The paper offers a model of language development, first and second, within a processing perspective. We first sketch a modular view of language, in which competence is embodied in the processing mechanisms. We then propose a novel approach to language acquisition (Acquisition by Processing Theory, or APT), in which development of the module occurs as a natural product of processing activity, without any acquisition mechanisms as such. The approach is illustrated and explicated through examples of the development of content words, derivational morphology, the functional category I with its variable features, and Case and thematic roles, as well as apparent cross-linguistic variation in processing strategies and the status of bootstrapping in the model. We then examine some possible applications to issues in second language acquisition – noticing the gap, the initial state, transfer, and the apparent limits of SLA – and finally offer a broader perspective on the model: its scope, its relations to other approaches, and its possible limits.

As an occasional visitor to the land of SLA, I found myself somewhat mystified by the approach of Truscott and Sharwood Smith (henceforth TSS). Unless I have totally misunderstood them, they are arguing against the separate existence of both a Language Acquisition Device (LAD) and a Universal Grammar (UG). But who ever thought they were two? Occasionally linguists may write about one rather than the other, but I have always assumed them to be the same thing appearing in different guises (like Christ and the Holy Ghost in Christian theology). If you have UG, why would you need LAD? Processing with the aid of UG seems the only way to go – to the extent that UG is available to SLA, an issue TSS does not directly address.

Truscott and Sharwood Smith (henceforth T&SS) propose a novel theory of language acquisition, ACQUISITION BY PROCESSING THEORY (APT), designed to account for both first and second language acquisition, monolingual and bilingual speech perception and parsing, and speech production. This is a tall order. Like any theoretically ambitious enterprise, the APT shares certain properties with much that has gone before. Like the Competition Model (CM; MacWhinney, 1987, 1997; MacWhinney and Bates, 1989, inter alia) and other associative network connectionist learning models, the APT eschews a Language Acquisition Device (LAD) by treating acquisition as the strengthening of levels of representation activation. A parser can produce multiple representations of a parse string in parallel, which then ‘compete’ as analyses for an input string. Unlike the CM, however, the APT is not motivated by a solid program of empirical studies in language acquisition or cross-language processing. Nor does it strike me as theoretically coherent, for the APT, unlike the CM, assumes that knowledge of language involves knowledge of grammatical structure and that the parser makes deterministic use of Universal Grammar in the form of a Minimalist grammar. The determinism is important here; the claim to eliminate LAD hinges on it.

In their keynote contribution, Truscott and Sharwood Smith offer a general model of language development from a processing perspective. As they state, their model is very ambitious: Their ‘acquisition by processing’ theory (APT) aims not only at explaining both first and second language acquisition but also real-time processing in language comprehension and production. APT takes a cross-disciplinary approach that intends to bring together research on linguistic structure and on general cognition. However, the joint contribution of linguistic and psycholinguistic approaches is mainly evident in terms of theoretical concepts (e.g. UG, syntactic rules, modules, activation) because the presented empirical evidence is limited in scope.

Acquisition by Processing Theory (APT) is a unified account of language processing and learning that encompasses both L1 and L2 acquisition. Bold in aim and broad in scope, the proposal offers parsimony and comprehensiveness, both highly desirable in a theory of language acquisition. However, the sweep of the proposal is accompanied by an economy of description that makes it difficult to evaluate the validity of key learning claims, or even how literally they are to be interpreted. Two in particular deserve comment; the first concerns the learning mechanisms responsible for adding new L2 grammatical information, and the second the theoretical and empirical status of the activation concept used in the model.

This keynote article proposes a new model of language development based on processing, the sole mechanism of acquisition for the Acquisition by Processing Theory (APT). The language module – adapted from Jackendoff's distinction between integration (building complex structures) and interface (facilitating information transfer at the intersections of the sub-modules) – has phonological, syntactic and conceptual processors that interact with each other, working memory and lexical entries. On the one hand, the application of Jackendoff's model to language learning bridges a gap between syntactic theory and processing. On the other hand, its breadth sometimes brings together strange bedfellows, as, for example, minimalist syntax and APT's ‘clear affinities with connectionism’ (p. 17). In order to explore this hybrid model within a focused area, I will limit my comments to a discussion of features of I(nflection), Case and thematic (theta) roles, highlighting the question of how much syntactic theory APT needs. I will suggest that the model could profit from a greater exploitation of morphological input and its role in parameter setting.

Truscott and Sharwood Smith (henceforth T&SS) attempt to show how second language acquisition can occur without any learning. In their APT model, change depends only on the tuning of innate principles through the normal course of processing of L2. There are some features of their model that I find attractive. Specifically, their acceptance of the concepts of competition and activation strength brings them in line with standard processing accounts like the Competition Model (Bates and MacWhinney, 1982; MacWhinney, 1987, in press). At the same time, their reliance on parameters as the core constructs guiding learning leaves this model squarely within the framework of Chomsky's theory of Principles and Parameters (P&P). As such, it stipulates that the specific functional categories of Universal Grammar serve as the fundamental guide to both first and second language acquisition. Like other accounts in the P&P framework, this model attempts to view second language acquisition as involving no real learning beyond the deductive process of parameter-setting based on the detection of certain triggers. The specific innovation of the APT model is that changes in activation strength during processing function as the trigger to the setting of parameters. Unlike other P&P models, APT does not set parameters in an absolute fashion, allowing their activation weight to change by the processing of new input over time. The use of the concept of activation in APT is far more restricted than its use in connectionist models that allow for Hebbian learning, self-organizing features maps, or back-propagation.

Truscott and Sharwood-Smith's (henceforth T&SS's) paper offers an interesting set of hypotheses about one possible processing perspective in research on language acquisition. What is striking about this exposition of their model is that it ignores almost entirely the context of previous research on this issue. Embedding their exposition in its historical context would have in no way diminished the innovation entailed in their proposal. Instead it could have served as an instrument for ensuring conceptual parsimony, which in my view is flawed in this exposition.

This essay by Truscott and Sharwood-Smith is a valiant attempt with a laudable goal. It seeks to show how different perspectives and disciplines can capture what is happening in acquisition and notably in L2 acquisition. Nonetheless, I think that the results are much closer to an elaborated grammatical theory than an elaborated processing theory (thus it seems less important to me than to the authors to choose one label over the other). Their essential, sensible idea is that where different grammars provide independent analyses of constructions, both are computed (even where the lexical items belong to only one grammar). Therefore conflict is experienced which produces computational demands. However this is pure grammar – just a more sophisticated situation where, as I like to argue, the L2 person has Multiple Grammars (Roeper, 1999), creating more computational work as well. In this respect, the situation is no different from L1 acquisition, where the course of acquisition involves generation and maintenance of multiple grammars, some of which are shed and some retained in the Final State.

To begin with, we thank the commentators for their remarks, which both intrigued and challenged us in various ways. Time and space limits prevent us from doing justice to all the points they raised, so we will focus on those we consider most significant and are best prepared to deal with. One of these points, of course, is the overall character and status of our proposal. Pienemann terms Acquisition by Processing (APT) ‘an interesting point of departure’ (p. 38) for the development of a theory of acquisition and processing. Harrington and Carroll say in effect that it can't be done; others seem more open to the possibility. In any case, we quickly accept Roeper's suggestion that what we have presented is more a ‘prole-gomena’ (p. 21) to a theory than a full-fledged theory in itself.

We present a hypothesis for a specific kind of code-mixing in young bilingual children, during the development of their two first languages, one of which is considerably weaker than the other. Our hypothesis, which we label the Ivy Hypothesis, is that, in the interaction meant to be in the weaker language, the child uses portions of higher syntactic structure lexically instantiated in the stronger language combined with lower portions in the weaker language. Code-mixing patterns were studied in five Swedish-French/Italian children aged 2–4. The parts of the code-mixed utterances reflected as much syntactic structure of each language as was used in monolingual utterances in the same recording of each child. This uneven development, which is due to different amounts of input of the two languages, can be accounted for by assuming that syntactic structure is acquired by building each language from the bottom up through lexical learning.

This study examines the acquisition of the voicing contrast in German-Spanish bilingual children, on the basis of the acoustic measurement of Voice Onset Time (VOT). VOT in four bilingual children (aged 2;0–3;0) was measured and compared to VOT in three monolingual German children (aged 1;9–2;6), and to previous literature findings in Spanish. All measurements were based on word-initial stops extracted from naturalistic speech recordings. Results revealed that the bilingual children displayed three different patterns of VOT development: 1. Delay in the phonetic realization of voicing: two bilingual children did not acquire long lag stops in German during the testing period; 2. Transfer of voicing features: one child produced German voiced stops with lead voicing and Spanish voiceless stops with long lag voicing; and 3. No cross-language influence in the phonetic realization of voicing. The relevance of the findings for cross-linguistic interaction in bilingual phonetic/phonological development is discussed.

This note points out an incorrect quotation of the theory of codeswitching outlined in various publications of Carol Myers-Scotton as cited by Boussofara-Omar (2003). In addition, the issue is raised whether the type of multi-lectal situation encountered in the varieties of Arabic spoken in Tunisia should be analyzed with a model designed to deal with ‘classic’ codeswitching.