Re: how does a language develop gender??

On 30 Mar 2005 04:13:26 -0800, sanlosinst@xxxxxxxxxxx wrote:

>sonjaaa@xxxxxxxxx wrote:
>> I'd love to examine cases where a language has developped gender
>where
>> it didn't have it before. How does a language develop gender?
>
>I'm taking a wild guess, but I imagine that a word with a
>gender-specific connection is grammaticised and becomes an affix.
>
>Hamito-Semitic typically has feminines in _-t_. One could wildly
>speculate that in Pre-Proto-Hamito-Semitic, a noun something like
>_*atu_ with a meaning something like '*woman' was grammaticised as a
>feminine suffix, so _*ba?al atu_ '*lord woman' could become _*ba?alatu_
>'*lady'.

This sort of thing can happen: PIE, for instance, probably
had a word *sor- "woman", which is found used in Hittite as
a means of deriving feminine nouns from masculines (e.g.
hassu- "king", hassusara- "queen"), and has left a few
traces elsewhere in IE (*swe-sor "sister", perhaps Lat. uxor
"wife"). However, it is not a true feminine marker in
Hittite, merely a word-formation device. The PIE feminine
marker, which the Anatolian languages (such as Hittite) may
have lost, or which was only created after the break-off of
Anatolian, is athematic *-ih2, thematic *-e-h2. This suffix
is in my view connected with the diminutive suffix *-ik-.

I think Afro-Asiatic offers a parallel case. The suffix
*-at- has three basic uses in the Afro-Asiatic languages:
1) singulative (e.g. Arabic s^ajaru "trees", s^ajaratu one
tree")
2) diminutive (in e.g. Chadic and Cushitic)
3) feminine

If the original sense of the feminine suffix *-at- was a
diminutive/singulative, that would explain some of the
peculiarities of the feminine in Semitic.

>(I stress that these are not supposed to accurate
>reconstructions of PPHS forms.) Of course, this would not explain
>Hebrew _hem_ 'you plural masc' versus _hen_ 'you plural fem'. I forget
>my Semitic verb inflections, but I seem to recall that 3S fem has _-t_
>opposed to 3S masc _-0_, indicating the same origin.

The 3sg.f. ending -t belongs in the stative/permansive
suffix-conjugation, which is nominal in origin (still so in
Akkadian, e.g. s^arr-a:ku "I am king").

The distinction in the plural personal pronouns between
masculine -m and feminine -n is secondary. The true
difference is in the vocalism, as can be seen in the
Akkadian forms 2pl.m.(obl.) kunu-, 2pl.f.(obl.) kina-,
3pl.m. s^unu, 3pl.f. s^ina. The ending -un of the
masculine forms gave -um, while feminine -in kept the
original nasal. The plural marker (independent of gender)
in the personal pronouns is *-n, which is added to a nom.pl.
base (-u) in the masc. (ku-n, s^u-n), but to an oblique
plural base (-i) for the feminine (ki-n, s^i-n).

For the further understanding of the following, there are
two essential clarifications to make:
1) PAA had two distinct suffixes *-at-, one a singulative
(diminutive, feminine) originally added to substantives
only; the other a plural suffix *-at- (*-ut-), added to
pronouns, adjectives and numerals.
2) In absolute final position, *-t gave *-n (and, we can
assume, *-p > *-m, and *-k > *-ng). For parallels, cf. the
same soundlaw which is found independently in Eskimo-Aleut
and independently in Northern Samoyed.

The peculiarities of the feminine in Semitic that I alluded
to above are in the first place the lack of nunation (c.q.
mimation) in the feminine plural definite *-a:tu, obl.
*-a:ti. The nunation is derived from an old definite
article of pronominal origin, featuring the same suffix *-n
that is found in the plural of the personal pronouns (s^un,
s^in). Besides masc. -u:n, -i:n, one would expect feminine
*-a:tun, *-a:tin, but the plural marking is absent. That
suggests that the creation of a feminine plural postdates
the agglutination of the plural article.

The second peculiarity is the so-called "gender inversion"
with numerals 3-10, e.g. Arabic xamsatu rija:lin "5 men" vs.
xamsu bana:tin "five girls". If the feminine was originally
a singulative, it had no plural, of course. That explains
the lack of nunation in the definite, and the use of an
unextended (singular) numeral, used as noun, followed by the
feminine noun in the genitive (obviously a genitive singular
originally), literally "five (instances) of girl". The
apparently "feminine" suffix -atu on the numeral used in
conjunction with a masculine noun is in fact the pronominal
plural suffix *-at-/*-ut- (the numeral used as an
adjective), followed by the masculine noun in the plural
(originally not necessarily in the genitive).

The general trend in the further developments in Semitic was
for adjectives and numerals to become less like pronouns and
more like substantives, and for the feminine to acquire a
full-fledged inflection, also in the plural (and dual). In
Arabic, adjectives have by and large maintained their old
plural in -atu, -ati (reinterpreted as feminine singular),
always so in concord with inanimate plurals, optionally with
animate plurals. With animate plurals, the adjectives can
also show a substantival plural in -u:/-i:, a "broken"
plural, or the new feminine plural in -a:tu, -a:ti. The use
of -atu as a plural marker on nouns (e.g. di:ku => di:katu
"rooster"), is perhaps attributable to substantivized
adjectives. Many feminine nouns in Arabic drop -at- in the
plural, a direct reflection of the former singulative
semantics of the suffix.

Akkadian adjectives maintain the pronominal plural, in its
*-ut-variant, but have adapted the suffix to the
substantival plurals by lengthening the vowel: -u:ti, obl.
-u:ti, construct *-u:n > *-u:m > -u:. (It is perhaps
interesting to note that these "new" adjectival and feminine
plurals (-a:tu, -u:tu) are formed according the same
procedure as the -u:/-i: plurals and the majority of
"broken" plurals, which is by lengthening the vowel of the
"posttonic" syllable. For bi-consontal roots:
CVCu => CVCu: [besides other patterns]
For most tri-consonantal nominal roots:
CV'C(V)Cu => C(V)CV:Cu [wáladu => (?a)wlá:du]
For "end-stressed" tri-consonantal roots:
C(V)CV(:)'Cu => C(V)CVCu: [kabi:ru => kabi:ru:]

In North-West Semitic, there is to my knowledge no
distinction anymore between substantival and adjectival
plurals.

In the numerals, the replacement of original *<xamsatu
rija:lu(m)> by <xamsatu rija:lin> is also indicative of a
shift from adjective-like to noun-like behaviour of the
numerals, even if the archaic "gender polarity" (as it was
reinterpreted) was largely maintained.

The feminine acquired a plural -a:tu, -a:ti (construct *-a:n
[attested in Aramaic, and secondarily giving the Akkadian
and Arabic plural patterns in -a:nu, -a:ni]), and a dual
-ata:, -atay, although its former lack of plural forms (as
expected for a singulative) is still to be seen in the lack
of nunation on the plural, and in the lack of plural concord
(reinterpreted as gender polarity) with numerals.

The other Afro-Asiatic data confirm the original lack of a
plural feminine. Chadic and Berber have generalized the
construct plural in *-n in the pronouns, and the pattern is
masc.sg. (zero marking), fem. sg. -t (t-), pl. -n. There is
no separate feminine plural. In Egyptian, the feminine
plural is either identical to the masculine plural (dropping
-t) or is made by tacking -t onto the masculine plural in
-(a)w, clearly a secondary development. Cushitic has
generalized the pronominal plural marker *-t, and the
pattern is masc.sg. (zero marking), fem.sg. -t, pl. -t,
again without a separate feminine plural (the confusion of
nominal singulative *-at- and pronominal plural *-at- of
course gave rise to "polarity" phenomena similar to the ones
in the Semitic numerals).

=======================
Miguel Carrasquer Vidal
mcv@xxxxxx
.

Quantcast