<OT> The integrated responses of the /r/~/z/ alternation

[Lee, Amy P]

Dear all,

First we would like to thank those who have contributed what they know
to this topic.  All the responses we have received so far is now
integrated as follows:
	Based on the responses this alternation, or the similar ones,
can also be found in some other Indo-European languages (Latin,
Germanic, French, Slavic, Irish, etc.) and perhaps Mandarin.  Most of
the cases are involved with sound change.
	In the history of French, as Kathryn Tippetts notes, there was a
time when the /r/, a uvular fricative, was replaced by the dental
fricative, such as 'Paris' [pazi] instead of [pari].  Another example is
'chaise' (chair), showing the historical interaction between English and
French.  English borrowed the word before the sound change occurred.
Modern French has retained the /z/ form.
	Rob O'Connor mentions about the 'broad/slender /r/' distinction
in Irish.  The /r/ in Irish is trilled, unlike the flap or silent /r/ in
English.  The slender 'r' approaches the sound of 'z' and in some
dialects it has almost become 'z'.  He suggests the acoustic similarity
between the trill /z/ and /r/.  The slender 'r' only occurs in
word-medial and final position, and is in complementary distribution
with the broad 'r', which occurs in word-initial position.  This
broad/slender 'r' is also considered as velarised/palatalised 'r',
respectively, in e.g. Antony Dubach Green (2000).  Andrew Spencer,
however, regards this as a common result of palatalization.
	The Slavic examples in Czech and Polish were given by Maria
Gouskova, Nathan Senders, and Andrew Spencer.  In Polish the [r] becomes
a palatal fricative [zh] before a front/high vowel.  N. Sanders provides
this example: [mur] 'wall (nom sg)' and [muZe] '(loc sg)'.  In Czech, as
Andrew Spencer notes, 'palatalized trilled /r/ becomes a strange trilled
alveopalatal fricative without the trill component', which is what
Townsend and Janda describe as a 'zh and r rendered simultaneously':
reference given by M. Gouskova: (p. 236, Townsend and Janda 1996 Common
and Comparative Slavic: Phonology and Inflection. Columbus: Slavica).
	Latin is mentioned by Elizabeth Pyatt, David Stampe, and
Christian Uffmann.  Examples such as 'flos' (flower-nom) and 'floris'
(flower-gen), 'plus' and 'plural' (*plus-al in pre-Latin, intervocalic
s> *z > r), 'vetusia' and 'vetoria' (an example of vowel lowering).
This 'rhotacism' is seen in Italic dialects affecting coda: Latin 'quis'
(who) > Umbrian 'pir'.  Latin displays different direction of sound
change from that of Gothic: 'fiskaz' < fiskr (Old Icelandic) (r >z) (D.
Stampe).
	Also mentioned is the 'were' ~ 'was' alternation in English,
which is connected with the Germanic 'Verner's law' (E. Pyatt).  The
English example is derived as 'were' > OE waere > *waese.  And in 'us'~
'our', where 'our' > OE ure > *use (C. Uffmann).
	Geoffrey S. Nathan suggests that the English hypocoristic names,
e.g. Barry~Baz, may not be phonological, as 'British English has a
tendency to attach a plural'.  Examples include 'Babs', 'Mags', 'Chaz',
and the pet rat/familiar in Harry Potter: Scabbers.  This 'plural'
hypocoristic is also seen in 'walkies' and 'preggers'. (I'm not
convinced by this suggestion--notice that the /z/</r/ forms may also be
extended by schwa e.g. Sharon --> shaz --> shazza etc.  Babza sounds
totally unnatural to me.  Only names with intervocalic /r/ are affected
by this and the process is productive.--WJ)
	This alternation is also found in Spanish/Catalan: dos besos >
dor besos (two kisses) -- reference to Joaquin Romero by C. Uffmann.
Another example is free variation in Bolivian Spanish [pizara]~[pizaza]
(jg).
	Moira Yip and A. Spencer mentions the possible influence in
Mandarin, where the initial /r/ as in 'ren' is fricated and is now being
a fricative.  (a retroflex /z/?)
	C. Uffmann gives a Niger-Congo example: 'white person' [musungu]
in Kiswahili and [murungu] in ChiShona.  He also suggests that the
process, as discussed so far, is more diachronic than synchronic.
	A final note. Paul Boersma explains the articulatory similarity
between the two segments, and provides examples regarding sound change
between Latin and Germanic languages.  For example, in early Latin the
intervocalic <s> is realized as [r] in classical Latin: kwaiso > quaero
and Latin 'soror' > English 'sister'.  He suggests that sound change in
both direction (/r/ > /z/ or /z/ > /r/) is possible. 

Wyn Johnson (wyn@essex.ac.uk)
Amy Peijung Lee (aplee@essex.ac.uk)