LINGUIST List 5.920

Fri 26 Aug 1994

Sum: Frequency of Coronals, Relative Clauses, Two Hands, Shoebox

Editor for this issue: <>

Directory

1. , Summary: Frequency of Coronals (and Nostratic)
2. Simon Kirby, Sum (and Q): Relative Clauses
3. , Addendum: Two Hands
4. "RANDY J. LAPOLLA", IT, Shoebox summary

In response to my query about whether there are languages with more
velars and labials than coronals, I received some encouraging
information, which is summarized below:

Lance Eccles, Macquarie University (lance.ecclesmq.edu.au)
suggested that (some) Australian languages might have this
property. And, indeed, Susan Fitzgerald, University of Victoria
(susanfsol.uvic.ca) counted the approx. 1550 entries in the
Ngarluma dictionary and came up with the following:

k 581
p 483
rt 264 (retroflex t)
t 38

(These figures are approximate. She didn't count two common
suffixes (-ku and -lku), which would have made k appear even more
frequent.)

She also tells me that preliminary counts by Geoff O'Grady for
small samples of four other Australian languages gave similar
results, although of the size of the sample may make these less
than certain.

Don Churma, Ball State (00DGCHURMAbsuvc.bsu.edu) suggested that
the total absence of coronal stops in Hawaiian would in itself
constitute an example, but I am not sure that typologically it is
correct to count nonoccurrence of a sound as (nearly) the same
thing as low frequency, which is why I did not think Hawaiian (and
Samoan) were good enough examples of what I was looking for.

As to what I was looking for, I recently counted the occurrences
(in initial position only) of the different stops in the Nostratic
reconstructions of Illich-Svitych and came up with the fact that
coronals are less frequent that labials and much less frequent than
velars (Manaster Ramer, in press). At the time I thought it was a
universal that coronals are always the most frequent class of stops
in a language (unless they do not occur at all, as in Hawaiian or
Samoan), and so I thought this might be an embarrassment for the
Nostratic reconstruction. It is still not clear to me that
Australian languages (where only one stop is found at each point of
articulation) are a good enough parallel for Nostratic, where three
phonation types (=airstream mechanisms) need to be distinguished at
each point of articulation (although it is controversial what these
were). But at least the frequencies I found are NOT in direct
violation of a universal. So perhaps Nostratic is not all that bad
a theory.


Hale, Kenneth L. (1982) _Ngarluma dictionary_, MIT (computer
printout))

Manaster Ramer, A. (1993) On Illi~-Svity~'s Nostratic Theory.
Studies in Language 17: 205-249.


Manaster Ramer, A. (in press) A "glottalic theory" of Nostratic.
In: Vitaly Shevoroshkin (ed.) Studies on Nostratic. Bochum:
Brockmeyer.

Manaster Ramer, A. (in preparation) Nostratic, an outsider's view.
Nostratic: Evidence and Status (eds. Brian Joseph and Joe Salmons).
John Benjamins.

Some time ago (5.482), I posted a query regarding constraints on
relative clauses across languages. I was particularly interested to
see if -- as I expected -- any constraints on the role of the coindexed
noun in the subordinate clause are independent of the role of the head
noun in the matrix. In other words, I was looking for confirmation of
the non-existence of a particular type of language (not an easy
task!). No such language has come to light, but I would like to thank
the following people for kindly replying, and commenting on my
request: Hubert Truckenbrodt, Kevin Donelly, Mihiko Kubota, Dave
Britain and Alan Hyun-Oak Kim.

Following on from this request, I would like to pose another question
and try to give a brief summary of my reasons for asking. (As a
footnote, if anyone who is interested is going to be at the LAGB
meeting in September, I will be talking about this stuff then.)

The question, simply put is:

*** Are there any languages in which the available types of FREE
*** RELATIVES (also known as 'nominal' or 'headless' relatives) is more or
*** less constrained dependent on the grammatical function of the relative
*** in the matrix? (Where "type" refers to the position of the trace in
*** the relative clause.)

An exemplifying language might, for example, have the following sort
of grammaticality pattern:

 Example Matrix Func. Relative Type

 [Who likes me] is well known. S S
* [Who I like] is well known. S O
 John hates [who likes me]. O S
 John hates [who I like]. O O

(This hypothetical language allows object free-relatives only in object
position.)

OK, so why do I want to know? Well, I'm particularly interested in
where the adaptive link between function and cross-linguistic
universal breaks down. I hypothesise that these limits on the
adaptability of language are imposed by innate meta-constraints on
acquisition. (This approach, therefore, relies both on functional
and innatist explanations for universals.)

One such example is provided by relative clause data: several
psycholinguistic studies of relative clauses suggest at least two
factors determining the ease of processing --- accessibility and
parallel function. The former is well known from cross-linguistic
studies (eg. by Keenan and Comrie). The latter one refers to the
function of the RC in the matrix. Any such constraint on relatives that is
dependent on their function in the matrix is inexpressible in the
current innatist grammatical theory. This explains why parallel
function does not show up in grammars, whereas accessibility does. Any
FREE-relative data would shed further light on the matter, since it is
not so clear that there is an expressibility constraint for these
sorts of relatives.

Anyway, thank you for any help, and sorry for the delay in replying,

Simon Kirby (simonling.ed.ac.uk)
Department of Linguistics
University of Edinburgh

Some time ago I asked for examples of languages with a word
denoting the two hands held together. I have since found out
that there are several examples in Dravidian (see Burrow and
Emeneau's Dravidian Etymological Dictionary), some in Uralic
(see Redei's Uralisches Etymologisches Worterbuch), and in
Sanskrit.

Many thanks to all those who responded to my question about FTP sites
for IT and Shoebox. Here is the response I got from SIL's Evan L. Antworth
(evan.antworthSIL.ORG):

>IT and Shoebox, as well as other SIL software is available by anonymous FTP
>from our own site, sil.org [198.213.4.1]. When you connect, I suggest you
>get the files 00readme.txt and 00index.txt first.
>
>Macintosh IT is available on our FTP site. However, there is no Mac version
>of Shoebox. There is a program called MacLex which manages lexicon files.
>However, I have not seriously tried to use it, and my initial impression is
>that it is a bit quirky. Also, it is only author-supported; i.e. it is not
>supported by SIL.
>
>If you have not discovered Conc, our Macintosh concordance program, I urge
>you to get it and try it. By far it's our most widely-used Mac program.

Randy LaPolla
Institute of History and Philology
Academia Sinica