LINGUIST List 2.798

Sat 16 Nov 1991

Disc: Inflection/Derivation; Finite Languages

Editor for this issue: <>

Directory

1. , Re: 2.770 Queries
2. , Inflection/derivation
3. Rick Wojcik, Re: Inflection/Derivation
4. , Infinite Language

Message 1: Re: 2.770 Queries

To John Nerbonne about the order of inflection and derivation in words:
I am not sure about Russian s'/s'a. Maybe it is an enclitic of some
kind. The clearest counterevidence to derivation occurring closer to the stem
and inflection at the edges seems to me to occur in some Native American
languages, particularly in Athapaskan, e.g. Navajo, and in Siouan, e.g.
Lakota. It might not be insignificant that both these language families are
heavily prefixing. Even with a very unsophisticated theory of what the
difference between inflection and derivation is, the counterexamples are
quite striking. In the 1992 LSA meeting there will be several papers on
Inflection inside Derivation, one by Philip LeSourd on Passamaquoddy, and
one on Bengali by Shila Baksi; I assume this recent interest in the
phenomenon has to do with one's precise theory of the distinction between
derivation and inflection. I don't know the facts in these languages, but
they might or might not be as crystal-clear as in Athapaskan or Siouan.
Willem J. de Reuse
Dept. of Anthropology
University of Arizona
Tucson, AZ 85721

I think that John Nerbonne is right in suspecting that the Russian Reflexive
suffix -s'/-sja is an exception to Greenberg's universal (derivation inside
inflection). If it were a true reflexive, one might say it is inflectional
after all (as David Stampe suggests). However, -sja is used as a reflexive
only in typically reflexive situations like 'wash oneself, turn (oneself)
around', etc. (Cf. on this topic Emma Geniusiene. 1987. Typology of Reflexives.
Berlin: Mouton, and Suzanne Kemmer's forthcoming book The Middle Voice,
Amsterdam: Benjamins) Russian -sja has often anticausative meaning, as in
otkryt'-sja 'open (intr.)' (from otkryt' 'open (tr.)'), and quite commonly
the meaning is lexically idiosyncratic.
 However, such an exception to Greenberg's universal is devastating only if
one's explanation of the universal requires it to be an absolute universal.
Bybee 1985 (Morphology. Amsterdam: Benjamins) proposes to account for this
universal in diachronic terms: Derivation is inside inflection because it
arises from the grammaticalization of elements that are closest to the host
(there's more to this explanation, but I won't go into details here). The
grammaticalization of a reflexive pronoun as a verbal category is something
rather unusual (most of the time verbal derivational categories come from
auxiliary verbs, apparently), so the result is also unusual. Incidentally, this
is catured by Slavists by means of the term postfix (post-inflectional suffix),
which is applied most prototypically to -sja.
 If the diachronic explanation is correct but the derivation-inside-inflection
constraint is also synchronically valid, then one might expect speakers to
do something about this unfavorable situation. Although I know nothing about
such a development in Russian, there are some interesting facts from the
genetically and (even more) typologically closely related Lithuanian. The
Lithuanian postfix corresponding to Russian -sja is -si, e.g. kelia 'raises',
kelia-si 'rises', kelia-me-s(i) 'we rise'. In some Lithuanian dialects the
order of -si and subject agreement affixes has been reversed, so that we
get kelia-si-me 'we rise'. See further Bybee 1985:40 on the reordering of
derivational postfixes, and my article "The grammaticization of passive
morphology" (Studies in Language 14 (1990):25-72, esp. p.42-46 and 52-53).
Martin Haspelmath, Free University of Berlin

This is just to point out that the so-called reflexive s'a in Russian has both
inflectional and derivational interpretations. I.M. Pulkina's A Short Russian
Reference Grammar (3rd ed.) lists 6 classes:
 1) true reflexive: odevat' "to dress" --> odevat's'a "to dress oneself"
 2) reciprocal: vstretit' "to meet (tr.)" --> vstretit's'a "to meet (intr.)"
 3) passive: stroit' "to build" --> stroit's'a "to be built"
 4) new word: dobit' "to finish off" vs. dobit's'a "to achieve"
 5) not used without s'a: *bojat' vs. bojat's'a "to fear"
 6) impersonal: mne xoc^ets'a "I would like..." (lit. "to me is wanted...")
One can dispute these classifications, but it is clear that s'a qualifies as
a true derivational suffix in Russian in at least some of its uses. So it
appears to contradict Greenberg's universal (although I don't think that it
represents a counterexample in any theoretical sense, since Greenberg was not
proposing any theoretical generalizations).
			-Rick Wojcik (rwojcikatc.boeing.com)

I wonder if it would help people who are puzzled by the issue of
the size of natural language (like Bruce Nevin, in his latest posting)
if we considered analogies to other areas of human knowledge and
behavior.
On the one hand, it seems to me that I know what makes a natural number
in the decimal notation, no matter how long, even in practice there
may all sorts of limits on the exercise of this knowledge. So, I would
say that in theory I know an infinite set of natural numbers.
On the other hand, there is a strict limit on the lengths of the
words I know of an extinct Uto-Aztecan language variously called
Giamina and Omomil, because there is just a brief list recorded
by Kroeger and another one by Harrington, and there will never be
any more.
Now, it seems to me perfectly reasonable to inquire (although
difficult to answer) whether my knowledge of various aspects
of English is more like the first case or more like the second.
Or something else entirely or something in between or ...
And note that the answer may be different for different aspects
of a language and perhaps for different speakers.