LINGUIST List 7.280

Thu Feb 22 1996

Sum: R/L Change

Editor for this issue: Ljuba Veselinova <lveselinemunix.emich.edu>

Directory

1. mcconvell_puncl04.ntu.edu.au, R/L change: summary

 Late last year I posted a query about r's changing to l's and
vice versa, and environments favouring r or l in such changes. The
following responded: Heriberto Arelino, Ken Beesley, James Fidelholtz,
Spike Gildea, Luise Hercus, Frank Lichtenberk, Carl Mills, Rebecca
Larch Moreton, Niels Schiller, Jane Simpson, Larry Trask, and Jeroen
van der Weijer.
 The query arose from work Mary Laughren and I are doing about an
apparent change from *r to *rl in Ngumpin-Yapa, a Pama-Nyungan
subgroup in Australia. (r is a retroflex glide, distinct from the
tap/flap rr also found in most Australian languages, and rl is a
retroflex lateral distinct from l, an alveolar lateral; see also
McGregor 1988 for an argument against treating r and rr as co-members
of a "rhotic" class). It seems that the Ngumpin-Yapa r>rl change
occurred in all environments except a preceding i (i/a/u 3 vowel
systems are found in most of the relevant languages).
 In general r>l and l>r seem to be among the commonest sound
changes in language, and varieties of r and l act as allophones of a
single phoneme in various languages. In Australia too, the r/rl change
we are working on is far from isolated, and appears to go in the
opposite direction too. In the non-Pama-Nyungan language family
Tangkic (Gulf of Carpentaria), for instance, proto-Tangkic *rl has
been retained in Yukulta, but replaced by r in Kayardild and Lardil,
except where it is a coda preceding d (in both languages), and
initially (in Lardil); in Lardil, even recent loanwords are affected
by this change and the situation approaches phonemic merger of r and
rl (these are my own interpretations based on data from Ken Hale, Nick
Evans, Sandra Keen et al).
 Proto-Romance *l appears as r occasionally in some Romance
languages e.g. in Portuguese following bilabial stops. In Otomapean
languages l systematically corresponds to r in Chichimeco. The list
could go on, no doubt.
 As to the affinity of i and r, Jeroen van de Weijer's article
(1995) comes closest to investigating the problem I had in mind with
comparative data. He argues that the contrast between liquids
(laterals and rhotics) is one of continuancy: l's pattern with
non-continuants, and r's with continuants; as the phonological data do
not appear to bear this out - there is not evidence of continuancy
values spreading to neighbouring segments - he proposes a different
solution, which will not be explored here as it seems to have little
bearing on the present question.
 However the article does present a range of data which seem to be
relevant. In a number of Bantu languages, according to Bhat (1976), r
has a lateral allophone before either i, or front vowels in general,
and in Usurufa, after i and before i,e,a. On the other hand in
Maddieson's example languages (1984) the lateral allophone occurs
following central and back vowels - more in accord with the Australian
examples I raised. The Ganda case of r appearing after front vowels
and l elsewhere, also conforms to this pattern, which is regarded by
van der Weijer as "typical". Other cases show r intervocalically, l
initially and/or as coda (recalling Lardil above). Kwaio was also
drawn to my attention: it has l and r (alveolar flap) in complementary
distribution, with r following the high vowels i and u - a somewhat
different picture from the examples in van der Weijer.
 While the data are not uniform, there seems on balance to be a
tendency for laterals to favour a preceding relatively low/back vowel
environment, and rhotics a preceding relatively high/front environment
across languages, paralleling the Ngumpin-Yapa situation. If so it
would be good to find a reason for it, and articulatory phonetics
might be one place to look. Respondents did not directly address this
question and my (admittedly limited and inexpert) reading of phonetics
texts did not yield much either. I suspect that high front tongue
position (in the preceding vowel) acts to curb the full opening of a
lateral airway.
 Niels Schiller who is working on phonetic variation in German /R/
provided an interesting draft paper and bibliography of which I have
selected three items which may be relevant to my query for the
references below, but I cannot vouch for it as I have not seen them
yet.

References

 Bhat, D.N.S. (1978) A general study of palatalization. In
J.H.Greenberg ed. Universals of Human language. 47-92.
Stanford UP.
 McGregor, W.(1988) On the status of the feature rhotic in
some languages of the north-west of Australia. Aboriginal
Linguistics 1: 166-187.
 Maddieson, I (1984) Patterns of sound. Cambridge UP.
 Miyawaki, K, A. Liberman, O. Fujimura, W. Strange and J.
Jenkins (1975) Cross-language study of the perception of the
F3 cue for [r] versus [l] in speech and nonspeech-like
patterns. In G. Fant and A.Tatham eds. Auditory Analysis and
perception of speech. Academic Press.
 Stemberger, J.P. (1983) The nature of /r/ and /l/ in
English: evidence from speech errors. Journal of Phonetics.
11:139-147.
 van der Weijer, J. (1995) Continuancy in liquids and in
obstruents. Lingua 96: 45-61.
 Walsh, L. (1995) The phonology of liquids. Ph.D.
dissertation, U. Massachusetts Amherst.