Accurate grammar and spelling feedback for PTE exam candidates

Alfa PTE is a leading practice platform for PTE Academic, PTE UKVI and PTE Core preparation.

Its AI-powered scoring works much like the real exam and gives instant score feedback, so users can quickly see their strengths and the areas they need to work on. That makes study time more focused and preparation for the PTE exam more effective.

WProofreader sits inside that scoring flow. It checks what students write and highlights the grammar and spelling mistakes in their answers.

What Alfa PTE needed

The main users here are students, and the checking happens at a very specific moment. Alfa PTE wanted to show students what went wrong in the answer they submitted, not to correct them while they were still writing it.

So the platform needed a proofreader that could:

• check a student’s response after submission rather than during the test
• highlight grammar and spelling errors, and show how many of each there are
• get it right, because a wrong highlight costs a student marks

Grammar and spelling are the two error types that matter most. Style suggestions are useful sometimes, but not always.

Accuracy is the part that really counts, because everything on the platform feeds into a score. Users come to Alfa PTE to see exactly where they stand. If something gets highlighted incorrectly, the platform counts it as a grammar error and deducts marks, and the student ends up with a lower score than they earned. Once that happens, they start to feel the system is not accurate, and that works against the platform and its reputation.

Challenge

Alfa PTE had used a different solution for this before, and it was not giving them the results they needed.

There were plenty of things being marked that should not have been marked, and plenty of real mistakes that were not being marked at all. Both were landing on Alfa PTE’s reputation, so the team started looking at alternatives.

When they came across WProofreader, they tested it properly. The team took hundreds of questions and hundreds of responses they already had from real users, ran them through both tools, and compared what each one flagged.

The analysis was clear. WProofreader was giving them a better response and highlighting the issues that were genuinely there. The previous software was not doing that. It was flagging random words that were not actually wrong.

Because the test used actual end user responses, the team could see exactly how each tool would behave in their own product.

Solution

After that testing, the team was in no doubt about which way to go.

Alfa PTE integrated WProofreader gradually, one set of questions at a time, rather than switching everything over at once.

Here is how it works today. The writing part of the platform has different sets of questions, including ‘summarize written text’, ‘write essay’ and ‘write email’. A student picks a question, writes their answer and submits it. They then choose whether the answer should be checked as American or British English. The response goes through WProofreader, and whatever WProofreader highlights is shown back on the student’s own answer.

It is a simple setup on purpose. The team is not using the deeper capabilities, just the part that matters for scoring.

Benefits

Used across the whole writing experience

WProofreader now runs on every writing module on the platform. Wherever the team can use it, they already do.

Proofreading questions while they’re written

The team also uses the WProofreader extension for Google Chrome when they add new questions to the platform. If there is an error in the content, it gets highlighted while they write, and one click swaps in the suggested correction on the spot.

It does not slow anyone down. If anything, it is the quickest way to do the job.

A better option than free AI tools

Alfa PTE does not use free AI writing tools for this work, and does not encourage users to either. The reason is the same one that brought them to WProofreader in the first place: those tools are not accurate enough. They highlight parts that should not be highlighted, and mark things as errors when they are not.

What the team would like next

One thing on the wish list is a breakdown of the errors WProofreader catches, showing how many are spelling errors and how many are grammar errors. That would help the team see where the checker is doing the most work on their platform.

Conclusion

WProofreader is now part of how Alfa PTE works. It checks every writing module on the platform, shows students the grammar and spelling errors in the answers they submit in either American or British English, and keeps new questions clean while the content team writes them in the browser. The setup is deliberately simple, and it does the one job the platform depends on: giving users a score they can believe in.

The partnership has a working rhythm to it as well. When a highlight looks wrong, Alfa PTE’s expert trainers review it, the operations manager sends us an email, and we come back with an answer quickly. That loop is a good part of why the team is comfortable building their scoring on top of the tool.

There is still room to do more. The clearest ask from Alfa PTE is better visibility into what WProofreader is catching: a breakdown of how many of the flagged errors are spelling and how many are grammar. That would show their product team where the checker is doing the most work across the platform, and it is the piece of feedback we are taking up next.