Re: [edict-jmdict] Re: いい/よい POS matters

On Sat, Jul 26, 2014 at 3:22 AM, Jim Breen jimbreen@********* [edict-jmdict] <edict-jmdict@***************> wrote:

(Re)joining this thread after some hours stuck in the medical system.
I'm going to split the emails into three threads as I think the topics
and solutions differ a bit.

On 25 July 2014 14:21, Stuart McGraw smcg4191@******** [edict-jmdict]
<edict-jmdict@***************> wrote:
> On 07/24/2014 09:37 PM, René Malenfant rene_malenfant@*********** [edict-jmdict] wrote:
>> I think it’s the overwhelming majority of Japanese sources that say
>> it don’t inflect, though (e.g., every dictionary I have access to).
>> I agree that it’s an “irregular adjective” (of course), but if it
>> really needs its own PoS to show that it is irregular, that PoS
>> should have no inflections at all, except for いい. よくない and the others
>> belong to よい.

I am with René on this point. I can see why JSL teachers say
いい is irregular; in fact I would probably do it myself if I were to
be teaching Japanese. But it's really a linguistic sleight-of-hand
because giving the real explanation about 終止形 and 連体形 forms
would be messy and confusing.

> If you are saying that the conjugation tables in the JMdictDB
> submission pages should not conjugate いい, I am fine with
> changing them (a trivial change) to make it so (assuming there
> is a consensus on this).

Having a POS other than "adj-i" for the ~いい entries would
certainly fix it.

> But いい (and friends) should still have a unique PoS because not
> every consumer of JMdict will agree that いい is not conjugatable.
> Someone who wishes to provide a web site for language learners
> that does provides its conjugations should be able to do so easily.
>
> Those who chose to present it as non-conjugatable can simply
> ignore the adj-ii (or whatever) tag, or map it to adj-f, or act
> only on the adj-f tag (if it has both).
>
> Giving it a unique PoS serves both points of view (with the
> appropriate explanatory text for the PoS tag.)

I had thought that tagging いい and ~いい entries as "adj-f" and
having xrefs to the matching ~よい entries would do it. The problem
is that we currently define our "adj-f" as "noun or verb acting prenominally",
which they are not. We could either broaden the adj-f definition, or bring
in a "adj-ix" POS for non-inflecting i-adjectives (of which there is just one.)
Maybe the latter is cleaner. It would signal to learners not to try for いかった.

I have proposed a split of the basic よい/いい entry with "adj-f" on the
いい part. I'll hold fire on other splits until we sort out the POS.

Jim

--
Jim Breen
Adjunct Snr Research Fellow, Japanese Studies Centre, Monash University

--
Francis Bond <http://www3.ntu.edu.sg/home/fcbond/>
Division of Linguistics and Multilingual Studies
Nanyang Technological University