David McCann, On 23/10/2010 13:14:
> On Fri, 2010-10-22 at 12:45 -0400, Jim Henry wrote:
>> In what grammatical category do "he", "she", and "it" differ, then?
>> They're the same in person, case, and number, and the difference
>> between them doesn't look like definiteness or any other category
>> usually associated with nouns/pronouns. Some linguists (I think I
>> first saw this in John Lyons, but I can't find the reference now)
>> consider gender in English a crypto-category, requiring pronoun
>> agreement but not adjective agreement and rarely marked on the noun
>> itself. But I gather that's a minority usage, given how few of the
>> Ghits for "crypto-category" have to do with linguistics rather than
> They differ in gender, and agree with the noun to which they refer.
> There's nothing odd about English here; it only looks odd compared to
> other European languages in that a noun's gender is almost always
> semantically determined. "Bull" is masculine, "cow" is feminine, "table"
> is neuter, "conglanger" varies. As I said previously, Dravidian
> languages are much the same, except that non-sentients are neuter. Pure
> semantic assignment can also be found in Africa and Australia. There are
> also a few other languages resembling English in only showing agreement
> in pronouns, such as Zande.
> There have been a few linguists who don't consider pronominal anaphora
> to be agreement, but they don't have any convincing arguments.
The answer to Jim's question depends on the larger question one is asking. The typological question is "What analytical framework best describes all languages?". The ontological question is "What knowledge must one have if one is to know language X?" You seem to be asking the typological question and providing an analytical apparatus intended to describe as many languages as possible. I ask the ontological question, and my answer to Jim was trying to establish what a knower of English must actually know. For English, under the ontological question, the arguments against gender and against pronominal anaphora being agreement boil down to the application of Occam's razor: there is simply no need for gender or agreement, and these things can be deleted from the inventory of what must be known.
There's a kind of ObConlang point to be made here too, which is that the two approaches to language description described above pertain to description of conlangs too. The typological approach yields a description of the conlang, where the description can be very detailed or very sketchy and impressionistic. The ontological approach yields a *definition* of the conlang (and the definition may differ in how complete it is). I think engelangers generally feel themselves to be creating definitions, whereas artlangers generally (but not universally) feel themselves to be creating descriptions.