SPTH2101 Chap.6 Word-Level Assessment Design
Word-Level Assessment Design
Word-level assessment is designed around opportunities. The tutorial comparison sheet asks what each tool samples across word length, syllable shape, vowels, consonants and consonant clusters. The Quick Screener form adds intelligibility, stimulability, PCC and pattern counts. Tool selection should therefore start with the referral question and required contrasts rather than the test name.
A list with many singleton onsets cannot efficiently answer a final-cluster question. A word obtained after imitation provides different evidence from spontaneous naming and must retain its cue level. Matched words help isolate a live variable, while connected speech checks whether the production survives outside a controlled task.
A screener can flag patterns and build a tentative inventory, but it cannot replace multilingual comparison, history, functional evidence or deeper sampling where diagnosis is at stake.
What this chapter covers
- 01
Assessment question before tool selection
- 02
Sound and vowel coverage across positions
- 03
Cluster, word-length and syllable-shape opportunities
- 04
Matched contrasts for isolating complexity
- 05
Cue levels and stimulability records
- 06
Screening signals compared with diagnostic evidence
Worked example · free
Select a tool from its coverage
- +1Choose the set that creates repeated opportunities relevant to cluster complexity.
- +1Add matched singleton controls so cluster load can be compared.
- +1Add missing word positions or cluster families that matter to the referral.
- +1Bridge to connected speech and record any cueing required during elicitation.
Key terms
- Elicitation opportunity
- A planned chance for the child to produce a specified sound, structure or contrast.
- Coverage matrix
- A table mapping candidate items against the sounds, positions, lengths and syllable shapes they sample.
- Matched contrast
- A pair of items designed to differ in one important feature while holding other demands as similar as possible.
- Cue level
- The amount and type of support used to obtain a response, such as naming, choice or delayed imitation.
- Screener
- A brief task designed to flag possible concerns and guide further assessment rather than complete diagnosis alone.
Word-Level Assessment Design FAQ
How do I compare two speech assessment tools?
Start with the clinical question, then audit each tool's opportunities across sounds, word positions, clusters, vowels, word length and syllable shape. Also compare cueing, scoring and language fit. Choose the tool that best tests the hypothesis and name its remaining gaps.
Does an imitated word count as accurate?
It counts as an accurate production under the recorded imitation condition and can inform stimulability. It should not silently replace spontaneous naming performance. Preserve the cue level so the reader understands what support was necessary.
Why add connected speech after a word list?
A list maps controlled coverage. Connected speech tests sequencing, rate, linguistic load and functional intelligibility. Comparing them shows whether a target is stable in communication and may reveal patterns that the planned list did not sample naturally.
How do I know whether an elicitation set fits the referral?
Translate the referral into required sounds, positions, cluster types, word lengths, syllable shapes and language conditions before reviewing the items. Mark repeated and missing opportunities, cueing demands and unfamiliar vocabulary. Prefer a set that creates interpretable contrasts, then add only the gaps relevant to the question. Plan a connected-speech bridge and preserve cue level beside every response.
Fit is demonstrated by the decisions the set supports, not by its total number of words.
Exam move
Take a blank grid with columns for sound, position, cluster, vowel, word length and syllable shape. Audit an invented word list before imagining any child response. Then add a referral question and remove every word that does not help answer it. Practise recording cue level beside each production and explaining why imitation supports a different inference from spontaneous naming.
Finish by designing one connected-speech task that tests whether the controlled result generalises. Begin each design problem by turning the referral into required opportunities: name the target feature, word position, structural complexity, number of contrasts and relevant language. Compare two candidate lists against that specification rather than choosing the longer or more familiar form.
Mark repeated coverage, genuine gaps and items that confound the target with unfamiliar vocabulary or several simultaneous complexity changes. Create matched contrasts that hold word length or vowel context as stable as possible while changing the feature under investigation. Pilot each picture or prompt for recognisability and separate a naming problem from a speech response.
Write an administration ladder that distinguishes spontaneous naming, neutral question cue, choice, delayed imitation and direct imitation. Record where the response first appears and avoid collapsing all levels into one score. After the controlled task, plan a play, narrative or conversation sample containing natural opportunities for the same structure. Predict what agreement and disagreement across formats would mean.
Finish with an audit trail containing target, child form, opportunity type, cue level, consistency and transcription uncertainty. The aim is not maximal item count; it is a compact set in which every item has a reason and every missing opportunity is visible before administration.
