In the past century major changes in the study of language learning, and its attendant effects on language instruction, derived from the espousal of behavioral psychology by structural linguistics of the Bloomfieldian persuasion. Did not the founder of this strand of structuralism, who collaborated in the Army language manual for the teaching of Russian under the revealing pseudonym of I. M. Lisnin, declare: “Language learning is overlearning, nothing else is of any use”? Noam Chomsky's (1959) brilliant demonstration of the reductionism of the most extreme form of behaviorism—Skinner's operant conditioning (Review of B. F. Skinner's Verbal Behavior, Language, 35, 26–58)—triggered the demise of the influence of behavioristic views of language acquisition. This demise was accompanied by that of the structural approach to second language teaching associated with such applied linguists as Robert Lado, although it must be acknowledged that the combination of these two theoretical strands did lead to significant changes in foreign language classrooms—for example, the abandonment of grammar translation and a shift of focus from written texts to speech. However, what applied linguists of that generation failed to do was to observe how actual second language learners in both naturalistic and instructed contexts process and reorganize linguistic input, how they intake it, and how they turn it into output in communicative interactions. It was not until the impact of Chomskian-inspired studies of first language acquisition and Pit Corder's (1967) seminal “The significance of learners' errors” (International Review of Applied Linguistics, 5, 161–170) that the second language learner came into focus and that the field of second language acquisition research began to flourish. In this connection, it is noteworthy that bringing to a wider international audience the proceedings of the Neuchâtel colloquia led by Corder served as a catalyst for the launching of SSLA.

This article shows how language processing is intimately tuned to input frequency. Examples are given of frequency effects in the processing of phonology, phonotactics, reading, spelling, lexis, morphosyntax, formulaic language, language comprehension, grammaticality, sentence production, and syntax. The implications of these effects for the representations and developmental sequence of SLA are discussed. Usage-based theories hold that the acquisition of language is exemplar based. It is the piecemeal learning of many thousands of constructions and the frequency-biased abstraction of regularities within them. Determinants of pattern productivity include the power law of practice, cue competition and constraint satisfaction, connectionist learning, and effects of type and token frequency. The regularities of language emerge from experience as categories and prototypical patterns. The typical route of emergence of constructions is from formula, through low-scope pattern, to construction. Frequency plays a large part in explaining sociolinguistic variation and language change. Learners' sensitivity to frequency in all these domains has implications for theories of implicit and explicit learning and their interactions. The review concludes by considering the history of frequency as an explanatory concept in theoretical and applied linguistics, its 40 years of exile, and its necessary reinstatement as a bridging variable that binds the different schools of language acquisition research.

The use of formulaic expressions by second language learners has received little attention from second language acquisition research investigating interlanguage temporal systems. Instead, this field of inquiry has emphasized the productive use of verbal morphology by employing type-token analyses. This paper considers the proposed developmental sequence of formula > low-scope pattern > construction in the emergence of future expression in a longitudinal study of 16 adult learners of English as a second language. The findings suggest that the use of formulaic expressions may be subject to individual variation and that learners may use formulaic expressions to different degrees when developing form-meaning associations even in the same grammatical subsystem, such as the tense-aspect system. The findings also suggest, however, that the practice of favoring type over token analysis as a matter of course may eliminate valuable information about the emergence and development of temporal expression.

Using frequency findings from corpus linguistics, this paper explores the relationship between the information presented in ESL-EFL materials and what is known about actual language use based on empirical studies. Three aspects of materials development for grammar instruction are discussed: the grammatical features to be included, the order of grammatical topics, and the vocabulary used to illustrate these topics. For each aspect, we show that there are often sharp contrasts between the information found in grammar materials and what learners encounter in the real world of language use. In our conclusion, we argue that a selective revision of pedagogy to reflect actual use, as shown by frequency studies, could result in radical changes that facilitate the learning process for students.

Although there are certainly observable frequency effects in language, in most cases, there are alternative approaches to explanation that more directly relate to the essential characteristic of language—that it is a system relating form to meaning. For example, for both word choice in production and ambiguity resolution in comprehension, meaning-based approaches can often provide equally satisfying, or more satisfying, explanations. In the meaning-based approach, the statistical structure of the language can affect the development of linguistic knowledge (for example, by influencing acquisition order or providing evidence for developing grammars); however, linguistic knowledge is not itself knowledge of the statistical structure of language. An example is provided of how frequency may relate to grammaticality judgments of nonnative speakers acquiring multiple wh-questions.

Phonological evidence supports the frequency-based model proposed in the article by Nick Ellis. Phonological reduction occurs earlier and to a greater extent in high-frequency words and phrases than in low-frequency ones. A model that accounts for this effect needs both an exemplar representation to show phonetic variation and the ability to represent multiword combinations. The maintenance of alternations conditioned by word boundaries, such as French liaison, also provides evidence that multiword sequences are stored and can accrue representational strength. The reorganization of phonetic exemplars in favor of the more frequent types provides evidence for some abstraction in categories beyond the simple registration of tokens of experience.

The extent to which form-focused instruction contributes to the acquisition of second language implicit knowledge is controversial. Whereas Krashen (1993) has argued that the effects of FFI on acquisition are peripheral, N. Ellis (this issue) sees FFI as facilitative and even necessary for developing implicit L2 knowledge. This article examines the role of FFI in developing implicit knowledge by reviewing 11 studies that have examined the effect of FFI on learners' free production. The review suggests that FFI can contribute to the acquisition of implicit knowledge and points to two variables that appear to influence its success—the choice of the target structure and the extent of the instruction. FFI involving extensive instruction directed at “simple” structures was more likely to succeed. However, limited instruction directed at complex structures also proved effective, provided that the target structures are readily available in noninstructional input.

Ellis refers to a wide range of data supporting the uncontroversial claim that there are frequency effects in linguistic and other behavior. He further resurrects the long-discredited claim that language acquisition consists of frequency-based abstraction of regularities from input. What he fails to do is to show how the former claim leads to the latter, or indeed to show any evidence for this claim. Ellis ignores fundamental and well-known problems—the poverty of the stimulus, cases of instantaneous acquisition, evidence for innate knowledge, evidence that even children regularly ignore perceptual input, the compositionality of concepts, and the systematicity of language—that make his frequency-based account of acquisition a nonstarter.

In this response to Ellis's target article on frequency in language processing, language use, and language acquisition, we argue in favor of a role for frequency in several areas of second language acquisition, including interactional input and output and speech processing. We also discuss areas where second language acquisition appears to proceed along its own route and at its own pace regardless of the frequency of the input, as well as areas where input is infrequent but acquisition appears to be unimpeded. Our response is intended to highlight the complexity of the task of deciphering the role and importance of frequency.

Input-driven models provide an explicit and readily testable account of language learning. Although we share Ellis's view that the statistical structure of the linguistic environment is a crucial and, until recently, relatively neglected variable in language learning, we also recognize that the approach makes three assumptions about cognition and language learning that are not universally shared. The three assumptions concern (a) the language learner as an intuitive statistician, (b) the constraints on what constitute relevant surface cues, and (c) the redescription problem faced by any system that seeks to derive abstract grammatical relations from the frequency of co-occurring surface forms and functions. These are significant assumptions that must be established if input-driven models are to gain wider acceptance. We comment on these issues and briefly describe a distributed, instance-based approach that retains the key features of the input-driven account advocated by Ellis but that also addresses shortcomings of the current approaches.

This peer commentary emphasizes the importance of placing frequency in an overarching theoretical framework of language acquisition. Three issues are raised that appear to be both important and timely: (a) the question of how innate, or initial, cognition can deal with stimulus frequency; (b) the likelihood that frequency has a differential impact depending on the type of knowledge concerned; and (c) preliminary evidence that frequency affects receptive language knowledge more than productive knowledge, which raises the issue of a possible dissociation between the two knowledge types.

Nick Ellis's article is impressive for its breadth of scholarship and the cogent case made for frequency as an important factor in second language acquisition (SLA). This response begins with a brief historical sketch, which aims to contextualize the frequency factor in terms of the evolution of SLA research. Although researchers have known about a frequency effect for some time, until recently, we have lacked neurologically plausible models and technologically convenient means of measuring and testing frequency effects in input. Still, as relevant and important as a frequency factor is, it requires greater definition and qualification. For instance, L2 learners are agents of their own learning process. They do not merely record frequency; they categorize frequently occurring patterns, abstract, and generalize from them. Higher level generalizations emerge from interactions, impose top-down expectancies on future data, and, in turn, are altered by the data in a perpetually dynamic interplay. In a similar fashion, it is incumbent on SLA researchers not to be satisfied with a frequency explanation, but to interpret it—in short, to make sense of frequency.

Ellis's target article suggests that language processing is based on frequency and probabilistic knowledge and that language learning is implicit. These findings are consistent with those of SLA researchers working within a variationist framework (e.g., Tarone, 1985; Bayley & Preston, 1996). This paper provides a brief overview of this research area, which has developed useful models for dealing with frequency effects in language use, and describes a psycholinguistic model of language variation currently being proposed by Preston (2000a, 2000b) that dovetails nicely with Ellis's proposals. The present commentary addresses the question “To what extent is the learner's interlanguage passively and unconsciously derived from input frequencies?” Ellis does state that factors other than frequency are also important for SLA—specifically, conscious noticing and social context. A related factor is the learner'screativity, revealed when learners' noticing leads them to view utterances not just as potential objects of analysis but as potential objects of language play. Noticing results in the selective internalization of language input in interactions with various L2 speakers, and creativity occurs in the learner's consequent production of any of a range of different voices thus internalized for the purposes of expressing social identity and of language play (Tarone 2000).

This response addresses the following points raised in the commentaries: (a) complementary learning mechanisms, the distinction between explicit and implicit memory, and the neuroscience of “noticing”; (b) what must and what need not be noticed for learning; (c) when frequency fails to drive learning, which addresses factors such as failing to notice cues, perseveration, transfer from L1, developmental readiness, thinking too hard, pedagogical input, and practicing; (d) attention and form-focused instruction; (e) conscious and unconscious knowledge of frequency; (f) sequences of acquisition—from formula, through low-scope pattern, to construction; (g) the Fundamental Difference hypothesis; (h) the blind faith of categorical grammar; (i) Labovian variationist perspectives; (j) parsimony and theory testing; (k) universals and predispositions; and (l) wanna-contractions. It concludes by emphasizing that language acquisition is a process of dynamic emergence and that learners' language is a product of their history of usage in communicative interaction.