Yi Ting Huang

Visual-world eye-tracking has long been a useful tool for measuring young children’s real-time interpretation of words and sentences. Recently, researchers have extended this method to virtual platforms to reduce equipment costs and recruit more diverse participants. However, there is currently limited guidance on best practices, which require individual researchers to invent their own methodologies and may prevent broader adoption. Here, we present three broad approaches for implementing nine remote visual-world eye-tracking studies, and show that this method is highly feasible for assessing fine-grained language processing across populations of varying ages, clinical statuses, and socioeconomic status backgrounds. We outline strategic methods for conducting this research effectively, including strategies for experimental design, data collection, and data analysis given the variable conditions outside of a lab setting. We adopt four criteria for evaluating success for this method: 1) Minimal subject attrition relative to in-person studies, 2) Minimal track loss relative to conventional eye-tracking, 3) Conceptual replication of previous findings, and 4) Evidence of broadening participation. These case studies provide a thorough guide to future researchers looking to conduct remote eye-tracking research with developmental populations. Ultimately, we conclude that visual-world eye-tracking using internet-based methods is feasible for research with young children and may provide a relatively inexpensive option that can reach a broader, more diverse set of participants.

During sentence comprehension, young children anticipate syntactic structures using early-arriving words and have difficulties revising incorrect predictions using late-arriving words. However, nearly all work to date has focused on syntactic parsing in idealized speech environments, and little is known about how children's strategies for predicting and revising meanings are affected by signal degradation. This study compares comprehension of active and passive sentences in natural and vocoded speech. In a word-interpretation task, 5-year-olds inferred the meanings of novel words in sentences that (1) encouraged agent-first predictions (e.g., The blicket is eating the seal implies The blicket is the agent), (2) required revising predictions (e.g., The blicket is eaten by the seal implies The blicket is the theme), or (3) weakened predictions by placing familiar nouns in sentence-initial position (e.g., The seal is eating/eaten by the blicket). When novel words promoted agent-first predictions, children misinterpreted passives as actives, and errors increased with vocoded compared to natural speech. However, when familiar words were sentence-initial that weakened agent-first predictions, children accurately interpreted passives, with no signal-degradation effects. This demonstrates that signal quality interacts with interpretive processes during sentence comprehension, and the impacts of speech degradation are greatest when late-arriving information conflicts with predictions.

Children often interpret first noun phrases (NP1s) as agents, which improves comprehension of actives but hinders passives. While children sometimes withhold the agent‐first bias, the reasons remain unclear. The current study tests the hypothesis that children default to the agent‐first bias as a “best guess” of role assignment when they face uncertainty about sentence properties. Thus, rather than always relying on early‐arriving cues, children can attend to different sentence cues across communicative contexts. To test this account, we manipulated interpretive uncertainty by varying cues to the discourse status of initial arguments (referring to new vs. given entities) and measured interpretation accuracy for active (where the agent‐first bias predicted verb morphology) and passive sentences (where the two conflicted). Across three experiments, we found that children relied on the agent‐first bias more when new discourse entities were signaled by definite NP1s, reference to unmentioned entities, and novel words. This, in turn, led to higher accuracy for actives relative to passives. In contrast, when given entities were implied through pronoun NP1s, reference to mentioned entities, and known words, children avoided the agent‐first bias and instead assigned roles using more reliable but later‐arriving verb morphology. This led to similar comprehension accuracy across constructions. These findings suggest that children simultaneously interpret relations between sentences (e.g., discourse continuity) and within sentences (e.g., role assignment), such that commitments to the former can influence parsing cues for the latter.

Children acquire argument structure through distributional evidence, but how does this interacts with event semantics and existing verb knowledge? The current study compares verb learning in adult speakers of Japanese (where lexical causatives span wider semantic categories) and English (where alternation is more restricted). In the Fully alternating familiarization, internally caused events (similar to descend) and externally caused events (similar to rock) occurred with novel verbs in transitive and intransitive frames. In the Semi-alternating familiarization, internally caused events occurred with intransitives only while externally caused events occurred with both frames. During the test, participants rated the naturalness of transitive and intransitive descriptions for all events. For internally caused events, transitives were rated equivalently low across speakers after the Semi-alternating familiarization (which featured only intransitives), while intransitives were more felicitous for English compared to Japanese speakers. For externally caused events, all participants rated transitives higher than intransitives, despite equal occurrence across familiarizations. This may be related to the presence of salient animate agents in scenes. Together, this suggests that syntactic and semantic biases concurrently influence the interpretation of distributional evidence. Input statistics are interpreted through existing syntactic representations, and salient semantic distinctions may exert a bias for syntactic instantiations.

Reference production is often studied through single dimensions of contrast (e.g., “tall glass” when there are one or two glasses of varying height). Yet real-world communication is rarely so simple, raising questions about the factors guiding more complex referents. The current study examines decisions to mention set relations (e.g., using quantity-denoting expressions like “some of the houses” to refer to two of five houses) versus object categories only (e.g., using bare plurals like “houses”). Two experiments used vignettes to vary discourse focus on objects (prominent vs. nonprominent) and scenes to vary the set type described (subset vs. total set). Speakers were more likely to communicate set relations of prominent objects, particularly when they elicit high name agreement in the case of total sets. Speakers’ use of quantity-denoting expressions also increased listeners’ sensitivity to set relations in an object-matching task. This suggests that unlike simpler forms of modification that often decrease with greater focus, quantity-denoting expressions provide additional information about the set relations of prominent referents.