Theoretical Foundations of First Language Acquisition in the Toddler Years
The Theoretical Problem of Early Language Acquisition
Between a child's first birthday and her third, something extraordinary occurs that no engineered system has replicated and no contemporary theory has fully explained. A twelve-month-old with a productive lexicon of perhaps three to five conventionalized words, no demonstrable syntax, and no metalinguistic awareness becomes, by thirty-six months, a child who comprehends thousands of lexical items, produces structured multiword utterances, marks tense and number systematically (and productively overregularizes them), participates in extended conversational exchange, and uses language to comment on absent referents, hypothetical states, and her own past and future mental states. The transition spans roughly seven hundred and thirty days. Within that window, the average child acquires vocabulary at rates between five and nine new words per day after the so-called vocabulary spurt, masters the basic argument-structure templates of the ambient language, and converges on adult-like phonological, morphological, and pragmatic norms. This is the phenomenon that any theory of early language acquisition must explain, and it is the phenomenon that, despite sixty years of concentrated investigation, continues to resist adequate characterization.
The central question this chapter addresses is not which theoretical tradition is "correct," but how the commitments of the major traditions can be organized into a coherent architecture that respects the empirical phenomena characterizing the one-to-three-year window. We argue that the apparent competition between nativist, usage-based, social-pragmatic, and emergentist accounts dissolves once each tradition's explanatory locus is properly identified. Innate constraints, statistical mechanisms operating on structured input, and socially scaffolded interaction are not rival explanations of the same data; they are components of a multi-level system whose acquisition dynamics cannot be reduced to any single component alone. Our aim in this opening chapter is therefore to make these components explicit, to compare their explanatory reach against the empirical record, and to propose a working framework that will orient the remainder of the volume.
Before turning to the theoretical traditions themselves, however, we must fix the operational definitions that will be used throughout the book. Without terminological precision, the comparison of theories is a comparison of incommensurable objects, and much of the polemical energy in the field derives from exactly this kind of conceptual slippage.
Operational Definitions for the Volume
The term *language* is used in this volume in a deliberately restricted sense. We do not use it to denote the abstract structural object posited by generative grammar (an idealized internalist object specified by formal principles), nor do we use it to denote the open-ended population of utterances produced by a speech community (a usage-based externalist object). Instead, we use *language* to refer to the internalized knowledge system that a child constructs between birth and roughly six years of age, including its lexical, morphological, syntactic, phonological, and pragmatic-pragmatic components, and the procedures by which that system is deployed in comprehension and production. This knowledge system is *generative* in the weak sense that it supports the production and comprehension of utterances that have never been encountered before; it is *constrained* in that not all logically possible utterances are grammatical or interpretable; and it is *gradient* in that probabilistic preferences overlay categorical restrictions. When we speak of "the child's language" at age two, we mean this internal system as it can be inferred from spontaneous production, comprehension probes, elicited imitation, and experimental tasks validated for that age range.
*Acquisition* refers to the developmental process by which this internal system is constructed. We deliberately avoid the stronger term *learning* in many contexts, because *learning* carries connotations of explicit instruction, conscious hypothesis-testing, or reward-driven modification that do not characterize the bulk of early language development. Acquisition in our sense is largely implicit, often unconscious on the part of the child, and not reducible to any single mechanism. It proceeds through exposure to linguistic input, through interaction with caregivers and peers, through the child's own productive attempts, and through the operation of domain-specific and domain-general cognitive capacities whose precise contribution remains contested.
*Competence*, the third operational term, requires careful handling because it has been used in sharply different ways across traditions. We follow a tripartite distinction introduced by Bates and MacWhinney and elaborated by several subsequent researchers. *Linguistic competence* refers to the abstract knowledge of phonological, lexical, morphological, and syntactic structure that the child has internalized. *Pragmatic competence* refers to the knowledge of how language is used in context, including the conventions of turn-taking, the marking of given versus new information, the management of joint attention, and the selection of appropriate registers. *Functional competence* refers to the integration of linguistic and pragmatic competence in real-time comprehension and production, including the ability to interpret intended meaning in the face of ambiguity, ellipsis, and non-literal usage. This tripartite division is not uncontested — particularly the boundary between linguistic and pragmatic competence — but it allows us to be explicit about what each theoretical tradition is claiming to explain. A nativist account of competence emphasizes linguistic competence narrowly construed; a social-pragmatic account emphasizes pragmatic competence as primary; an emergentist account often claims that the distinction between linguistic and pragmatic competence is itself a post-hoc artifact of mature systems.
A fourth operational term, *input*, denotes the totality of linguistic stimulation available to the child, including child-directed speech, overheard adult-directed speech, ambient television and recorded media, and the child's own vocalizations. Throughout this volume we will treat input as a structured rather than a uniform quantity. Child-directed speech differs systematically from adult-directed speech in prosody, lexical frequency distribution, syntactic complexity, and discourse function; the input the child receives is therefore not a random sample of the ambient language, and any account of acquisition must take this distributional structure seriously.
Finally, *developmental window* refers to the one-to-three-year age range that constitutes the empirical focus of this volume. This window is bounded on one side by the canonical first-word milestone at around twelve months and on the other by the emergence of complex syntax, narrative competence, and metalinguistic awareness between ages two-and-a-half and three-and-a-half. Within this window the four phenomena on which we will focus most intensively — vocabulary growth rate, syntactic productivity, overregularization, and the role of joint attention — are most clearly observable and most clearly differentiated across theoretical accounts.
The Major Theoretical Traditions and Their Commitments
Six theoretical traditions have shaped contemporary work on early language acquisition. Each makes distinctive commitments regarding what is innate, what is learned, what role input plays, and what cognitive and social mechanisms are implicated. We treat them in roughly chronological order of their dominance in the field, but the chronology of emergence is not a chronology of merit, and several of the later traditions have absorbed and reformulated insights from the earlier ones.