A Bayesian framework for word segmentation: Exploring the effects of context

Since the experiments of Saffran et al. [Saffran, J., Aslin, R., & Newport, E. (1996). Statistical learning in 8-month-old infants. Science, 274, 1926-1928], there has been a great deal of interest in the question of how statistical regularities in the speech stream might be used by infants to begin to identify individual words. In this work, we use computational modeling to explore the effects of different assumptions the learner might make regarding the nature of words - in particular, how these assumptions affect the kinds of words that are segmented from a corpus of transcribed child-directed speech. We develop several models within a Bayesian ideal observer framework, and use them to examine the consequences of assuming either that words are independent units, or units that help to predict other units. We show through empirical and theoretical results that the assumption of independence causes the learner to undersegment the corpus, with many two- and three-word sequences (e.g. what’s that, do you, in the house) misidentified as individual words. In contrast, when the learner assumes that words are predictive, the resulting segmentation is far more accurate. These results indicate that taking context into account is important for a statistical word segmentation strategy to be successful, and raise the possibility that even young infants may be able to exploit more subtle statistical patterns than have usually been considered.     

Finding one's meaning: a test of the relation between quantifiers and integers in language development

We explored children’s early interpretation of numerals and linguistic number marking, in order to test the hypothesis (e.g., Carey (2004). Bootstrapping and the origin of concepts. Daedalus, 59-68) that children’s initial distinction between one and other numerals (i.e., two, three, etc.) is bootstrapped from a prior distinction between singular and plural nouns. Previous studies have presented evidence that in languages without singular-plural morphology, like Japanese and Chinese, children acquire the meaning of the word one later than in singular-plural languages like English and Russian. In two experiments, we sought to corroborate this relation between grammatical number and integer acquisition within English. We found a significant correlation between children’s comprehension of numerals and a large set of natural language quantifiers and determiners, even when controlling for effects due to age. However, we also found that 2-year-old children, who are just acquiring singular-plural morphology and the word one, fail to assign an exact interpretation to singular noun phrases (e.g., a banana), despite interpreting one as exact. For example, in a Truth-Value Judgment task, most children judged that a banana was consistent with a set of two objects, despite rejecting sets of two for the numeral one. Also, children who gave exactly one object for singular nouns did not have a better comprehension of numerals relative to children who did not give exactly one. Thus, we conclude that the correlation between quantifier comprehension and numeral comprehension in children of this age is not attributable to the singular-plural distinction facilitating the acquisition of the word one. We argue that quantifiers play a more general role in highlighting the semantic function of numerals, and that children distinguish between numerals and other quantifiers from the beginning, assigning exact interpretations only to numerals.     

EFFECTS OF RESPONSE SPACING ON ACQUISITION AND RETENTION OF CONDITIONAL DISCRIMINATIONS

Pigeons were exposed to a repeated acquisition procedure in which no delays were imposed and rate of responding was relatively high. They also were exposed to conditions in which delays were arranged between trials within chains or between completed chains, and rates of responding were lower. Number of trials, rate of reinforcement, difficulty of the discrimination, and motivating operations were held constant. Terminal accuracy was highest under the no‐delay condition, in which rate of responding was highest. Effects of trial spacing on retention were mixed and depended on whether delays were imposed between trials within chains or between completed chains. These findings provide basic‐research support for the rapid presentation of trials in direct instruction and for rate building in precision teaching.     

Is banara really a word?

Bowers, Davis, and Hanley (Bowers, J. S., Davis, C. J., & Hanley, D. A. (2005). Interfering neighbours: The impact of novel word learning on the identification of visually similar words. Cognition, 97(3), B45-B54) reported that if participants were trained to type nonwords such as banara, subsequent semantic categorization responses to similar words such as banana were delayed. This was taken as direct experimental support for a process of lexical competition during word recognition. This interpretation assumes that banara has been lexicalized, which predicts that masked form priming for items such as banara-banana should be reduced or eliminated. An experiment is reported showing that the trained novel words produced the same amount of priming as untrained nonwords on both the first and the second day of training, suggesting that the interference observed by Bowers et al was not due to word-on-word competition.     

Learning words’ sounds before learning how words sound: 9-Month-olds use distinct objects as cues to categorize speech information

One of the central themes in the study of language acquisition is the gap between the linguistic knowledge that learners demonstrate, and the apparent inadequacy of linguistic input to support induction of this knowledge. One of the first linguistic abilities in the course of development to exemplify this problem is in speech perception: specifically, learning the sound system of one’s native language. Native-language sound systems are defined by meaningful contrasts among words in a language, yet infants learn these sound patterns before any significant numbers of words are acquired. Previous approaches to this learning problem have suggested that infants can learn phonetic categories from statistical analysis of auditory input, without regard to word referents. Experimental evidence presented here suggests instead that young infants can use visual cues present in word-labeling situations to categorize phonetic information. In Experiment 1, 9-month-old English-learning infants failed to discriminate two non-native phonetic categories, establishing baseline performance in a perceptual discrimination task. In Experiment 2, these infants succeeded at discrimination after watching contrasting visual cues (i.e., videos of two novel objects) paired consistently with the two non-native phonetic categories. In Experiment 3, these infants failed at discrimination after watching the same visual cues, but paired inconsistently with the two phonetic categories. At an age before which memory of word labels is demonstrated in the laboratory, 9-month-old infants use contrastive pairings between objects and sounds to influence their phonetic sensitivity. Phonetic learning may have a more functional basis than previous statistical learning mechanisms assume: infants may use cross-modal associations inherent in social contexts to learn native-language phonetic categories.     

RESPONSE ACQUISITION BY HUMANS WITH DELAYED REINFORCEMENT

The present experiment examined whether a response class was acquired by humans with delayed reinforcement. Eight white circles were presented on a computer touch screen. If the undergraduates touched two of the eight circles in a specified sequence (i.e., touching first the upper‐left circle then the bottom‐left circle), then the touches initiated an unsignaled resetting delay culminating in point delivery. Participants experienced one of three different delays (0 s, 10 s, or 30 s). Rates of the target two‐response sequence were higher with delayed reinforcement than with no reinforcement. Terminal rates of the target sequence decreased and postreinforcement pauses increased as a function of delay duration. Other undergraduates exposed to yoked schedules of response‐independent point deliveries failed to acquire the sequence. The results demonstrate that a response class was acquired with delayed reinforcement, extending the generality of this phenomenon found with nonhuman animals to humans.     

When vaccinating against information reduction works and when it does not work

Practice not only affects how information is processed, but also which information is processed. The Information-Reduction Hypothesis (i.e. Haider & Frensch, 1996) holds that — with practice — irrelevant task information (i.e. information that is not logically needed to correctly perform the task) is discarded from processing. Recently, Gaschler and Frensch (2007) have demonstrated that Information Reduction is not affected by the frequency with which individual task configurations are presented: well-practiced and little-practiced irrelevant task configurations are discarded at the same point in time during practice; Information Reduction is thus an item-general phenomenon. These findings suggest that Information Reduction is at least in part a consequence of top-down, voluntary control. In the present research, we ask how tasks can be constructed such that Information Reduction is avoided. Our results show that item-general Information Reduction is observed even when it leads to severe processing costs (i.e. errors). On the whole, the present results are in line with models of skill acquisition incorporating top-down modulation (e.g. Haider & Frensch, 2002) and are incompatible with purely data-driven accounts of skill acquisition (e.g. Logan, 1988; Nosofski & Palmeri, 1997; Palmeri, 1997; Rickard, 2004).     

Solution Strategies and Achievement in Dutch Complex Arithmetic: Latent Variable Modeling of Change

In the Netherlands, national assessments at the end of primary school (Grade 6) show a decline of achievement on problems of complex or written arithmetic over the last two decades. The present study aims at contributing to an explanation of the large achievement decrease on complex division, by investigating the strategies students used in solving the division problems in the two most recent assessments carried out in 1997 and in 2004. The students’ strategies were classified into four categories. A data set resulted with two types of repeated observations within students: the nominal strategies and the dichotomous achievement scores (correct/incorrect) on the items administered.It is argued that latent variable modeling methodology is appropriate to analyze these data. First, latent class analyses with year of assessment as a covariate were carried out on the multivariate nominal strategy variables. Results showed a shift from application of the traditional long division algorithm in 1997, to the less accurate strategy of stating an answer without writing down any notes or calculations in 2004, especially for boys. Second, explanatory IRT analyses showed that the three main strategies were significantly less accurate in 2004 than they were in 1997.     

In search of a science‐based application: A learning tool for reading acquisition

This is a story about the fate of a psychological application: from its conception to the optimistic vision surrounding its future. We hope that this application – an enjoyable learning game (www or mobile phone‐based, available free of charge to the end users) for children – can at best help millions of children in their reading acquisition in the future. Its basis was created by following intensively the development of children with (N = 107) and without (N = 92) genetic (familial) risk for dyslexia from birth to puberty in the Jyväskylä Longitudinal study of Dyslexia (JLD)‐project. We summarize some of the major findings of the JLD in order to facilitate understanding of the reasons and logic behind the development of the game. Originally intended as a research tool for reading acquisition, its potential for prevention of reading difficulties was quickly recognized.     

EMBODIED TRANSCENDENCE: BONOBOS AND HUMANS IN COMMUNITY

Multiple dimensions and textures of transcendence are evoked not just by reflection on humans in their relationship with God and community but also by encounter with bonobos—primates that are very close genetic kin with humans. The promise for theological reflection is rooted in bonobo social adaptation as a highly cooperative species. Bonobo sexual behavior accompanies and expresses a high level of social intelligence. The point of my project is not a scientific one intended to argue persuasively for individual self-awareness or self-transcendence in bonobos. Instead, it emphasizes connectedness, interdependence, and sociality as windows on transcendence. Such a view does not require consciousness or intellectual recognition of self-in-relation, but it certainly presumes embodiment of self-in-relation. Various textures of transcendence reflect multidirectional relationships among Pan paniscus (bonobos), Homo sapiens , and the Sacred.