Loading field map
The study of language acquisition is driven by a single, persistent question: how do human beings come to know and use language? The answers have divided the field for decades, with some researchers arguing that language depends on specialized, innate knowledge, others that it emerges from general learning mechanisms applied to rich input, and still others that social interaction is the essential engine. This tension—between nature, environment, and social experience—has shaped every major framework in the subfield, whether the focus is first language (L1) acquisition in children or second language (L2) learning in older learners.
The first systematic framework for language acquisition in the mid-20th century was Behaviorism, dominant from roughly 1957 to 1965. Drawing on B. F. Skinner's operant conditioning, behaviorists argued that children acquire language through imitation, reinforcement, and habit formation. A child hears a word, repeats it, and is rewarded; incorrect forms are extinguished through lack of reinforcement. Language was treated as a set of verbal behaviors, no different in principle from any other learned behavior. This view was decisively challenged in 1959 by Noam Chomsky, who argued that behaviorism could not explain the creativity of language—children produce and understand sentences they have never heard—nor the speed and uniformity of acquisition across diverse environments. Chomsky's critique did not merely point out gaps; it reframed the entire object of study. Language, he insisted, was not a set of habits but a rule-governed system, and the child's task was not to imitate but to infer those rules from limited evidence.
Chomsky's alternative, Nativism (1959–present), proposed that humans are born with a Language Acquisition Device (LAD)—a specialized, innate module dedicated to grammar. The core argument was the poverty of the stimulus: the linguistic input available to children is fragmentary, noisy, and underdetermines the complex grammatical knowledge they eventually attain. Therefore, much of that knowledge must be built in. Nativism dominated L1 acquisition research for decades, especially through the theory of Universal Grammar (UG), which claimed that all human languages share a deep structural blueprint. Over time, however, the framework narrowed. Early strong claims that UG specified the full range of possible grammars gave way to more constrained proposals about core syntactic principles, with peripheral and lexical knowledge left to experience. This narrowing opened space for alternative accounts that did not reject nativism outright but argued that domain-general learning mechanisms could handle much of what UG was invoked to explain. In SLA research, nativism was applied more cautiously: some researchers argued that adult L2 learners no longer had full access to UG, while others maintained that UG remained available but was filtered through the L1.
In second language acquisition, the first framework to offer a systematic picture was Stephen Krashen's Monitor Model (1977–1990). It consisted of five hypotheses: the Acquisition-Learning distinction (unconscious acquisition vs. conscious learning), the Monitor (conscious editing of output), the Natural Order (grammatical structures are acquired in a predictable sequence), the Input Hypothesis (acquisition occurs when learners understand input at level i+1, just beyond their current competence), and the Affective Filter (anxiety blocks input from reaching the acquisition device). The Monitor Model was enormously influential in pedagogy, but it drew sharp criticism. Researchers pointed out that the acquisition-learning distinction was difficult to test empirically, that the i+1 concept was vague, and that the model placed almost all explanatory weight on comprehensible input while ignoring the learner's own production. These limitations did not kill the framework; rather, they prompted two complementary correctives that refocused attention on interaction and output.
The Interaction Hypothesis (1980–present), associated primarily with Michael Long, argued that input alone is insufficient. What matters is negotiated interaction: when a learner and an interlocutor work together to repair a communication breakdown—through clarification requests, confirmation checks, or recasts—the learner's attention is drawn to form-meaning connections in ways that raw input cannot achieve. This framework did not replace the Monitor Model so much as absorb its insight about input while adding a social-interactive mechanism. The Output Hypothesis (1985–present), proposed by Merrill Swain, went further. Swain argued that producing language—being pushed to say something accurately—forces learners to process syntax more deeply than comprehension alone requires. Output serves noticing, hypothesis testing, and metalinguistic reflection. Together, the Interaction and Output Hypotheses transformed SLA research by shifting the unit of analysis from the individual learner's mind to the conversational dyad. Both remain active today, often combined in studies of task-based interaction and corrective feedback.
A more radical departure from cognitivist frameworks came with Sociocultural Theory (1980–present), grounded in the work of Lev Vygotsky. Where the Interaction Hypothesis treated conversation as a source of input and feedback, Sociocultural Theory saw it as the very medium of cognitive development. Language is acquired through mediated activity: a learner first performs a task with the help of a more capable peer or teacher (the zone of proximal development, or ZPD), and over time internalizes that collaborative performance as independent ability. This framework does not deny the role of input or output, but it reinterprets them as tools for co-constructing knowledge rather than as data for an internal processor. Sociocultural Theory coexists with interactionist and output-based approaches in SLA, but its assumptions about the social origin of higher cognition set it apart from frameworks that treat the individual mind as the primary locus of learning.
While SLA researchers were developing interactional and sociocultural accounts, a different challenge to Nativism emerged from cognitive science. Connectionism (1986–present), inspired by parallel distributed processing models, proposed that language acquisition could be explained by domain-general associative learning mechanisms operating over large networks of simple units. The landmark demonstration was Rumelhart and McClelland's 1986 model of past-tense acquisition, which learned to produce both regular and irregular forms without any explicit rules—simply by adjusting connection weights based on input frequency and pattern similarity. Connectionism directly challenged the nativist claim that rule-governed behavior requires innate grammatical knowledge. If a network could learn to produce "goed" and then "went" without a rule, perhaps children could too. Critics argued that connectionist models were too sensitive to training data and did not scale to full grammatical complexity, but the framework forced nativists to sharpen their arguments about what exactly is innate.
Emergentism (1990–present) and Usage-Based Linguistics (1990–present) grew out of the connectionist challenge but broadened it into a comprehensive alternative to nativism. Emergentism holds that linguistic structure is not pre-specified but emerges from the interaction of simpler, domain-general processes—pattern detection, categorization, statistical learning—applied to massive amounts of input. Grammar is not a set of rules in the mind but a dynamic, probabilistic system that crystallizes from usage. Usage-Based Linguistics, closely related but with roots in functional and cognitive linguistics, emphasizes that knowledge of language is knowledge of specific constructions (form-meaning pairings) that are learned from exposure and stored as exemplars. Frequency, entrenchment, and preemption are the key mechanisms. Together, these frameworks have become the leading alternative to Nativism in L1 acquisition research. They do not deny that humans have powerful learning abilities, but they argue that those abilities are domain-general rather than language-specific. The two frameworks overlap heavily, but Emergentism tends to foreground the self-organizing, dynamic character of the system, while Usage-Based Linguistics focuses on the concrete, item-based nature of linguistic knowledge.
The most recent major framework, Complexity Theory (2000–present), extends emergentist thinking by importing concepts from dynamic systems theory. Language acquisition, in this view, is a complex, non-linear system in which multiple variables—input frequency, learner motivation, social context, cognitive resources—interact in ways that cannot be predicted by simple cause-effect models. Development is characterized by variability, phase transitions, and sensitive dependence on initial conditions. A learner's interlanguage is not a steady progression toward a target but a dynamic system that can show sudden restructuring, backsliding, and U-shaped behavior. Complexity Theory does not replace Emergentism or Usage-Based Linguistics; rather, it provides a meta-theoretical vocabulary for describing the behavior of emergent systems. Its main contribution has been methodological: it encourages longitudinal, dense data collection and analysis of variability rather than group averages.
Today, no single framework commands the field. In L1 acquisition, the main division is between Nativism and the emergentist/usage-based camp. Nativists continue to argue that certain structural properties of language—such as hierarchical phrase structure and displacement—cannot be learned from input alone, while emergentists counter that statistical learning, combined with social-cognitive skills, is sufficient. This is a living disagreement, with both sides producing empirical studies and computational models. In SLA, the landscape is even more pluralistic. The Interaction Hypothesis, Output Hypothesis, and Sociocultural Theory all remain active, often combined in studies of classroom interaction and task-based learning. Connectionism has been largely absorbed into Emergentism, which now provides the broader theoretical umbrella for domain-general accounts. Complexity Theory adds a dynamic-systems perspective that is compatible with both emergentist and sociocultural approaches. What the leading frameworks agree on is that acquisition is a gradual, input-driven process shaped by frequency, attention, and social interaction—a position that would have been unthinkable during the heyday of Behaviorism or early Nativism. Where they disagree is on whether the learning mechanisms are language-specific or domain-general, and on whether social interaction is a source of data or the very process of development itself.