Linguistic Inference & Emergent Language
What lies beneath the surface of every lesson — the meaning students must derive, and the language that arises when no one planned for it.
The Implicit Dimension of Language
Both inference and emergent language live below the surface — in what isn’t said, and what wasn’t planned. Understanding both is essential to responsive teaching.
Linguistic Inference
The ability to derive meaning that is not explicitly stated — reading between the lines, guessing a word from context, understanding what a speaker really meant.
A learner reads “the film was no masterpiece” and must infer it was bad without the word “bad” appearing anywhere.
Emergent Language
Language that arises from communication tasks rather than lesson plans — vocabulary students reach for, errors that reveal gaps, structures students attempt unprompted.
A student mid-discussion says: “What’s the word for when you feel nervous… but for someone else?” — the word anxious on someone’s behalf has emerged.
The Connection
Both are about language that is implicit rather than explicit. Both require the teacher to respond in the moment. The bridge is noticing: Schmidt (1990) showed that learners acquire what they consciously notice — and inference is fundamentally a noticing skill.
Why This Matters
Most real communication relies heavily on inference. A speaker who says “It’s getting late” at a party means “I’d like to leave now” — but only listeners who can infer the implication will respond appropriately. For L2 learners, inference in a second language is slower, less automatic, and more culturally dependent.
Emergent language is one of the richest sources of learning in any classroom — yet it is routinely ignored or suppressed. When a student reaches for a word they don’t yet have, they are operating exactly in their Zone of Proximal Development.
Emergent language is not a disruption to the lesson plan — it is the lesson plan revealing what it should have been. Every unexpected word a student reaches for, every error that surfaces, every moment of pragmatic confusion is a data point telling the teacher exactly what the learner is ready to learn next.
What teachers typically do
- Pre-teach vocabulary before a text (removing the need for inference)
- Correct emergent errors mid-task (suppressing communication)
- Move on when a student reaches for an unknown word
- Treat unexpected language as off-topic
What skilled teachers do
- Teach strategies for dealing with the unknown, not removal of the unknown
- Note emergent language, return to it purposefully
- Use the board as a living record of the lesson’s real language
- Treat unexpected language as the most valuable content in the room
Linguistic Inference
Reading between the lines — the types of inference, why L2 inference is harder, and what it looks like in class.
Inference is not guessing. It is a systematic, evidence-based process of deriving meaning beyond what is literally stated. Skilled readers and listeners infer constantly. For second language learners, this automaticity is missing, and the effort required is significantly higher.
Lexical Inference
Guessing the meaning of an unknown word from surrounding context — syntax, semantics, and word parts.
“The ornithologist spent hours in the hide, binoculars raised.” → ornithologist = bird watcher.
Referential Inference
Understanding what a pronoun or ellipsis refers to — tracking entities across a text.
“Maria called Sophie but she didn’t answer.” → Who didn’t answer? Requires inference from context — it’s Sophie.
Bridging Inference
Connecting information across two sentences that are related but not explicitly linked.
“Sarah grabbed her coat. The taxi was waiting.” → Inference: Sarah is about to leave.
Elaborative Inference
Filling gaps using world knowledge and schema — adding information the text implies but never states.
“He struck a match and held it to the candle.” → Inference: the room was dark.
Pragmatic Inference
Understanding what a speaker meant vs. what they said — intent, implication, and social context.
“It’s a bit warm in here, isn’t it?” → Inference: please open a window.
The L2 Inference Challenge
Processing Speed
L2 readers are slower at the word recognition level, leaving fewer cognitive resources for inference.
Schema Mismatch
Elaborative inferences depend on shared cultural knowledge. A student from a different cultural context may not have the world-knowledge schema a native speaker takes for granted.
L1 Transfer
Students often import inferencing strategies from their first language — which may produce errors when L1 and L2 inferencing conventions diverge.
Classroom Scenarios
A student reads a film review: “The director’s debut was certainly ambitious. The performances were committed. The cinematography had its moments.”
The student’s response: “This review is positive.”
“The manager told the assistant that she needed to improve her punctuality.”
Student translation: “The manager is always late.”
A student encounters “the punishing schedule left them all haggard.” They don’t know “haggard.”
Student: “I think haggard means very tired or worn out — because the schedule is punishing.”
Research Background — Linguistic Inference in SLA
Nation (2001) established that readers need to know approximately 95–98% of words in a text for comfortable comprehension, and that above a certain vocabulary threshold, lexical inference becomes possible.
Nassaji (2003) found that L2 learners with higher proficiency used significantly more inference strategies and were more accurate.
Bartlett (1932) — the foundational schema theory study — showed that readers actively reconstruct meaning using prior knowledge. L2 readers from different cultural contexts may have schemas that actively mislead inference.
Cain & Oakhill (1999) showed that inference-making ability is distinct from decoding ability — inference is a teachable skill, not simply a byproduct of comprehension.
Pragmatic Inference
The gap between what is said and what is meant — Grice’s maxims, implicature, presupposition, and the cultural dimension.
Pragmatic inference is the most socially complex form of inference — and the one most likely to cause real communicative failure. A student who misunderstands a word can look it up. A student who misunderstands an implication may never know they missed it.
“What a man says and what he means can be enormously different things — and the gap between them is navigated entirely by inference.”After H. P. Grice, Logic and Conversation (1975)
Grice’s Cooperative Principle
H. P. Grice (1975) proposed that all communication rests on a tacit agreement to cooperate governed by four maxims. When speakers appear to violate a maxim, listeners infer an additional, implied meaning — a conversational implicature.
Quantity
Be as informative as required — no more, no less.
Violation: “Can you speak Spanish?” / “I lived in Mexico for three years.” → Implicature: yes, I can.
Quality
Say only what you believe to be true.
Violation (irony): “Oh brilliant — another Monday morning.” → The speaker means the opposite.
Relevance
Be relevant.
Violation: “How’s John at school?” / “He’s been on his Xbox a lot.” → Implicature: not well.
Manner
Be clear, brief, and orderly. Avoid ambiguity.
Violation: “The student was not entirely without merit.” → Elaborate phrasing signals hedging — implicature: mediocre.
Implicature in the Classroom
A student reads a reference: “Mr García was always punctual and his desk was very tidy.”
Student: “This is a good reference.”
In an office: “It’s a bit warm in here, isn’t it?”
Student responds: “Yes, it is.”
Presupposition
A presupposition is something a speaker takes for granted — assumed to be shared knowledge. Unlike implicature, presupposition persists even when the sentence is negated or questioned.
“Have you stopped lying to me?”
Presupposition: you were lying to me. Assumed whether the answer is yes or no.
“Even Maria got it right.”
Presupposition: Maria is not usually good at this.
“The president’s lies shocked everyone.”
Presupposition: the president lied. This cannot be questioned.
Japanese business communication often relies on heavy implicature — a vague “that might be difficult” conventionally means “no.” A Dutch interlocutor operating in a low-context tradition may interpret the same phrase as “maybe, let’s discuss.” Neither will know a miscommunication has occurred.
Research Background — Pragmatic Inference
Grice (1975) — the foundational statement of the Cooperative Principle and the four maxims. Grice distinguished between conventional implicature and conversational implicature.
Sperber & Wilson (1986) — Relevance Theory — all communication is governed by a single principle of relevance. This better explains why inference is automatic in L1 and costly in L2 (where processing costs are higher).
Brown & Levinson (1987) — Politeness Theory — many indirect speech acts are face-saving devices that L2 learners must learn to read as such.
Taguchi (2009) — even advanced L2 users process pragmatic implicature more slowly than L1 speakers, suggesting it requires explicit instruction and exposure.
Teaching Inference
Strategies, tasks, what to do, what not to do, and what to do when it all goes sideways.
Inference is a teachable skill. Research consistently shows that explicit instruction in inferencing strategies produces measurable gains in both reading comprehension and vocabulary acquisition.
Core Inferencing Strategies to Teach
Context before the word
Teach students to read the sentence containing the unknown word, then the sentence before and after it. What topic is being discussed? What semantic field are we in?
Grammar clues (syntax)
Is the unknown word a noun, verb, or adjective? Where it sits in the sentence tells you what kind of thing it is before you know what it means.
Word-part analysis (morphology)
Prefixes and suffixes carry meaning: un- (not), -ology (study of), pre- (before). Teaching 20–30 high-frequency word parts unlocks hundreds of lexical inferences.
Semantic logic
If the surrounding text is negative (crisis, disaster, failed), an unknown word is likely negative too. Teach students to read the tone and valence of a passage.
The “good enough” principle
Students often don’t need a precise definition — they need a functional understanding sufficient to continue reading. Teach “continue with a working guess” as a valid strategy.
Teach the process, not the answer
When a student asks “what does X mean?”, ask: “What do you think it might mean? Look at the sentence around it.” Only confirm after they’ve attempted an inference.
Use “thinking aloud” texts
Read a short text aloud and narrate your own inference process: “I don’t know this word — let me look at what’s around it…”
Set a “no dictionary for 5 minutes” rule
For the first five minutes of any reading task, students must attempt inferences before consulting any resource.
Celebrate good inference, not just right answers
If a student makes a logical inference that is slightly off, acknowledge the quality of reasoning: “That’s excellent reasoning — the actual meaning is close to what you said…”
Pre-teaching every unknown word
It removes the need for inference before the task begins. Students arrive at the text with everything explained.
Accepting “I don’t know” without scaffolding
The intervention should be a guiding question, not the answer: “What kind of word is it? What’s the sentence about?”
Treating inference errors as failures
A wrong inference made through good reasoning is far more valuable than a right answer obtained by looking up the word.
Checking comprehension without checking inference
“Did you understand the text?” reveals nothing. “What do you think the writer’s attitude is — and what in the text tells you?” forces inference.
What-ifs
Don’t fix the answer — fix the strategy. Ask them to talk through their reasoning aloud. Usually the error traces to ignoring syntax, importing an L1 schema, or not reading enough surrounding context.
Address it explicitly: explain why inference matters. Then use time-pressure tasks where dictionary use is physically impossible. Make inference the only option available, then reflect on the experience.
Give them a sentence where their interpretation would not work: “If it means X, what would this sentence mean?” Let the contradiction surface through reasoning.
Research Background — Teaching Inference Skills
Rosenshine & Meister (1994) — reciprocal teaching, which includes explicit inference instruction, produced effect sizes of 0.32–1.36 on standardised comprehension tasks.
Pressley et al. (1992) showed that students who received explicit strategy instruction significantly outperformed control groups on comprehension measures.
Nassaji (2006) found that successful inference depends on depth of processing: learners who engaged with syntax, morphology, and semantic coherence outperformed those who used context alone.
Emergent Language
Language that arises from communication rather than lesson plans — what it looks like, why it matters, and the theory behind it.
Emergent language is any language — vocabulary, grammar, pronunciation, discourse — that arises naturally from a communication task and was not part of the lesson plan. It can emerge from the learner (reaching for language they don’t yet have), from the task (triggering unanticipated needs), or from the input (a text containing language students notice).
The Vocabulary Gap
A student reaches for a word they don’t yet have.
“What’s the word for… when you feel nervous but for someone else?” → They need anxious on someone’s behalf.
The Revealing Error
An error that exposes a systematic gap in the learner’s interlanguage.
“I made my homework.” → Reveals a lexical collocation gap: do homework.
The Noticed Item
A student reads or hears something and asks about it — triggering unplanned language focus.
“In the article it said ‘nevertheless’ — is it the same as ‘however’?”
The Approximation
Language that is almost right — the learner has grasped the concept but not the exact form or register.
“I’m very embarrassing about it.” → They mean embarrassed.
“Materials light, conversation heavy, language emergent.” Dogme does not mean no planning — it means holding your plan lightly enough to follow where the learners’ language needs lead.
Students discuss travel. The teacher circulates with a notepad.
Student A: “When I was in Japan I was very… impressed shocked? By how clean everything is.”
Student B: “And my wallet was very… bad?”
The teacher notes: impressed shocked → awestruck / taken aback; wallet was bad → took a hit / suffered
Research Background — Emergent Language
Thornbury & Meddings (2009) — Teaching Unplugged. The founding text of Dogme ELT. Argues that conversation is both the medium and content of language learning, and that emergent language should form the basis of a lesson’s language focus.
Swain (1985, 2000) — the Output Hypothesis. Producing language forces learners to notice gaps between what they want to say and what they can say. Emergent language moments are precisely these gaps made audible.
Ellis (2001) — Language-Related Episodes (LREs). Defined as moments in a task where learners stop to focus on language. Research shows that items discussed in LREs have significantly higher retention rates than items encountered passively.
Capturing & Using Emergent Language
What to do when language appears — the board, the notepad, hot vs. delayed correction, and the decisions no coursebook can make for you.
The Teacher’s Notepad
While students are working on a task, the teacher circulates and records emergent language. The goal is not to catch errors — it is to curate. Selection criteria:
High frequency
If three students produced the same error independently, it is systematic and worth addressing collectively.
Fossilisation risk
Some errors — if not addressed — become permanent features of a learner’s interlanguage.
Achievable target form
The learner almost had it right. A small intervention produces a disproportionate result.
Lesson relevance
The emerged item connects to the lesson’s theme or a previously taught item.
The Board as a Living Record
The whiteboard during an emergent-language lesson is a living record of the lesson’s language. The teacher writes emerged target forms (not errors) on the board with minimal explanation. The board at the end of a lesson is a snapshot of what that particular group of learners needed on that day. No coursebook contains it.
When writing emergent errors on the board for correction, never attribute them to a specific student. Write: “Someone said this: I made my homework.” The class corrects it collectively.
Hot vs. Delayed Correction — The Decision
Is the error blocking communication right now?
Recast immediately and gently, then continue. Don’t stop the task.
Is this a self-correction opportunity? Did the student pause and look uncertain?
Use a rising intonation prompt or gesture — invite self-repair before intervening.
Is the error systematic, shared by multiple students, and non-urgent?
Note it. Return after the task. Address on the board anonymously.
Is this a high-fossilisation risk error the student has made before?
Intervene — but choose recast or prompt over explicit correction.
Recycle emerged language
An item that emerged in speaking can reappear in a writing task or the next lesson’s warm-up. Recycling dramatically increases retention.
Ask students to keep a “language caught today” log
Learners record emerged items from board work in a vocabulary journal — organised by context, not alphabetically.
Use reformulation for vocabulary gaps
Give the target form, use it in a sentence, then invite the student to use it: “So the word is anxious on someone’s behalf — can you use it to finish your sentence?”
Select ruthlessly
Five emerged items treated well is better than fifteen items listed and forgotten.
Interrupting mid-task to correct
This stops communication and teaches learners that accuracy matters more than meaning.
Ignoring everything and moving on
The task ends, the emergent language disappears, and the lesson contains only what was pre-planned.
Writing errors on the board
Always write the target form, not the error. Writing errors reinforces them in memory.
Attributing errors to individuals
Use anonymous collective feedback: “Someone said X — what should it be?”
What-ifs
This is diagnostic. Either the task is too controlled (students only producing language they already know) or the class culture is inhibiting authentic communication. If it’s the task: add information gaps, opinion demands, or role constraints. If it’s the culture: smaller groups and explicit permission to “not know the right word” reduce inhibition over time.
Triage using the selection criteria (frequency, fossilisation risk, achievable correction, lesson relevance). Deal with three to five items thoroughly. Photograph the rest of your notepad and use them in the next lesson’s warm-up.
Acknowledge it briefly and defer: “Great attempt — that’s a subtle point in English. Let me write it here and we’ll come back to it.” Students feel seen without the lesson being derailed.
Noticing
The mechanism that turns input into acquisition — how conscious attention to language forms, gaps, and meaning shapes what learners retain, produce, and ultimately own.
Richard Schmidt argued that conscious attention is a necessary condition for language acquisition: learners cannot acquire what they do not consciously notice. Encountering a form in the input is not enough — it must register in awareness for it to become available for acquisition.
Schmidt later distinguished between noticing (attentional registration) and understanding (grasping why the form works). Only noticing is claimed to be strictly necessary.
Four Things Learners Need to Notice
Noticing the form
The learner consciously registers a lexical item, structure, or sound in the input.
Reading “the policy was rescinded” — the learner flags rescinded rather than reading past it.
Noticing the hole (Swain)
During production, the learner realises they cannot say what they want to say.
“I want to say she was being dishonest but in a subtle way… what’s that word?” The hole is noticed; the search begins.
Noticing the difference
The learner compares what they produced with what the target language actually says — and registers the mismatch.
After recasting “I am here since Monday” → “I’ve been here since Monday,” the learner notices: something changed.
Noticing the pattern
Beyond individual instances, the learner perceives a recurring regularity — the beginnings of implicit rule formation.
“It’s always ‘have been’ with time expressions… is that a rule?”
What Makes Language Noticeable?
Frequency
Forms that appear repeatedly are more likely to be noticed. After initial registration, repeated exposure deepens processing.
Implication: One encounter is rarely enough. Three to seven encounters across varied contexts is a common rule of thumb.Salience
Perceptually prominent forms — those that carry communicative stress or appear in stressed positions — are more easily noticed.
Implication: English grammar endings (plural -s, third person -s) are notoriously low-salience and chronically under-noticed.Communicative need
When a learner needs a form to express a meaning they actually want to convey, noticing is dramatically more likely.
Implication: Tasks that create genuine communicative need produce richer noticing than decontextualised grammar drills.Prior knowledge
Learners notice what is slightly beyond their current interlanguage — within their Zone of Proximal Development.
Implication: Input should be pitched at i+1 — just one step ahead — for optimal noticing.Input Enhancement — Teaching Noticing
Sharwood Smith (1993) called this input enhancement — deliberately manipulating input to increase the salience of target forms:
Typographic marking
Bold, underline, colour, or italicise target forms in reading texts.
Oral stress & repetition
Slightly emphasise and repeat a target form: “She has been — has been — waiting for three hours.”
Board recording
Writing an emerged form on the board is a noticing prompt — students who copy it down have processed it twice.
Input flooding
Saturate a text with multiple instances of a target form without explicit instruction.
Dictogloss
Learners hear a text at natural speed, take brief notes, then reconstruct it collaboratively. Gaps in reconstruction make them notice what they missed.
Consciousness-raising tasks
Present learners with language data and ask them to discover the pattern themselves.
Corrective Feedback as a Noticing Tool
Recast
Teacher reformulates the error correctly while maintaining conversational flow. High fluency, lower noticing — learners often process recasts as comprehension checks rather than corrections.
S: “I goed to the shops.” T: “Oh, you went to the shops — what did you get?”
Elicitation
Teacher pauses and signals that repair is needed without supplying the form. Forces the learner to produce the correction — which deepens noticing significantly.
S: “I goed—” T: “You… ?” (pause, expectant look)
Clarification request
Teacher signals non-understanding (“Sorry?”), prompting the learner to reformulate.
Explicit correction
Teacher directly supplies the correct form with a metalinguistic signal. Highest noticing, lower fluency.
S: “I goed.” T: “No — the past of go is went. Say it: I went.”
Input: Students read an article containing the word haggard.
Inference: The teacher instructs them not to look it up — to infer from context.
Noticing (of form): The teacher confirms the meaning and writes haggard on the board.
Emergence: In the subsequent speaking task, a student tries: “I was feeling very… haggard? After the trip.”
Noticing (of gap): Teacher recasts: “Yes — you looked haggard! It describes how someone looks, not just feels.”
Research Background — Noticing in SLA
Schmidt (1990) — “The Role of Consciousness in Second Language Learning.” His diary studies of L2 Portuguese acquisition in Brazil. The strong claim is that noticing is necessary for acquisition; the weaker, that it facilitates it.
Swain (1993, 2000) — Pushed output forces noticing of gaps in ways that input alone does not. Her concept of “noticing the hole” is the production-side complement to Schmidt’s claim.
Lyster & Ranta (1997) — Analysed 18,000 turns in French immersion classrooms. Found that recasts produced the lowest rates of learner uptake and repair. Elicitation and explicit correction prompted significantly more learner self-correction.
Long (1991) — Focus on Form (FonF): brief, reactive attention to form in the context of meaning-focused activity — essentially engineering noticing opportunities without abandoning communicative goals.
Theory & Research
The full academic grounding — key theorists, foundational studies, and the debates that matter for practice.
H. P. Grice (1913–1988)
Oxford philosopher. His 1975 paper “Logic and Conversation” introduced the Cooperative Principle and four maxims. Grice showed that communication is always richer than what is literally said.
Key text: Grice, H.P. (1975). Logic and conversation. In Cole & Morgan (eds.), Syntax and Semantics Vol. 3.
Sperber & Wilson (Relevance Theory)
Proposed in 1986 that all communication is governed by a single principle: be as relevant as possible. In L2, higher processing costs mean inference is less efficient.
Key text: Sperber, D. & Wilson, D. (1986). Relevance: Communication and Cognition. Blackwell.
Richard Schmidt (Noticing)
Applied linguist at University of Hawai’i. His 1990 Noticing Hypothesis is one of the most cited ideas in SLA: conscious registration of a form is a necessary (though not sufficient) condition for acquisition.
Scott Thornbury (Dogme ELT)
Applied linguist and teacher educator. Thornbury’s Dogme ELT movement argued for teaching from emergent language rather than coursebooks. His concept of the “language-rich classroom” places student-generated language at the centre.
Key text: Thornbury, S. & Meddings, L. (2009). Teaching Unplugged. Delta Publishing.
Key Research Findings
| Study | Focus | Key Finding |
|---|---|---|
| Schmidt (1990) | Role of noticing in acquisition | Features attended to consciously were acquired; features not noticed were not. Noticing is necessary for acquisition. |
| Taguchi (2009) | Pragmatic inference in L2 | Even advanced L2 users process conversational implicature more slowly than L1 speakers. Pragmatic competence requires explicit attention. |
| Ellis (2001) | Emergent language in tasks | Items discussed in LREs had significantly higher retention rates at delayed post-test than items encountered passively. |
| Nassaji (2003) | L2 learner inferencing strategies | Strategic use of context was the key differentiator, not proficiency alone. |
| Lyster & Ranta (1997) | Response to emergent errors | Recasts (most common teacher response) produced the least learner uptake. Prompts produced significantly more uptake. |
| Swain (2000) | Pushed output and noticing | When pushed beyond their comfort zone, learners notice gaps between their interlanguage and the target. This is the mechanism that makes emergent language pedagogically powerful. |
| Donato (1994) | Peer interaction and emergence | Learners working together co-construct forms they could not produce individually — and these show up in subsequent individual production. |
The Live Debates
Is noticing really necessary for acquisition?
Schmidt’s Noticing Hypothesis has strong empirical support but faces a principled objection: much of what we know in our first language was acquired without conscious attention — particularly phonology and syntax. Critics (notably Krashen) argue that unconscious acquisition is primary. The current consensus leans toward Schmidt’s view for L2 acquisition in classroom contexts — but the debate is not closed.
Does explicit error correction actually help?
Truscott (1996) controversially argued that written error correction has no positive effect and may be harmful. Ferris (1999) disputed this vigorously. For oral correction specifically, Lyster & Ranta (1997) showed that recasts produce the least uptake. The practical implication: correction works, but how you correct matters enormously.
Can pragmatics be taught, or only acquired through immersion?
Kasper & Rose (2002) reviewed substantial research and concluded that pragmatic competence can be developed through instruction — but instruction must be explicit and sustained. Implicit instruction (exposure alone) produces slower and less reliable gains.
Is Dogme practical in exam-focused contexts?
Dogme has been criticised for being impractical in syllabus-driven contexts. Thornbury’s response has consistently been that Dogme is not a method requiring no planning — it is an attitude. The practical synthesis most teachers reach is a planned lesson framework into which space for emergent language is deliberately built.