gum icon indicating copy to clipboard operation
gum copied to clipboard

Sentence initial w tag

Open amir-zeldes opened this issue 4 years ago • 0 comments

w tags can't indicate sentences beginning with an unspaced word fused to the preceding period, due to XML nesting conflict. Currently this is marked by a single token <w> tag on the first word of the sentence not separated by a space:

<!-- text = events.I -->
<s>
...
events
.
</s>
<s>
<w>
I
</w>

The build bot should interpret such single token w spans correctly for NoSpace and text generation.

amir-zeldes avatar Oct 21 '19 14:10 amir-zeldes