Publishing Partner: Cambridge University Press CUP Extra Publisher Login
amazon logo
More Info

New from Oxford University Press!


Cognitive Literary Science

Edited by Michael Burke and Emily T. Troscianko

Cognitive Literary Science "Brings together researchers in cognitive-scientific fields and with literary backgrounds for a comprehensive look at cognition and literature."

New from Cambridge University Press!


Intonation and Prosodic Structure

By Caroline Féry

Intonation and Prosodic Structure "provides a state-of-the-art survey of intonation and prosodic structure."

Review of  Comparative Historical Dialectology

Reviewer: Herbert John Izzo
Book Title: Comparative Historical Dialectology
Book Author: Thomas D. Cravens
Publisher: John Benjamins
Linguistic Field(s): Historical Linguistics
Issue Number: 16.2749

Discuss this Review
Help on Posting

Date: Mon, 19 Sep 2005 20:08:36 -0400
From: Herbert Izzo
Subject: Comparative Historical Dialectology: Italo-Romance Clues to
Ibero-Romance Sound Change

AUTHOR: Cravens, Thomas D.
TITLE: Comparative Historical Dialectology
SUBTITLE: Italo-Romance Clues to Ibero-Romance Sound Change
SERIES: Amsterdam Studies in the Theory and History of Linguistic
PUBLISHER: John Benjamins Publishing Company
YEAR: 2002

Herbert J. Izzo, Professor Emeritus, Department of Linguistics,
University of Calgary, Canada

Neither the title nor the subtitle of this work gives a real indication of
its content. It is not about dialectology in the usual meaning of the
term, for it is concerned chiefly with standard Romance
languages. 'Two Problems in Romance Historical Phonology:
Intervocalic Voicing in Italy and Initial Palatalization in Spain' would
have been a somewhat more informative title, especially since
phonology alone is dealt with.

As his Introduction (Chapter I) accurately states, Cravens sets out to
explore two sets of sound changes in the Romance languages: the
voicing vs. retention of intervocalic Latin /p, t, k/ in Italy and the
palatalization of initial /l, n/ in Iberia. This exploration leads us through
quantities of relevant data, (many of them beyond the usual repertory
familiar to every Romanist) and many theoretical considerations, both
old and new.

Chapter II is a 25-page excursus on substratum influence, mostly
concerning the probability of Celtic and Basque influence on voicing in
Western Romance. Cravens is generally skeptical but reaches no firm
conclusions, which makes the whole discussion seem otiose,
considering that Chapter III argues that intervocalic voicing occurred
early in Latin itself. The evidence that Cravens adduces for early
voicing (mostly misspellings in inscriptions, graffiti, and early Medieval
texts) is not negligible and must be considered, but he takes no notice
of the strong counterevidence we find in Latin loanwords in Germanic
(cf. Latin catillus: Engl. kettle) and Latin loans both to and from Greek.

Chapter IV's insistence that the threat of homophony seems never to
have impeded a sound change is, I think, excessively prolix since it is
a long-established fact. Cravens' rejection -- correct in my opinion --
of various diachronic-structural arguments of Martinet and Weinrich
seems a trifle ironic in view of the fact that his whole approach
depends on the assumption that phonological pressures underlie
sound changes. ("This book attempts to demonstrate that [all the
sound shifts it discusses] are ultimately attributable to the loss of early
pan-Romance consonant gemination." [p.1])

Finally, although it may be only as a lone voice crying in the
wilderness, I must express my objection to the use of the
term "variable rule". If we say that Italian has 'strada' (< via strata)
but 'aneto' (< anetu) because the voicing rule is variable, we have
uttered an empty tautology: /-t-/ is voiced in the words in which it is
voiced and it remains voiceless in the words in which it is not voiced.
If, however, we try to find the conditions or reasons for different
outcomes of the same sound in different words (as in fact Cravens
does) we are not making the rule variable but rather more precise.
And this is simply to do normal historical phonology.
English 'was'/'were' do not show that rhotacism was variable in
Germanic. The outcome depended on the placement of stress. Nor
do 'gero'/'gestum' show that rhotacism was a variable rule in Latin.
Latin rhotacism occurred only intervocalically. But why did final /s/
become /r/ in Old Latin honos? By analogy to all the other forms of its
paradigm. That French 'peine' (< pena) and 'avoine' (< avena) show
two different results of Latin /e:/ was not the result of a variable rule
but the result of interdialectal borrowing. The apparent change of /r/
to /l/ in English 'belfry' was due not to a variable rule but to popular
etymology. Then there are cases where words do not undergo sound
changes because they were not in that language when the change
occurred. In Spanish the voicing of Latin intervocalic /p, t, k/ is a
completely regular, not variable change, yet there are hundreds of
Spanish words in which voicing did not occur because those words
were adopted long after voicing had occurred. The rule was not
variable but is simply no longer in effect, like a law that has been
repealed. A new and opposite change can occur, which may (in part
at least and without, of course, returning the whole system to its
previous state [consider the devoicing of Medieval Spanish voiced
sibilants, which restored the /s/ of 'casa', 'mesa', etc. but also
destroyed the contrast that had originally existed between Latin /s/
and /ss/ and was preserved in Medieval Spanish as a contrast
between /z/ and /s/.

So then, a regular sound shift may appear to be variable because it
operates only in certain environment (i.e., the statement of the
change, the "rule", is incorrect because it is too general [cf. Latin and
Germanic rhotacism or the Germanic Sound Shift]); or irregularities
may be introduced by analogy (Old Italian 'veggio' < 'VIDEO replaced
by 'vedo' on analogy to 'vede' < VIDIT, etc.) or anomalous forms may
be borrowed from a dialect that made a different (or no) change
(Fr. 'aveine/avoine').

Seeking such explanations for apparent exceptions to "sound laws"
has been fundamental to historical phonology for many decades. The
term "variable rule" seems like an abandonment of responsibility: "A
sound may change or not; there is no reason why it does one or the
other in different cases, the rule is variable" (i.e., there is no rule).

Cravens has been using "variable rule" in regard to the voicing of /p, t,
k/ in Tuscan for 25 years (the length of time he and I have disagreed
about it), but he has not used "variable rule" as a pretext to avoid
investigation of the causes of differing outcomes of the same sounds;
on the contrary, this book investigates carefully and creatively the
problems it attacks. Although I disagree with it in certain respects, I
consider this work essential reading for all who are seriously
concerned with Romance historical linguistics. Personally, it has
caused me to modify my own view of intervocalic voicing in Tuscan.


Herbert J. Izzo is Professor Emeritus of the University of Calgary. He
studied Romance languages and linguistics at the University of
Michigan as well as at the University of New Mexico and in Mexico
and Italy. He has been Visiting Professor at the University of
Michigan, Stanford, and University of Bucharest and has done dialect
research in Italy and Spain. He is currently a Visiting Scholar in
Classics at the University of Michigan.