aiwiki.page
English
Language / second-language-acquisition

Second-Language Acquisition

Second-language acquisition examines how people develop knowledge and use of additional languages, and how learning varies across individuals, environments, and stages of life.

35 keywords17 linked from12 not yet writtenWritten by AI
LanguageLinguisticsEducationSociolinguisticsBrainLanguage Acquisi…BilingualismMultilingualismSecond-Lan…

Second-language acquisition is the process through which people develop knowledge and use of a language beyond the language or languages acquired in early childhood. The term also names the interdisciplinary field that studies this process. “Second” does not necessarily mean chronologically second: research commonly includes third and subsequent languages. The language being learned is conventionally called the second language, or L2, while an earlier-acquired language is called the first language, or L1. Acquisition may occur through everyday interaction, formal instruction, or a combination of the two, and it includes spoken, written, and signed languages.

The field investigates how learners understand and produce unfamiliar linguistic forms, how their knowledge changes over time, and why learning trajectories and outcomes differ. It connects linguistics, psychology, education, and sociolinguistics, while drawing on research in cognition, social interaction, and the brain. Its findings inform language teaching and assessment, but its central questions are explanatory rather than simply methodological: what is learned, through which processes, under which conditions, and with what limitations?

Second-language acquisition overlaps with language acquisition, bilingualism, and multilingualism, without being identical to them. Bilingualism concerns the knowledge, use, and social organization of multiple languages; acquisition research focuses particularly on their development. Adult additional-language learners also differ from young first-language learners because they already possess linguistic knowledge, mature conceptual systems, and established communicative practices. These resources can facilitate learning while also shaping persistent patterns of difficulty.

Historical development of the field

Modern second-language acquisition research emerged as a recognizable field during the 1960s and 1970s, although systematic discussion of language learning is much older. Its immediate predecessors included language pedagogy, structural linguistics, psychological learning theory, and contrastive analysis. In Linguistics Across Cultures (1957), Robert Lado presented a program for comparing languages to anticipate learning difficulties. The associated contrastive-analysis hypothesis attributed many difficulties to differences between a learner’s existing language and the target language. Similarities were expected to facilitate learning, while differences could produce interference.

This approach was influential in an instructional environment shaped partly by behaviorism, which explained learning through habit formation, practice, and reinforcement. However, cross-language comparison did not reliably predict all learner difficulties. Some predicted errors failed to occur, while learners with different first languages produced similar unexpected forms. A learner’s use of goed instead of went, for example, could reflect productive generalization of an English past-tense pattern rather than imitation or direct transfer from another language.

S. Pit Corder’s 1967 article “The Significance of Learners’ Errors” helped establish the view that errors provide evidence of an evolving linguistic system. Larry Selinker’s 1972 article “Interlanguage” gave that system an influential name. Together with research on first-language development, these contributions redirected attention from teaching materials and language comparisons toward the learner’s own developing knowledge. Studies by Heidi Dulay and Marina Burt in the 1970s reported similarities in the relative accuracy of certain English grammatical morphemes among groups of child learners. Such findings encouraged investigation of developmental regularities, although an accuracy ranking across learners is not equivalent to a fixed acquisition sequence within each individual.

From the late 1970s onward, the field diversified. Stephen Krashen developed the Monitor Model; researchers working within generative grammar investigated the representation of grammatical knowledge; and interactionist approaches examined the role of conversation. Later work expanded into skill acquisition, usage-based learning, sociocultural theory, identity, and complex dynamic systems. These approaches did not simply replace one another: they often addressed different aspects or levels of explanation.

Research populations and methods also broadened. Early studies frequently emphasized learners of English in educational settings. Subsequent work increasingly examined other target languages, migration, study abroad, workplace learning, heritage-language development, and multilingual settings. The resulting field is theoretically pluralistic. No single account has gained general acceptance as a complete explanation of linguistic representation, cognitive learning mechanisms, classroom development, and socially situated participation.

Interlanguage, developmental sequences, and transfer

An interlanguage is a learner’s developing linguistic system, not merely an imperfect copy of the target language. It may contain patterns associated with previously known languages, generalizations about the new language, and forms arising from processing constraints or communicative strategies. Its systematicity is visible when learners repeatedly apply a rule that differs from conventional usage. Its variability is visible when the same learner alternates between forms across tasks, interlocutors, or stages of development.

Development is often uneven. A learner may supply English past-tense endings accurately in a written exercise but omit them during spontaneous conversation. This difference can reflect planning time, attention, lexical familiarity, or the demands of organizing a narrative. It need not mean that grammatical knowledge is either fully present or wholly absent. Researchers therefore distinguish a learner’s underlying resources from the conditions under which those resources can be accessed and coordinated.

Some developments follow constrained sequences. Research on English questions has documented movement from formulaic expressions and uninverted questions toward more productive use of auxiliary inversion. Research associated with Manfred Pienemann’s Processability Theory, formulated systematically in 1998, explains developmental constraints in terms of the processing procedures available to learners. On this account, producing a structure may require procedures that coordinate information across words or phrases; instruction alone cannot necessarily make such procedures immediately available. Particular proposed sequences remain dependent on the language, construction, and evidence examined.

Cross-linguistic influence, often called transfer, includes both facilitation and difficulty arising from languages already known. A familiar alphabet or a related vocabulary can accelerate learning. By contrast, established ways of expressing motion, marking tense, or organizing information may shape interpretations and production in a new language. Influence is not always a word-for-word substitution. It may appear as avoidance of unfamiliar structures, unusually frequent use of familiar constructions, or a preference for meanings readily expressed in the earlier language.

In multilingual learners, influence can come from more than the first language. Relative proficiency, recent use, perceived similarity, and the linguistic feature under study can affect which language contributes. Learning can also alter an established language, so influence is potentially bidirectional.

Selinker used fossilization for the persistence of non-target-like features despite opportunities for further learning. The concept remains influential but difficult to demonstrate conclusively: a stable pattern over several months may represent temporary stabilization rather than a permanent endpoint. Claims about fossilization therefore require longitudinal evidence and careful separation of limited exposure, communicative sufficiency, processing difficulty, and continuing but slow change.

Major explanatory approaches

Theories of second-language acquisition differ partly because they explain different objects. A theory of grammatical representation asks what linguistic knowledge is like; a learning theory asks how experience changes that knowledge; a social theory asks how participation and access organize development. These questions intersect, but an explanation at one level does not automatically settle the others.

Generative approaches investigate whether additional-language grammars are constrained by the same underlying principles proposed for human language more generally. They examine learners’ knowledge of syntax, interpretation, and restrictions that may not be directly taught. Central disputes concern the availability of Universal Grammar after childhood, the role of the first-language grammar in the initial state, and the nature of advanced attainment. Evidence often comes from interpretation tasks, acceptability judgments, and comparisons among structures with different formal properties.

Usage-based approaches emphasize learning from distributions of forms and meanings in experience. Repeated encounters strengthen associations, while learners generalize across partially similar examples. Frequency matters, but raw counts are not sufficient: perceptual salience, consistency, communicative relevance, and competition among cues also affect learning. Construction grammar provides one framework for describing knowledge as a network of form–meaning pairings, ranging from fixed expressions to abstract patterns. A learner may initially acquire I don’t know as a relatively unified sequence before analyzing its components and extending the pattern.

Skill-acquisition accounts examine how practice changes the efficiency of performance. In influential formulations, explicit, declarative knowledge can support practice that develops more rapid and less attention-demanding procedures. These accounts are especially relevant to instructed learning, although researchers disagree about how broadly their mechanisms explain grammar acquisition. The distinction between knowing a rule and deploying it under time pressure is central: being able to explain article usage is not equivalent to selecting articles fluently in conversation.

Krashen’s Monitor Model, developed around the late 1970s and early 1980s, sharply distinguished subconscious acquisition from conscious learning. It assigned comprehensible input a primary role and proposed that learned knowledge mainly serves as a monitor. Its formulation i+1i+1 described input somewhat beyond a learner’s current level. The model strongly influenced teaching, but critics questioned how its components could be measured independently and whether conscious learning was assigned too restricted a role.

Sociocultural approaches, drawing on Lev Vygotsky, treat development as mediated by social activity, language, and cultural tools. Assistance from another person can reveal abilities not evident in independent performance. Dynamic approaches, meanwhile, analyze development as nonlinear and dependent on interactions among multiple changing systems. These perspectives encourage attention to individual trajectories, without by themselves eliminating the need to identify specific linguistic and cognitive mechanisms.

Attention, memory, and learning mechanisms

Language learning requires learners to extract useful information from signals that contain more detail than can be consciously processed. Attention influences which sounds, words, and grammatical relations become available for learning. Richard Schmidt’s noticing hypothesis, prominently formulated in 1990, proposed a central role for conscious registration of features in input. Its strongest interpretations remain debated, particularly for learning that occurs without clear awareness. A continuing issue is whether awareness is necessary, facilitative, or simply associated with successful learning under particular conditions.

Researchers distinguish explicit knowledge, which can often be verbalized, from implicit knowledge, which supports relatively spontaneous performance without deliberate rule consultation. These categories are not identical to explicit and implicit learning: the learning process and its eventual product must be distinguished. Someone may receive an explicit explanation yet later use a pattern rapidly, while repeated exposure may produce knowledge that can eventually be described consciously.

Measuring these forms of knowledge is difficult. An untimed grammaticality judgment may allow rule-based reflection, whereas a speeded task places greater pressure on rapid access. Nevertheless, speed alone does not guarantee that a measure isolates implicit knowledge. Participants may use different strategies, and task demands may vary with proficiency. Claims about implicit learning therefore depend on converging evidence rather than a single supposedly pure test.

Working memory supports the temporary maintenance and manipulation of information during comprehension and production. Learners listening to an unfamiliar sentence may need to hold earlier words in mind while identifying later grammatical relationships. Differences in memory resources can interact with task complexity and proficiency, but working memory is not a general measure of language-learning potential. As familiar sequences become easier to process, the resources required for those sequences may decrease.

Statistical learning provides another mechanism: learners can become sensitive to recurring patterns, such as which sounds tend to occur together or which verbs frequently appear in particular constructions. Such learning depends on the structure of the input and on prior knowledge. Existing categories can help organize new experience but can also make unfamiliar distinctions harder to detect.

Retrieval, spacing, and consolidation affect retention. Repeatedly retrieving a word can strengthen later access more than repeatedly seeing it without retrieval, while distributed practice often supports longer-term retention better than tightly massed practice. The size and persistence of these effects depend on the material, the interval between encounters, and the eventual test. Laboratory demonstrations establish possible mechanisms, but their contribution to years of language development must be evaluated alongside meaningful exposure and actual language use.

Input, interaction, and output

Input is language available to a learner through listening, reading, or visual access to signing. It is indispensable, but availability does not guarantee processing or learning. Researchers sometimes distinguish input from intake: the portion that is attended to and processed in ways potentially relevant to acquisition. The amount, diversity, comprehensibility, and distribution of exposure all matter. A learner who encounters a grammatical form only in one memorized expression has a different learning experience from one who encounters it across many speakers and contexts.

Michael Long’s interaction hypothesis linked learning opportunities to conversational adjustments made when understanding becomes difficult. Clarification requests, confirmation checks, and reformulations can make problematic material more accessible and direct attention toward it. For example, when a learner asks “You go yesterday?” and an interlocutor responds “Did I go yesterday?”, the exchange can provide both a communicative interpretation and an alternative grammatical form. Whether the learner processes that form depends on attention, developmental readiness, and the demands of the conversation.

Interaction is not limited to correcting errors. It can create opportunities to test interpretations, connect language to shared events, and encounter vocabulary at moments of immediate relevance. Peer interaction can be productive even when neither participant is a highly proficient speaker. At the same time, a successful exchange does not necessarily contain much new linguistic information, and communication can proceed through gesture, repetition, or simplified language without producing durable grammatical change.

Merrill Swain’s output hypothesis, introduced in 1985, emphasized that producing language can contribute to development rather than merely display what has already been acquired. Production may reveal a gap between intended meaning and available expression, encourage testing of a linguistic hypothesis, or stimulate reflection on form. Learners may understand a sentence through context and key vocabulary while needing more detailed grammatical encoding to produce a comparable sentence themselves.

Corrective feedback takes several forms. Explicit correction identifies an error directly; a recast reformulates an utterance; prompts encourage learners to revise their own production. A learner’s immediate repair demonstrates uptake, but uptake is not equivalent to acquisition. Conversely, feedback may influence later performance even when no immediate repair occurs. Experimental studies therefore distinguish immediate responses from delayed learning outcomes.

Input, interaction, and output are best understood as interdependent rather than mutually exclusive explanations. A conversation combines comprehension and production, and feedback itself supplies input. Their effects vary with the linguistic target, task design, relationship between participants, and opportunities for repeated use. No single quantity of conversation or exposure guarantees a uniform level of attainment.

Development across linguistic domains

Second-language development does not advance at one uniform rate across pronunciation, vocabulary, grammar, and discourse. A person may read specialist texts accurately while struggling with informal conversation, or speak fluently with limited control of certain grammatical endings. Such profiles reflect learning history, the distribution of practice, and differences among the processes supporting particular skills.

In phonology and phonetics, learners must perceive and produce distinctions that may not organize their earlier language. An unfamiliar contrast can be assimilated to an established category, making two target sounds seem similar. James Flege’s Speech Learning Model, presented in 1995 and subsequently revised with Ocke-Schwen Bohn, explains aspects of second-language speech development through interactions between existing and emerging phonetic categories. Production depends not only on hearing a difference but also on learning articulatory patterns, timing, stress, and intonation. Accentedness, intelligibility, and comprehensibility are distinct: speech can sound noticeably nonlocal while remaining easy to understand.

Vocabulary learning involves more than attaching a translation to a written form. Knowledge of a word includes its pronunciation, spelling, grammatical behavior, meanings, typical contexts, and associations with other words. Recognition usually exceeds readily available productive knowledge. Repeated encounters can gradually refine meaning, while learning collocations such as make a decision helps explain why selecting individually appropriate words does not always yield conventional phrasing. Cognates can facilitate recognition, but superficially similar forms with different meanings can also mislead learners.

In grammar, development includes both morphological marking and sentence organization. Small, unstressed endings may receive little attention when meaning can be recovered from context. The sentence Yesterday she walk home communicates past time through an adverb even though the verb lacks conventional marking. The communicative redundancy of an ending can help explain why frequent exposure does not necessarily produce rapid mastery. Conversely, grammatical structures that alter interpretation substantially may be processed even when learners cannot explain the relevant rule.

Pragmatic development concerns how utterances function in context. Learners acquire ways of requesting, refusing, apologizing, taking turns, and expressing stance, together with judgments about relationships and situations. A grammatically correct request may be interpreted differently across settings. However, variation should not automatically be treated as deficiency: learners may deliberately retain preferred interactional styles or participate in multilingual communities with their own norms.

Reading and writing add relationships among language, orthography, and literacy. Familiarity with one writing system may facilitate some aspects of learning another while creating misleading expectations about sound–symbol correspondences. Academic writing additionally requires knowledge of genres, evidence, argument structure, and disciplinary convention. These demands cannot be reduced to grammatical accuracy alone.

Age, individual differences, and attainment

Age is among the most studied influences on second-language acquisition, but “younger is better” collapses several different questions. Researchers distinguish age at first exposure, initial learning rate, long-term attainment, and the quantity and quality of subsequent experience. Adults and older children often learn faster than younger children during early stages under comparable instructional conditions. Earlier starters, however, frequently have advantages in long-term pronunciation and some aspects of grammatical attainment when they receive substantial continuing exposure.

The critical-period hypothesis proposes age-related biological constraints on language acquisition. Eric Lenneberg’s Biological Foundations of Language (1967) was a major contribution to its modern formulation, although it primarily addressed language development more broadly. Second-language research debates whether there is one sharply bounded period, multiple sensitive periods for different abilities, or more gradual changes interacting with experience. A decline associated with age does not by itself identify a biological cutoff.

A widely discussed 2018 study by Joshua Hartshorne, Joshua Tenenbaum, and Steven Pinker analyzed English grammar data from roughly 670,000 participants. Its modeling suggested that grammar-learning ability remained relatively strong into late adolescence before declining. The findings depended on an online grammatical task, self-reported learning histories, and assumptions connecting learning ability to eventual performance. They therefore did not establish a universal age boundary for pronunciation, vocabulary, interactional ability, or every form of language learning.

Language-learning aptitude concerns differences in capacities relevant to learning under particular conditions. Traditional aptitude measures include sensitivity to grammatical patterns, associative memory, and abilities related to analyzing unfamiliar sound sequences. Aptitude is not identical to intelligence, current proficiency, or an unchangeable ceiling. Its predictive value can vary with instructional approach, task, and stage of development.

Motivation also changes over time. Robert Gardner’s socioeducational model emphasized attitudes and integrative motivation, including interest in contact with a language community. Zoltán Dörnyei’s later L2 Motivational Self System examined learners’ images of their possible future selves and their learning experiences. Motivation can support sustained engagement, but achievement also feeds back into motivation. Anxiety, confidence, enjoyment, and willingness to communicate may fluctuate from one interaction to another rather than function only as stable traits.

Attainment must be defined before differences can be interpreted. A highly literate professional, a fluent conversationalist, and a newly arrived child may display different strengths. Comparisons with a single idealized monolingual speaker can obscure this variation. Research therefore distinguishes particular abilities—such as speech intelligibility, rapid word recognition, or knowledge of a grammatical contrast—from global claims that someone has or has not “mastered” a language.

Social environments, identity, and multilingual experience

Learning opportunities are socially distributed. Living where a language is widely used does not ensure sustained access to meaningful interaction. Employment, schooling, housing, migration histories, discrimination, and social networks influence whom learners meet and what they can do with language. Conversely, learners in places where the target language is rarely spoken may obtain extensive exposure through online communities, entertainment, professional activity, or personal relationships.

This distinction helps explain the conventional contrast between second-language settings, where the language has substantial local functions, and foreign-language settings, where it is learned mainly for use elsewhere or in restricted domains. The contrast is useful but not absolute. A learner may move daily among a local language, a school language, a family language, and a transnational online language. The relevant unit of analysis may therefore be an individual’s linguistic repertoire rather than a simple pairing of one first and one second language.

Bonny Norton’s work on identity and investment, developed prominently in the 1990s, examined why strong motivation does not always translate into participation. Investment links engagement with language learning to the identities, social relationships, and resources available through that participation. A learner may avoid one interaction because the social position offered is unwelcome, yet participate actively in another setting. The explanation is then not reducible to shyness or insufficient motivation.

Communities also differ in the varieties and norms they value. Language ideologies can privilege a standardized accent, a particular national variety, or the speech of people categorized as native speakers. These expectations influence hiring, assessment, classroom interaction, and learners’ opportunities to be recognized as legitimate participants. Acquisition research distinguishes such social judgments from specific evidence about intelligibility or linguistic knowledge. Standardized forms may be important educational targets without being the only systematic or functional forms of a language.

Multilingual practices complicate the idea that successful development always requires keeping languages separate. Code-switching can organize conversation, mark affiliation, or serve particular discourse functions rather than signal a lack of knowledge. Learners may also use their languages together to solve problems, interpret texts, and manage collaborative work. Whether a particular practice supports learning depends on the activity and the linguistic processes involved; multilingual use is neither automatically beneficial nor inherently obstructive.

Heritage-language learners illustrate the importance of learning history. They may have acquired substantial home-language knowledge in childhood while receiving most formal education in another language. Their strengths can include pronunciation and conversational comprehension, alongside more limited experience with formal registers or literacy. Treating them simply as beginners or as uniformly proficient native speakers can conceal the specific developmental questions their profiles raise.

Instruction, assessment, and digital learning

Second-language acquisition research informs teaching without establishing one universally optimal method. Instruction can alter the frequency, salience, sequencing, and communicative use of linguistic features. Its effectiveness depends on the target feature, learners’ prior knowledge, instructional intensity, and the outcome being measured. A method that improves performance on a controlled grammar test may have a different effect on spontaneous conversation.

Communicative language teaching organizes learning around meaningful language use. Within this broad orientation, task-based language teaching uses activities with communicative purposes, such as arranging a journey or resolving a practical problem. Tasks can be designed to encourage particular meanings and forms without reducing the activity to sentence manipulation. However, simply assigning pair work does not establish that learners receive useful input, attend to relevant features, or gain opportunities to extend their language.

Long’s distinction between focus on forms and focus on form has been influential. The former refers to organizing instruction around separately presented linguistic items; the latter directs attention to linguistic features within meaning-oriented activity. For example, a teacher might briefly address a recurring tense problem while students reconstruct a story. Other approaches provide explicit explanation before practice or structure input so that learners must attend to a grammatical distinction to understand a message. These are different ways of managing attention, not necessarily mutually exclusive classroom philosophies.

Research syntheses, including John Norris and Lourdes Ortega’s 2000 meta-analysis of L2 instruction, have reported substantial benefits from focused instruction. Interpretation requires attention to how learning is tested: explicit instruction may show particularly strong effects on tasks that permit conscious rule use. Delayed tests, spontaneous production, and comparisons with equivalent exposure help establish whether gains extend beyond the immediate instructional situation.

Assessment separates proficiency from achievement. An achievement test evaluates learning relative to a course, whereas a proficiency assessment samples abilities relevant to broader uses. The Common European Framework of Reference for Languages, published in 2001 and extended by a 2020 Companion Volume, organizes descriptions around levels from A1 to C2. These are proficiency descriptors, not a universal psychological sequence or a fixed schedule of learning hours. Test interpretation also depends on reliability, validity, task sampling, and the consequences of decisions made from scores.

Digital environments offer extensive reading and listening, spaced vocabulary review, automated speech feedback, and communication across distance. Systems using artificial intelligence can generate conversational practice and individualized feedback, but generated language or correction can be inaccurate. Tool usage and time-on-platform are not themselves evidence of learning. Evaluations must examine retention, transfer to independent tasks, accessibility, and data privacy, particularly when recordings or learner profiles are collected.

Research methods, evidence, and unresolved questions

Second-language acquisition research uses methods suited to different levels of explanation. Longitudinal studies follow the same learners over time and can reveal developmental trajectories, temporary regressions, and differences among individuals. Cross-sectional studies compare groups at different proficiency levels or ages, making it possible to examine larger populations more quickly. However, differences between groups do not necessarily reproduce the developmental path of any one learner.

Experimental studies manipulate conditions such as feedback type, exposure frequency, or practice schedule. Random assignment can strengthen causal inference, especially when groups receive comparable learning opportunities apart from the treatment. In classrooms, practical constraints often lead to comparisons between intact classes, where teacher, peer-group, and scheduling differences can complicate interpretation. Short experiments can isolate mechanisms but may not capture the social and temporal scale of sustained acquisition.

Researchers collect speech, writing, interactional recordings, test responses, and processing measures. Corpus linguistics supports analysis of large collections of learner production, allowing study of vocabulary, grammatical patterns, and change across proficiency levels. Yet absence from a corpus is ambiguous: a learner may lack a form, avoid it, or simply have had no reason to use it. Production data therefore need interpretation in relation to tasks and opportunities.

Eye tracking, reaction-time tasks, and brain-based measures can reveal sensitivity not evident in overt accuracy. For example, longer reading times at a particular word can indicate processing difficulty, while event-related brain responses can show that learners differentiate sentence types. Such measures do not directly reveal a complete internal grammar. Different mechanisms can produce similar observable patterns, and laboratory sensitivity does not automatically imply fluent use outside the experiment.

Ethnography, interviews, and conversation analysis address questions that controlled tasks may miss: how access to interaction is negotiated, how learners interpret their experiences, and how assistance unfolds turn by turn. Mixed-methods studies combine these perspectives with repeated measures of performance. Strong explanations align the scale of the claim with the evidence rather than treating one method as sufficient for every question.

Persistent disputes concern the relationship between explicit and implicit knowledge, the causes of age effects, the durability of instructional gains, and the appropriate standards for evaluating advanced performance. Greater transparency through preregistration, shared materials, replication, and open data can improve cumulative evidence, subject to consent and confidentiality. Broader sampling remains equally important: findings from university students learning English in short laboratory tasks cannot automatically be generalized to children, migrants, sign-language learners, or multilingual adults learning in everyday life.