ArticlePDF Available

Aspects of a theory of language acquisition

Authors:

Abstract

This paper presents a hypothesis-testing theory of syntax acquisition. The first section presents our model. We claim that: (1) children learn a transformational grammar, including a set of phrase structure and transformational rules; (2) linguistic universals and Occam's razor constrain the initial hypothesis space available to the device; (3) hypotheses tested by the device consist of candidate phrase structure and transformational rules; (4) linguistic evidence confirms or disconfirms hypotheses. Specific examples of incorrect phrase structure and transformational hypotheses are presented. The second section briefly surveys other approaches to language acquisition – both syntactic and non-syntactic – and compares them to our model. In the third section, we address several methodological issues: (1) the relevance of linguistic theory to the model; (2) how the model is tested; (3) the domain of the theory.
... These errors are not found in adult speech and, it must be assumed, generally go uncorrected in the speech of children. Theories in the standard linguistic tradition have offered differing accounts of some of these errors (Hurford (1975) ;Erreich, Valian and Winsemer (1980); Pinker (1984); Radford (1994); Guasti, Thornton and Wexler (1994a); Rowley (1998)) but each has the underlying assumption that the child is undergoing processes of hypotheses and rule-formation which, in later accounts, is a reflection of the constraints of the principles and parameters of (an innate) Universal Grammar. This therefore poses a particular challenge to connectionism, which must account for these errors strictly in terms of the way a child analyses what it hears. ...
... Double-tense and double-aux marking has been noted since, for example Hurford (1975), and accounted for theoretically (e.g. Erreich et al. (1980)), and the latter phenomenon has recently been picked up and investigated carefully by Guasti et al. (1994a,b), which will provide the empirical child data for this study. A summary of this data follows, accompanied by the theories of Guasti et al. and Erreich et al. as representative of the generative tradition. ...
... This theory also predicts that AUX...AUX doubling should never occur, as the spell-out of trace exists to support clitic n't. Erreich et al. (1980) however, cite an example of this kind of doubling from Bellugi (1971): ...
Article
This paper demonstrates the successful modelling of the attested developmental error of auxiliary doubling in question forms in first language acquisition of English, such as do you don't like bananas, generally considered as spell-out of trace. Through distributional analysis of the 18 clause types presented word-by-word (though not coded for word category) as input, a connectionist SRN network learns to predict the correct word category at each instant for each clause type, with a low percentage error (mean error < 2%). The network does, however, exhibit a marked tendency for auxiliary doubling in questions, found to be a consequence of the presence of particular input clause types. It is argued that the different frequencies of different types of aux-doubling error found in the literature can be only be explained in these terms; no rule-based model can account for them. Since the network has no innate linguistic competence, these findings therefore undermine the arguments that aux-doubling necessarily reveals the workings of (a) any innate linguistic component and (b) processes of rule-formation, and challenge the assumption that other spell-out-of-trace errors are evidence for a Chomskian Universal Grammar.
... The use of the dummy auxiliary do, however, is a little different as it is not associated with the appearance of any aspectual morpheme. Erreich et al. (1980) argue that all rules for moving a question word include two operations: copy and delete. They report the output of the double marking of the auxiliary verb. ...
Article
Full-text available
The present article is a corpus-based case-study of the role of using patterns for creating online headlines as a means of ensuring textual efficiency. The corpus consists of over 130 headlines from the Love category on the American website of the Cosmopolitan women’s magazine (http://www.cosmopolitan.com), uploaded between 2010 and 2016. An important aspect of ensuring the quality of web headlines is the existence of patterns for creating headlines with analogous structures. Patterns in headlines serve as templates or models applied in the creation of particular headline occurrences. Firstly, these patterns comprise specific sets of constant elements which build up the basis of the pattern itself. Secondly, they include empty “slots” or positions which are open to be filled by various elements in order to produce specific occurrences or, in this case, headlines following the particular pattern in question. In addition to the constant elements which build it, each pattern contains information about the structural relations among its constant as well as its variable constituents. As a result, once a user recognizes a pattern, they can reactivate previous knowledge and experience regarding the processing operations this particular pattern involves and apply it to the occurrence at hand. Approaching actual occurrences in this manner enables users to economize on the processing resources which would have otherwise been necessary in establishing a new structure. Instead, they can focus only on the variables, i.e. the new elements which fill the slots provided by the recognized pattern. The use of patterns thus enhances efficiency by allowing for analogical processing of numerous actualizations and specific occurrences.
... Etablissement du niveau de difficulté des productions dans une langue en cours d'apprentissage 1. La caractéristique structurale -ou type -de l'énoncé telle qu'elle a été définie par la grammaire générative (Chomsky, 1965). Tout énoncé, compris ou produit, se verra conférer un premier niveau de difficulté sur la base du critère de transformation de la phrase-noyau P introduit dans le modèle chomskien. ...
Article
Full-text available
L’enseignement traditionnel des langues ne produisant pas les résultats escomptés, diverses solutions sont volontiers proposées pour tenter d’y remédier. Cependant, il est très difficile de comparer la pertinence et le succès de ces diverses propositions, car on ne dispose pas d’instruments de mesure aisément transposables. Deux obstacles majeurs peuvent être repérés. Premièrement, comment comparer les performances d’élèves recueillies dans des circonstances et par des procédures d’évaluation différentes, des situations de test et des situations d’expression spontanée, par exemple? En second lieu, comment établir de l’extérieur le niveau de difficulté des énoncés compris ou produits par les élèves lorsque la nature exacte des situations travaillées en classe n’est pas la même et de plus, souvent décrite de manière insuffisamment explicite? Cet article présente une réflexion sur ces obstacles et la solution que nous proposons pour les surmonter. Nous avons développé un instrument permettant de définir le niveau de difficulté des énoncés quel que soit le contexte d’évaluation et les situations travaillées en classe. Un premier test de cet instrument, élaboré au moyen du paradigme de la généralisabilité figure dans la seconde partie.
... Arris & Wexler suggest that forms like (1) and (2) together show that at this particular stage children regard Tense marking as optional (optional infinitive stage). While they did not find any evidence that English children do not attempt to move full verbs over Negation, Radford cites an example by Ehreich et al. (1980:163) where a child produced an inverted use of the verb go in a yes/no question. (3) goes paci in mouth? ...
Chapter
Full-text available
This volume explores the implications of cross-linguistic structures in simultaneous bilingualism. It aims to find cognitive explanations for the presence or absence of cross-linguistic structures that go beyond the debate of ‘one system or two’. The contributors present syntactic, morphological and phonological features that are found in bilingual children, but are untypical of monolingual development, and discuss pertinent methodological issues. The orientation of this volume stands out from competing volumes in the field in that the focus is not limited to similarities between monolingual and bilingual first language acquisition. The volume will be of interest to researchers in the field of bilingualism and primary language acquisition, language theorists, and professionals working with bilingual populations.
Article
Di fatto, oggi sono in voga delle teorie che collegano la psicopatologia grave dell'adulto all'incapacità del bambino nella fase preverbale di formulare o rappresentare le esperienze traumatiche. A questo proposito, vengono analizzati i lavori di Howard B. Levine e Donnel B. Stern, partendo dall'idea che i risultati delle ricerche empiriche siano decisamente rilevanti per la teorizzazione psicoanalitica. Da questi risultati emergono scarse evidenze a supporto dell'ipotesi di un neonato con stati mentali "primitivi e non rappresentati". Di fatto, i risultati delle ricerche suggeriscono proprio il contrario: l'idea di un "neonato competente", dotato di una capacità di discrimina-zione percettiva estremamente accurata e di una capacità innata di registrare e rappresentare l'esperienza soggettiva, a livello di memoria sia procedurale sia dichiarativa, persino nel periodo prenatale. Non sembra quindi plausibile sostenere che i deficit di rappresentazione siano il nucleo della psicopatologia grave nell'adulto, la quale invece è il frutto di manovre difensive nei confronti di verità inconoscibili e indicibili e non dell'assenza di una capacità rappresentativa preverbale. Le ricerche pongono un'importante sfida alle teorie psicoanalitiche che postulano cosiddetti "stati mentali primitivi", esperienze "non formulate", "non rappresentate", "non simbolizzate" o stati "non-consci".
Article
The author cites the prominence of theories that locate serious adult psychopathology in the preverbal infant’s inability to formulate or represent traumatic experience. The work of two such authors, H. Levine and D. B. Stern, is briefly considered. The frame of reference for this investigation is that clinical and academic research findings are highly relevant to psychoanalytic theorizing. It is argued that when such findings are considered, a view of the infant with “primordial and unrepresented” states of mind has little evidence to support it. In fact, research findings summarized herein point to an opposite view: that of the “competent infant,” one with highly accurate perceptual discrimination capacities and an innate ability to register and represent subjective experience in both procedural and declarative memory, even prenatally. Given the infant’s competencies, it seems implausible to hold that representational deficits are at the heart of serious adult psychopathology, which is instead seen to be the result of defensive maneuvers against unknowable and unspeakable truth rather than the absence of a preverbal representational capacity. Current research findings seem to pose a significant challenge for psychoanalytic theories that espouse “primitive mental states”; “unrepresented,” “unformulated,” or “unsymbolized” experience; or “nonconscious” states.
Chapter
This paper is concerned with the acquisition of the morphosyntax of finite verbs by monolingual children acquiring British English as their first language. The primary data are drawn from a large naturalistic sample of more than 100,000 early child utterances (based on the corpus of 39 cross-sectional and 94 longitudinal studies described in Radford, 1990a:11-13). The theoretical framework used will be that of Government and Binding Theory (=GB), in the version outlined in Chomsky (1986), with modifications introduced by Abney (1987) and Pollock (1989). In order to clarify some of the descriptive assumptions made here, I shall beg in by providing a brief outline of the morphosyntax of finite verbs in adult English.
Article
Full-text available
Effective communication instruction has drawn heavily from applied behavior analysis (ABA). Although intervention practices have evolved over time, technologies developed through the adherence to ABA principles and techniques continue to flourish. This article examines ABA-based communication instruction for persons with autism by providing a historical context for ABA as an instructional methodology and by reviewing ABA-related treatment options.
Article
This article reports the latest phase of a research programme whose long-term goal is the creation of a comprehensive theory of linguistic/cognitive development. A partial theory, and computer models embodying aspects of it, were developed in previous phases of the programme (Wolff, 1982) and tested in restricted domains using artificial languages in the main. The aim here is to show how the previously established framework of ideas may be extended to explain the learning of more realistic cognitive structures.
Article
Full-text available
-
Article
A descriptive analysis is presented of the syntactic patterns in 16 corpora of word combinations from 11 children learning either English (six children), Samoan, Finnish, Hebrew, or Swedish. The mean utterance lengths range up to about 1.7 morphemes. There are both reanalyses of corpora in the literature and new corpora. The data indicate that each child has learned a number of positional formulae that map components of meaning into positions in the surface structure. Each formula expresses a specific, often quite narrow, range of relational conceptual content. In each corpus, the bulk of the combinations are generated by a small number of such formulae; the differences between one corpus and another are considerable, and their nature indicates that the formulae are independent acquisitions. The formulae are not broad rules of the kind usual in transformational grammars; and the semantic categories are usually much more specific than those of case grammars or those proposed by Schlesinger (although Schlesinger's views are supported in other respects). Also, there is no evidence for grammatical word classes. In general, the evidence indicates less grammatical competence at this stage of development than children are being credited with in much current work. Two kinds of phenomena involving free word order are noted. One kind, not previously reported, is called a "groping pattern": positional formulae are sometimes preceded by an earlier stage in which the components are unordered. The lack of order is due to the child groping to express a meaning before he has learned a rule that determines the position of the elements. The other kind is due to the learning of two formulae, one for each order: longitudinal study of some cases indicates that the two orders were learned at separate times and that they may have subtly different semantic content.
Article
Errors in child speech show that some children initially formulate tense-hopping and subject-auxiliary inversion as copying without deletion. Other errors suggest that some children may formulate other movement rules as deletion without copying. A claim about the nature of the language acquisition device is made on the basis of our analysis of these errors: the language acquisition device formulates hypotheses about transformations in terms of basic operations. The basic-operations hypothesis predicts that for any transformation which is composed of more than one basic operation, there exists a class of errors in child speech correctly analyzed as failure to apply one (or more) of the operations specified in the adult formulation of the rule.RésuméLes erreurs rencontrées dans le discours des enfants montrent que certains enfants formulent d'aboard la transformation affixe et l'inversion sujet-auxiliaire comme copie sans effacement. D'autres erreurs suggèrent que certains enfants peuvent formuler d'autres règles de mouvement comme l'effacement sans copie. En se fondant sur l'analyse de ces erreurs on fait une proposition sur le mécanisme de l'acquisition du langage: ce mécanisme d'acquisition du langage formule les hypothèses sur les transformations en terme d'opérations fondamentales. Cette hypothése d'opérations fondamentales prédit que pour chaque transformation composée de plus d'une opération fondamentale, il existe une classe d'erreur dans le langage de l'enfant. Cette classe d'erreur peut s'analyser comme unéchecàappliqueràune (ou plus) des opérations spécifiées par la formulation adulte de la régle.
Article
The assumption that language acquisition is relatively independent of the amount and kind of language input must be assessed in light of information about the speech actually heard by young children. The speech of middle-class mothers to 2-year-old children was found to be simpler and more redundant than their speech to 10-year-old children. The mothers modified their speech less when talking to children whose responses they could not observe, indicating that the children played some role in eliciting the speech modifications. Task difficulty did not contribute to the mothers' production of simplified, redundant speech. Experienced mothers were only slightly better than nonmothers in predicting the speech-style modifications required by young children. These findings indicate that children who are learning language have available a sample of speech which is simpler, more redundant, and less confusing than normal adult speech.
Article
Some aspects of a theory of grammar are presented which derive from a formal theory of language acquisition. One aspect of the theory is a universal constraint on analyzability known as the Freezing Principle, which supplants a variety of constraints proposed in the literature. A second aspect of the theory is the Invariance Principle, a constraint on the relationship between semantic and syntactic structure that makes verifiable predictions of syntactic universals. The relationship between the notion of 'explanatory adequacy' of a theory of grammar and the learnability of a class of transformational grammars is discussed.
Article
Coordinate conjunction was evaluated in early child language with regard to its structural properties. In a series of four studies, 60 two- and three-year-olds grouped by MLU were studied in an elicited imitation task where in the linguistic form of sentences was varied according to conjunction structure (whether sentential or phrasal) and according to pattern of redundancy deletion in conjunction reduction (whether forward or backward in directionality). Both factors were found to affect children's imitation. The results suggested specific constraints on the structure of coordination in child language.