LINGUIST List 6.685

Sun 14 May 1995

Disc: Indefinite Pronouns/Relativizers

Editor for this issue: <>

Directory

1. benji wald, indefinite pronouns
2. wachal robert s, relatives

I'm trying a new policy of actually trying to understand what a posting
says before responding to it. One drawback that has become obvious to
me, given my reading skills, is that the responses are hopelessly late.
Even so, I thought I'd give it a go.

I would like to make some comments on Harris's summary of his
relativizer/interrogative issue (posted sometime at the end of April).
I had thought about the issue for particular contexts. Since I
don't think I will ever address the issue as a whole in another
context, I would like to share my ideas with interested list members.

The basic issue, introduced by Harris, is: how widespread is the
formal identity between relative markers and analogous interrogatives?
I couldn't figure out WHY he wants to know, but no matter...

Harris states:
"I do know that those relativizers which relate to nouns and pronouns
often are not based on interrogatives."

To consider relativizers to be BASED on interrogatives, where there is
formal identity, is taking too much for granted. The alternative is that
there is a kind of INDEFINITE PRONOMINAL BASE from which both proceed.
In questions a more definite referent is asked for of that base. Otherwise
it remains indefinite, as in someone, something (= Northern British
sum'at ( someWHAT), someWHERE, etc. (I don't think someWHO ever
 developed in English, but the older forms are WHO etc. without the
SOME- prefix, cf. German WER kommt "someone/who's coming./?,
but etwas "something" ( it-what vs. was "what?")

When I last looked into the problem, a while ago, I noticed that Langacker's
de scriptions of Aztec fits this phenomenon. It also fits Indo-European,
and I think Semitic. I don't know the origin of the change in accent
in Greek to distinguish the two uses. It seems to be just a
difference in prosody which arose in Greek WH questions (as
opposed to, say, English where there is no necessary prosodic
difference between statements and WH questions).

As far as the evolution of this base into a general relativizer which can have
a DEFINITE head, the case I think Harris wants, I did not look into that for
Aztec. In Eurasia I concluded that such an extension from the indefinite use
was an areal phenomenon, affecting the Mediterranean at least, and spreading
out from there. It is neither Proto-Indo-E or Proto-Semitic, but certainly
it developed in Latin and Arabic to some extent, particularly in the context of
*complementisers*, e.g., post-QUAM and ante-QUAM in Latin (after/before +
adver bial clause), which are paralleled by Arabic ba?ada MA
and qabla MA (where Latin QU and Classical Arabic MA are also parts of the
interrogative WHAT).

The germ of the development of WH and its cognates into general relativisers
is in the I-E correlative constructions, still characteristic of the Indic
languages when relative clauses are called for, e.g., Hindi:

 JIS-ne yah pya:la: tor.a diya US lar.ke ko bulao
 WHO-by this cup broken given THAT boy Acc call
 "call the boy WHO(*ever) broke this cup"

cf. German wer...der... =
"WHOever plays, THAT one pays". Modern descendants have "the one WHO(*ever)
plays pays". Germanic used the *t correlative to *kw, and developed the
"the one THAT/WER (( *t) plays pays". The relationship between these
phenomena and word order changes is interesting for the I-E languages
affected. Even in Hindi (at least in some varieties) it is possible
to shuffle the correlatives in various ways, so that the "head" actually
becomes an antecedent:

 US lar.ke ko JIS-ne .... (that boy Acc WHO-by...)

I think this was already possible in Vedic. (a y-base was the
correlative corresponding to Hindi j-, it's not the IE *kw base).

There's something vaguely related to this in Lehmann's discussion of
relativisation in "Proto-Indo-European Syntax", but I haven't had the
patience yet to read it with understanding.

In the history of English, the development of WHO as a relativiser is usually
discussed as a generalisation from the INDEFINITE meaning, as in:
 WHO durst do this is a blankety-blank
More modern would be "ANYONE WHO ..." (more colloquial) or "WHOEVER..."
(The semantic development of this -EVER parasite is also interesting.)
)From such indefinite heads as "THE ONE WHO ..." then comes the generalisation
to definite heads "THE MAN WHO lives in that house" (we all may know as
well as not know WHO he is, whatever that means; is he still "indefinite" if
we don't know anything other than he's a man and he lives in that house?). I
forget whether any accounts of the development of English relate it to Latin,
where such relativisation developed earlier, but it has often been observed
that colloquial English prefers THAT as the relativiser with definite heads
 "THE MAN THAT..." (e.g., Suzanne Romaine's teething opus on Scots English).

WH adverbial complementisers remained indefinite in English until
Middle English, and remain so in German. So WHEN would mean WHENEVER,
just as WENN still does in German. A specific time would be THEN, as in:
 THEN (=WHEN) I played, (THEN) paid my backers (i.e., my backers paid)
This is German DA, still from the *t correlative, meaning "on THAT
particular occasion". Same for THERE () WHERE)... THERE... (where I played,
(there) I paid). Many German dialects, however, have generalised WO
(WHERE) not only as a relativiser for WHERE but for a definite nominal head.

In Harris' summary Ingria asserts:
"Note that even in English, interrogatives and relatives do not exactly
overlap in usage. ``what'' is used for interrogatives but not for
relatives; ``which'' is used in relatives but not interrogatives (it...."

Ignoring the hair-splitting on WHICH (( WHAT-LIKE correlative of
SUCH ( SO-LIKE), this takes a narrow view of English, to the extent
that it is not simply false. Generalised use of WHAT (for THAT)
as a relativiser is still a feature of some American dialects, though
rural and receding, and is very much alive in many many British
non-standard dialects, urban as well as rural,
"the man/house WHAT got wet..."

Harris presents the more conservative type of standard example:

"The house *where* Jack lives. The moment *when* she arrives. I will
show you *how* to do it. I will explain *why* it is done that
way. This is *what* we want."

The last example has the WHAT relativiser. It also shows that Harris'
interest goes beyond the "adverbial" contexts which he claimed to be
exclusively interested in, when complaining about the
off-the-mark-edness of most of the responses.

An issue which has long intrigued me is the role (indirect) embedded
WH questions play as a link between interrogative and relativiser uses
of WH in English. So alongside pairs like
WHO did it? and THAT'S WHO did it
we have
HE asked WHO did it and I (don't) know WHO did it
He asked WHAT I did (nonstandard)) he asked WHAT did I do

English (among other languages) makes a distinction between:
 he knows WHO did it and he knows the ONE WHO/THAT did it
The second does not necessarily mean that "he knows WHO did it", he
just knows that person -- I know that the person he knows did it, he
may not know. A lot of languages don't make that distinction; they
just use the "demonstrative pronoun" for embedded Qs, not the
interrog one.

I leave it to the expert syntacticians to try to figure out what I'm saying
about linking embedded questions to relative clauses, if anything.

To use a WH word in so-called embedded WH questions is far from universal.
However, it is much more widespread than formal identity between
interrogatives and relativizers. I think the necessary condition for
development of interrogative/relative formal identity is simply the
existence of formal identity between the interrogative and the corresponding
indefinite non-specific pronoun. The semantics relating the two seems so
transparent to me that I don't see what the issue is that various unrelated
languages could have this animal. Or that they could develop it.
That's as far as I want to take the issue now. Benji

I always though that 'what' was a nominal relative. The others cited by
Harris are all adverbial relatives, aren't they?
Bob Wachal