Assessing two methods of webcam-based eye-tracking for child language research.

We assess the feasibility of conducting web-based eye-tracking experiments with children using two methods of webcam-based eye-tracking: automatic gaze estimation with the WebGazer.js algorithm and hand annotation of gaze direction from recorded webcam videos. Experiment 1 directly compares the two methods in a visual-world language task with five to six year-old children. Experiment 2 more precisely investigates WebGazer.js' spatiotemporal resolution with four to twelve year-old children in a visual-fixation task. We find that it is possible to conduct web-based eye-tracking experiments with children in both supervised (Experiment 1) and unsupervised (Experiment 2) settings - however, the webcam eye-tracking methods differ in their sensitivity and accuracy. Webcam video annotation is well-suited to detecting fine-grained looking effects relevant to child language researchers. In contrast, WebGazer.js gaze estimates appear noisier and less temporally precise. We discuss the advantages and disadvantages of each method and provide recommendations for researchers conducting child eye-tracking studies online.

The effects of maternal input on language in the absence of genetic confounds: Vocabulary development in internationally adopted children.

Previous studies have found correlations between parent input and child language outcomes, providing prima facie evidence for a causal relation. However, this could also reflect the effects of shared genes. The present study removed this genetic confound by measuring English vocabulary growth in 29 preschool-aged children (21 girls) aged 31-73 months and 17 infants (all girls) aged 15-32 months adopted from China and Eastern Europe and comparing it to speech produced by their adoptive mothers. Vocabulary growth in both groups was correlated with maternal input features; in infants with mean-length of maternal utterance, and in preschoolers with both mean-length of utterance and lexical diversity. Thus, input effects on language outcomes persist even in the absence of genetic confounds.

Word vs. World Knowledge: A developmental shift from bottom-up lexical cues to top-down plausibility.

Both 5-year-old children and adults infer the structure of a sentence as they are hearing it. Prior work, however, has found that children do not always make use of the same information that adults do to guide these inferences. Specifically, when hearing ambiguous sentences like "You can tickle the frog with the feather," children seem to ignore the aspects of the referential context that adults rely on to resolve the ambiguity-e.g., are there two frogs in the scene, one with a feather and one without? Or is there only one frog to be tickled by using a feather? The present study explored two hypotheses about children's failure to use high-level, top-down context cues to infer the structure of these ambiguous sentences: First, children may be less likely to use any top-down cue during comprehension. Second, children may only have difficulties with top-down cues that are unreliable predictors of which syntactic structure is being used. Thus, to disentangle these hypotheses, we conducted a visual world study of adults' and children's ambiguity resolution, manipulating a more reliable top-down cue (the plausibility of the interpretation) and pitting it against a robust bottom-up cue (lexical biases). We find that adults' and children's final interpretations are influenced by both sources of information: adults, however, give greater weight to the top-down cue, whereas children primarily rely on the bottom-up cue. Thus, children's tendency to make minimal use of top-down information persists even when this information is highly valid and dominates adult comprehension. We propose that children have a systematic propensity to rely on bottom-up processing to a greater degree than adults, which could reflect differences in the architecture of the adult and child language comprehension systems or differences in processing speed.

Spared bottom-up but impaired top-down interactive effects during naturalistic language processing in schizophrenia: evidence from the visual-world paradigm.

BACKGROUND: People with schizophrenia process language in unusual ways, but the causes of these abnormalities are unclear. In particular, it has proven difficult to empirically disentangle explanations based on impairments in the top-down processing of higher level information from those based on the bottom-up processing of lower level information. METHODS: To distinguish these accounts, we used visual-world eye tracking, a paradigm that measures spoken language processing during real-world interactions. Participants listened to and then acted out syntactically ambiguous spoken instructions (e.g. 'tickle the frog with the feather', which could either specify how to tickle a frog, or which frog to tickle). We contrasted how 24 people with schizophrenia and 24 demographically matched controls used two types of lower level information (prosody and lexical representations) and two types of higher level information (pragmatic and discourse-level representations) to resolve the ambiguous meanings of these instructions. Eye tracking allowed us to assess how participants arrived at their interpretation in real time, while recordings of participants' actions measured how they ultimately interpreted the instructions. RESULTS: We found a striking dissociation in participants' eye movements: the two groups were similarly adept at using lower level information to immediately constrain their interpretations of the instructions, but only controls showed evidence of fast top-down use of higher level information. People with schizophrenia, nonetheless, did eventually reach the same interpretations as controls. CONCLUSIONS: These data suggest that language abnormalities in schizophrenia partially result from a failure to use higher level information in a top-down fashion, to constrain the interpretation of language as it unfolds in real time.

It takes a village: The role of community size in linguistic regularization.

Studies of artificial language learning provide insight into how learning biases and iterated learning may shape natural languages. Prior work has looked at how learners deal with unpredictable variation and how a language changes across multiple generations of learners. The present study combines these features, exploring how word order variation is preserved or regularized over generations. We investigate how these processes are affected by (1) learning biases, (2) the size of the language community, and (3) the amount of input provided. Our results show that when the input comes from a single speaker, adult learners frequency match, reproducing the variability in the input across three generations. However, when the same amount of input is distributed across multiple speakers, frequency matching breaks down. When regularization occurs, there is a strong bias for SOV word order (relative to OSV and VSO). Finally, when the amount of input provided by multiple speakers is increased, learners are able to frequency match. These results demonstrate that both population size and the amount of input per speaker each play a role in language convergence.

Some inferences still take time: Prosody, predictability, and the speed of scalar implicatures.

Experimental pragmatics has gained many insights from understanding how people use weak scalar terms (like some) to infer that a stronger alternative (like all) is false. Early studies found that comprehenders initially interpret some without an upper bound, but later results suggest that this inference is sometimes immediate (e.g., Grodner, Klein, Carbary, & Tanenhaus, 2010). The present paper explores whether rapid inferencing depends on the prosody (i.e., summa rather than some of) or predictability of referring expressions (e.g., consistently using some to describe subsets). Eye-tracking experiments examined looks to subsets (2-of-4 socks) and total sets (3-of-3 soccer balls) following some and found early preferences for subsets in predictable contexts but not in less predictable contexts (Experiment 1 and 2). In contrast, there was no reliable prosody effect on inferencing. Changes in predictability did not affect judgments of the naturalness of some, when a discourse context was available (Experiment 3). However, predictable contexts reduced variability in speakers' descriptions of subsets and total sets (Experiment 4). Together, these results demonstrate that scalar inferences are often delayed during comprehension, but reference restriction is rapid when set descriptions can be formulated beforehand.

Getting to No: Pragmatic and Semantic Factors in Two- and Three-Year-Olds' Understanding of Negation.

Although infants say "no" early, older children have difficulty understanding its truth-functional meaning. Two experiments investigate whether this difficulty stems from the infelicity of negative sentences out of the blue. In Experiment 1, given supportive discourse, 3-year-olds (N = 16) understood both affirmative and negative sentences. However, with sentence types randomized, 2-year-olds (N = 28) still failed. In Experiment 2, affirmative and negative sentences were blocked. Two-year-olds (N = 28) now succeeded, but only when affirmatives were presented first. Thus, although discourse felicity seems the primary bottleneck for 3-year-olds' understanding of negation, 2-year-olds struggle with its semantic processing. Contrary to accounts where negatives are understood via affirmatives, both sentence types were processed equally quickly, suggesting previously reported asymmetries are due to pragmatic accommodation, not semantic processing.

Priming is swell, but it's far from simple.

Clearly, structural priming is a valuable tool for probing linguistic representation. But we don't think that the existing results provide strong support for Branigan & Pickering's (B&P's) model, largely because the priming effects are more confusing and diverse than their theory would suggest. Fortunately, there are a number of other experimental tools available, and linguists are increasingly making use of them.

The logic in language: How all quantifiers are alike, but each quantifier is different.

Quantifier words like each, every, all and three are among the most abstract words in language. Unlike nouns, verbs and adjectives, the meanings of quantifiers are not related to a referent out in the world. Rather, quantifiers specify what relationships hold between the sets of entities, events and properties denoted by other words. When two quantifiers are in the same clause, they create a systematic ambiguity. "Every kid climbed a tree" could mean that there was only one tree, climbed by all, or many different trees, one per climbing kid. In the present study, participants chose a picture to indicate their preferred reading of different ambiguous sentences - those containing every, as well as the other three quantifiers. In Experiment 1, we found large systematic differences in preference, depending on the quantifier word. In Experiment 2, we then manipulated the choice of a particular reading of one sentence, and tested how this affected participants' reading preference on a subsequent target sentence. We found a priming effect for all quantifiers, but only when the prime and target sentences contained the same quantifier. For example, all-a sentences prime other all-a sentences, while each-a primes each-a, but sentences with each do not prime sentences with all or vice versa. In Experiment 3, we ask whether the lack of priming across quantifiers could be due to the two sentences sharing one fewer word. We find that changing the verb between the prime and target sentence does not reduce the priming effect. In Experiment 4, we discover one case where there is priming across quantifiers - when one number (e.g. three) is in the prime, and a different one (e.g. four) is in the target. We discuss how these findings relate to linguistic theories of quantifier meaning and what they tell us about the division of labor between conceptual content and combinatorial semantics, as well as the mental representations of quantification and of the abstract logical structure of language.

The use of prosody during syntactic processing in children and adolescents with autism spectrum disorders.

In this study, we employed an eye-gaze paradigm to explore whether children (ages 8-12) and adolescents (ages 12-18) with autism spectrum disorders (ASDs) are able to use prosodic cues to determine the syntactic structure of an utterance. Persons with ASD were compared to typically developing (TD) peers matched on age, IQ, gender, and receptive language abilities. The stimuli were syntactically ambiguous but had a prosodic break that indicated the appropriate interpretation (feel the frog … with the feather vs. feel … the frog with the feather). We found that all groups were equally sensitive to the initial prosodic cues that were presented. Children and teens with ASD used prosody to interpret the ambiguous phrase as rapidly and efficiently as their TD peers. However, when a different cue was presented in subsequent trials, the younger ASD group was more likely to respond in a manner consistent with the initial prosodic cue rather than the new one. Eye-tracking data indicated that both younger groups (ASD and TD) had trouble shifting their interpretation as the prosodic cue changed, but the younger TD group was able to overcome this interference and produce an action consistent with the prosodic cue.

Development of the first-mention bias.

In many contexts, pronouns are interpreted as referring to the character mentioned first in the previous sentence, an effect called the 'first-mention bias'. While adults can rapidly use the first-mention bias to guide pronoun interpretation, it is unclear when this bias emerges during development. Curiously, experiments with children between two and three years old show successful use of order of mention, while experiments with older children (four to five years old) do not. While this could suggest U-shaped development, it could also reflect differences in the methodologies employed. We show that children can indeed use first-mention information, but do so too slowly to have been detected in previous work reporting null results. Comparison across the present and previously published studies suggests that the speed at which children deploy first-mention information increases greatly during the preschool years.

Love is hard to understand: the relationship between transitivity and caused events in the acquisition of emotion verbs.

Famously, dog bites man is trivia whereas man bites dog is news. This illustrates not just a fact about the world but about language: to know who did what to whom, we must correctly identify the mapping between semantic role and syntactic position. These mappings are typically predictable, and previous work demonstrates that young children are sensitive to these patterns and so could use them in acquisition. However, there is only limited and mixed evidence that children do use this information to guide acquisition outside of the laboratory. We find that children understand emotion verbs which follow the canonical CAUSE-VERB-PATIENT pattern (Mary frightened/delighted John) earlier than those which do not (Mary feared/liked John), despite the latter's higher frequency, suggesting children's generalization of the mapping between causativity and transitivity is broad and active in acquisition.

How words can and cannot be learned by observation.

Three experiments explored how words are learned from hearing them across contexts. Adults watched 40-s videotaped vignettes of parents uttering target words (in sentences) to their infants. Videos were muted except for a beep or nonsense word inserted where each "mystery word" was uttered. Participants were to identify the word. Exp. 1 demonstrated that most (90%) of these natural learning instances are quite uninformative, whereas a small minority (7%) are highly informative, as indexed by participants' identification accuracy. Preschoolers showed similar information sensitivity in a shorter experimental version. Two further experiments explored how cross-situational information helps, by manipulating the serial ordering of highly informative vignettes in five contexts. Response patterns revealed a learning procedure in which only a single meaning is hypothesized and retained across learning instances, unless disconfirmed. Neither alternative hypothesized meanings nor details of past learning situations were retained. These findings challenge current models of cross-situational learning which assert that multiple meaning hypotheses are stored and cross-tabulated via statistical procedures. Learners appear to use a one-trial "fast-mapping" procedure, even under conditions of referential uncertainty.

Understanding words in context: A naturalistic EEG study of children's lexical processing.

When listening to speech, adults rely on context to anticipate upcoming words. Evidence for this comes from studies demonstrating that the N400, an event-related potential (ERP) that indexes ease of lexical-semantic processing, is influenced by the predictability of a word in context. We know far less about the role of context in children's speech comprehension. The present study explored lexical processing in adults and 5-10-year-old children as they listened to a story. ERPs time-locked to the onset of every word were recorded. Each content word was coded for frequency, semantic association, and predictability. In both children and adults, N400s reflect word predictability, even when controlling for frequency and semantic association. These findings suggest that both adults and children use top-down constraints from context to anticipate upcoming words when listening to stories.

It's All in the Interaction: Early Acquired Words Are Both Frequent and Highly Imageable.

Prior studies have found that children are more likely to learn words that are frequent in the input and highly imageable. Many theories of word learning, however, predict that these variables should interact, particularly early in development: frequency of a form is of little use if you cannot infer its meaning, and a concrete word cannot be acquired if you never hear it. The present study explores this interaction, how it changes over time and its relationship to syntactic category effects in children acquiring American English. We analyzed 1461 monolingual English-speaking children aged 1;4-2;6 from the MB-CDI norming study (Fenson et al., 1994). Word frequency was estimated from the CHILDES database, and imageability was measured using adult ratings. There was a strong over-additive interaction between frequency and imageability, such that children were more likely to learn a word if it was both highly imageable and very frequent. This interaction was larger in younger children than in older children. There were reliable differences between syntactic categories independent of frequency and imageability, which did not interact with age. These findings are consistent with theories in which children's early words are acquired by mapping frequent word forms onto concrete, perceptually available referents, such that highly frequent items are only acquired if they are also imageable, and vice versa.

The truth about chickens and bats: ambiguity avoidance distinguishes types of polysemy.

Words mean different things in different contexts, a phenomenon called polysemy. People talk about lines of both people and poetry, and about both long distances and long times. Polysemy lets a limited vocabulary capture a great variety of experiences, while highlighting commonalities. But how is this achieved? Are polysemous senses contextually driven modifications of core meanings, or must each sense be memorized separately? We show that participants' ability to avoid referentially ambiguous descriptions of pictures named by polysemous words provides evidence for both possibilities. When senses followed a regular pattern (e.g., animals and the foodstuffs derived from them; noisy chicken, tasty chicken), participants avoided using ambiguous labels in referentially ambiguous situations (e.g., both types of chicken were present), a result indicating that they noticed a common meaning. But when senses were idiosyncratically related (e.g., sheet of glass, drinking glass), participants frequently produced ambiguous labels, a result indicating that the meanings were separately stored. We discuss implications for the relationship between word meanings and concepts.

Potentially recursive structures emerge quickly when a new language community forms.

Human languages can express an infinite number of thoughts despite having a finite set of words and rules. This is due, in part, to recursive structures, which allow us to embed one instance of a rule inside another. We investigated the origins of recursion by studying the development of Nicaraguan Sign Language (LSN), which emerged in the last 40 years and is not derived from any existing language. Before this, deaf individuals in Nicaragua lacked access to language models and each individual created their own gestural system, called homesign. We tested four groups: homesigners, who represent the point of origin, and the first three generations of LSN signers, who represent consecutive stages in the language's development. We used a task that was designed to elicit sentences with relative clauses, a device that allows for the recursive embedding of sentences inside of sentences (e.g., [the girl [who was drawing] removed the picture]). Signers in all three LSN cohorts consistently produced utterances that appeared to have embedded predicates (girl draw remove picture) which served the function of a relative clause (picking out the correct member of a set, based on previously mentioned information). Furthermore, in these utterances, the first verb was shorter than the second and shorter than the same verb in parallel unembedded structures. In contrast, homesigners produced similar utterances in embedded and unembedded contexts. They did not reintroduce previously mentioned information or produce reduced verb forms in the embedded context. These results demonstrate that syntactic embedding that is potentially recursive can emerge very early in a language. These embedded predicates, however, may not be widespread, or systematically marked, in homesign systems. This raises the possibility that the emergence of recursive linguistic structure is a consequence of interaction within a language community. These findings pave the way for future work which investigates the syntactic form of these embedded predicates and explores whether multiple levels of embedding are possible.

Disentangling the effects of cognitive development and linguistic expertise: a longitudinal study of the acquisition of English in internationally-adopted children.

Early language development is characterized by predictable changes in the words children produce and the complexity of their utterances. In infants, these changes could reflect increasing linguistic expertise or cognitive maturation and development. To disentangle these factors, we compared the acquisition of English in internationally-adopted preschoolers and internationally-adopted infants. Parental reports and speech samples were collected for 1 year. Both groups showed the qualitative shifts that characterize first-language acquisition. Initially, they produced single-word utterances consisting mostly of nouns and social words. The appearance of verbs, adjectives and multiword utterances was predicted by vocabulary size in both groups. Preschoolers did learn some words at an earlier stage than infants, specifically words referring to the past or future and adjectives describing behavior and internal states. These findings suggest that cognitive development plays little role in the shift from referential terms to predicates but may constrain children's ability to learn some abstract words.

Counting the nouns: simple structural cues to verb meaning.

Two-year-olds use the sentence structures verbs appear in--subcategorization frames--to guide verb learning. This is syntactic bootstrapping. This study probed the developmental origins of this ability. The structure-mapping account proposes that children begin with a bias toward one-to-one mapping between nouns in sentences and participant roles in events. This account predicts that subcategorization frames should guide very early verb learning, if the number of nouns in the sentences is informative. In 3 experiments, one hundred and thirty-six 21- and 19-month-olds assigned appropriately different interpretations to novel verbs in transitive ("He's gorping him!") versus intransitive sentences ("He's gorping!") differing in their number of nouns. Thus, subcategorization frames guide verb interpretation in very young children. These findings constrain theoretical proposals about mechanisms for syntactic bootstrapping.

The Emergence of Natural Language Quantification.

Classical quantifiers (like "all," "some," and "none") express relationships between two sets, allowing us to make generalizations (like "no elephants fly"). Devices like these appear to be universal in human languages. Is the ubiquity of quantification due to a universal property of the human mind or is it attributable to more gradual convergence through cultural evolution? We investigated whether classical quantifiers are present in a new language emerging in isolation from other languages, Nicaraguan Sign Language (NSL). An observational study of historical data collected in the 1990s found evidence of potential quantifier forms. To confirm the quantificational meaning of these signs, we designed a production study that elicited, from three age cohorts of NSL signers (N = 17), three types of quantifiers: universal (all), existential (some), and negative (none). We find evidence for these classical quantifiers in the very first generation of signers, suggesting they may reflect a universal property of human cognition or a very rapid construction process.