Re: [edict-jmdict] POS tags

On Fri, Jun 9, 2017 at 1:56 AM, René Malenfant rene_malenfant@*********** [edict-jmdict] <edict-jmdict@***************> wrote:

I’m a little worried about this, because I think it will make the glosses a bit of an unwieldy (and redundant) mess, particularly for entries that already have multiple senses. (Here, “senses” meaning that the term means different things in different contexts, not that “senses” have been erected to match multiple PoS.)

And don’t forget that if you add “に” instead of “な” it turns into an adverb... It doesn’t seem consistent to argue that--if the noun is “haste”--the dictionary needs an additional adj-na sense to account for “hasty” but it doesn’t need an adv-ni sense to account for “hastily”. (It is assumed that the reader is smart enough to figure that adverb out for him/herself based on the adjective, but not smart enough to figure out the adjective based on the noun?)

I don’t know that the structure of entries is well-suited to implementing this well... Really, we shouldn’t be creating new “senses” for parts of speech at all. For instance, I think if you look at a typical GG5 entry, both 〜な and 〜に usages are mentioned within a sense, not given their own sense. I mean, ideally I think we should be trying to use “senses” to split up an entry based on the meaning of the word. Not PoS.

To be honest, I like the current system of only providing multiple senses based on PoS when it is not obvious how you would convert the noun to an adjective (etc.) Otherwise, perhaps changing the structure of the db so that PoS can be made subordinate to sense?

Rene

On Jun 7, 2017, at 11:02 PM, Jim Breen jimbreen@********* [edict-jmdict] <edict-jmdict@***************> wrote:

Something I have never been happy with is the way the glosses
and POS tags have worked out with 形容動詞. Of course I'm very
much to blame here for having set out on a course of having
minimal glosses back on the 90s, and never really setting up
clear guidelines.

In many cases the term is plainly an adjective and there is no problem
just giving it an "adj-na" POS and adjectival glosses. 静か is a good
example, and here 大辞林 (one of the few references with any POS
indications) says: （形動）［文］ナリ.

More of an issue are the terms that can be either nouns or adjectives.
In the past I've favoured giving them noun glosses and leaving it to
the user to derive adjectival ones where appropriate. Lately Robin
Scott has been pushing quite a few over to adjectival glosses, and in
some cases breaking them into two senses, and I have to say I rather
like the results.

A recent case is 柔和 which we currently have as: "(adj-na,n) gentleness;
mildness; meekness". Robin has proposed changing the glosses to:
"gentle/mild/meek/tender", which makes sense, but doesn't it mean
gentleness, etc. as well?

Looking at the references, 大辞林 says: "（名・形動）［文］ナリ". The JEs
mostly only mention adjective glosses (GG5 & 中辞典), but ルミナス
breaks it out into distinct sets of adjective and noun glosses.

I think it's time to bite the bullet, call a spade a spade, etc. and where
a 形容動詞 can function as both a noun and an adjective:
- break the glosses into two senses (or groups of senses); noun glosses
and adjectival glosses;
- order them according to which is used more commonly.

Thus for 柔和 we'd have:

(adj-na) gentle; mild; meek; tender
(n) gentleness; mildness; meekness

How does that sound? If people agree, I'll embed that in the
Editorial Policy page, and it can guide future edits. We have
about 5k entries with "adj-na" tags, and 4,300 have only one sense.

Cheers

Jim

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

Re: [edict-jmdict] POS tags - keiyoudoushi