|
Post by dreamingfifi on Jan 20, 2015 20:21:36 GMT
I keep meaning to mention this, but I keep forgetting, so while I remember I'll mention this. It's a pretty minor thing, but I think it needs to be mentioned.
Spaces between words. Not all writing systems use them. Japanese and Chinese don't, for example. There's also affixes to consider. In Sindarin, do we use รบ- as a separate word or as part of another word?
I think that there's a pretty easy way to get around this problem. We can just have the translator put in spaces before and after the words that will have spaces around them, and none if it's an affix. So, for example, this sentence in Sindarin: Pedithangid n'aduial. 'The two of us will speak at evening twilight."
The morpheme entries for the Sindarin part would be:(using underscores to show spaces to make them more visible) [_Pedi] = [_speak_] [tha] = [_will_] [ng] = [_us_] [id_] = [_The two of_] [_n'] = [_at_] [aduial._] = [_evening twilight._]
I don't know if you thought of this beforehand or not, but I figured I'd mention it just to be safe.
|
|
|
Post by Auburn on Jan 20, 2015 22:11:49 GMT
(you also reminded me, ...another thing i keep meaning to mention is paragraph breaks! so everything isn't one long unbroken wall of text)
Hum, in terms of mass-scale production of the tool, I could see adding spaces all the time getting really tedious for some, especially if a language doesn't require exceptions much. So maybe an on/off option can exist in the admin panel for "insert spaces automatically"?
Maybe something like [ br ] could be used to cause line breaks. Hmmm.
|
|
|
Post by Auburn on Jan 20, 2015 22:16:57 GMT
Btw, That's a really curious way to break down a word! I hadn't thought about that. I was thinking you'd use the morphology menus for that. Like designate a verb as "verb -> conjugation -> future -> 2nd familiar dual". Otherwise, wouldn't you run into a problem of creating slots for words without roots? like "ng" and "id", if they had their own row then they'd need to link up to a root word. which, of couse, we can make "ng" and "id" roots but i thought they were suffixes.
But to take Spanish for example: sufriremos = we will suffer
root: sufre dropdown -> tense: future (description is added of what happens to future tense words) dropdown -> person: 1st person plural (description is added for how this ending is "-mos")
I could see both methods working great though. One or the other will probably be more beneficial depending on the language.
|
|
|
Post by dreamingfifi on Jan 21, 2015 20:38:02 GMT
Btw, That's a really curious way to break down a word! I hadn't thought about that. I was thinking you'd use the morphology menus for that. Like designate a verb as "verb -> conjugation -> future -> 2nd familiar dual". Otherwise, wouldn't you run into a problem of creating slots for words without roots? like "ng" and "id", if they had their own row then they'd need to link up to a root word. which, of couse, we can make "ng" and "id" roots but i thought they were suffixes. But to take Spanish for example: sufriremos = we will suffer root: sufre dropdown -> tense: future (description is added of what happens to future tense words) dropdown -> person: 1st person plural (description is added for how this ending is "-mos") I could see both methods working great though. One or the other will probably be more beneficial depending on the language. Well they're different levels of translation, so they'd have separate morphology trees. I think of translation as falling into three types: Loose, Literal, and Linguistic. (Yay! Alliteration!) Loose translations are what most novels, movies, TV shows and such go through when being localized. This can include changing the setting of a story to make it more familiar, or changing the foods into foods that are more familiar. The idea is to capture the tone and intent to communicate the original emotional meaning of the text, rather than the individual words. Literal translations are what are found usually in a more scholarly text, because they're heavily annotated to explain the cultural context of something. A lot of fan translations of Manga and Anime are like this - focused on communicating the author's intent in the context of the author's culture. This is what I imagine most of the translations in our library will be like. Linguistic translations are morpheme-by-morpheme translations, like what I did with "Pedithangid n'aduial." They aren't very easily read, and they're used for scholarly analysis. I could see this tool being used to analyze and present morphological datasets (comparing similar phrases to figure out what the morphemes are and what they mean.) I'll give it a try with the beta to show you what I mean. I composed a dataset for analyzing Japanese verb-classes and conjugation that I can use as an example.
|
|