Loading field map
Language acquisition is the study of how human beings come to know and use a language. Although the term can refer to the learning of any language at any age, the subfield within linguistics focuses overwhelmingly on first language acquisition—the process by which infants and children develop the ability to understand and produce their native language—and on the comparison between that process and second language acquisition in older learners. The field asks a deceptively simple question: what exactly does a child know when they know a language, and how do they get from not knowing it to knowing it?
The stakes of this question extend well beyond the classroom. Language is the most complex cognitive skill that every typically developing human acquires, and it is acquired without formal instruction, on the basis of noisy and incomplete input, in a remarkably short time. Understanding how this happens bears on fundamental questions about the nature of the human mind: whether knowledge is learned from experience or constrained by innate structures, whether there is a critical period for learning, and whether the mechanisms that support language are specific to language or general to all learning.
Any serious account of language acquisition must confront what philosophers and linguists call the poverty of the stimulus. Children hear a finite, messy set of sentences, many of them incomplete or containing errors, and from this they derive a grammar that generates an infinite set of novel sentences they have never heard. They also acquire knowledge that goes beyond anything directly observable in the input. For example, English-speaking children learn that the sentence "John is too stubborn to talk to" means something different from "John is too stubborn to talk," even though the surface difference is just one word. They are not taught this distinction, and it is not present in any obvious way in the sentences they hear. The puzzle is how they arrive at such abstract and subtle knowledge so reliably.
This problem is not merely philosophical. It shapes the entire field because different answers to it lead to radically different research programs. If the input is rich enough to support general learning mechanisms, then language acquisition can be studied as a special case of learning. If the input is genuinely insufficient, then the child must bring substantial prior structure to the task, and the field becomes the study of that innate endowment.
The most influential answer to the poverty of the stimulus was formulated by Noam Chomsky from the 1950s onward. Chomsky argued that the gap between input and knowledge is so large that children must be born with a dedicated biological faculty for language—a "language organ" or Universal Grammar. This faculty, on the nativist account, does not contain the rules of any particular language. Rather, it contains a set of abstract principles that constrain what a possible human language can be, plus a set of parameters that vary within fixed limits. The child's task is to set these parameters on the basis of the ambient language. For example, a principle might state that all languages have subjects at some level of structure, while a parameter might determine whether the subject must be overtly pronounced (as in English) or can be dropped (as in Italian or Spanish).
The nativist program transformed the study of language acquisition from a descriptive enterprise into a theoretical one. Instead of cataloguing the errors children make, researchers began to look for evidence that children's errors are principled—that they reflect the operation of an innate grammar rather than imperfect imitation. A classic example is the phenomenon of overregularization. English-speaking children go through a stage in which they say "goed" and "foots" instead of "went" and "feet." They have never heard these forms from adults. The nativist interpretation is that the child has extracted a rule ("add -ed for past tense") and is applying it too broadly—evidence that the child is constructing a grammar, not copying surface forms.
The nativist approach has been enormously productive, generating detailed predictions about the order in which children acquire specific structures, the kinds of errors they do and do not make, and the ways in which language acquisition breaks down in cases of deprivation or impairment. Its central limitation is that Universal Grammar has proven difficult to specify precisely. The theory has undergone multiple major revisions, and critics argue that the content of the innate endowment has been adjusted to fit whatever phenomena are currently under investigation. The strongest evidence for nativism remains the logical argument from the poverty of the stimulus, which is compelling but indirect.
A major alternative tradition, often called usage-based or constructivist, rejects the premise that the input is impoverished. This approach, associated with researchers such as Michael Tomasello and Joan Bybee, argues that children do not need an innate grammar because they are equipped with powerful general learning mechanisms—statistical pattern detection, intention reading, and analogy—and because the input, when properly analyzed, is far richer than the nativist argument assumes.
On the usage-based account, children begin with specific, concrete utterances they have heard and only gradually abstract general patterns from them. A child might first learn "I wanna go" as an unanalyzed chunk, then later learn to substitute elements ("I wanna eat," "Daddy wanna go"), and only much later acquire a general rule for forming such sentences. The order of acquisition is therefore driven by frequency and communicative function: children learn the constructions that are most common and most useful in their daily lives, not the ones that are most structurally basic.
This tradition has produced a large body of corpus studies showing that children's early language is heavily tied to specific words and contexts, and computational models demonstrating that statistical learning from input can account for many phenomena that were once thought to require innate knowledge. Its central limitation is that it has struggled to explain the more abstract and subtle aspects of adult grammar—the kind of knowledge exemplified by the "too stubborn to talk to" distinction—where the relevant evidence is vanishingly rare in the input. Usage-based researchers respond that these phenomena are acquired later and through more general cognitive processes, but the debate remains unresolved.
The disagreement between nativist and usage-based approaches is often framed as a debate over domain specificity: is the mechanism that acquires language specialized for language, or is it the same mechanism used for other kinds of learning? This question has driven much of the empirical research in the field.
One influential line of evidence comes from the study of statistical learning. In the 1990s, researchers showed that infants can track the statistical regularities in a stream of artificial speech sounds and use those regularities to segment words from continuous speech. This finding was initially taken as support for the usage-based position, since it demonstrated a powerful general learning mechanism operating in infancy. However, subsequent research showed that statistical learning is itself constrained in ways that look language-specific: infants are better at tracking statistics over speech-like sounds than over other kinds of stimuli, and they apply different statistical computations to different types of linguistic structure. The finding thus did not settle the debate but rather refined it.
A related line of research concerns the critical period hypothesis—the claim that language must be acquired within a certain window of development to be acquired fully. Evidence for a critical period comes from studies of individuals who were deprived of language input in childhood, such as the case of a girl named Genie, who was isolated from language until age thirteen and never acquired full grammatical competence. Additional evidence comes from second language acquisition: adults who learn a new language rarely achieve native-like proficiency, and their performance declines with age of onset. The critical period is often cited as evidence for a specialized language faculty, since it suggests that the mechanism for language is available only during a specific developmental window. However, the evidence is complicated by the fact that general learning abilities also change with age, and some researchers argue that the apparent critical period reflects motivational, social, or cognitive factors rather than a dedicated language-specific window.
The study of second language acquisition (SLA) is a distinct but closely related subfield. It asks how learners acquire a language after the first has been established, and it addresses questions that do not arise in first language acquisition: What is the role of the first language? Why do learners at the same stage make similar errors regardless of their native language? Why do most adult learners fail to achieve native-like competence?
SLA research has been organized around several competing frameworks. The earliest influential framework, contrastive analysis, held that errors in second language learning could be predicted by comparing the structures of the first and second languages: where the languages differ, learners would struggle. This approach was largely abandoned when empirical research showed that many predicted errors do not occur and many actual errors are not predicted. It was replaced by error analysis, which focused on the systematic patterns in learners' errors, and then by interlanguage theory, which treats the learner's developing system as a language in its own right, with its own rules and logic, rather than as a defective version of the target language.
A major debate in SLA concerns the role of explicit instruction and conscious learning. Stephen Krashen's influential but controversial monitor model distinguished between acquisition, which occurs unconsciously through meaningful communication, and learning, which occurs consciously through instruction, and argued that learning can only "monitor" or edit output, not contribute to fluency. This distinction has been criticized as unfalsifiable, but it stimulated a large body of research on the effects of instruction, feedback, and practice. The current consensus is that explicit instruction can be effective, particularly for structures that are difficult to notice in the input, but that its effects are constrained by the learner's developmental readiness—a finding that echoes the nativist claim that acquisition follows an internally driven sequence.
SLA also engages with the critical period hypothesis more directly than first language research. The finding that age of onset strongly predicts ultimate attainment in a second language is robust, but its interpretation is contested. Some researchers argue that it reflects a biologically determined window for language acquisition; others argue that it reflects the different circumstances of child and adult learners, such as the amount and type of input, the social pressures, and the entrenchment of the first language. The debate remains active, and it has practical implications for language teaching policy.
A third major tradition, which overlaps with both nativist and usage-based approaches, focuses on the role of the linguistic environment. This tradition, associated with researchers such as Catherine Snow and Michael Long, examines the ways in which caregivers modify their speech when talking to children and the ways in which learners and their interlocutors negotiate meaning in conversation.
Research on child-directed speech has shown that adults do not typically teach language explicitly but do adjust their speech in systematic ways: they speak more slowly, use shorter utterances, exaggerate intonation, and repeat more often. The nativist tradition initially dismissed this input as too degenerate to matter, but subsequent research showed that child-directed speech is actually well-formed and that its properties correlate with the rate of language development. The usage-based tradition has made child-directed speech central to its account, arguing that the frequency and distribution of constructions in the input directly shape the order and manner of acquisition.
In SLA, the interaction hypothesis holds that conversation is not merely a context for practice but the primary site of acquisition. When learners encounter communication breakdowns, they are pushed to notice gaps in their knowledge, to modify their output, and to attend to input that they might otherwise ignore. This hypothesis has generated a large body of experimental research on the effects of negotiation, recasts (reformulations of a learner's erroneous utterance), and other forms of feedback. The findings are broadly positive: interaction does promote acquisition, and feedback does help, but the effects are variable and depend on the learner's level, the type of structure, and the timing of the feedback.
The field today is characterized less by a single dominant paradigm than by a set of overlapping research programs that share methods and data while disagreeing on interpretation. The nativist program continues to be influential, particularly in the study of syntax and in the investigation of language disorders, but it has become more modest in its claims and more open to input-driven explanations. The usage-based program has grown substantially, driven by advances in corpus linguistics and computational modeling, but it has not succeeded in eliminating the need for some account of how abstract knowledge arises. The interaction tradition has become the dominant framework in SLA research, but it is increasingly integrated with cognitive and social approaches rather than standing alone.
Several developments have reshaped the field in recent decades. The rise of computational modeling has allowed researchers to test claims about learnability with explicit simulations, though the relevance of these models to real children remains contested. The growth of research on bilingualism has complicated the traditional distinction between first and second language acquisition, since many children acquire two languages from birth and show patterns that differ from both monolingual acquisition and sequential second language acquisition. The increasing availability of large corpora of child speech, such as the CHILDES database, has made it possible to test claims about input frequency and distribution with unprecedented precision. And the study of sign language acquisition has provided a crucial test case: deaf children acquiring sign language from their deaf parents show the same milestones and the same kinds of errors as hearing children acquiring spoken language, which strongly suggests that the underlying mechanism is not tied to speech but to language itself.
The most durable contribution of the field may be its demonstration that language acquisition is neither simple imitation nor simple instruction. Children acquire language through a complex interaction between their own cognitive capacities, the structure of the input they receive, and the social contexts in which they use language. The relative weight of these factors remains the central unresolved question, and it is likely to remain so for the foreseeable future. What has changed is the precision with which the question can be asked: researchers now have detailed models of the input, detailed descriptions of the developmental sequence, and increasingly powerful tools for testing hypotheses about the mechanisms that connect the two.