Variability in SLA

Definition:

Variability in SLA refers to the phenomenon that L2 learners produce different forms for the same linguistic target across different contexts, tasks, interlocutors, and time points—sometimes using target-like forms, sometimes non-target-like forms, and sometimes forms transitional between both. Variability is simultaneously a challenge to acquisition models (which forms does the learner “know”?) and a window into the dynamic architecture of interlanguage. Ellis (1987, 1994) distinguishes systematic variability (predictable variation conditioned by linguistic, psycholinguistic, or social factors) from unsystematic (free) variability (apparently random fluctuation), and argues that systematic variability is itself a linguistic competence—not simply noise around a stable knowledge state.

In-Depth Explanation

Sources of systematic variability:

1. Linguistic context:

Learner forms vary systematically with phonological, morphological, and syntactic environment. Labov’s (1966) classic finding in L1 sociolinguistics—that deletion of word-final /t,d/ is conditioned by the following phonological segment—has parallels in L2: learners may correctly use past tense -ed before pauses but delete it before consonant-initial words following English phonotactic patterns.

2. Task type:

Ellis (1987) proposed that learners perform more target-like grammar in careful, monitored tasks (written composition, controlled grammar tests) than in spontaneous conversation. This is predicted by Krashen’s Monitor Hypothesis: explicit knowledge (the Monitor) is most accessible in low-time-pressure, form-focused conditions. The learner “knows” the rule in one sense (can access it with monitoring) but does not automatically apply it in spontaneous conditions.

3. Planning time:

Skehan and Ellis’s research shows that more planning time produces more accurate, complex output. Accuracy rates for the same grammatical structures vary systematically with the amount of pre-task planning available.

4. Interlocutor:

Learners produce more formal or more target-like language with perceived authority figures (teachers, native speakers in formal contexts) than with peers. This mirrors L1 sociolinguistic style-shifting (Labov’s attention-to-speech model), applied to L2 learners.

5. Topic familiarity:

When discussing familiar topics (high schema activation), learners can allocate more processing resources to form; output is more target-like. Unfamiliar topics deplete available processing resources, increasing variability and error.

Systematic versus free variability:

The distinction is theoretically important:

Systematic variability: Can be accounted for by identifiable conditioning factors — is part of the learner’s interlanguage grammar.
Free (unsystematic) variability: No identifiable conditioning factor — often signals a developmental transition when two competing forms coexist before one wins out.

Tarone (1982, 1988) argued that free variability is real and reflects an inherent feature of the learner’s variable competence — not measurement error. This challenged Chomskyan models that treated underlying competence as stable and unitary.

Dynamic Systems Theory and variability:

Dynamic Systems Theory (DST; Larsen-Freeman & Cameron, 2008) treats variability not as noise to be explained away but as a fundamental property of complex adaptive systems. In DST:

Variability is evidence of the system’s dynamic development — the amount of variability in a learner’s linguistic behavior may itself be a developmental indicator (high variability signals a system in transition; low variability signals stabilization or fossilization).
Individual learners have unique attractor states — preferred forms they gravitate toward — and variability reflects the non-linear competition between attractors.
Context is a genuine component of the system, not a confound to be controlled.

Variability in Japanese:

Japanese L2 learner variability has several characteristic patterns:

Desu/masu vs. plain form: Learners may variably use desu/masu and plain form across contexts inappropriately — overgeneralizing teineigo into casual conversation, or underusing it in formal contexts. This register variability is both systematic (conditioned by social context) and sometimes unsystematic during developmental transitions.
Kanji reading: Learners may correctly read a kanji in one context and fail to recognize it in another — variability in lexical access for partially-acquired kanji forms.
Morphological endings: -te form vs. plain form dependency marking may vary systematically with attention to form and task complexity.
SOV compliance: Learners may variably produce Japanese-order and English-order sentences under different processing load conditions.

History

1966: Labov’s sociolinguistic variable competence in L1.
1978: Corder’s early work on systematic interlanguage; variability partially acknowledged.
1982: Tarone proposes variable competence model for L2; free variability acknowledged.
1986: Skehan and Ellis task-type variability studies.
1987: Ellis’s framework of systematic vs. free variability in Interlanguage Variability.
1993: Bayley & Preston’s Second Language Acquisition and Linguistic Variation volume.
2008: Larsen-Freeman & Cameron integrate variability into Dynamic Systems Theory of SLA.

Common Misconceptions

“Variability means the learner doesn’t know the rule.” Variability may reflect conditioned knowledge — the learner knows the rule in some contexts and not in others. The pattern of variability reveals the conditions under which knowledge is accessible.

“L2 variability is just random errors.” Systematic variability is patterned and conditioned; analysis reveals the factors governing it. Even ‘free’ variability may reflect developmental dynamics rather than pure noise.

“When a learner gets something right in a test, they know it.” Controlled test conditions favor Monitor-mediated performance; the same learner may not spontaneously use that form in conversation. Test performance is a variability-context measure, not a direct read on underlying competence.

Criticisms

The distinction between systematic and free variability is difficult to make empirically; what appears free may be systematically conditioned by unmeasured factors.
DST’s reframing of variability as fundamental rather than noise is theoretically productive but has generated little clear predictive or pedagogical application.
The variability literature has been largely descriptive; causal accounts of when and why learners shift forms remain incomplete.

Social Media Sentiment

Language learners regularly observe variability in their own performance: “I could produce that Japanese grammatical pattern perfectly yesterday but today I can’t.” This variability is real and reflects task, cognitive load, and attention factors. L2 learner communities sometimes misinterpret it as “not really knowing” — but the research suggests correct performance under some conditions constitutes real knowledge that can be stabilized through practice toward automatization.

Last updated: 2026-04

Practical Application

Diagnose your variability pattern: Do you use a Japanese form correctly in writing but not in speaking? In slow, careful speech but not in fast conversation? This pattern reveals whether your knowledge is explicit and monitor-dependent or not yet automatized.
Targeted automatization practice: Forms you use correctly only in monitored conditions need fluency practice — timed speaking tasks, Shadowing under time pressure, or high-speed oral reading that forces automaticity without Monitor access.
Track variability in output: Review your Japanese writing and speaking output across multiple tasks and times. Forms with high variability are actively developing; forms with consistent errors may be fossilizing.
Variable input contexts: Produce target forms in multiple task types and contexts — written, oral, fast, slow, formal, informal. Broadening the contexts in which correct production occurs is how variable competence becomes stable competence.

Related Terms

Research

Tarone, E. (1982). Systematicity and attention in interlanguage. Language Learning, 32(1), 69–84. [Summary: Proposes variable competence model; argues interlanguage has a stylistic continuum with attention to form as the varying dimension; challenges uniform underlying competence models; foundational variability paper.]

Ellis, R. (1987). Interlanguage variability in narrative discourse: Style shifting in the use of the past tense. Studies in Second Language Acquisition, 9(01), 1–20. [Summary: Documents systematic variability in L2 English past tense use across narrative and conversational tasks; links task-type variability to Krashen’s Monitor; key empirical variability study.]

Labov, W. (1966). The Social Stratification of English in New York City. Center for Applied Linguistics. [Summary: Foundational sociolinguistic variability study; establishes that L1 variation is systematic and socially conditioned; theoretical model applied to L2 variability research by Tarone and others.]

Larsen-Freeman, D., & Cameron, L. (2008). Complex Systems and Applied Linguistics. Oxford University Press. [Summary: Applies Dynamic Systems Theory to SLA; reframes variability as fundamental to complex adaptive systems; variability patterns as developmental indicators rather than noise; influential reframing of the variability construct.]

Skehan, P. (1998). A Cognitive Approach to Language Learning. Oxford University Press. [Summary: Cognitive framework for task-based variability; accuracy-complexity-fluency trade-offs under different task conditions; provides psycholinguistic account of why variability occurs across task types.]