Re: More flexible line length in browsing mode


Tony Malykh
 

I believe in browse mode lines are defined visually, I don't recall NVDA setting that would define line length.

If you would like to read by sentences, you can install my SentenceNav add-on.

HTH

--Tony

On 11/18/2021 11:07 PM, Martin J. Dürst wrote:
Hello everybody,

I have been using NVDA on and off for a few weeks. It's really a great help. I'm new to this mailing list, so please forgive me if I'm asking something old.

When reading text from a Web page, the text is read in "lines", and the user presses arrow-down for each line. NVDA has a setting for line length, which is at 100 characters originally. So well, so good.

What I find somewhat confusing, and possibly a place for improvement, is that often a "line" ends a word or two before the end of a sentence, or includes a word or two of a new sentence. I suspect that quit a bit of thought must have gone into this, but I haven't found any details yet.

I would really appreciate if somebody could explain why "lines" end at arbitrary positions in sentences, and are not done a bit more flexibly so that they more often end at the end of a sentence.

If this has been discussed already, I would appreciate pointers. Also if there's some scientific paper about the issue.

I have tried to think about why things are as described above, and have come up with various possible reasons. If any of these reasons applies, please just tell me.

- There is already a setting/add-on for this, just use it.

- Having more variable line lengths would make it more difficult to read
  Web pages (e.g. because the intervals between the presses of the down
  arrow would be more irregular). If that's the case, then I haven't yet
  had enough practice to notice it.

- Finding better positions to split text into lines is a much harder
  problem than it looks. It is difficult to find actual sentence
  boundaries in text (not all periods are sentence endings), and
  long sentences without punctuation are also difficult to split.

- Finding better positions to split is possible, but good algorithms
  are too slow. Text-to-speech conversion already uses quite a bit
  of processing power.

- Finding sentence boundaries is quit language dependent, and therefore
  difficult to implement in a general way.

- The overall architecture of NVDA (and other screen readers) makes
  it too difficult to implement such a feature.

- Some other screen readers already do a better job at this, but we
  at NVDA just have not had time to get around to do something here.
  Help is appreciated. (I might want to help.)

- That's how screen readers always have done it, and everybody is
  used to it, and so changing it isn't a good idea.

If there are any other actual or potential reasons, please tell me.

Many thanks in advance for your help.

With kind regards,   Martin.



Join nvda@nvda.groups.io to automatically receive all group messages.