Presentations

Abstract

Usage-based approaches to language acquisition (e.g. Tomasello, 2003) propose that children use multi-word utterances – chunks – to build up grammatical knowledge from recurring patterns in their linguistic input. We investigate the changing influence of this statistical, chunk-based learning on children’s language production over time using the CAPPUCCINO model (McCauley & Christiansen, 2011). This model simulates child language production using chunks extracted from caregivers’ speech.
We selected orthographic transcriptions of conversations between 6 North American children and their caregivers, by sampling transcripts at 6-month intervals between 1;0 and 4;0 (Providence; Demuth, Culbertson, & Alter, 2006). The model parsed caregivers’ utterances for each child by comparing the transitional probabilities between words to a running average transitional probability, making splits between word chunks when the transitional probability between two words dropped below the current average. At the same time, the model also tracked the transitional probabilities between these discovered chunks. After training the model, we simulated children’s sentence production by reconstructing the utterances they actually used in the transcript from the chunk-to-chunk probabilities detected in the caregivers’ speech.
The number of child utterances that were reconstructed correctly based on transitional probabilities between chunks in the caregivers’ speech decreased over time (β = - 0.720, SE = 0.157, p < 0.001). However, the number of child utterances that contained words or chunks the caregivers did not use, increased (β = 0.547, SE = 0.064, p < 0.001). In other words, these results indicate that, over time, children’s speech less directly imitates chunk sequences in their caregivers’ speech, partly because their chunk combinations become more inventive. We discuss how these findings fit within broader usage-based approaches to language acquisition.

Abstract

Transitions between speakers in conversation are usually smooth, lasting around 200 milliseconds. Such rapid response latencies suggest that, at least sometimes, responders must begin planning their response before the ongoing turn is finished. Indeed, evidence from EEG suggests that listeners start planning their responses to questions as soon as they can, often midway through the incoming turn [1]. But given substantial overlap in the neural hardware for language production and comprehension, early response planning might incur a cost on participants’ concurrent comprehension of the ongoing turn. Do early responses come at the expense of less careful listening? We performed an EEG study in which participants played an interactive game with a confederate partner. Participants saw two pictures on their screen (e.g., a banana and a pineapple), then heard a (prerecorded) question from their partner, and then responded verbally by naming the correct picture. Participants were made to believe that their partner spoke to them live. Examples of the conditions in the experiment: 1. Early planning: 'Which object is curved and is considered to be fruit/healthy?'; 2. Late planning: 'Which object is considered to be fruit/healthy and is curved?' (response: 'the banana'). The questions were designed such that participants could start planning their response early (Example 1) or late (Example 2) in the turn. Crucially, in another part of the turn, we included either an expected word (e.g., 'fruit') or an unexpected one (e.g., 'healthy') to elicit a differential N400 effect. Our aims were two-fold: replicating the prior planning effect [1] and testing the effect of planning on comprehension. First, our results largely replicated the earlier study [1], showing a large positivity in the ERPs and an alpha/beta reduction in the time-frequency domain, both immediately following the onset of the critical information when participants could have first started planning their verbal response (i.e., 'curved'). As before [1], we interpret these effects as indicating the start of response planning. Second, and more importantly, we hypothesized that the N400 effect (the ERP difference between 'fruit' and 'healthy') would be attenuated when participants were already planning a response (i.e., in early vs. late planning). In contrast, we found an N400 effect of similar size in both the early and late planning conditions, although a small late positivity was only found in the late planning condition. Interestingly, we found a positive correlation between participants' overall response time and the size of the N400 effect after planning had started (i.e., in early planning), illustrating a trade-off between comprehension and production during turn taking. That is, quick responders showed a smaller N400 effect. We argue that their focus on production planning reduced their attention to the incoming audio signal and probably also their predictive processing, leading to a smaller N400 effect. Slow responders focused instead on the audio signal, preserving their N400 effect but delaying their response. Reference [1]: Bögels, S., Magyari, L., & Levinson, S. C. (2015). Neural signatures of response planning occur midway through an incoming question in conversation. Scientific Reports, 5: 12881. Topic Area: Meaning: Discourse and Pragmatics

Abstract

Speaker transitions in conversation are often brief, with minimal vocal overlap. Signed languages appear to defy this pattern with frequent, long spans of simultaneous signing. But recent evidence suggests that turn boundaries in signed language may only include the content-bearing parts of the turn (from the first stroke to the last), and not all turn-related movement (from first preparation to final retraction). We tested whether signers were able to anticipate “stroke-to-stroke” turn boundaries with only minimal conversational context. We found that, indeed, signers anticipated turn boundaries at the ends of turn-final strokes. Signers often responded early, especially when the turn was long or contained multiple possible end points. Early responses for long turns were especially apparent for interrogatives—long interrogative turns showed much greater anticipation compared to short ones.

Abstract

In spoken interactions, interlocutors carefully plan and time their utterances, minimising gaps and overlaps between consecutive turns.1 Cross-linguistic comparison indicates that spoken languages vary minimally in their turn timing.2 Pre-linguistic vocal turn taking has also been attested in the first six months of life.3 These observations suggest that the turn-taking system provides a universal basis for our linguistic capacities.4 It remains an open question, however, whether precisely-timed turn taking is solely a property of speech. It has previously been argued that, unlike speakers, signers do not attend to the one-at-a-time principle, and instead form a collaborative turn-taking floor with their interlocutors, thus having a higher degree of social tolerance for overlap.5 But recent corpus analyses of Sign Language of the Netherlands (NGT) have revealed that, although simultaneous signing is more frequent in NGT than overlapping speech in spoken languages, the additional overlap may come as a consequence of having larger and thus slower articulators.6 The beginnings and ends of signed utterances are bookended by preparatory and retractive movements — phonetically necessary articulations that do not add to the interpretation of the utterance.7 When turn timing is calculated on the basis of stroke-to-stroke turn boundaries, NGT turn timing and turn overlap are consistent with documented averages for spoken turn taking.6 This paper presents new experimental evidence supporting the psychological reality of stroke-to-stroke turn boundaries for signers by using an adapted button-press paradigm, originally developed for measuring spoken turn prediction.8 Our results indicate that signers indeed anticipated turn boundaries at the ends of turn-final strokes. These findings are the first to experimentally support the idea that signers use something like stroke-to-stroke turn boundaries to coordinate conversational turns. They also suggest that linguistic processing, represented by participant age and age of acquisition, plays a role in the ability to use precisely-timed turns in conversation.

Abstract

In spontane gesprekken wisselen gebaarders steeds vlug van beurt. Gebarentaalgebruikers moeten daarom steeds op het juiste moment naar de juiste persoon kijken. Hoe voorspellen gebaarders wanneer de beurt gaat wisselen en wie deze overneemt? Wij hebben de eerste vraag onderzocht door verschillende groepen gebarentaalgebruikers (doven en horenden, jong en oud, verschillende regios) te testen. Omdat er in Nijmegen weinig (dove) gebaarders wonen, hebben we dit gedaan in ons lab op wielen: de Gebarenbus.

Abstract

We investigate children’s online predictive processing as it occurs naturally, in conversation. We showed 129 children (1;0–7;0) short videos of improvised conversation between puppets, controlling for available linguistic information through phonetic manipulation: normal, prosody only (lowpass filtered), lexical only (rhythm controlled and pitch flattened), and none (multi-talker babble). We tracked their eye movements during the videos, measuring their anticipatory looks to upcoming speakers at points of turn switch (e.g., after a question and before an answer). Even one- and two-year-old children made accurate and spontaneous predictions about when a turn-switch would occur: they gazed at the upcoming speaker before they heard a response begin. By age three, children distinguished between different types of response-eliciting speech acts, looking faster to question- than non-question responses—but only when all linguistic information was available. By age seven, children’s gaze behavior also distinguished between rising and non-rising turns in the prosody only condition. These predictive skills rely on both lexical and prosodic information together, and are not tied to either type of information alone. We suggest that children integrate prosodic, lexical, and visual information to effectively predict upcoming linguistic material in conversation.

Abstract

Combining data from two longitudinal studies of young children, we track the development of turn-timing in spontaneous infant-caregiver interactions. We focus on three aspects of timing: overlap, gap, and delay marking. We find evidence for early development of turn-timing skills, in-line with the Interaction Engine Hypothesis. (see attached .pdf for our 2-page abstract)