Transliteration, the は / wa particle

[Sixtease]

Hi everyone. I'm programming a tool for transliterating Hiragana and Katakana into Czech transliteration (which is a bit different from English transliteration as Czech has a richer letter repository). The task is quite simple (or so it seems so far), but there is one problem I don't know how to solve.
The particle は is read as 'wa' although in other occurances, the は syllable is really 'ha'. Is there an algorithm or some heuristics for determining when it is that particle and when not?
The method I use so far is that I check for は to be a separate word but that depends on writing spaces, which I know is something I can't rely on in Japanese text.
Any ideas?

 

[Mike Cash]

Don't forget you're also going to have the same problem with the particle へ.

 

[undrentide]

Ahoj Sixtease!
I'm now trying to think how we recognize the particle は when reading Japanese texts.
Particle は is placed immediately after noun or nominal, which should not be too difficult, but then it can also be used with another particle (e.g. とは のは) which can be confusing...
へ is the same "problem", as Mike Cash pointed out.

Does this kind of idea help at all?

 

[hkBattousai]

1) You must check all は, if the preceding word is a noun. To do that, you must have a word dictionary for all nouns.

2) Put spaces between all words before the entire progress, then convert all 'は's to 'wa'. Again you must have a dictionary to recognize meaningful words.

 

[Nall-ohki]

I'm afraid you've got your work cut out for you - a primitive tool such as this is not going to be able to parse a language syntactically.

What you are trying to do is something that is still extremely hard to do for natural language even for advanced projects. Seriously, we're talking about REALLY difficult.

At best, you can do something like try to transliterate a は that does not appear as part of another noun, but this will fail in a lot of real-world cases. If you want, I can craft some for you.

 

[Sixtease]

Thanks for the answers.
So it's not trivial to do, that's what I get... I hoped for a simple heuristic that would succeed in most cases and it may be the preceding-noun thing but I don't have a dictionary of nouns and I think it's way beyond this primitive tool... What crosses my mind is that the tool is designed for the people attending a certain Japanese class, so I can safely restrict myself to the nouns we learned so far
Thanks again.

 

[lonesoullost3]

It's important to note that it may not always be a noun (or nominal depending what terminology you use - the latter coming from from Eleanor Jorden's BJ and JSL textbook series). The は may also come after a verb~て if it's more formal writing (while in speech this will be converted into a ちゃ). Sorry to make things harder. Best of luck on this project though.

 

[Elizabeth]

It's important to note that it may not always be a noun (or nominal depending what terminology you use - the latter coming from from Eleanor Jorden's BJ and JSL textbook series). The は may also come after a verb~て if it's more formal writing (while in speech this will be converted into a ちゃ). Sorry to make things harder. Best of luck on this project though.

Contracted in casual speech and writing, not all by any extent.

は is used in describing comparisons as well so theoretically it can be used after any part of speech. [Suru] ha [desu] yori yoku tsukaimasu. (I use "suru" more than "desu") as one example.

 

[qianyuxuan]

1) You must check all は, if the preceding word is a noun. To do that, you must have a word dictionary for all nouns.
2) Put spaces between all words before the entire progress, then convert all 'は's to 'wa'. Again you must have a dictionary to recognize meaningful words.

Why don't you do the reverse?
Assume that は is wa unless you can find it in your noun, verb, adj and others list, that proved it as ha?
And if you use tree, think it should be quite ok?
just a 3-yen suggestion!