English Child-directed Speech Research Articles

Conceptual Hierarchy in Child-Directed Speech: Implicit Cues are More Reliable

ABSTRACT In order for children to understand and reason about the world in an adult-like fashion, they need to learn that conceptual categories are organized in a hierarchical fashion (e.g., a dog is also an animal). While children learn from their first-hand observation of the world, social knowledge transmission via language can also play an important role in this learning. Previous studies have documented several cues in parental talk that can help children learn about conceptual hierarchy. However, these studies have used different datasets and methods that have made it difficult to compare the relative usefulness of various linguistic cues to conceptual knowledge and to test whether they scale up to naturalistic speech. Here, we study a large-scale corpus of English child-directed speech – collected in North America and the UK – and used a unified classification-based evaluation method which allowed us to investigate and compare cues that vary in terms of how explicit the information they offer is. We found the more explicit cues to be too sparse or too noisy in child-directed speech, making them unlikely to support robust learning. In contrast, the implicit cues offered a more reliable source of information. Further, we investigated developmental changes from 3 to 6 years of age, and we found no differences in the availability of these cues in the input. Our work confirms the utility of caregiver talk for conveying conceptual information and supporting the development of early taxonomic knowledge. It provides a first step toward a cognitive model that would combine perceptual- and language-based mechanisms, leading to a more comprehensive account of children’s conceptual development.

Sociophonetic variation in English /l/ in the child-directed speech of English-Malay bilinguals

Three realisations of syllable-final /l/ have been described in previous work on Singapore English: vocalised-l (or deleted-l in some phonetic contexts; the local norms), dark-l (a form associated with the exonormative standards), and clear-l (a Malay-derived phonetic trait observed in the speech of some English-Malay bilinguals). This study examined whether, how and why Singaporean English-Malay bilinguals vary their English /l/ in their child-directed speech, and whether the phonetic variation, if any, could be socially-conditioned. The laterals in the English child-directed speech of ten father-mother dyads with their preschoolers were analysed using auditory and acoustic methods. All participants were simultaneous or early English-Malay bilinguals. The findings revealed that in informal contexts, both mothers and fathers used a relatively clearer /l/ in all syllable positions. Contrastingly, in formal contexts that involved teaching and learning, the coda laterals of mothers were significantly darker, thereby exhibiting positional contrast between onset and coda laterals. They also produced significantly more vocalised-l in these contexts. Fathers, however, did not show differentiation in the darkness of the laterals, nor did their laterals show significant positional differences in formal contexts, although some fathers of younger children did produce more vocalised-l than they did in informal contexts. The variation observed was discussed by exploring the potential socio-indexical meanings of these variants of /l/ within the context of variationist accounts of Singapore English and by drawing parallels with socially-conditioned variation in bilectal monolinguals and ethnolect speakers. Differences between maternal and paternal CDS patterns could be attributed to gender roles and cultural expectations of mothers’ dominant role in child-rearing, and may also be a result of and enabled by Malay women’s potentially more complex repertoire range.

The strength of phonological cues for noun categorization in child-directed speech

Abstract One of the most important tasks for language learning children is the identification of the grammatical category to which words belong. This is essential in order to be able to form grammatically correct utterances. The present study investigates how phonological information might help English-learning infants in the categorization of nouns. We analyze four different corpora of English child-directed speech in order to explore the reliability with which words are represented in mothers’ speech based on several phonological criteria. The results of the analysis confirm the prediction that most of the nouns to which English-learning children are exposed share several phonological characteristics, which would allow their early classification in the same grammatical category.

Crosslinguistic Differences in the Encoding of Causality: Transitivity Preferences in English and Japanese Children and Adults

ABSTRACTLanguages differ in how they encode causal events, placing greater or lesser emphasis on the agent or patient of the action. Little is known about how these preferences emerge and the relative influence of cognitive biases and language-specific input at different stages in development. In these studies, we investigated the emergence of sentence preferences to describe causal events in English- and Japanese-speaking children (aged three and five years) and compared this to preferences displayed by adults. We studied two factors suggested to influence this choice: Language (Corpus study & Experiment 1) and Intentionality (Experiment 2). Participants watched videos depicting familiar and novel causal actions, and made a best-match choice between a transitive and intransitive description. We found a stronger preference for intransitive sentences with causal verbs and more frequent argument omissions in Japanese child-directed speech than in English child-directed speech. The trajectory of acquisition in the selection of transitive sentences for causal events differed between languages. For intentionality, with familiar verbs both Japanese and English speakers selected fewer transitives for accidental than intentional scenes, but this pattern was more pronounced in Japanese speakers. However, with novel verbs, only five-year-olds and adults showed this preference. These data provide important new information to constrain theories about the process of learning to map event structure to language, and its interdependence with intentionality and the distributional properties of linguistic input to children.

Adjective forms and functions in British English child-directed speech.

Adjectives are essential for describing and differentiating concepts. However, they have a protracted development relative to other word classes. Here we measure three- and four-year-olds' exposure to adjectives across a range of interactive and socioeconomic contexts to: (i) measure the syntactic, semantic, and pragmatic variability of adjectives in child-directed speech (CDS); and (ii) investigate how features of the input might scaffold adjective acquisition. In our novel corpus of UK English, adjectives occurred more frequently in prenominal than in postnominal (predicative) syntactic frames, though postnominal frames were more frequent for less-familiar adjectives. They occurred much more frequently with a descriptive than a contrastive function, especially for less-familiar adjectives. Our findings present a partial mismatch between the forms of adjectives found in real-world CDS and those forms that have been shown to be more useful for learning. We discuss implications for models of adjective acquisition and for clinical practice.

Lexical category acquisition is facilitated by uncertainty in distributional co-occurrences.

This paper analyzes distributional properties that facilitate the categorization of words into lexical categories. First, word-context co-occurrence counts were collected using corpora of transcribed English child-directed speech. Then, an unsupervised k-nearest neighbor algorithm was used to categorize words into lexical categories. The categorization outcome was regressed over three main distributional predictors computed for each word, including frequency, contextual diversity, and average conditional probability given all the co-occurring contexts. Results show that both contextual diversity and frequency have a positive effect while the average conditional probability has a negative effect. This indicates that words are easier to categorize in the face of uncertainty: categorization works best for words which are frequent, diverse, and hard to predict given the co-occurring contexts. This shows how, in order for the learner to see an opportunity to form a category, there needs to be a certain degree of uncertainty in the co-occurrence pattern.

An argument from acquisition: Comparing English metrical stress representations by how learnable they are from child-directed speech

ABSTRACTIt has long been recognized that there is a natural dependence between theories of knowledge representation and theories of knowledge acquisition, with the idea that the right knowledge representation enables acquisition to happen as reliably as it does. Given this, a reasonable criterion for a theory of knowledge representation is that it be useful for acquisition, particularly in nontrivial learning situations. We propose quantitative learnability metrics meant to capture how useful a representation is for acquisition. We then apply these metrics to the case study of English metrical stress, a language that is notorious for having nonproductive aspects in its grammar and so being nontrivial to learn a productive grammar for. We examine three theories of metrical stress representation, assessing their learnability via these metrics from English child-directed speech at different stages of linguistic development. We find that while all three theories are only somewhat useful at the initial stages of stress acquisition, they are far more useful at later stages and define a grammar able to capture the vast majority of English children’s acquisitional intake. Interestingly, we also find that the proposed English grammars in each representational theory are not the grammars most easily learnable from English child-directed speech. Instead, minor changes to each English grammar yield an English-like grammar that is far more learnable, suggesting that these alternative English grammars should be given further theoretical, experimental, and computational consideration. We discuss implications for both theories of representation and theories of acquisition.

Comprehension without segmentation: a proof of concept with naive discriminative learning

ABSTRACTCurrent theories of auditory comprehension assume that the segmentation of speech into word forms is an essential prerequisite to understanding. We present a computational model that does not seek to learn word forms, but instead decodes the experiences discriminated by the speech input. At the heart of this model is a discrimination learning network trained on full utterances. This network constitutes an atemporal long-term memory system. A fixed-width short-term memory buffer projects a constantly updated moving window over the incoming speech onto the network's input layer. In response, the memory generates temporal activation functions for each of the output units. We show that this new discriminative perspective on auditory comprehension is consistent with young infants' sensitivity to the statistical structure of the input. Simulation studies, both with artificial language and with English child-directed speech, provide a first computational proof of concept and demonstrate the importance of utterance-wide co-learning.

Morphosyntactic cues to noun categorization in English child-directed speech

Mainstream research in Linguistics claims that grammatical regularities are scarcely represented in the linguistic input to which children are exposed. However, recent empirical research shows that child-directed speech contains a series of reliable cues that might assist young language learners in language development. The present study aims at testing whether English child-directed speech contains morphosyntactic regularities which might be robust enough for infants to group nouns in their grammatical category. The results from the study show that, in fact, the kind of input available to English-learning infants contains reliable and consistent distributional cues to account for most of the nouns to which children are exposed.

Constructional ‘scene encoding’ and acquisition: Mothers’ use of argument structure constructions in English child-directed speech

AbstractConstruction-based language models assume that grammar is meaningful and learnable from experience. Focusing on five of the most elementary argument structure constructions of English, a large-scale corpus study of child-directed speech (CDS) investigates exactly which meanings/functions are associated with these patterns in CDS, and whether they are indeed specially indicated to children by their caretakers (as suggested by previous research, cf. Goldberg, Casenhiser and Sethuraman 2004). Collostructional analysis (Stefanowitsch and Gries 2003) is employed to uncover significantly attracted verb-construction combinations, and attracted pairs are classified semantically in order to systematise the attested usage patterns of the target constructions. The results indicate that the structure of the input may aid learners in making the right generalisations about constructional usage patterns, but such scaffolding is not strictly necessary for construction learning: not all argument structure constructions are coherently semanticised to the same extent (in the sense that they designate a single schematic event type of the kind envisioned in Goldberg’s [1995] ‘scene encoding hypothesis’), and they also differ in the extent to which individual semantic subtypes predominate in learners’ input

“Frequent Frames” in German Child-Directed Speech: A Limited Cue to Grammatical Categories

Mintz (2003) found that in English child-directed speech, frequently occurring frames formed by linking the preceding (A) and succeeding (B) word (A_x_B) could accurately predict the syntactic category of the intervening word (x). This has been successfully extended to French (Chemla, Mintz, Bernal, & Christophe, 2009). In this paper, we show that, as for Dutch (Erkelens, 2009), frequent frames in German do not enable such accurate lexical categorization. This can be explained by the characteristics of German including a less restricted word order compared to English or French and the frequent use of some forms as both determiner and pronoun in colloquial German. Finally, we explore the relationship between the accuracy of frames and their potential utility and find that even some of those frames showing high token-based accuracy are of limited value because they are in fact set phrases with little or no variability in the slot position.

When Unbiased Probabilistic Learning Is Not Enough: Acquiring a Parametric System of Metrical Phonology

Parametric systems have been proposed as models of how humans represent knowledge about language, motivated in part as a way to explain children's rapid acquisition of linguistic knowledge. Given this, it seems reasonable to examine if children with knowledge of parameters could in fact acquire the adult system from the data available to them. That is, we explore an argument from acquisition for this knowledge representation. We use the English metrical phonology system as a nontrivial case study and test several computational models of unbiased probabilistic learners. Special attention is given to the modeled learners' input and the psychological plausibility of the model components in order to consider the learning problem from the perspective of children acquiring their native language. We find that such cognitively inspired unbiased probabilistic learners uniformly fail to acquire the English grammar proposed in recent metrical studies from English child-directed speech, suggesting that probabilistic learning alone is insufficient to acquire the correct grammar when using this parametric knowledge representation. Several potential sources of this failure are discussed, along with their implications for the parametric knowledge representation and the trajectory of acquisition for English metrical phonology.

Listening Preference for the Native Language Compared to an Unfamiliar Language in Hearing and Hearing-Impaired Infants After Cochlear Implantation

The goal of this study was to investigate the preference for the native language compared with an unfamiliar language in normally hearing (NH) and hearing-impaired infants with cochlear implants (CIs). Preference for the native language is an important step in the process of language acquisition because it helps infants to attend to the important signals of their environment. This preference has been shown very early in the infants' life. In the case of infants with CI, it is of interest to determine whether they show similar perceptual biases as NH infants despite their impaired auditory system and impoverished input provided by the CI device. Nine hearing-impaired infants with CI (most with 1-2 mo of CI use) and 19 NH infants were tested on their preference for their native language (Hebrew child-directed speech) compared with a nonnative language (English child-directed speech). The central fixation preference procedure was used in which listening times were measured via orientation responses of the infant to visual stimuli. 1) Normal hearing and infants with CI had perceptual bias for their native language (Hebrew) when compared with a nonnative language (English). 2) Infants with CI have shorter attention to speech stimuli compared with NH. The findings of the present study are the first to show that hearing-impaired infants with CI bring the same perceptual biases to the task of language learning as NH infants. These have important implications on understanding the process by which infants with CI acquire language via the CI device.

Online Learning Mechanisms for Bayesian Models of Word Segmentation

In recent years, Bayesian models have become increasingly popular as a way of understanding human cognition. Ideal learner Bayesian models assume that cognition can be usefully understood as optimal behavior under uncertainty, a hypothesis that has been supported by a number of modeling studies across various domains (e.g., Griffiths and Tenenbaum, Cognitive Psychology, 51, 354–384, 2005; Xu and Tenenbaum, Psychological Review, 114, 245–272, 2007). The models in these studies aim to explain why humans behave as they do given the task and data they encounter, but typically avoid some questions addressed by more traditional psychological models, such as how the observed behavior is produced given constraints on memory and processing. Here, we use the task of word segmentation as a case study for investigating these questions within a Bayesian framework. We consider some limitations of the infant learner, and develop several online learning algorithms that take these limitations into account. Each algorithm can be viewed as a different method of approximating the same ideal learner. When tested on corpora of English child-directed speech, we find that the constrained learner’s behavior depends non-trivially on how the learner’s limitations are implemented. Interestingly, sometimes biases that are helpful to an ideal learner hinder a constrained learner, and in a few cases, constrained learners perform equivalently or better than the ideal learner. This suggests that the transition from a computational-level solution for acquisition to an algorithmic-level one is not straightforward.

Categorizing words using 'frequent frames': what cross-linguistic analyses reveal about distributional acquisition strategies.

Mintz (2003) described a distributional environment called a frame, defined as the co-occurrence of two context words with one intervening target word. Analyses of English child-directed speech showed that words that fell within any frequently occurring frame consistently belonged to the same grammatical category (e.g. noun, verb, adjective, etc.). In this paper, we first generalize this result to French, a language in which the function word system allows patterns that are potentially detrimental to a frame-based analysis procedure. Second, we show that the discontinuity of the chosen environments (i.e. the fact that target words are framed by the context words) is crucial for the mechanism to be efficient. This property might be relevant for any computational approach to grammatical categorization. Finally, we investigate a recursive application of the procedure and observe that the categorization is paradoxically worse when context elements are categories rather than actual lexical items. Item-specificity is thus also a core computational principle for this type of algorithm. Our analysis, along with results from behavioural studies (Gómez, 2002; Gómez and Maye, 2005; Mintz, 2006), provides strong support for frames as a basis for the acquisition of grammatical categories by infants. Discontinuity and item-specificity appear to be crucial features.

The secret is in the sound: from unsegmented speech to lexical categories

When learning language, young children are faced with many seemingly formidable challenges, including discovering words embedded in a continuous stream of sounds and determining what role these words play in syntactic constructions. We suggest that knowledge of phoneme distributions may play a crucial part in helping children segment words and determine their lexical category, and we propose an integrated model of how children might go from unsegmented speech to lexical categories. We corroborated this theoretical model using a two-stage computational analysis of a large corpus of English child-directed speech. First, we used transition probabilities between phonemes to find words in unsegmented speech. Second, we used distributional information about word edges--the beginning and ending phonemes of words--to predict whether the segmented words from the first stage were nouns, verbs, or something else. The results indicate that discovering lexical units and their associated syntactic category in child-directed speech is possible by attending to the statistics of single phoneme transitions and word-initial and final phonemes. Thus, we suggest that a core computational principle in language acquisition is that the same source of information is used to learn about different aspects of linguistic structure.

Modeling the Development of Children's Use of Optional Infinitives in Dutch and English Using MOSAIC

In this study we use a computational model of language learning called model of syntax acquisition in children (MOSAIC) to investigate the extent to which the optional infinitive (OI) phenomenon in Dutch and English can be explained in terms of a resource-limited distributional analysis of Dutch and English child-directed speech. The results show that the same version of MOSAIC is able to simulate changes in the pattern of finiteness marking in 2 children learning Dutch and 2 children learning English as the average length of their utterances increases. These results suggest that it is possible to explain the key features of the OI phenomenon in both Dutch and English in terms of the interaction between an utterance-final bias in learning and the distributional characteristics of child-directed speech in the 2 languages. They also show how computational modeling techniques can be used to investigate the extent to which cross-linguistic similarities in the developmental data can be explained in terms of common processing constraints as opposed to innate knowledge of universal grammar.

Early Finiteness in Inuktitut: The Role of Language Structure and Input

A stage of optional infinitive (OI) production has been identified in typically developing (TD) children learning languages that do not permit null subjects (Wexler (1994; 1998; 1999)), and this stage has been shown to be extended in at least English- and German-speaking children with specific language impairment (SLI; Rice, Noll, and Grimm (1997), Rice, Wexler, and Cleave (1995)). Although TD children learning null subject languages do not go through an OI stage (Bar-Shalom and Snyder (1997), Guasti (1993)), reports differ concerning whether children with SLI learning these languages go through this stage (Bortolini, Caselli, and Leonard (1997), Bottari, Cipriani, and Chilosi (1996)). In this article, we present evidence from Inuktitut, a null subject language not yet investigated with respect to OIs. We show that although TD children learning Inuktitut do not go through an OI stage, one child with SLI does go through an OI stage. In addition, the percentage of finite verb forms marked with an overt verbal inflection in Inuktitut child-directed speech (CDS) is strikingly high compared with that in English CDS. We discuss the implications of these results for theories of continuity, the initial stage of child grammar, and the effect of language structure and input on language acquisition.