally. Potential explanations are discussed. | Gwendolyn Rehrig, Nicolaus Schrum, Sten Knutsen and Karin Stromswold | Conference Room B | Session 6 | 2:00 PM - 6:15 PM |
Submission | Title | Authors | |
Chair: | Tamara Swaab | ||
Invited Speaker | What Event-related neural responses can tell us about language comprehension: A dynamic generative framework | Gina Kuperberg | 2:00 PM |
177(abstract) | Plausibility is not reducible to predictability Effects of plausibility on reading are often thought to reduce to the predictability of words. Our experiments show that they do not, when the ranges of both variables are not artificially restricted and when plausibility varies across all levels of predictability: (1) Predictability does not explain how plausible participant fillers are when plausibility is high and explains very little of the variance when it is low; (2) plausibility affects reading time independently from and later than predictability. The later effect of plausibility suggests that plausibility-sensitivity is part of a discourse integration process that is not associated with prediction. | Hong Mo Kang, Jean-Pierre Koenig and Gail Mauner | 2:40 PM |
Coffee | 3:10 PM | ||
Chair: | Jennifer Arnold | ||
26(abstract) | Predictability effects in reading require valid parafoveal preview Two eye movement experiments directly confirm a pattern suggested by previous studies: Effects of a word's predictability on first fixation and gaze duration are eliminated by invalid parafoveal preview, while word frequency effects are not. The experiments also show that the elimination of predictability effects does not depend on the lexical status of the invalid preview string. This data pattern is interpreted in terms of a modified Bayesian account of lexical processing in context. The persistence of the predictability-based N400 in the absence of parafoveal preview is also discussed. | Adrian Staub and Kirk Goddard | 3:30 PM |
96(abstract) | A multi-dimensional view of NPI licensing This study uses negative polarity item (NPI) licensing to test the timing of incremental logical-semantic and pragmatic processing. Results show that NPI licensing is sensitive to fine-grained logical-semantic properties of the licensing context beyond the presence of a licensor (self-paced reading, Experiment 1), and that there are two mechanisms acting in parallel to license NPIs based on (i) the logical-semantic properties of the licensor, and (ii) contextual pragmatic inferencing (eye-tracking, Experiment 2). These results shed new light on how different levels of representation interact to compute meaning relations during real-time comprehension. | Dan Parker | 4:00 PM |
Break | 4:30 PM | ||
Chair: | Eva Wittenberg | ||
298(abstract) | A cognitively realistic left-corner parser with visual and motor interfaces We introduce a Python3 reimplementation of ACT-R (Anderson and Lebiere 1998) in which we build an end-to-end simulation of syntactic parsing in a typical self-paced reading experiment. The model uses a left-corner parsing strategy implemented as a skill in procedural memory (following Lewis and Vasishth 2005), makes use of independently motivated components of the ACT-R framework (content-addressable declarative memory, Wagers and Phillips 2009, goal and imaginal buffers etc.), and explicitly models the motor and visual processes involved in self-paced reading. The ACT-R model can be embedded in a (Bayesian) statistical model to estimate its sub-symbolic parameters and do model comparison. | Adrian Brasoveanu and Jakub Dotlacil | 4:45 PM |
161(abstract) | Correlative adverbs mark not only scope but also contrast: Corpus and eye tracking data I report results from corpus and reading studies on the correlative construction not only-but also, and propose that the processor anticipates conjunction size and upcoming focus structures on the basis of the location of the focus sensitive adverb not only. Corpus results showed a bias for the adverb to mark the size of the conjunct transparently, unless it shifts rightward to mark focal sub-contrast. Eye tracking results showed a reading penalty only when the position of the adverb was uninformative for both syntax and focus marking, suggesting that focus structure is computed rapidly from focus sensitive elements in silent reading. | Jesse Harris | 5:15 PM |
75(abstract) | A context constructivist account of contextual diversity Word frequency effects have long served as an empirical and theoretical test bed for theories of language processing. Recently, Contextual Diversity (CD) has been proposed as a better metric of retrieval processes than word frequency. Motivated by these findings, we sketch an active account of lexical access during sentence processing: language users store statistics about contextualized lexical representations and use lexical-contextual relations to both construct context and predict words given the context. In support of this hypothesis, we demonstrate that language users store contextualized word representations and that CD effects on reading time attenuate as discourse context is constructed. | Francis Mollica, Shaorong Yan and Mike Tanenhaus | 5:45 PM |