LINGUIST List 6.914

Thu Jun 29 1995

Sum: Representing retroflex

Editor for this issue: Ann Dizdar <>


  1. , Sum: Representing Retroflex

Message 1: Sum: Representing Retroflex

Date: Tue, 27 Jun 1995 21:13:00 Sum: Representing Retroflex
From: <>
Subject: Sum: Representing Retroflex

 Summary: Representing Retroflex

 At the beginning of the month I posted a query that asked for
 information on an alternative way of representing retroflexed segments
 as dorsalized coronals instead of [-anterior]. Warm thanks to those
 who responded:

 Philip Hamilton
 Richard Desrochers desrochrERE.UMontreal.CA
 Suzanne Urbanczyk
 Mark Verhijde
 Stig Eliasson

 Here is a list of annotated references followed by contributors'
 excerpts on the following topics:

 1. Against Dorsalization
 2. Origins of Coronal Domination of Retroflex: Sanskrit evidence
 3. Phonetic features of Retroflex

 1994. Course notes from the 1994 Australian Linguistic

 CHO, Y.
 1990. Parameters of Consonantal Assimilation.
 PhD thesis Stanford

 About Sanskrit retroflexed elements. The claim made here is that
 retroflexation equals the formation of segments that contain two place
 nodes, COR and DOR, i.e. what some have defined as a "complex" place.
 Interestingly, in assuming retroflexed segments as having two place
 nodes, some neutralization effects at right word edges fall out quite
 naturally. (Verhijde)

 No title given.
 1980. Languages of Australia

 Assumes a feature [+retroflex], as does Hamilton's 1993 Toronto paper.

 1986. Sandhi in Peninsular Scandinavian.
 In: Henning Andersen (ed.), Sandhi phenomena in the languages of Europe
 271-300. Berlin: Mouton de Gruyter.

 Postalveolarization or retroflexion is a most important sandhi process
 in Swedish and Norwegian, and the major part of the above article
 is devoted to that problem. (Eliasson)

 No title given.
 NELS 24

 1993. The feature geometry of coronal subplaces.
 University of Massachusetts Occasional Papers in Linguistics 1993

 She argues against the feature [anterior] for defining coronals.
 I believe that retroflexes are represented as [-distributed]
 [+back] where [back] is dorsal. (Urbanczyk)

 1993. No title given

 Paper on Coronal articulation

 1993. No title given
 Toronto Working Papers in Linguistics 1993.

 1991. Coronal places of articulation
 in The Special Status of Coronals, Paradis and Prunet, eds.

 Phonetic clues on coronal articulations

 1993:179, citing Kirchner's University of Maryland MA thesis.

 1. Against Dorsalization

 Hamilton: I am currently working on a paper where I argue against
 dorsalisation more fully, based on a variety of evidence: retroflexes
 are transparent to +back vowel harmony; all of the evidence for
 interaction between retroflexes and back vowels is from very low level
 phonetic facts (there are no lexical alternations backing front
 vowels: /rti/ going to [rtu]) and there is never _neutralisation_ of a
 lexical back/front contrast conditioned by retroflexes, all that is
 attested is that front vowels have backed allophones when beside a
 retroflex; retroflexes are based represented with a feature dependent
 on an apical node, since the lack of heteroganic apical clusters may
 be elegantly expressed with an OCP constraint on adjacent apical

 2. Origins of Coronal domination of retroflex: Sanskrit

 Wechsler: The presently-orthodox account of retroflection being
 dominated by the coronal node receives a lot of its support from the
 Sankrit "rnati" rule. In this rule, if I recall it correctly, n -> rn
 anywhere to the right of a retroflex consonant, but intervening
 non-nasal alveolars block the rule. This kind of interaction between
 retroflection and the coronal node appears in other places as well;
 the one I've studied is in Warlpiri, where historically there was a
 rule that partially unretroflexed a retroflex stop unless it was
 closely followed by another retroflex.

 There are other reasons why you might want to avoid involvement
 with the dorsal tier -- all the vowels live there, and you would have
 to explain why they are transparent to assimilations involving
 anteriority. a

 3. Phonetic origins

 Desrocher: Ladefoged (1974 [1971]: Preliminaries...) speaks of
 retroflexes (RXs) as apical postalveolar and gives the example of Ewe.
 He adds: "In some South Asian languages the retroflex consonant
 involve only the tip of the tongue and the back of the alveolar ridge,
 whereas in others there is contact between a large part of the
 underside of the tongue tip and much of the forward part of the hard
 palate" and elsewhere, he speaks of the "extremely retroflex sounds
 which occur in some Indo-Aryan languages" (Hindi, Gujerati,
 Penjabi,and so on, I guess) and when characterizing everything with
 the SPE features, describes RXs as [-ant, +cor, +high, -back, -low,

 SPE refers to Zwicky (1965, his Dissert.) as describing
 convincingly Sanskrit s. as [-ant] (actually, [-comp]) and SPE seems
 to favor the natural class apicals + RXs [-dist] as opposed to
 laminals + non-RXs [+dist]. They refer for these matters to Ladefoged
 1964 A phonetic study of W-Afr Languages, and maintain that
 distinction between dentals and RXs support a [dist] feature.

 Malmberg (1974, Manuel de phonetique generale) writes that RXs
 are produced with the tongue markedly curved backwards towards the hard
 palate, but his diagram, as Ladefoged's, indicates that this the very
 front of the palate, or the back of the ridge, that is touched by the
 apex, and mentions South-Italians dialects and of course, India.

 Hockett 1958 makes an interesting comment: he says that the same
 acoustical effect than in the RX in "bird" is achieved by some English
 speakers not by curling back the tip, but by a "peculiar contour of the
 central part of the tongue, the tip being held behind the lowe teeth".

 Further comments, corrections, and questions welcome
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue