it seems like soft Gs are an issue

[jseakle]

Among the examples in the paper is "higil", which I imagine most English speakers would pronounce to rhyme with "sigil" (and indeed, "sigil" itself is a proquint). This seems identical to how the same speakers would pronounce the proquint "hijil" (or "sijil").

Obviously an enfranchised user will understand that they need to use a hard G, but this (and possibly other similar cases, maybe "lasar" and "lazar"?) seems like a real impediment to general usability. I wonder if anyone has thought about solutions to this, and whether there have been any real-world deployments that have encountered this issue.

[dsw]

I'm not sure there is any real help for this other than to use slightly artificial regular pronunciation rules where you map each letter to a single sound, even where it produces results that are not the usual but irregular pronunciations of English. So for "lasar", using "s" in its usual pronunciation, even in the contest of "lasar", where it would normally sound like a "z".

…

On Sun, Apr 7, 2019 at 2:44 AM jseakle ***@***.***> wrote: Among the examples in the paper is "higil", which I imagine most English speakers would pronounce to rhyme with "sigil" (and indeed, "sigil" itself is a proquint). This seems identical to how the same speakers would pronounce the proquint "hijil" (or "sijil"). Obviously an enfranchised user will understand that they need to use a hard G, but this (and possibly other similar cases, maybe "lasar" and "lazar"?) seems like a real impediment to general usability. I wonder if anyone has thought about solutions to this, and whether there have been any real-world deployments that have encountered this issue. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#11>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAC_XPR-zyberDvNkau2HEZdISG_vuilks5veb4VgaJpZM4cgqWi> .