epepke wrote:I've always been a fan of deep case grammars (a la Fillmore) and HPSG to represent them (though I think the HPSG people have gotten a bit, shall we say, elaborate.) I'm not so convinced by the Language of Thought hypothesis. It seems to me that is of the class of ideas that sounds good in general, but the proof of the pudding is in the eating.

Of course, these tools may just have been the best I could find for what I was trying to do, which was language-based user interfaces. In a computer, the problem of internal representation of knowledge has already been solved (with any luck), and it's a matter of translating it to and from a human language. So I viewed language generation and understanding as an algorithmic process, with a conversational context. This seems to me to work well to describe some fairly common things about language, such as when pronouns are used. For example:

1) Fred opened the door, and a bucket of water fell on his head.
2) A bucket of water fell on his head, and Fred opened the door.

are both fine syntactically, but in terms of conveying meaning, they're rather different.

Anyway, our friend mawanli has caused me to wonder if deep cases and HPSG would be good for Chinese, or if something else might be better.

epepke wrote:That's actually quite easy. Time-stamping and -tracking are pretty common these days.

mawanli wrote:

epepke wrote:That's actually quite easy. Time-stamping and -tracking are pretty common these days.

Is what you mean that computer begin to understand the tense?

epepke wrote:I was interested in translating from the information stored in a computer to English (and Spanish and German) utterances for the purposes of user interfaces. Mostly it was for automatically generating responses and menu choices for adventure games. The time information is quite easy to store in a computer. Translating to tenses is another issue, but the basic information is there.

I established some deep tenses and translated them to the target language. Note that English has more ways of expressing time than tense. (Of course, actually, there is no future tense in English; we use the helper verbs "will" and "shall.")

Also note that in English there are lots of ways of dealing with time. For example:

1) I go to the store. (from time to time, or habitually)
2) I am going to the store. (present continuous or immanent or specific future)
3) I went to the store. (simple past)
4) I had gone to the store. (I went to the store before doing something else)
5) I have gone to the store. (completion of an action)
6) I just went to the store. (very recent past)
7) I will/shall be going to the store. (future, maybe continuous, maybe coinciding with something else)
8) I will/shall go to the store (future)

Of course, that's not an exhaustive list, and there are many ways to indicate time that don't have much to do with the verb (such as "Tomorrow I go to the store" or "So, last week, I'm going to the store, when I see a spotted dog.")

I had to use deep tenses and translating them because I needed the program to say different things. For example, if you try to open a door and it is locked, the program would say "The door is locked." If, however, you just locked it, it would say "the door is now locked." (The word "say" was sometimes literal, as I used the language of a speech synthesizer. The ability to do emphasis, which is better than plain text interfaces, as I could indicate emphasis from the deep structure.)

In any event, representing the basic information in the computer was easy. It's information that is readily available or adaptible that you need to write the program anyway. Writing the language generator was a lot harder.

Loren Michael wrote:I think that, to the extent that a language is good or bad depending on its efficiency in accurate communication, Chinese is extremely bad except insofar as it's relatively compact on a page relative to non-ideogrammatic languages. I think Korean is about as compact on a page, but they actually use an alphabet. Korean gets the best of both worlds. If only their social hierarchy wasn't so thickly and inextricably integrated into their language.

I don't know that it's possible to actually quantify how much brainspace a language uses relative to its effectiveness in communication, but Chinese can't be worth it.

Zwaarddijk wrote:

Loren Michael wrote:I think that, to the extent that a language is good or bad depending on its efficiency in accurate communication, Chinese is extremely bad except insofar as it's relatively compact on a page relative to non-ideogrammatic languages. I think Korean is about as compact on a page, but they actually use an alphabet. Korean gets the best of both worlds. If only their social hierarchy wasn't so thickly and inextricably integrated into their language.

I don't know that it's possible to actually quantify how much brainspace a language uses relative to its effectiveness in communication, but Chinese can't be worth it.

Of course, one needs to keep in mind that the written form of a language is a later thing, an accident of the language. The Chinese spoken language quite likely is no better or worse than any other language. The written language has some extra challenges that few other written languages need to deal with - it really is a written language for a huge number of divergent dialects. Most other written languages basically just correspond (to some extent) to one or a few prestige forms of the language. (English might be an exception in that its written form doesn't really have a one-to-one correspondence to any form of English any longer.)