Language wrongly reported as unsupported

[CyrilleB79]

Steps to reproduce:

• In the input gesture dialog, assign a gesture to "Reports the language for the text under the caret. If pressed twice, presents the information in browse mode"
• Switch to eSpeak
• Open the following URI:
  data:text/html,<p>Text in <span lang="fr-ca">Francais</span> and in <span lang="it">Italiano</span><p>
• In the page that open, go on "Francais" and execute the command to know the language of the text
• Go on "Italiano" and execute the command to know the language of the text

Actual behavior:

• French (Canada) is reported as unsupported
• Italiano is normally reported, i.e. supported

Expected behavior:

• Italiano should still be normally reported, i.e. supported
• French (Canada) should be normally reported, i.e. do not indicate "unsupported".

Indeed when reading the sentence, the synthesizer switches language to French and Italian when needed. Reporting that French (Canada) is unsupported is quite misleading. eSpeak has no French (Canada) voice, but it supports reading "fr-ca" text falling back to generic French (French France).

I guess that all synth fall back switching to the non-dialect version of the language in case the language with dialect is not supported.

To be precise, we should report "Supported as a fallback". But this is probably unuseful and makes the feature much more complex.

NVDA logs, crash dumps and other attachments:

N/A

System configuration

NVDA installed/portable/running from source:

Installed

NVDA version:

2025.3beta3

Windows version:

Windows 11 24H2 (AMD64) build 26100.4946

Name and version of other software in use when reproducing the issue:

N/A

Other information about your system:

N/A

Other questions

Does the issue still occur after restarting your computer?

Yes

Have you tried any other versions of NVDA? If so, please report their behaviors.

No.

If NVDA add-ons are disabled, is your problem still occurring?

Not tested. I only know IBMTTS as TTS using voices with language without country specified. On the contrary, TTS available without add-on (OneCore, eSpeak) have voices language including the country code.

Does the issue still occur after you run the COM Registration Fixing Tool in NVDA's tools menu?

No tested, unrelated

[CyrilleB79]

@nvdaes, you have implemented this feature recently. Do you agree with my reasoning?

[nvdaes]

Cyrille wrote:

you have implemented this feature recently. Do you agree with my reasoning?

I'm not sure.

• If the main language, for example, French, is supported, it should be well pronnounced by the syntesizer.
• • In the other hand, users have ability to toggle dialect switching, and they may decide not to use this option if a dialect is not fully supported.

So at this moment I don't have a strong opinion about this.

[CyrilleB79]

The reason I had opened this issue in the first place, is that I did as follows:

1. report text language command let me know that the language of the text (e.g. was unsupported).
2. Then when reading the same text, the synth switched language anyway.

Hearing "not supported" for a specific text and having the synth switching language when reading the same text was quite surprising and confusing to me.

[nvdaes]

I've readed again relevant code of languageHandling.py and the script in globalCommands, and for me this is by design. In languageHandling, there is a function to know the language which should be reported, i.e., if autodialectSwitching is on, the full language is reported, and just the main language otherwise, for example, French cCanada or French. According to this, if autodialect switching is on, French Canada should be reported, in this case, as not supported (according to configuration for reporting not supported, if this is enabled. And French would be reported normally, without adding "Not supported".
Perhaps we should add a check to inform if the main language is supported, and not the dialect, in the script of globalCommands?

[CyrilleB79]

@nvdaes this issue has just been triaged.

If you wish/can work on this, it may be a good opportunity to ask or discuss clarifications if needed.

[nvdaes]

OK Cyrille. I'll take this, though perhaps I'll need clarifications, since I did this intentionally thinking that this would be better, not a bug.

[nvdaes]

@CyrilleB79 , I've created a Pen with your sample for testing:

https://codepen.io/nvdaes/pen/WbrXxZL