Learnability Theory Research Articles

Learnability Theory and Rural Roma Children’s Knowledge of Languages: An Empirical Psycholinguistic Comparative Study

Learnability Theory and Rural Roma Children’s Knowledge of Languages: An Empirical Psycholinguistic Comparative Study

Psychological & linguistic approaches to language acquisition

Language acquisition research experienced a boom following the Chomskyian revolution. The focus of attention centred primarily on the English child's acquisition of syntax. In the seventies, the range of problem areas in language acquisition began to diversify and alternative perspectives (non-nativist) on how the child acquires language beganto emerge. It is argued that socio-cognitive approaches to language acquisition, though providing an important prerequisite for the acquisition of linguistic structure, cannot themselves account for the acquisition of the complex mapping relation between grammar and meaning that is required for full-blooded linguistic communication. Recenttrends in language acquisition research including Learnability theory, Individual differences and Cross-linguistic approaches are reviewed. The article concludes with speculation about the future role of non-nativist approaches in language acquisition research. Although much current detailed work would seem to point to the existence of a Language Acquisition Device that is specifically tuned to the processing of linguistic information, it is premature to conclude that amore general cognitive learning mechanism that is able to account for both universal and particular properties of linguistic development, cannot provide a more parsimonious explanation of acquisition.

Efficient Hardware Malware Detectors That are Resilient to Adversarial Evasion

Hardware Malware Detectors (HMDs) have recently been proposed to make systems more malware-resistant. HMDs use hardware features to detect malware as a computational anomaly. Several aspects of the detector construction have been explored, leading to detectors with high accuracy. In this article, we explore whether malware developers can modify malware to avoid HMDs detection. We show that existing HMDs can be effectively reverse-engineered and subsequently evaded. Next, we explore whether retraining using evasive malware would help and show that retraining is limited. To address these limitations, we propose a new type of Resilient HMDs (RHMDs) that stochastically switch between different detectors. These detectors can be shown to be provably more difficult to reverse engineer based on recent results in probably approximately correct (PAC) learnability theory. We show that indeed such detectors are resilient to both reverse engineering and evasion, and that the resilience increases with the number and diversity of the individual detectors. Furthermore, we show that an optimal switching strategy between the RHMDs base detectors not only reduces misclassification on evasive malware but also maintains high classification accuracy on non-evasive malware. Our results demonstrate that these HMDs offer effective defense against evasive malware at low additional complexity.

Variability in L2 Acquisition: Investigating the Role of Formal Features

Following the development of the minimalist program in the 1990’s, Second Language Acquisition (SLA) researchers’ interest shifted from a focus on functional categories to a focus on formal features which, thereafter, came to be viewed as the ‘center of learnability theory’. Subsequently, Chomsky’s distinction of feature selection and feature assembly, divided SLA researchers among those who take L2 impairment to result from a failure in the selection of parametrized features, and those who believe that the problem is caused by a failure in feature assembly. The present study examines the validity of the feature (re)-assembly account by providing arguments from the second language acquisition of English by speakers of two varieties of Arabic. In particular, the findings from the comparison of the written samples of two groups of students are argued to support the claim that L2 impairment can be explained by a failure to (re)-assemble features. However, certain problems in L2 acquisition are left unaccounted for, suggesting that more is involved in SLA than just feature selection and feature assembly.

Learnability theory.

Learnability theory is a body of mathematical and computational results concerning questions such as: when is learning possible? What prior information is required to support learning? What computational or other resources are required for learning to be possible? It is therefore complementary both to the computational project of building machine learning systems and to the scientific project of understanding learning in people and animals through observation and experiment. Learnability theory includes work within a variety of theoretical frameworks, including, for example, identification in the limit, and Bayesian learning, which idealize learning in different ways. Learnability theory addresses one of the foundational questions in cognitive science: to what extent can knowledge be derived from experience? WIREs Cogn Sci 2013, 4:299-306. doi: 10.1002/wcs.1228 For further resources related to this article, please visit the WIREs website.

Learning How to License Null Noun-Class Prefixes in Sesotho

Noun-class prefixes are obligatory in most Bantu languages. However, the Sotho languages (Sesotho, Setswana, Sepedi) permit a subset of prefixes to be realized as null at the intersection of ‘unmarked’ phonological, syntactic, and discourse conditions. This raises the question of how and when the licensing of null prefixes is learned. Using longitudinal data from three Sesotho-speaking children, this article shows that the conditions needed to license null prefixes have been learned before the age of three, suggesting early abilities for grammatical generalization even at the intersection of different levels of linguistic structure. The implications for learnability theory and Bantu linguistic structure more generally are discussed.

Clinical Implications of Dynamic Systems Theory for Phonological Development

To examine treatment outcomes in relation to the complexity of treatment goals for children with speech sound disorders. The clinical implications of dynamic systems theory in contrast with learnability theory are discussed, especially in the context of target selection decisions for children with speech sound disorders. Detailed phonological analyses of pre-and posttreatment speech samples are provided for 6 children who received treatment in a previously published randomized controlled trial of contrasting approaches to target selection (Rvachew & Nowak, 2001). Three children received treatment for simple target phonemes that did not introduce any new feature contrasts into the children's phonological systems. Three children received treatment for complex targets that represented feature contrasts that were absent from the children's phonological systems. Children who received treatment for simple targets made more progress toward the acquisition of the target sounds and demonstrated emergence of complex untreated segments and feature contrasts. Children who received treatment for complex targets made little measurable gain in phonological development. Treatment outcomes will be enhanced if the clinician selects treatment targets at the segmental and prosodic levels of the phonological system in such a way as to stabilize the child's knowledge of subcomponents that form the foundation for the emergence of more complex phoneme contrasts.

The Acquisition of Subset and Superset Phonotactic Knowledge in a Second Language

Can second language (L2) learners acquire a grammar that allows a subset of the structures allowed by their native grammar? This question is addressed here with respect to acquisition of phonotactics. On the assumption that the L2 initial state equals the native grammar's final state, learnability theory would predict that a lack of negative evidence for phonotactic structures that are illegal in the target language precludes acquisition of the target grammar. This prediction is tested for L1-Russian (superset) and L1-Spanish (subset) L2 learners of Dutch by means of word-likeness judgments and lexical decision experiments. Participants responded to nonwords containing consonant clusters in onsets and codas that are legal (1) only in Russian, (2) only in Russian and Dutch, or (3) in all three languages. The results converge to show that advanced L1-Russian and L1-Spanish L2 learners possess native-like phonotactic knowledge. Analysis shows that this knowledge cannot be attributed to transfer of lexical statistics from the native language. The results suggest that L2 phonotactic acquisition is not affected by subset/superset relations between the native language and target language. Some possible explanations for our findings are discussed.

The Complexity of Ranking Hypotheses in Optimality Theory

Given a constraint set with k constraints in the framework of Optimality Theory (OT), what is its capacity as a classification scheme for linguistic data? One useful measure of this capacity is the size of the largest data set of which each subset is consistent with a different grammar hypothesis. This measure is known as the Vapnik-Chervonenkis dimension (VCD) and is a standard complexity measure for concept classes in computational learnability theory. In this work, I use the three-valued logic of Elementary Ranking Conditions to show that the VCD of Optimality Theory with k constraints is k-1. Analysis of OT in terms of the VCD establishes that the complexity of OT is a well-behaved function of k and that the ‘hardness’ of learning in OT is linear in k for a variety of frameworks that employ probabilistic definitions of learnability.

Phonological Complexity and Language Learnability

To extend formal models of language learnability to applications in clinical treatment of children with functional phonological delays. The focus of the narrative review is on phonological complexity. This follows from learnability theory, whereby complexity in the linguistic input to children has been shown to trigger language learning. Drawing from the literature, phonological complexity is defined from epistemic, ontological, and functional perspectives, with specific emphasis on the application of language universals in the selection of target sounds for treatment. The cascading effects of phonological complexity on children's generalization learning are illustrated, and frequently asked questions about complexity in treatment are addressed. The role of complexity in cognitive development is introduced to demonstrate the apparent robustness of effects.

Applications of learnability theory to clinical phonology

This paper provides a tutorial on the selection of complex target sounds for treatment following from known principles of language learnability. The focus is on syllable structure in recommending onset consonant clusters for treatment. The step-by-step procedure of cluster selection is illustrated for one child, Jarrod, who presented with a phonological disorder. Target selection procedures are guided by universal principles that govern the phonotactics of onset clusters and experimental evidence that supports the efficacy of phonologically complex targets. The prediction is that treatment of onset clusters will facilitate Jarrod's learning of both complex and simple properties of the sound system.

Language Acquisition and Learnability (review)

Reviewed by: Language acquisition and learnability ed. by Stefano Bertolo Lisa DeWaard Dykstra Language acquisition and learnability. Ed. by Stefano Bertolo. Cambridge: Cambridge University Press, 2001. Pp. viii, 247. ISBN 0521646200. $24. Stefano Bertolo’s collection of articles on language acquisition and learnability theory attempts to bridge three disciplines to better approach the problem of second language acquisition. In his opening statement, the editor provides a rationale for combining linguistics, psychology, and learnability theory into a theory of natural language learnability, claiming that each discipline contributes a different and necessary piece of the puzzle. The introduction further develops this rationale and provides the framework upon which the remaining articles are based, with three main goals: (1) It acknowledges the role that the principles and parameters hypothesis plays in the development of theory in the subsequent articles; (2) It identifies five essential components of the learning problem (presented as questions); (3) It provides the working definitions upon which the rest of the book is based. Martin Atkinson continues this dialogue with his contribution on the learnability of syntax. In it, he provides a thorough overview of the development of learnability theory, providing support for his argument by framing it in the context of children’s linguistic environments (e.g. negative evidence, input complexity). He goes on to discuss subset principles and parameters and then concludes with the TLA (triggering learning algorithm) and its applicability to the setting of parameters. In the third chapter, ‘Language change and learnability’, Ian Roberts is concerned with the synchronic and diachronic dimensions of language change. He asserts that ‘parameter values can change as a function of time’ (81) and that ‘the standard paradigm for language acquisition is not immediately compatible with [this] observation’ (83). He then proceeds to develop a parameter-setting algorithm which he tests in several case studies involving language change, concluding that language change provides additional evidence for learnability theory. Robin Clark, in the chapter entitled ‘Information theory, complexity and linguistic descriptions’, uses a variety of mathematical techniques (probability theory, information theory, and other computational approaches to linguistic description) to examine syntactic variation, specifically the question of what constitutes a parameter. Finally, William G. Sakas and Janet Dean Fodor examine the question of language acquisition difficulties, especially for adult learners. An analysis of parameter setting (specifically the elements of trigger recognition, value adoption, and value resetting) yields the conclusion that triggering cannot be effectively implemented in natural languages. The book is a thorough investigation of learnability theory and its application to language acquisition theory. It is wide in scope and methodology. Due to the sophisticated nature of the terminology used, as well as the reliance on complex theories of learnability, this book is most useful for researchers and advanced graduate students in linguistics who have a thorough grounding in learnability theory. Exercises are provided throughout each chapter, highlighting important concepts and providing opportunity for the application of presented theory. Lisa DeWaard Dykstra University of Iowa Copyright © 2003 Linguistic Society of America

Second language learnability and the acquisition of the argument structure of English locative verbs by Korean speakers

The present research focuses on Korean English as a foreign language (EFL) learners’ knowledge of the locative alternation (e.g., John loaded hay onto the wagon/John loaded the wagon with hay) and its relationship to theories of language-particular and language-universal properties. Korean, the native language of the participants, has a locative alternation resembling that of English. However, although Korean and English are similar in terms of broad-range constraints, they are dissimilar in terms of narrow-range constraints for locative alternations. This study investigates whether the acquisition of such constraints in English locatives by Korean speakers, and whether the first language (L1) influences the second language (L2) acquisition of locative alternations. Two instruments are used in the experiment: a forced-choice picture-description task and a forced-choice sentence selection task. The study investigates an experimental group of Korean learners of English and a control group of native speakers of English. The results are discussed with reference to universality of linking, to the transfer of argument structure and to Pinker’s learnability theory. The primary results are: • The Korean learners of English had acquired the constructional meaning of the locative construction (which is related to Pinker’s (1989) concept of broad-range rules and broad conflation classes), a property claimed to be universal. • They had not achieved native-speaker knowledge of language-particular properties - which narrow conflation class verbs belong to - so that they did not reject ungrammatical sentences; and • Significant L1 transfer effects were not found.

A grounded theory approach to modelling learnability of hypermedia authoring tools

One potential barrier to the development of educational hypermedia is the design of current hypermedia authoring tools (HATs) that unfortunately require higher knowledge and skills levels than possessed by most academics. Whilst the usability of hypermedia has been extensively researched, the usability of the tools required to build hypermedia has not. Learnability of HATs, an associated factor of usability, has been similarly neglected. Analysing contemporary approaches to the study of human computer interaction, this paper concludes that they do not support the kind of ‘theory building’ required to study and describe the learnability of HATs. Grounded theory (GT) is posited as an alternative approach, which if applied correctly can provide explanatory theory to inform HAT design. The paper describes the application of GT to two studies of the ease-of-learning of HATs. The first study uses quantitative and qualitative data to explore the experiences of 16 subjects learning to use HATs. In the second study, key HATs are demonstrated to a focus group of IT trainers to analyse their observations of users learning HATs. From these studies a causal model of learnability of HATs that is more detailed and complete than that offered by other contemporary theories of learnability was developed. The paper concludes that applying a GT approach can enhance HCI research through the development of explanatory, extensible and evolutionary theory to inform HAT design.

Syntactic development By William O' Grady (review)

REVIEWS171 Syntactic development. By William O' Grady. Chicago: University ofChicago Press, 1997. Pp. ix, 409. Reviewed by Stanley Dubinsky, University of South Carolina O'Grady's volume on the syntactic aspects of first language acquisition is a welcome addition to the field. It provides a thorough empirical and theoretical overview of research on the acquisition of syntax that is useful to anyone who desires to understand how Ll acquisition has informed syntactic theory and how syntactic theories have primed research in Ll acquisition. The text does not oversimplify any of the experimental research and theoretical concepts brought up for discussion, but it is still clear enough to be accessible to anyone with a basic background in linguistics. For this reason, among others, the volume is useful both as a source book for the various issues it covers as well as a good text for a course in Ll acquisition. Having used it as a text for a directed reading course on Ll acquisition of syntax, I am convinced that it works quite well as a course book. Following a short introductory chapter, the book is divided into two parts, with the first (twothirds ) consisting often chapters on 'The developmental facts' and the second (one-third) dealing with "Theories of learnability and development'. The issues discussed in Part 1 are introduced topically and, for the most part, in the temporal order oftheir relevance to the acquisition process. They include one-word utterances, early multiword utterances, word order and case, subject drop, embedded clauses, w/t-questions, inversion, relative clauses and clefts, passives, and constraints on coreference. Each of the chapters in Part 1 surveys the important (and often conflicting) claims made by psycholinguistic researchers regarding acquisition of the phenomena in question and reviews the experiments that have been brought to bear in these debates. For instance, the chapter on early multiword utterances (Ch. 3) examines four important approaches to the distribution of these—thematic analysis (Bowerman 1973 and many others), pivot analysis (Braine 1963), limited scope analysis (Braine 1976, Maratsos 1983), and syntactic analysis (including Brown & Fraser 1963 and Pinker 1984). In discussing each approach, O provides a brief description ofLl data and experimental evidence used to support it. In this particular chapter, the section on 'syntactic analysis' is quite a bit longer than the others and reveals the author's preference for this approach. In the chapter on subject drop (Ch. 5), O discusses three types of theories explaining this stage of acquisition: processing theories (which attribute subject drop to processing considerations), grammar-based theories (which associate it with 'confusion over a language's typological status'), and tensed-based theories (which ascribe subject drop to 'a temporary morphological deficit'). The third theoretical position, O's own, is one that he argues for in this chapter. I found the 'facts first' approach to the subject, as laid out in Part 1, to be quite appealing. In the first place, it provides a valuable overview of the range and depth of research in nearly every comer of the field. O's inclusive discussion allows the reader to get a sense of the variety of viewpoints on all of the topics covered. Further, the review of empirical studies helps to make the theoretical discussion in Part 2 much more accessible to the reader. Of course one cannot have things both ways as regards breadth and depth of discussion, and the empirical section of the book sometimes errs on the side of breadth. Although O includes a very helpful summary box for each experiment of special interest, some of the discussions were too abbreviated to be of much use to this reader. For instance, the chapter on embedded clauses (Ch. 6) races through fifteen experimental studies (covering a period from C. Chomsky 1969 to Eisenberg & Cairns 1994) in just 27 pages, with some articles summarized in less than a single page. In discussions such as this, the reader gains an overview of the literature (on embedding in this case) and an understanding of each study's place in this literature but is left without significant insights into the details of the individual articles themselves. In this respect, the volume is quite a good sourcebook...

Morphological smoothing

Utilising concepts from mathematical morphology and learnability theory, the authors demonstrate a well-behavedness result pertaining to smoothing — which has fundamental ramifications ranging from physics to cognition — that states that the number of instantiated points needed to adequately reconstruct the underlying finite-sized Euclidean set is tractably large.

Musicology and the problem of ‘Musical Ear’

What constraints are brought to the task of the acquisition, comprehension, and use of musical knowledge? This question raises the problem of ‘musical ear’, solutions to which have been given chiefly by philosophers and by a variety of psychologists, rather than by musicologists. This essay seeks to expose the problem as a problem for musicology, by showing how, historically and contemporaneously, solutions have been guided by two traditional theories of knowledge and learnability.

VC-Dimension Analysis of Object Recognition Tasks

We analyze the amount of data needed to carry out various model-based recognition tasks in the context of a probabilistic data collection model. We focus on objects that may be described as semi-algebraic subsets of a Euclidean space. This is a very rich class that includes polynomially described bodies, as well as polygonal objects, as special cases. The class of object transformations considered is wide, and includes perspective and affine transformations of 2D objects, and perspective projections of 3D objects. We derive upper bounds on the number of data features (associated with non-zero spatial error) which provably suffice for drawing reliable conclusions. Our bounds are based on a quantitative analysis of the complexity of the hypotheses class that one has to choose from. Our central tool is the VC-dimension, which is a well-studied parameter measuring the combinatorial complexity of families of sets. It turns out that these bounds grow linearly with the task complexity, measured via the VC-dimension of the class of objects one deals with. We show that this VC-dimension is at most logarithmic in the algebraic complexity of the objects and in the cardinality of the model library. Our approach borrows from computational learning theory. Both learning and recognition use evidence to infer hypotheses but as far as we know, their similarity was not exploited previously. We draw close relations between recognition tasks and a certain learnability framework and then apply basic techniques of learnability theory to derive our sample size upper bounds. We believe that other relations between learning procedures and visual tasks exist and hope that this work will trigger further fruitful study along these lines.

The neural basis of cognitive development: a constructivist manifesto.

How do minds emerge from developing brains? According to "neural constructivism," the representational features of cortex are built from the dynamic interaction between neural growth mechanisms and environmentally derived neural activity. Contrary to popular selectionist models that emphasize regressive mechanisms, the neurobiological evidence suggests that this growth is a progressive increase in the representational properties of cortex. The interaction between the environment and neural growth results in a flexible type of learning: "constructive learning" minimizes the need for prespecification in accordance with recent neurobiological evidence that the developing cerebral cortex is largely free of domain-specific structure. Instead, the representational properties of cortex are built by the nature of the problem domain confronting it. This uniquely powerful and general learning strategy undermines the central assumption of classical learnability theory, that the learning properties of a system can be deduced from a fixed computational architecture. Neural constructivism suggests that the evolutionary emergence of neocortex in mammals is a progression toward more flexible representational structures, in contrast to the popular view of cortical evolution as an increase in innate, specialized circuits. Human cortical postnatal development is also more extensive and protracted than generally supposed, suggesting that cortex has evolved so as to maximize the capacity of environmental structure to shape its structure and function through constructive learning.

The learnability of the alphabetic principle: Children's initial hypotheses about how print represents spoken language

ABSTRACTThis research examines the hypotheses about how print represents the speech that preliterate children select when they receive input compatible with several such hypotheses. In Experiment 1, preschoolers were taught to readhatandhatsandbookandbooks. Then, in generalization tests, they were probed for what they had learned about the letters. All of the children were able to transfer to other plurals (e.g., to decide thatbikessaid “bikes” rather than “bike,” and thatdogsaid “dog” and not “dogs”), but only those who knew the sound of the letter s prior to the experiment were able to decide, for example, thatbussaid “bus” and not “bug.” The failure to detect the phonemic value of s on the part of alphabetically naive children was replicated in Experiments 2, 3, and 4, which instituted a variety of controls. In Experiment 5, it was found that, although preschoolers who had been taught to read pairs of words distinguished by the comparative affixer(such assmall/smaller) were able to generalize to other comparatives (e.g.,mean/meaner), they could not generalize to pairs whereerhad no morphemic value (e.g.,corn/corner). A similar failure by alphabetically naive children to detect the syllabic, as compared with the morphemic, status of the superlative affixestwas found in Experiment 6. Overall, the results indicate that most preliterate children fail to select phonologically based hypotheses, even when these are available in the input. Instead, they focus on morphophonology and/or semantic aspects of words' referents. The research is couched in terms of the Learnability Theory (LT) (Gold, 1967), which provides a convenient framework for considering a series of interrelated questions about the acquisition of literacy. In particular, it is argued that if the data available to the child includes the pronunciation of written words, the alphabetic principle may be unlearnable, given the hypothesis selection procedures identified in these experiments.