Skip to content
Back to search results

Elicited Imitation Task

Introduced by
Applied linguists and child-language researchers (overview by Thora Vinther)
Year
2002
Type
Proficiency/implicit-knowledge measure via sentence repetition
Family
Second-language acquisition / child language assessment
Font size

What does it do?

In the elicited imitation task, participants listen to spoken sentences — typically of increasing length and grammatical complexity — and repeat each one back.

The key insight is that when a sentence exceeds short-term verbatim memory, accurate reproduction is impossible by rote echoing; the listener must comprehend the sentence and reconstruct it through their own grammar. Reproduction accuracy therefore indexes implicit linguistic proficiency rather than parroting. Widely used in second-language acquisition as an efficient proficiency measure and in child-language research to gauge developing grammar, it has been validated as a window onto implicit knowledge.

The intuition behind the method

You cannot accurately repeat a sentence you do not understand. Hold a string of words in mind long enough and rote memory fails — to give it back, you have to grasp its structure and rebuild it from your own linguistic system. That is why imitation is not mere echoing: a learner whose grammar lacks a structure will quietly 'repair' it, dropping the third-person -s or regularizing an irregular verb, while a learner who has internalized it reproduces it faithfully. By stretching sentences past the limit of verbatim memory and watching what survives the reconstruction, the task turns repetition into a sensitive, fast probe of what grammar the speaker actually has.

History

Elicited imitation has roots in 1960s–1970s psycholinguistics and first-language acquisition research, where it was used to probe whether children could repeat structures beyond their productive grammar, and the finding that they systematically reconstructed sentences toward their own grammar established its diagnostic logic. It became a staple of language assessment, and Thora Vinther's 2002 overview synthesized its varied uses and forms in applied linguistics. Rosamond Erlam's 2006 validation study was pivotal for the modern second-language field: by designing the task with timing and meaning-focus controls and showing it loaded on an implicit-knowledge factor distinct from explicit metalinguistic measures, she helped re-establish elicited imitation as a credible measure of implicit second-language knowledge, spurring its wide adoption in proficiency testing.

When to use it

Use elicited imitation when you need an efficient, scalable measure of overall language proficiency or of a learner's implicit grammatical knowledge — for placement and proficiency testing in second-language research, for tracking development longitudinally, and for assessing emerging grammar in children including in clinical (e.g., language-disorder) screening. It is well suited when you want a global proficiency index quickly or a probe of specific structures under time pressure. It is less appropriate when the sentences are short enough to be echoed verbatim (defeating the reconstruction logic), when you need to separate explicit metalinguistic knowledge, or when articulation, hearing, or memory impairments confound reproduction.

Limitations and considerations

  • If sentences are too short, participants echo them verbatim, so the task measures memory span rather than grammar.
  • Reproduction is confounded by working-memory capacity, attention, articulation, and hearing, not grammar alone.
  • Scoring is labor-intensive and requires reliable rubrics and trained raters to remain consistent.
  • What exactly it measures (implicit knowledge versus processing or memory) remains debated and design-dependent.

Source conditions do not replace independent scientific review or validation by an analysis engine.

Strengths

  • Efficient and scalable — quick to administer and yielding a reliable global proficiency estimate.
  • Taps implicit, automatized knowledge by forcing reconstruction through the learner's own grammar rather than rote echo.
  • Flexible: can yield an overall proficiency score or target specific grammatical structures.
  • Validated against other proficiency and implicit-knowledge measures and able to discriminate developmental and proficiency levels.

Applications

  • Measuring overall second-language proficiency for placement and research stratification.
  • Assessing learners' implicit knowledge of specific target structures (e.g., agreement, tense, articles).
  • Tracking grammatical development longitudinally in learners and in children.
  • Screening for developmental language disorder, where sentence repetition is a sensitive clinical marker.
  • Comparing instructional treatments by detecting gains in reconstructable grammatical knowledge.

How it works

How it works

1

A set of stimulus sentences is constructed, graded so that length and syntactic complexity increase beyond the span of rote verbatim memory, ensuring that successful repetition requires comprehension and reconstruction rather than echoing. In targeted designs, sentences are built around specific grammatical structures (and sometimes grammatical/ungrammatical pairs) so that the task probes those features; in proficiency designs, the set samples a broad range of structures to yield an overall score, as Vinther's overview describes.

2

Each sentence is presented aurally, and the participant is prompted to repeat it. To prevent surface acoustic rehearsal and push processing through the grammar, designs often interpose a brief filled delay or a comprehension judgment between hearing and repeating — for example, having participants first indicate whether they agree with the sentence's meaning. Erlam's validation work uses such timing and content-focus manipulations to ensure the task taps reconstruction rather than parroting.

3

The participant's spoken reproduction is audio-recorded for each item. Designs specify whether participants may attempt only once, and how non-responses and partial attempts are treated. The recordings preserve exactly what the learner produced — including substitutions, omissions, additions, and self-corrections — which are the raw material for scoring and for diagnosing which structures the grammar did and did not support.

4

Reproductions are scored against the target with a defined rubric, ranging from a graded scale per sentence (e.g., 0–4 for exact, accurate-with-minor-change, partial, repeats little, no response) to targeted scoring of whether the critical structure was reproduced correctly, with attention to whether learners restructure ungrammatical items toward their own grammar. Scores are aggregated into a proficiency estimate or a structure-specific accuracy measure; Erlam shows that, designed this way, the task correlates with other implicit-knowledge measures and discriminates proficiency levels, supporting its interpretation as an index of implicit knowledge.

Common pitfalls

  • Using sentences within verbatim memory span, allowing rote repetition and undermining the reconstruction rationale.
  • Omitting a delay or comprehension step, so participants rehearse the acoustic surface instead of processing meaning.
  • Applying an ill-defined or inconsistently trained scoring rubric, producing low inter-rater reliability.
  • Interpreting low scores as grammatical deficits without ruling out memory, attention, or articulation limits.

Frequently asked questions

Why is repeating a sentence a measure of grammar rather than memory?

When a sentence is short, listeners can hold and echo its exact wording from a phonological short-term buffer, which is memory, not grammar. But when sentences are made longer and more complex than that buffer can hold, verbatim echoing fails — to give the sentence back, the listener must understand it and regenerate it from their linguistic system. At that point what they reproduce reflects the grammar they actually have: structures they have internalized survive, while structures they lack are altered toward their own grammar. Adding a delay or comprehension step strengthens this by blocking surface rehearsal.

How does elicited imitation differ from a grammaticality judgment task?

Both probe grammatical knowledge, but in opposite modes. A grammaticality judgment task is receptive and metalinguistic: participants reflect on a sentence and judge its well-formedness, which can recruit explicit, learned rules, especially when untimed. Elicited imitation is productive and processing-based: participants must comprehend and regenerate the sentence under time and memory pressure, which is thought to tap automatized, implicit knowledge with little room for conscious rule application. Researchers often use both, treating elicited imitation as an implicit-knowledge measure and untimed judgments as more explicit.

How are reproductions scored?

Two main approaches are used. Holistic graded scoring assigns each item a value on a short scale (for example 0 for no usable response up to a top score for an exact or meaning-preserving accurate repetition), and these are summed for a proficiency estimate. Targeted scoring instead checks whether the specific structure under study was reproduced correctly, and notes whether learners restructured an ungrammatical stimulus toward grammaticality. Both require a clear rubric, trained raters, and inter-rater reliability checks, since consistent scoring is essential to the task's validity.

Related methods

Sources

  1. Vinther, T. (2002). Elicited imitation: A brief overview. International Journal of Applied Linguistics, 12(1), 54–73.

    Open DOI reference
    Copied.
  2. Erlam, R. (2006). Elicited imitation as a measure of L2 implicit knowledge: An empirical validation study. Applied Linguistics, 27(3), 464–491.

    Open DOI reference
    Copied.