Orthography-phonology consistency in English: Theory- and data-driven measures and their impact on auditory vs. visual word recognition.

Research on orthographic consistency in English words has selectively identified different sub-syllabic units in isolation (grapheme, onset, vowel, coda, rime), yet there is no comprehensive assessment of how these measures affect word identification when taken together. To study which aspects of consistency are more psychologically relevant, we investigated their independent and composite effects on human reading behavior using large-scale databases. Study 1 found effects on adults' naming responses of both feedforward consistency (orthography to phonology) and feedback consistency (phonology to orthography). Study 2 found feedback but no feedforward consistency effects on visual and auditory lexical decision tasks, with the best predictor being a composite measure of consistency across grapheme, rime, OVC, and word-initial letter-phoneme. In Study 3, we explicitly modeled the reading process with forward and backward flow in a bidirectionally connected neural network. The model captured latent dimensions of quasi-regular mapping that explain additional variance in human reading and spelling behavior, compared to the established measures. Together, the results suggest interactive activation between phonological and orthographic word representations. They also validate the role of computational analyses of language to better understand how print maps to sound, and what properties of natural language affect reading complexity.

Is retroflexion a stable cue for distributional learning for speech sounds across languages? Learning for some bilingual adults, but not generalisable to a wider population in a well powered pre-registered study.

Bilinguals are widely reported to have certain kinds of cognitive advantages, including language learning advantages. One possible pathway is a language-specific transfer effect, whereby sensitivity to structural regularities in known languages can be brought to to-be-acquired languages that share particular features. Here we tested for transfer of a specific linguistic property, sensitivity to retroflexion as contrastive phonemic feature. We designed a task for bilinguals with homogeneous language exposure (i.e., bilingual in the same languages) and heterogeneous feature representation (i.e., differing levels of proficiency). Hindi and Mandarin Chinese both have retroflexion in phoneme contrasts (Hindi: stop consonants, Mandarin: sibilants). In a preregistered study, we conducted a statistical learning task for the Hindi dental-retroflex stop contrast with a group of early parallel English-Mandarin bilinguals, who varied in their Mandarin understanding levels. We based the target sample size on power analysis of a pilot study with a Bayesian stop-rule after minimum threshold. Contrary to the pilot study (N = 15), the main study (N = 50) did not find evidence for a learning effect, nor language-experience variance within the group. This finding suggests that statistical effects for the feature in question may be more fragile than commonly assumed, and may be evident in only a small subsample of the general population (as in our pilot). These stimuli have previously shown learning effects in children, so an additional possibility is that neural commitment to adults' languages prevents learning of the fine-grained stimulus contrast in question for this adult population.

Is the Mind Inherently Predicting? Exploring Forward and Backward Looking in Language Processing.

Prediction is one characteristic of the human mind. But what does it mean to say the mind is a "prediction machine" and inherently forward looking as is frequently claimed? In natural languages, many contexts are not easily predictable in a forward fashion. In English, for example, many frequent verbs do not carry unique meaning on their own but instead, rely on another word or words that follow them to become meaningful. Upon reading take a the processor often cannot easily predict walk as the next word. But the system can "look back" and integrate walk more easily when it follows take a (e.g., as opposed to *make|get|have a walk). In the present paper, we provide further evidence for the importance of both forward and backward-looking in language processing. In two self-paced reading tasks and an eye-tracking reading task, we found evidence that adult English native speakers' sensitivity to word forward and backward conditional probability significantly predicted reading times over and above psycholinguistic predictors of reading latencies. We conclude that both forward and backward-looking (prediction and integration) appear to be important characteristics of language processing. Our results thus suggest that it makes just as much sense to call the mind an "integration machine" which is inherently backward 'looking.'

Can prediction and retrodiction explain whether frequent multi-word phrases are accessed 'precompiled' from memory or compositionally constructed on the fly?

An important debate on the architecture of the language faculty has been the extent to which it relies on a compositional system that constructs larger units from morphemes to words to phrases to utterances on the fly and in real time using grammatical rules; or a system that chunks large preassembled, stored units of language from memory; or some combination of both approaches. Good empirical evidence exists for both 'computed' and 'large stored' forms in language, but little is known about what shapes multi-word storage/ access or compositional processing. Here we explored whether predictive and retrodictive processes are a likely determinant of multi-word storage/ processing. Our results suggest that forward and backward predictability are independently informative in determining the lexical cohesiveness of multi-word phrases. In addition, our results call for a reevaluation of the role of retrodiction in contemporary language processing accounts (cf. Ferreira and Chantavarin, 2018).

Technology-Based Tools for English Literacy Intervention: Examining Intervention Grain Size and Individual Differences.

Technology plays an increasingly important role in educational practice, including interventions for struggling learners (Torgesen et al., 2010; de Souza et al., 2018). This study focuses on the efficacy of tablet-based applications (see Word Reading, Grapholearn, and an experimental word-level program) for the purpose of supplementing early English literacy intervention with primary grades 1 and 2 children. The children were identified for learning support programs within Singaporean schools, which follow a bilingual policy, meaning children were learning reading in English plus an additional language. One hundred forty-seven children across seven schools participated (Mean age = 6.66). Within learning support classrooms, triplets of students matched on basic reading skills were randomly assigned to one of three groups: (1) phoneme-level, (2) rime-level, or (3) word-level focused interventions. All groups performed reading skills activities on iPads, across two phases over a 14-week period. Assessments for word reading accuracy and fluency, pseudoword decoding accuracy and fluency, and spelling were administered at four time points, pre- and post-intervention. Additional baseline measures were taken to assess individual differences in phonological awareness, orthographic awareness, general cognitive ability, statistical learning, and bilingual vocabulary knowledge. Mixed model analysis was conducted on the pre- to post-test measures across the two phases of the intervention (focused on accuracy then fluency). All groups made gains across the different literacy measures, while the phoneme-level intervention showed an advantage over the rime-level intervention, but not the word-level intervention, for decoding. There were also moderating effects of individual differences on outcomes. The general pattern of results showed an advantage of the word-level intervention for those with poorer phonological awareness for reading fluency; and a phoneme-level intervention advantage for those with poorer statistical learning ability. Children's bilingual group (English plus Mandarin, English plus Malay, or English plus Tamil) also showed differential effects of the type of intervention (e.g., phoneme- or word-level) on different outcome measures. These results, along with data collected from the tablets during the intervention, suggest the need to examine the interplay between different types of technology-based interventions and individual differences in learning profiles.

Early developing syntactic knowledge influences sequential statistical learning in infancy.

Adults' linguistic background influences their sequential statistical learning of an artificial language characterized by conflicting forward-going and backward-going transitional probabilities. English-speaking adults favor backward-going transitional probabilities, consistent with the head-initial structure of English. Korean-speaking adults favor forward-going transitional probabilities, consistent with the head-final structure of Korean. These experiments assess when infants develop this directional bias. In the experiments, 7-month-old infants showed no bias for forward-going or backward-going regularities. By 13 months, however, English-learning infants favored backward-going transitional probabilities over forward-going transitional probabilities, consistent with English-speaking adults. This indicates that statistical learning rapidly adapts to the predominant syntactic structure of the native language. Such adaptation may facilitate subsequent learning by highlighting statistical structures that are likely to be informative in the native linguistic environment.

Social Network Limits Language Complexity.

Natural languages vary widely in the degree to which they make use of nested compositional structure in their grammars. It has long been noted by linguists that the languages historically spoken in small communities develop much deeper levels of compositional embedding than those spoken by larger groups. Recently, this observation has been confirmed by a robust statistical analysis of the World Atlas of Language Structures. In order to examine this connection mechanistically, we propose an agent-based model that accounts for key cultural evolutionary features of language transfer and language change. We identify transitivity as a physical parameter of social networks critical for the evolution of compositional structure and the hierarchical patterning of scale-free distributions as inhibitory.

Language development and disorders: Possible genes and environment interactions.

Language development requires both basic cognitive mechanisms for learning language and a rich social context from which learning takes off. Disruptions in learning mechanisms, processing abilities, and/or social interactions increase the risks associated with social exclusion or developmental delays. Given the complexity of language processes, a multilevel approach is proposed where both cognitive mechanisms, genetic and environmental factors need to be probed together with their possible interactions. Here we review and discuss such interplay between environment and genetic predispositions in understanding language disorders, with a particular focus on a possible endophenotype, the ability for statistical sequential learning.

Caregiver communication to the child as moderator and mediator of genes for language.

Human language appears to be unique among natural communication systems, and such uniqueness impinges on both nature and nurture. Human babies are endowed with cognitive abilities that predispose them to learn language, and this process cannot operate in an impoverished environment. To be effectively complete the acquisition of human language in human children requires highly socialised forms of learning, scaffolded over years of prolonged and intense caretaker-child interactions. How genes and environment operate in shaping language is unknown. These two components have traditionally been considered as independent, and often pitted against each other in terms of the nature versus nurture debate. This perspective article considers how innate abilities and experience might instead work together. In particular, it envisages potential scenarios for research, in which early caregiver verbal and non-verbal attachment practices may mediate or moderate the expression of human genetic systems for language.

The bottleneck may be the solution, not the problem.

As a highly consequential biological trait, a memory "bottleneck" cannot escape selection pressures. It must therefore co-evolve with other cognitive mechanisms rather than act as an independent constraint. Recent theory and an implemented model of language acquisition suggest that a limit on working memory may evolve to help learning. Furthermore, it need not hamper the use of language for communication.

Language experience changes subsequent learning.

What are the effects of experience on subsequent learning? We explored the effects of language-specific word order knowledge on the acquisition of sequential conditional information. Korean and English adults were engaged in a sequence learning task involving three different sets of stimuli: auditory linguistic (nonsense syllables), visual non-linguistic (nonsense shapes), and auditory non-linguistic (pure tones). The forward and backward probabilities between adjacent elements generated two equally probable and orthogonal perceptual parses of the elements, such that any significant preference at test must be due to either general cognitive biases, or prior language-induced biases. We found that language modulated parsing preferences with the linguistic stimuli only. Intriguingly, these preferences are congruent with the dominant word order patterns of each language, as corroborated by corpus analyses, and are driven by probabilistic preferences. Furthermore, although the Korean individuals had received extensive formal explicit training in English and lived in an English-speaking environment, they exhibited statistical learning biases congruent with their native language. Our findings suggest that mechanisms of statistical sequential learning are implicated in language across the lifespan, and experience with language may affect cognitive processes and later learning.

Similar Neural Correlates for Language and Sequential Learning: Evidence from Event-Related Brain Potentials.

We used event-related potentials (ERPs) to investigate the time course and distribution of brain activity while adults performed (a) a sequential learning task involving complex structured sequences, and (b) a language processing task. The same positive ERP deflection, the P600 effect, typically linked to difficult or ungrammatical syntactic processing, was found for structural incongruencies in both sequential learning as well as natural language, and with similar topographical distributions. Additionally, a left anterior negativity (LAN) was observed for language but not for sequential learning. These results are interpreted as an indication that the P600 provides an index of violations and the cost of integration of expectations for upcoming material when processing complex sequential structure. We conclude that the same neural mechanisms may be recruited for both syntactic processing of linguistic stimuli and sequential learning of structured sequence patterns more generally.

General cognitive principles for learning structure in time and space.

How are hierarchically structured sequences of objects, events or actions learned from experience and represented in the brain? When several streams of regularities present themselves, which will be learned and which ignored? Can statistical regularities take effect on their own, or are additional factors such as behavioral outcomes expected to influence statistical learning? Answers to these questions are starting to emerge through a convergence of findings from naturalistic observations, behavioral experiments, neurobiological studies, and computational analyses and simulations. We propose that a small set of principles are at work in every situation that involves learning of structure from patterns of experience and outline a general framework that accounts for such learning.

An empirical generative framework for computational modeling of language acquisition.

This paper reports progress in developing a computer model of language acquisition in the form of (1) a generative grammar that is (2) algorithmically learnable from realistic corpus data, (3) viable in its large-scale quantitative performance and (4) psychologically real. First, we describe new algorithmic methods for unsupervised learning of generative grammars from raw CHILDES data and give an account of the generative performance of the acquired grammars. Next, we summarize findings from recent longitudinal and experimental work that suggests how certain statistically prominent structural properties of child-directed speech may facilitate language acquisition. We then present a series of new analyses of CHILDES data indicating that the desired properties are indeed present in realistic child-directed speech corpora. Finally, we suggest how our computational results, behavioral findings, and corpus-based insights can be integrated into a next-generation model aimed at meeting the four requirements of our modeling framework.

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.

Learn locally, act globally: learning language from variation set cues.

Variation set structure--partial overlap of successive utterances in child-directed speech--has been shown to correlate with progress in children's acquisition of syntax. We demonstrate the benefits of variation set structure directly: in miniature artificial languages, arranging a certain proportion of utterances in a training corpus in variation sets facilitated word and phrase constituent learning in adults. Our findings have implications for understanding the mechanisms of L1 acquisition by children, and for the development of more efficient algorithms for automatic language acquisition, as well as better methods for L2 instruction.

Lexical categories at the edge of the word.

Language acquisition may be one of the most difficult tasks that children face during development. They have to segment words from fluent speech, figure out the meanings of these words, and discover the syntactic constraints for joining them together into meaningful sentences. Over the past couple of decades, computational modeling has emerged as a new paradigm for gaining insights into the mechanisms by which children may accomplish these feats. Unfortunately, many of these models assume a computational complexity and linguistic knowledge likely to be beyond the abilities of developing young children. This article shows that, using simple statistical procedures, significant correlations exist between the beginnings and endings of a word and its lexical category in English, Dutch, French, and Japanese. Therefore, phonetic information can contribute to individuating higher level structural properties of these languages. This article also presents a simple 2-layer connectionist model that, once trained with an initial small sample of words labeled for lexical category, can infer the lexical category of a large proportion of novel words using only word-edge phonological information, namely the first and last phoneme of a word. The results suggest that simple procedures combined with phonetic information perceptually available to children provide solid scaffolding for emerging lexical categories in language development.