This is Part 3 of our series. Your tools are connected and your template works. Now make the assessment fit the learners: choose a year-group blueprint, set the right support and check how the demand changes.
“Make it harder” is a vague brief. “Keep the taught words, add short contextual gaps and award two points for a sentence translation” gives the assistant something useful to work with.
Three choices that should stay separate
- Year-group blueprint: which task types and scoring structure suit what this class has been taught?
- Support: what help lets a pupil attempt those targets, such as first letters or a word box?
- A/B version: how should equivalent papers vary in order or layout?
A pupil in Year 9 may need extra support. A confident pupil in Year 6 may be ready for a more demanding task. Year labels organise the templates; they are not a diagnosis of individual ability or a guaranteed language level.
Our example progression: from words to short sentences
These are the blueprints used in our project, not a national requirement. Adapt them to your taught vocabulary, curriculum and assessment policy.
| Year | Task mix | Items / points |
|---|---|---|
| 5–6 | 10 German-to-English words; 5 word matches; 5 short contextual gaps | 20 / 20 |
| 7 | 10 German-to-English words; 2 English-to-German words; 4 sentence-half matches; 4 contextual gaps | 20 / 20 |
| 8 | 5 German-to-English words; 3 English-to-German words; 4 sentence-half matches; 8 contextual gaps | 20 / 20 |
| 9–10 | 5 German-to-English words; 5 contextual gaps; 5 short sentence translations | 15 / 20 |
In Years 9–10, the five word items and five gaps earn one point each. Each of the five sentence translations earns two points. That is 5 + 5 + 10 = 20 points. Fewer questions can still mean a greater demand.
Years 7 and 8 share a broad stage but have different blueprints. Give the exact year when generating a test; “7–8” alone should prompt the assistant to ask which format you want.
See the same target at three stages
To make the progression visible, keep one target constant: spoon. These are separate miniature examples, not three items to repeat within one test. Use each form only if the class has learned the surrounding language.
| Stage | Standard item | Supported item |
|---|---|---|
| Years 5–6: recall a word | der Löffel → ______ | der Löffel → s______ |
| Years 7–8: use context | I eat my soup with a ______. | I eat my soup with a s______. |
| Years 9–10: produce a short sentence | Translate: Ich esse meine Suppe mit einem Löffel. | Translate the same sentence. Frame: I eat my soup with a ______. |
Expected answers: spoon; spoon; “I eat my soup with a spoon.” Each supported item points towards the same answer as its standard partner.
The sentence frame removes much of the sentence-building demand. It therefore changes what the supported answer demonstrates. Use the adapted assessment according to your school policy rather than claiming both tasks measure independent sentence production in the same way.
Make sentences more demanding without making them muddy
A later-year vocabulary test does not need long sentences, unusual names or untaught grammar. Increase the required use of the target word while keeping the language clear. If pupils learned “spoon”, a short sentence using it is more useful than a complicated story about a restaurant inspection.
For sentence-half matching, make sure each beginning has one clear ending. Shuffle the endings, and do not let punctuation or grammar make the answer obvious before the vocabulary is read. In our template routine, a visible assessed target is bold; a target hidden in a gap stays hidden.
Choose support deliberately
| Support | What it helps with | What to check |
|---|---|---|
| First-letter cue | Starting to recall a word | The hint sits directly beside the correct gap |
| Shuffled word box | Recognising and choosing a word | Correct forms, no accidental answer order |
| Sentence frame | Reducing sentence-building load | Be clear about which grammar is now supplied |
| Extra writing space | Writing and page navigation | Keep targets and points unchanged |
Changing support does not automatically mean changing the learning target. Changing the target or reducing the points is a separate decision. Our supported page keeps the item numbers and points but omits the grade field and grading table.
Ask for an exact year and support profile
Use Vocab Test Assistant and Canva.
Create vocabulary test no. 4 for Year 8.
Use the confirmed source list and the Year 8 master.
Use 20 distinct taught targets:
- 5 German-to-English words
- 3 English-to-German words
- 4 sentence-half matches
- 8 contextual English gaps
One point per item: 20 points total.
Create Standard, Supported and Answer Key pages.
For Supported, add shuffled word-box help to word tasks
and first-letter cues to gaps. Keep targets and points.
Remove the supported grade field and grade table.
Use short sentences and only taught grammar.
Build and check the item map before filling the design.
If fewer than 20 suitable targets are supplied, ask me
whether to supply more words or change the test scope.For Year 9 or 10, request the 15-item, 20-point blueprint instead. For Year 7, specify its different translation and gap counts. Do not simply change the year in the heading and leave the old tasks underneath.
Mark sentence items transparently
For the two-point “spoon” sentence, a sample rule is one point for the target word in the correct form and one for an understandable, sufficiently accurate taught sentence. Record accepted alternatives before marking and avoid deducting twice for the same error.
Our miniature example uses: “I eat my soup with a spoon.” A missing full stop alone need not lose the sentence point under this example rule. Decide spelling and other grammar treatment in advance and use your school’s policy consistently.
Optional extension: make A and B versions
A/B versions and supported versions solve different problems. A/B papers vary the order or wording of comparable questions. Supported papers add scaffolding. For the same pupils, changing the order is the simplest first extension.
Our current Vocab Test Assistant defaults to three pages. The following is an extension brief, not a feature we have already tested in the plugin.
Extend this workflow to A/B papers.
Use the same confirmed targets, task types and points.
Create one item map with stable target IDs, then record
the printed question number for each version separately.
For B, reorder items only within their original task section.
Shuffle matching choices and word-box entries independently.
Keep the same sentence wording for this first A/B trial.
Do not create B merely by changing its heading.
Produce six clearly labelled pages:
1. Standard A
2. Supported A
3. Answer Key A
4. Standard B
5. Supported B
6. Answer Key B
Check each supported sheet against its matching standard
sheet. Check both keys against the printed numbering.
Report any page operation you cannot complete.For example, B’s first translation could be “das Wasser” instead of A’s “der Tisch”. That means B1’s key must be “water”. Reordering questions without rebuilding the key is an easy mistake to miss.
For later versions, you might use different simple contexts while keeping the same targets and points. Review them carefully: matching the number of questions does not prove equal difficulty. Reordering alone also does not prevent every form of copying.
Check the progression before classroom use
- Compare each task with what the class has actually learned.
- Check the exact year blueprint and total points.
- Compare standard and supported papers item by item.
- Solve the questions independently and check the key.
- For A/B, verify the key against each printed question number.
- Inspect every page and the exported file before printing.
Once this works, a routine request can be as short as “Year 8, test 4, the attached confirmed list, standard and supported, same template”. The detailed reusable rules still do the heavy lifting.
Prepared on 31 October 2026. Based on the private Vocab Test Assistant workflow. A/B support remains a proposed extension, not a tested plugin change.