Glossary issues, again

[credneb]

Hello again.

I use Nisus to translate Japanese to English, particularly patents. Patents have lots of boilerplate and odd, sometimes nonce, terms. Nisus' glossaries are terrific for handling such.

Be that as it may, the following sample entries work as defined.

the ave -> the average fiber length
the ave -> average fiber length

It _seems_ that Nisus goes back to the beginning of the string defined as the abbrev in the glossary. As the Nisus folk know, Japanese know does not use spaces btw words. This can cause problems. For example, the following Japanese terms do not expand correctly, in part. (these are ISO terms, so they may sound odd)

長さ−長さ加重平均繊維長 -> length−length-weighted mean length
長さ加重平均繊維長 -> length-weighted mean length

Specifically, the "−" in the first term seems to break the expansion. So the first term always expands as the second, shorter and incorrect, term (leaving the initial 長さー)

A further oddity: even without the second term being defined in the glossary, the first term expands to
長さ−length-weighted mean length

This is not limited to this character. If there is any Japanese string that is defined in the glossary but may be part of a longer string (term) that is not defined, that defined term will expand, leaving the original undefined portions of the string there and resulting in a (possibly) wrong, unintelligible (an un-useful) expansion.

For example, if the glossary has the term
管　-> pipe
and the text has the term 水土管 which you may prefer to call "water line", 水土管 becomes 水土pipe.

Perhaps I am trying to use the glossaries as not intended. If so, OK. But if a solution such as expanding in order of longest to shortest abbreviations is possible, that would be a nice improvement. Of course, these J terms are not "abbreviations" per se, but then an abbreviation is merely a shortened string.

PS. I have looked into CAT tools, but there are several problems, including virtually all being Windows programs, Euro-centric (with concomitant problems when working with J), and quite simply readability problems. I keep coming back to Nisus. (And not because of the high cost. )

[rwg]

PS. I have looked into CAT tools, but there are several problems, including virtually all being Windows programs, Euro-centric (with concomitant problems when working with J), and quite simply readability problems

Have you looked at the open-source OmegaT? Personally I find working with tags and so on too much of a hassle, but by all reports it works flawlessly with Japanese and would seem ideal for translating patents.

[credneb]

re: OmegaT

Thank you for the suggestion. Yes, I have tried it. But it suffers from the same readability problems (for me, anyway) as the others on Windows.

[Þorvarður]

Hello credneb,

the following Japanese terms do not expand correctly […]
長さ−長さ加重平均繊維長 -> length−length-weighted mean length
長さ加重平均繊維長 -> length-weighted mean length

Specifically, the "−" in the first term seems to break the expansion. So the first term always expands as the second, shorter and incorrect, term (leaving the initial 長さー)

I entered your examples into one of my glossaries. If I paste the first part into a Nisus document and then hit the spacebar, everything is expanded correctly.

if the glossary has the term
管　-> pipe
and the text has the term 水土管 which you may prefer to call "water line", 水土管 becomes 水土pipe.

Same here. 水土管 expands correctly as "water line".

Disclaimer:
I didn’t type these characters on the keyboard, because I don’t know how to do that, but pasting and then hitting the spacebar should produce the same results, I suppose.

Þorvarður
MacBook Pro mid 2010
OS 10.11.5

[phspaelti]

Disclaimer:
I didn’t type these characters on the keyboard, because I don’t know how to do that, but pasting and then hitting the spacebar should produce the same results, I suppose.

Hi Þorvarður,
I think you might be supposing wrongly. Unlike you, I have no experience with Nisus Glossaries, but my attempt to reproduce this gave results in line with credneb's.

First case
With the two glossary entries:
長さ−長さ加重平均繊維長 ->length−length-weighted mean length
長さ加重平均繊維長 -> length-weighted mean length

"長さ−長さ加重平均繊維長" expands to "長さ–length-weighted mean length"

I was able to fix this by adding the glossary entry:
長さ– -> length-

The expansion happens in two steps. Upon typing the full expression, "長さ–長さ加重平均繊維長" expands to "length-長さ加重平均繊維長" and then typing another space expands the rest.

Second case
With (only) the following glossary entry
管 -> pipe

"水土管" will expand to "水土pipe". (Of course?) if you also have an entry
水土管 -> water line

the latter expansion overrides, but I believe credneb is not considering that case. The problem is that the Nisus is essentially expanding part of a word. Of course since Japanese is written without actual word breaks, and spaces (etc.) are used to trigger conversion, the whole process seems like a nightmare for glossary expansion. However credneb is not really typing Japanese at all (merely using Japanese terms as "shortcuts" for English text). It would seem wise to at least stay away from single character expansions (like 管-> pipe).

[Þorvarður]

Hello Philip,

Thanks for your comment. As always, much appreciated.

I think you might be supposing wrongly.

Pasting works with Latin characters, which makes this method convenient when troubleshooting. Pasting also works with Japanese, and the examples from credneb are always expanded correctly when they are pasted, but I understand that this isn’t particularly consoling, because credneb wants to type in the characters.

I can’t type in those characters myself using the keyboard, so I can’t verify if *typing* (instead of pasting) actually makes a decisive difference in Japanese.

Maybe you or someone else who writes Japanese can test whether the same problem occurs in Typinator http://www.ergonis.com/products/typinator/ or TextExpander https://textexpander.com/textexpander-standalone-apps/ ?

If they also behave erratically, then this might be an error in Apple’s text engine and it should be reported to Apple.

Bug reports to Apple can be sent here:
http://www.apple.com/feedback/macosx.html


I noticed the following, which might be of importance. The abbreviation 管 by credneb is followed by a strange invisible character, which I think shouldn’t be there.

1.png (10.83 KiB) Viewed 12397 times

Maybe this character prevents expansion when you type “管"?
With Latin characters a successful expansion requires that ALL characters are being typed, and the typing is case-sensitive.

Why is this character there in the first place? Was it perhaps falsely inserted by credneb’s browser or the Nisus Forum software (?)

I also notice that the abbreviations in

2.png (23.93 KiB) Viewed 12397 times

have a trailing space. Trailing spaces do not affect expansion with Latin characters, but perhaps they do in Japanese. Just a thought. Test it.


Þorvarður

[credneb]

Thank you for your interest and comments, which sent me looking more.

I discovered that the problem was the hyphen in 長さ−長.
Somewhere in the trip from the original document in Nisus to Excel and back to Nisus glossary, that character was converted from U+30FC to U+2212 in the glossary. So it would not expand. Fixing that solved the problem. The characters look the same on screen, but are obviously different.

The problem I wrote poorly about using the example of

管 -> pipe

Assume the above entry in a glossary, and you have 水道管が土の中に in your sentence.

If you put the cursor at the end of that phrase and trigger expansion, you get

水道pipeが土の中に

The look back goes quite a way back.

I've only recently tried using glossaries for this type of expansion, and am discovering some quirks.