I'm returning here to a topic I have posted on before. This time I gave Chat GPT some notes to work on, it provided the bones of the blog and then I edited and supplemented it.
We know that language input that students understand is at the heart of language acquisition. If students don’t understand the input, learning doesn’t happen — it's mere exposire, not language that learners can process. The challenge for language teachers is to model input clearly, repeatedly, and meaningfully.
Below are five commonly used ways to model input, with concrete ideas on how to exploit each one, all classroom-tested and easy to adapt. AI makes creating these inputs faster than ever — but it’s how we use them that really matters.
1. Sentence builders
Maximum clarity, zero ambiguity
Originally known in audiolingual practice as substitition tables, sentence builders makes the target language completely transparent. Students know exactly what language they will use, and translations of every word and chunk ensure everyone can follow, regardless of confidence or prior attainment. This is input every student can understand and succeed with. The best sentence builders are not too busy or overwhelming, allowing the language they contain to be reused many times over, through listening, speaking, reading and writing. SBs are an intergral tool in the EPI approach developed by Gianfranco Conti. Check out his blog gianfrancoconti.com
The content of the SB can be closely related to the syllabus, it can be cultural input or just absurd combinations of chunks to amuse students.
Why they work
They force attention on to the form and meaning of the language being modelled.
They reduce cognitive load through their clarity amd use of translation.
They guarantee comprehensible input for all.
They are easy to exploit in multiple ways.
Ways to exploit a sentence builder
Guided sentence creation — pupils choose one element from each column to build accurate sentences orally or in writing.
Listening + pointing — the teacher reads a sentence; pupils point to or highlight the elements used in the builder.
Translation both ways — translate from L2 → L1 to confirm meaning, then rebuild from L1 → L2 using the builder.
Sentence expansion — start with a simple sentence, then require pupils to add one extra element (time phrase, opinion, reason).
Controlled manipulation — change one element only (e.g. verb tense or subject) to focus attention on a specific structure.
'Mind reader' — students must guess which sentence you are thinking of — this can be done in pairs.
Vocab learning — instead of learning words from lists or an app (boring!), students can attempt to memorise whole sentences from a SB to be reused at the start of the next lesson. This has more communicative power than learning isolated words.
Play read aloud games — Sentence Stealer and Find someone who... are good examples from the Conti repertoire.
2. Written texts
More natural, still controlled
A short written text provides a more authentic stream of language than a sentence builder, while still being manageable. With AI, teachers can now generate texts at exactly the right level, with the grammar and vocabulary they want. Just prompt AI with the CEFR or ACTFL level you want. Or specify individual words and chunks you want to appear. Only you know precisely what your class might be able to manage.
Why they work
They feel more “real” than isolated sentences.
They allow recycling of language in context.
They are easy to exploit in multiple ways.
They can an a storytelling, narrative aspect.
With a parallel translation the can be fully compehensible.
7 ways to exploit a written text
Question–answer comprehension — teachers can use the full range of question types (e.g. yes/no, either/or, multi-choice, opne-ended questions)
True / False / Not mentioned — forces close reading and discourages guessing. Be careful with 'not mentioned'!
Gap-fill using the original text — remove verbs, connectors, or key phrases learners must retrieve from memory. You can progressively remove words slide after slide to encourage retrieval and memorisation ('disappearing text'). Consider removing words you want students to notice most, maybe verbs.
Correct the false statements — students must correct false statements using language from the text. Can be done orally or in writing.
Text reconstruction — give the L1 version or jumbled sentences and ask learners to rebuild the original text.
Change the perspective — if the text is in the third person, get students to put it in the third person and vice versa.
'Complete my sentence' — hide the source text, then sy aloud the starts of sentences which studnets must complete. With weaker classes give them a lot, with stronger classes give them little.
3. Dialogues
Input that sounds like real language
Dialogues are one of the most natural sources of input: they model interaction, turn-taking, and spoken rhythm. AI makes it easy to create dialogues at the right level, with repeated structures and useful chunks that you can specify in advance.
Why they work
They model how language is actually used (up to a point - authentic speech with its hesitations, repetitions, use of slang and words like 'like' (In English) is often very different to written language).
With a parallel translation they can be completely comprehensible.
They are highly adaptable.
5 ways to exploit a dialogue
Read aloud with roles — teacher first, then confident pupils, then pairs. Use choral repetition or delayed choral repetition (leave a time gap after the teacher model)
Substitution practice — replace key details (time, place, activity) while keeping the structure. You can mark in bold items you want students to change.
Gap-fill — give students a gapped version to complete, fewer gaps for weaker clases, more gaps for stronger classes.
Questioning and other interactions — as with written texts above, do QA and other interactions to recycle language from the dialogue.
Writing a new dialogue — students can write, then read aloud their own dialogue with as much scaffolding as they need.
4. Images and words
Attention-grabbing, memory-friendly input
Images reduce reliance on translation and keep attention high. Whether it’s slides with single words, chunks, full sentences, or picture sequences, visuals help anchor meaning and support weaker readers. There was a time when teachers were encouraged to avoid the written word since the latter was thought to interfer with accurate pronunciation, but few teachers believe this now. On the contrary, showing the written word has many benefits, aiding memory, improving the grasp of phonics and supporting general literacy.
Why they work
They hold attention (my feeling in the classroom was that whenever there was an image to see, attention was likely to be better).
They reduces cognitive load (Dual Coding effect whereby it has been shown that allowing Working memory to process both images and language at the same time strengthens memory). Research studies show that when vocab is presented with a picture it is remembered better.
Where members of a class do not share the same L1 they provide a neutral stimulus.
5 ways to exploit images and words
Images/flashcards to supprt vocab learning — showing slides with images as well as words or phrases. use Kim's game for some gamification (recalling the hidden image(s))
Picture sequence narratives — use a sequence to tell a story, then use a repertoire of QA techniques and similar to build the story and maybe change perspective.
Story from a single picture — for advanced students use a single image to come up with a complex story behind the image.
Sentence generation — show an image and require a sentence using a specific structure.
Picture description — use complex images of scenes for description, e.g. classrooms, a beach, a town centre, a family.
5. Knowledge Organisers with parallel translations
Total comprehensibility, total confidence
A knowledge organiser with side-by-side translations removes uncertainty. Pupils know exactly what the language means and what they’re expected to learn. This is especially powerful for lower-attaining learners.
Why they work
As with a senetnce builder, everything is understandable.
They can support independent learning.
Each sentence can be adapted, using alternative vocabulary.
5 ways to exploit a knowledge organiser
Retrieval practice — cover the L1 side and recall meaning, then reverse the process.
Sentence mining — highlight useful chunks and reuse them in new contexts.
Translation races — short, timed translations to build fluency.
Reading aloud — KOs of this type often come with notes to aid pronunciation and reinforce knowledge of sound-spelling correspondebs (SSCs).
Writing support — use the organiser as a reference or literact mat for producing extended writing.
Conclusion
A good question to ask when providing modelled input is:
Can every learner understand this input, and can I exploit it multiple times in various ways to recycle the language?
If the answer is yes, learning will follow.
Comments
Post a Comment