OpenAccess

98% Oppose Plain Language Act Repeal

Callum McMenamin — 26 May 2025

Analysis of all written submissions to Parliament shows overwhelming public opposition to repealing the Plain Language Act — 96% say keep it.

A landslide of public opinion

Note: due to more submissions being uploaded by Parliament, the statistics have changed slightly — from 98% opposition to 96% opposition to the repeal.

According to analysis by OpenAccess, Parliament received 1,657 written submissions from the public for the Plain Language Act Repeal Bill.

To learn more about this repeal, OpenAccess has published two blog posts on it:

• National Party attacks accessibility for disabled people
• Plain Language Act repeal may breach UN human rights obligations

I classified every written submission using AI to understand how many submitters were for or against the repeal of the Plain Language Act:

• 1,598 were against the repeal (96.44%)
• 9 were for the repeal (0.54%)
• 43 were neutral or unknown (2.6%).
• 7 submissions were irrelevant, or corrupted (0.42%).

Conclusion: from 1,657 submissions, only 9 support the repeal

It is clear the public does not support the actions of the New Zealand Government. The public overwhelmingly wants to keep the Plain Language Act in place.

If the Select Committee ignores the opinion of the public, it will breach democratic values.

Perhaps those 9 submitters will be the only minority group they actually listen to.

Methodology

The classification of submissions was carried out using a combination of artificial intelligence and manual review, as follows:

1. I searched the New Zealand Parliament’s website for submissions on the Plain Language Act Repeal Bill.
2. For each search result, a JavaScript programme located the download link for the PDF labelled “Full evidence text”.
3. A Python application downloaded all identified PDFs.
4. gpt-4.1-mini was used to classify each submission.
5. Manual verification of results was performed on a sample of the results.

Limitations

1. While AI classification is reasonably accurate, it may make errors in classifying results.
2. Some submitters uploaded 2 submissions, causing duplicate submissions in the results. However, the number of duplicate submissions is negligible and is unlikely to materially impact the statistics.