Linguistic Change by Rule Generalization As Illustrated By
-- - National Li bcary- - E iblioth-e nationale - - - CANADIAN T-HEf ESP- TH&FS-&YAD~~S 1Je ofCanada . duCanada ON MtcRoFlcHE sua MicRoFIc \
-- J
TITLE OF ThESIS/?ITRE DE LA TH~SE Linqkist ic Change by Rule General izat ion as Illustrated by Portuguese ~Ssalization
i .P* IVERSITY/UNIVERSI T,! Simon Fraser University
; EAR THIS DEGREE CONFERRED/ANN,& D'OBTENTION D•’C€ GRADE 9
LAME ,OF SUPERVIStaih'OM DU DlREC TEUR DE THZSE D r . a F0 l e'~
Permission is hereby granted%to the NATIONAL LIBRARY OF L'autorisation est; par la prgsente, accdrdde 3 la 618~10~~~-
t CANADA to microfilm this thesis and to lend or sell copies QUE NATIONALE DU CANADA. de microfilmer cette these et of the film. -de pr&er ou de vendre des exemplaires du film.
Tne author reserves other publication rights, and neither the L'auteur se rgserve ' les autres droits de publicati~n;ni la " -_ raests nor extmsive extracts from it may be printed or other- th&seni,de longs extraits de celle-ci ne doivent stre ~mprimt+s
UYISF: reproduced without the author's written permission.' w eutrement reproduit sans l'autorisation Bcrite de /'auteur. 1 , t National L~braryof CaMda &Wioth&que nationale du -,a Collections Development Branch Direction du developpement des collections
NOTICE
"The quality bf this microfiche is heavily dep ent La qualit6 de cette microfiche depend grandement de upon the quality of the original thesis submi for la qualit6 de la the m~miseau microfilmage. Nous microfilming. Every effort has been made to ensure ' avons tout fait pour assurer une ' qualit6 suphrieure the highest quality of reproduction possible. de reproduction.
--ppp - - -- ppp------pp ------If pages are missing, contact the university which S'il manque des pages, veuillez communiquer granted the degree. avec I'universitk qui a confer6 le grade. - I Some pages may Rave indistinct print especially La qualite d'impression de certaines pages peut if the original pages were typed with a poor typewriter laisser a dksirer, surtout si les pages originales ont etk ribbon or if the university sent us a poor photocopy. dactylographiees B I'aide d'un ruban us6 ou si I'univer- site nous a fait parvenir une photocopie de mauvaise qualite.
Previously copyrighted materials (journal articles, Les documents qui font d6jB I'objet d'un droit
pubiished_teas,-excLarenatfilmed 3-p- -p-pp--d'auteur- (articles de revue, Emens publies, etc.) nee sont pas microfilmCs.
I Reproduction in full or in part of this film is gov- La reproduction, m6me partielle, de ce microfilm erned by the Canadian Copyright Act, R.S.C. 1970, est soumise a la Loi canadienne sur le droit d'auteur, c. C-30. Please read the authorization forms which SRC 1970, c. C-30. Veuillez prendre connaissance des accompany this thesis. formules d'autotkation qui accompagnent cette thkse.
THIS DISSERTATION LA THESE A' ETE HAS BEEN MICROFILMED MICROFILMEE TELLE QUE
EXACTLY AS RECEIVED NOUS L'AVONS RECUE *
Ottawa, Canada KIA ON4 , LINGUISTIC CHANGE BY -RULE GENERALIZATION AS ILLUSTRATED BY , PORTUGUESE NASALIZATION AND RADICAL VOWEL ALTERNATIONS
Paul HcFetridge a,
THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUlREMEWS FOR THE DEGREE OF
.'L MASTER-OF ARTS 8
in the. Department
7 I' , Languages, Literatures and L4nguist ics
@ Paul McPetridge 1981 SIMON FRASER UNIVERSITY
1 --reproduced in whole or' in part, by photocopy P or other means, without pemssion of the author. I hereby grant to Sirpdn Fraser university the right to lend
my thesis or dissertation (the title of which ahown i s below) _ to users \ - *. of :he Simcm Fraser University Library, and to make partial or single
copies only for such users or in response to a request from the library I - of any other university, or other educakional inLtitution, on its own I
- - - -A - - - behalf or for -one af-its users. I fu__r_ther.agree that permissi- for _ -- .- --
. t multiple copying of this thesis for scholarly purposes may be granted
by or the Dean of Graduate Studies. It is understood that copying
------or publication of this- thesis for ffnancial gain shall not be allowed
without my wit ten permission.
Title of Tnesis/~issertation:
Linguistic- Chancre by Rule Garali7lti'nn
and Radical Vowel A1 ternat ions,
Author : f / e / '(signature)
(date) Name: Paul Gil bert McFetri dge
Degree: Master of Arts
Title of Thesis: LingOistic Change by Rule Generalization as
Illustrated by Portuguese Nasalization
- - Vowel A1 ternat ions .
Chai man: Ross Saunders
Thomas A. Perry ,
Edward Finegan External Examiner Associate Professor Depa'rtment of Linguistics University of Southern California
Date approved : 3l @Wt" ", In the theory of phonoloa fondlated by J, Foley, it is,
maintained that !lphonologicai change is not caused by the < addition of rules to a gramnar, but by the repetition or
modif ication of rules already existing in the languagew. o ole^, , "The Latip origin of Romance rules". ) It is !maintained tha.t . ,,/ \
------linguistic chance, change inprocesses operatingin the - --
language, occurs in three ways; generalization of application to * I 5. include more elements, generalization of environment, and
-- - - - repe- pfureeSses.~rconcept- XrIlnguis tic change is t applied to the analysis of two aspects of Portuguese phonology:
nasalization and radical vowel a1,ternations.
Chapter I presents a brief outline of the conception of
linguistic change as generalization of rules which exist in the
language. It is demonstrated that denial of the traditional -- +------synchronic/diachronic distinction permits analyses which
simultaneously explain data traditionally classified as
synchronic as well as the linguistic change from historically
earlier stages of the language by reference to the same set of - - - 1) - processes and principles, I *
In Chapter 11, nasal diphthongs and the retention of
preconsonantal nasals are examined. The latter is theoretically
preconsonantaloef fare. Tt is nrvu- d retention is only apparent, the result of generalization of - --
Latin vocalization which causes a glide to remain after nasal
effacempt. This glide either fqrms a nasal diphthong withuthe
preceding vowel or,consonantalizes before a stop. This is further related to the vocalization of -1 and the palatalization of -s. 1 a - - -- In Chapter 3, the radical vowel aTternations of the t-----. Portuguese verb are examined. Analyses by Williams and Harris,
which propose rules of assfmilat ion, are considered, but
F because it uses morphologically sensitive-and language
particular rules. It iB argued that the alternations are caused
by the generalization of Latin contraction. c he contract ion of .-' - dissimilar vowels causes weakening of the radical vowel in
proportion to tJre strength required for contraction. ~vidence
from Old Portuguese is ad,duced in support of this claim.,
It is concluded that Portuguese rules are the result of
continuing linguistic change generalizing Latin rules. TABLE OF CONTENTS
Approval ...... ,...... ii
Abstract ......
a I. Introduction :.,...... l
11. Nasalization ...... ,.9
Summary ...... 25 . - - - 111. Radical Vowel Alternations in the - - - Portuguese Verb ...... 28 , Sumnafl ...... 55 . t
Bibliography ...... ,..'...... :.....65 - 1 a f - I. Znt roduct ion
Wrfting in 1906, Meillet argued that linguistic change is , -I neither a&aberr?tion nor a collectipn of arbitrary unrelated * -. .- f events,, but a, continuing coherent pfoce*ss.' He maintained that -- 4 ' i_ k explanation .of &his must refer to universal "t imeless,-'linguistic
laws".. i * It - -- I 13~-UC~e%ontinlti6-~~&volut is= -- -- Lnguistiqu=, et cette continuit6 r&v&lela - constance des causes qui dgterminent les i modalit& du changement. -"---- Les changements linguistiques ne prennent leur
sens que si l'on consid6re tout l'ensemble du # diiveloppement dont ils font. partie; un &me changement a une signification absolutkment differente suivant le proces dont il relave, et il n'est jamafs lggitime d'essayer d'expliquer un d6tail en dehors de la -- -- ;------consid6ration du syst&me g6n6ral de la langue oh il apparait.
D;?S lors la nCcessit6 stimposer%e chercher 2 formuler les lois suivant lesquelles sont susceptibles de s'opsrer les changements linguistiques. On determinera alnsi, -non plus ' Y des lois historiques,. telles que sont les <
, study of linguistic change and the study of a particular
synchronic stage on a language are separate fields of inquiry.
Ontological and methodological priority has been placed on m synchrony and diachrony has becdme the coAPpariaon of synchronic
continuing trends in language, their explanation, nor the - discovery of universal phonological. processes. The ------synchronic/diachronic distinction legitimizes the ahlysis of a
% particular language as an isolated entity, unrela-ted to other
languages or even to its histdrical' precursor. AS a result,
b rules formulated for a particular language are typically
unattested in languages of the world. Moreover, there-is .no
necessary reason why a particular language should share a cbrnmon
set of rules wi& its historical precusor.
% Theoretical Phonology, the theory of phonology proposed by
, . 2 Professor J. Foley(1970, 1971, 1972,. . 1975, 1977, 1979) , denies
the synchronic/diachronfc, distinction maintaining-that the
processes and principles which explaiq alternationsbperceived in .6 synchronic grarnnars are also those which explain historical
developrents. These processes are universal processes. They are,
------moreover, phpnologicalprocesTes, not 'To~~ques,".- The
heillet(1965: pg, 11). . 'see ekpecially Poley(1977) for the arguments for the parameters used in this work, -- Aements on w hi r h th~yowe--eaA-
3 elements, not phonetic elements.
The Inertial Development Principle (IDP), which governs the
application of phonological processes on phonological elements,
states that strong elements strengthen preferentially in strong
environrnerts and weak elements weaken preferentially in weakI I ------environments. It follows that the "primordial" rules are those - -. -
which strengthen the strongest element in a strong environment- - . oP weaken the weake%t element in a. weak environment. *.
The IDP predicts that if a language has the rule
then it also has the rule
------:see Foley(1977: pg. 28). The superscript '-' indicates a weakened element,
a because h is stronger than t. Therefore, it ig predicted that if . '-' - A the primordial rule k-->k 3eneralizes, it will generalize first - to include -t and-later to4 It will* never generalize to 2 5 unless it has first-generalized to -t . . J * Theoretical Phonology e&fimr*e-5- continuing trends of b
- - - -- linguistic change perceived thecoherent- - - - - &- \ g2neralization of rules governed by the IDP. Ln general, . .
strengthening ~rocessesgeneralize to-apply to weaker elemenm - - ---rc;- - y
-- - - - &ftR-*mm&Fii~sy'we;tk g-m-ra E------&
C 1 i a apply to stronger elements and/or in strong environments, i .: 4 As illustratiq of the difference between theories which V; maintain the synchronic/diachronic distinction and\ Theoretical-
Phonology, a simple example from Greek is considered.
22 Koutsoudas(l962) in ."
------a ' as [grafo] 'I write'
the underlying stem *~raf-and the rule Y 4 - !9 3 ------2C This prediction holds only for those languages in which the Alpha parameter is manifested as above. The Romance lan-wages are examples of such languages. In languages such as German and - 'English, the Alpha parameter is manifested as kPt
' -1 -23 -- In these languages, it is predicted,that the primordial rule will not generalize to include t unl-s it first generalizes to . include 1. See Foley(1977: pp. 48-52)for further discussion. A similar analysis & the ~nci6ntGreek 'glternation .of
9 [gra(p,h)o]6 #I,writet and "[grapso] '1 will write' would reveal ,' the underlying stem *~rg(p,h)- and the rule ..
i 0
>- - - I- The historical change of [gra(p,h)bl > [grafo] is
w' 4
- This rule never appears in the synchronic grammar of Modern * reek', I* must cAnsequently be conluded that it has 'been
"morphologized". Thus, the synchronic/diachronic distinction
necessitates postulating two different underlying-stems, a v "morphologized" diachronic rule ,-and two unrelated synchronic
rules.
The Theoretical solutlon maintains that the underlying stem
-. is *&ra(p,h)- in both languages. The Ancient Greek forms are
derived as
gra(p, h)so grapso (p, h)&>p/-~
~he'~odernGreek forms are derived as
-I------6 The symbol (p,h) indicates the aspirated labial stop. . 1 This process is interpreted as the conipression, of diphthongs and
-- $ ---"------
I
which ranges oGer. the Gamma parameter8. .
The process increases the bond strength of diphthongs from .- , Gamal to Ga1mna2. Ancie Greek contains a manifektation of the 5 4 same process in the rule . . *
(k ,w)->p. [ep-i] , Latin sequor Y s
The crucial dsfferencc between Akcient reek and Modern Greek is 1 'the application of this process in Ancient Greek to diphthongs n f -. containing u but diphthongs -cpntaining h in Modern Greek. As - $ - I ( represented by the Alpha parameter, -w is stronger than -h:
The process is reformulated 3s
M , Parochial Conditions: Am3 for Ancient Greek All far Modern Greek
In consonance with the IDP, this process applies preferentially
% B- to diphthongs containing the strongest glide. Represented in
4 0 this manner,+t is evident that the rule ("p,h)-->f is not an
\ of gen&alization of1a rule already in the language. In Ancient
Greek, it applies'to diphthongs containing the strongest glide. 1
B InoModern Greek, it has generalized to apply to diphthongs
- containing weaker glides. 9 Whereas Koutsoudas' solution accounted' for only an
alternation in Modern Greek and used a rule unrelated to th#t in
Ancient Greek, the Theoretical solution explains a constellation
------of facts; the alternations of Modern Greek and Ancient Greek,
and the change from Ancient Greek to Modem Greek. It represents ' . linguistic change as coherent and non-arbitrary, as the generalAation of a r~;e in the direction predict~dby the IDP.
The rules used in this solutions are also found in other - languages. The compression of diphthongs is found in Latin fero, 9 Sanskrit'bhara:mi, and in English father, Latin pater . The rule elidlng the glide element of a diphthong before a consonant is found in Latin vectum < *ve(g,h.)tum, first singular veho: , cf.
/ In the following chapters aspects of Portuguese nasalization/ and radica5b60wel alternations are analyzed. .
peculiar to Portuguese, it is argued that they are, in fact, the result of generalization of Latin rules. It is maintained that more general and explanatory analyses are made available when it
F. u is recognized that these phenomena are the result of = gen&alization of rules already existing in the language. To paraphrase Meillet, observed phonetic changes or alternations
4 have meaning only when their place in a gene~alsystem of,_ processes is understood. 11. Nasalization
This chapter presents an analysis of two aspects of
Portuguese nasalization: the formation of nasal diphthongs and .
* the differential application of nasal effazement. It is argued -- - f 6 *- that these seemingly disparate phenomena- are in fact s * manifestations of the same process.
- nasal effacement applies, preferkiTially to syllable final 11- nasals . For example; syllable final -m $effaces in French but A,
-7
L a intervocalic m does not: champ [W], Theoretical: phonology * - e., --i - ~xplains~thisfact as the preferential &facement of elements id
weak position over elements in Strong position. Syllable final
2 C before effacement in intervocalic (syllable initial) position . . . ^L According to the IDP, the following rule ~onfigurat~ionsare
allowed : - It is predicted by the IDP that the following rule configuration does not exist:
4g 1 .for this configuration maintains chat n is effaced in strong --. a ( I,i post-ion but not in a weak position.:
Q In'Portuguese, intervocalic nasals efface: *
However, pr,econsonantal nasals do not appear'to efface3. .
*
pont em ponte [pant@] '7 tantum tanto [tl!ntu] - ' : dentem dente . [dZnt@] t / I.
------. . e 3~hesibbol '@' is used throughout to designate schwa. . .. This rule configuration is the one which the theory predicts does not exist. It appears iniqially that this is a counterexample to the claims of the theory.
In response to this problem Foley writes 4 .
. - "sinc&.i&ervocalic (heterosyllabic) nasal . , effacement implies preconsonantal (tautosyllabic) effacement, if nasal effacement occurs in -manum > -m2o then it must also occur in centum > *s??ntu. Thebphonetic
nasal in cento [sZ%tuJ therefore cannot be the ------. ------7 okianaT, bii$-~Kst mesent post ef f acement epenthesisi" . I D
Po% this explanation to be complete, an explanation of epenthesis is necessary. This is especially so as epenthesis normally-occurs in consonant clusters: e.g. Latin simila:re,
Portuguese sembrar (< semblar'< *semlar). -T f
- - As-Mefl3~t-correctiy -no t ,-2-ppfrorveT1rc hangTZQihotbe -- -p-----p- understood when viewed as an isolated phenomenon. To understand the insertion of a nasal between as nasal vowel and a stop, it is neceksary to relate the development of nasals to tnat of other elements. In the following, the dev,elopments of Portuguese
A -1 and -s are considered. The,. motivation. for this procedure is
# their proximity on the Rho par~mter5.. where -t represents stops, -s sibilants, -n kasals, and -1 liquids.
proxidl elements are expected to +have similarly.
In ~ortuguese,as well as other Western Romance languages,
LATIN PORT. SPANISH FRENCH alterum outro otro autre f alee f ouci? hoz , faux ------enTwra =soscr' - ---= - palpa :re poupar PoPar
rple initially appears
that this rule is derived from a rule applying in Latin. $he evidence for this comes from the present
tense of the Latin verb velle:
volo : volumus I * vi :s vultis 4 vult volunt.
r- the r- and tafwel in -vi:s. ------6~hefollowing data is tak& from Boyd-Barman( 19%). --- Foley Wes tbit-6633hTre attributable to vocalization of -1 and 7 posits the rule
Z The character of the glide is apparently determined by the , preceding vowel. After vocalization,"€he radical vowel
assimilates to the epenthetie glide. The liquid onset -of (1,y) - is lost before -s; otherwise the glide elementx is elided. The length of the vowel is attributed to the subsequent contraction r :* ------pp----p - d ------?------of 9.The derivations of vi:s and -vult illustrate these rules:
ve l+s vol+t ve(l,y)+s vo(1 ,w)+t vocalization vi (1,y)+s vu(1 ,w)+t assirnibtion 11 t -- viys 1-->0/> s -= It - vult (1,w)-->l It vi :s contraction ,
The vocalization rule is reinterpreted as a strengthening process, followed by depotentiation of strengthened -1 as .a consonantal, dfphthong 8..
In this manner it-is possible to' relate Latin vocalization of -1
------with Englfsh vocalization of -1 between a consonant and a
h he symbol (1,y) indicates a segment with a liquid onset and glide offset. The superscript '+' indicates a strengthened element. ------of strengthening; however in Latin the strengthened -1 , depotentiates as a diphthokg, whereas in glish strengthened -1
depotentiates as [@1]: +' 1--1,w in iatin I -->@Iin English
- = C - The ~Lmancerule vocalizing -1 is the r&&t of
generalization of two rules. The first is the rule strengthening r 1 before a morpheme boundary and a consohant. This rule ------generaliz'es in Romance so that the morpheme boundary is nd , - longer required :
Second, the rule eliding the liquid onset of the resulting
as well as -s: L-
1-I- s ==> (1,w)->w/ -C
The relevant part of the derivation of Portuguese outro is:.
1 a4 tro a1 troL a(1,w)tro awtro ------In isolation, the vdcalization of -1 has no bearing on the -, problem of the retention of syllable final nasals. It is
necessary to consider the behavior of preconsonantals. In most Portuguese dialects, though in no other Romance language, -s palatalizes +fore consonants9:
'. - LATIN* PORT. - /----
vesti:r vestir [v@&tir] - --t4 s chola escola [@gkol@] audculto - escuto [@gkutu]
From the fact that intervocalic dental nasals efface, but
intervocalic palatal nasals do not, it is concluded that palatal
elements are stronger than their dental count-erparts:
LATIN PORTUGUESE Uao
he ref ore, the rule
describes a strengthening prqcess: i-
I The following data is from Williams(1962). According to
~illis(1962:pg. 447), in the dialect of Sao Paulo. s is-. not palatalized before consonants. The symbol '@' is,used throughout to t Finally, note that palatal -8 is he reflex of 9:
Latin russeum, Portuguese roxo *[ro•˜u]lo. This suggests that the * - vocalization of -1 and the palatalization of -s are the result of a the same process: strengthening before consonants with
I depotentiation as a diphthong: ------?
I The hifferent manifestations .of these diphthongs are due to i A the generalization of the Latin rule eliding 1 before a - 'i consonant. The diphthong (s,y) contracts to-the pala / \ Compare the derivations of vestir and outro. -- \;a -9-- -- I e B ve s$i r a4 tro* + I ves tir a4 tro CeC/ - C - ve(s,y)tir It s+->(s,Y) 3 I1 a(l ,w)tro 1 -->(l,w) 11 awtro (1 ,w)->w vegtir otro contraction
1
L
The theory 'predicts that if a weak element strengthens, \ I then all stronger elements strengthen in the same environment. a ------
------a. losee Foley ( 1977 : Chap 6) who explains the palatization of assibilated reflexes as contraction of z. Thus-- the-- rule- configufatih -
violates the IDP, for nasals are'stronger than s as . " - ' on the Rhp iarameteh f
It is concluded that Portuguese includes a single process 4 1 . which strengthens consonants of a Rho value greater than or L equal to 2 before other consonant?: i------
Parochial Condition:n=Z f6r Portpguese 1
.Since both -1 and -s depotentiate as consonantal diphthongs, -n also depotentiates as a consonantal diphthong, (n,y). Additional evidence for the strengthening of -n is found in the raising of the preceding vowel in, for example, Latin
C
- development-of stressed short 5 and 2 is Romance open -e and -o respectively: Latin petram, Portuguese pedra; ati in fortem, Portuguese fgrte, However, when short -e and -o are followed by a
------I, - _ tautosyllabic nasal they raise to cloaed -e and -o respectively".
L -, The vocalization of the nasal provides an immediate explanation
-for the raising of these vowels. As is the case in the
derivation of Latin vult, the preceding vowel assimilates to the 3 - 12 follo~in'~glide . b C ~rasington(l971), ~lmeida(l976)and ~illiams(l962)posit 2r: - * - - i -. A- - rules of progressive nasalization. This is especially evident in . - forms such as muito [mw&tu] from Latinamultum. ~rasingtoniand Almeida argue for such a rule on th grounds that the ------element of a nasal diphthong is als nasal, It is maintained
here that the phonetic nasal is the reflex of this nasalized
glide. That is, the glide consonantalizes when followed by a t stbp. The derivation of ponte illustrates the vocalization of y the nasal and the subsequent consonantalization of the nasalized 1 ;i ------glide, -- - -- 1 - - - . i . nasal+ assi n->n , + n -->(n,y> y-->w/o-* assimilation n-->8/-% *nasal assimilation consonantalization mt->nt Evidence 'for this claim is found in forms su
------[tPnd@s] 'you (pl.) have'. From the Latin
evident that Portuguese tendes is-not deri
8 vowel which is to be elided is stressed. Rather., after nasal '5' / effacement and contraction of the vowels, the iadical vowel, strengthened by nasalization, contraction and stress, n ,- - diphthongizes. The na&1 glide consonantalizes producing the /'
LL - phonetic nasal : % -
tijned@s nasal 'assimilation------p---p------L - -- - 7-7- T$ed@sP n-->0 n t&d@s contraction tZyd@s diphthongization ti5~ddZss nasalization t?hd@s c,onsonantalizati.on
Effacement of the nasal and contraction are also'found in the -
------t ractt-io~xp~a~i~t~rof reas the
traditional solution claims that the phonetic nasal was inserted A.p' JI after Antraction, the analysis presented* here dxplains why it
Mppears, A similar analysis seems available for forms such as -muito which have a nasal consonant where none previously existed. r 1 his anlysis illustrates how recognition of historical
developments provides solutions to theoretical problems, In ------p-T particular, it is argued that the process of vocalization has, ------generalized from application to only 1, as in Latin and early
*Romance, to application tor, nasals, and -s fn Portuguese. After- Z asal effacement, the remaining glide nasaliz,es and - 'p bnsonantalizes. -. 6 P In contrast to the analysis of Brasington(l971) wherein it
is sufficient to posit a rule which elides intervocalic nasals
but not preconsonantal nasals, a rule configuratbn predicted by ,
- L - - " ------
the IDP not- to eKtst--,--the is forced to - * . a ' explaic the anomalous retention of preconsdfantal nasals. The
.s . result is a deeper analysis yhich reveals the coherent and
. -- ,---7--- -- cdntinuing geaerali;ation of a Latin rule which by interaction '3. with other rules.produces the Portuguese phonetic nasals. L
'Although formulated to explain the apparent retention of
preconsonatal nasa-ls, this, analysis also explains the formation Y of nasal diphthongs. The phenomeqa,of nasal diphthongs is * .4 ------Z especially qident in'the formation of the n~ularand plural . nouns 14..
------141%e following data is taken from Brasington( 1971). It is representative of the dialect of Rio de Janeiro. I ' ------eao fkRil ' dq , caes [ leg1 .do+
m Pa0 1P-1 bread paes IP?B ? breads capital [kapitaw] capital capitais [kapitayg J capitals pais [paiBI . c9ntry paises [paizill ] countries la llm wool . - las [ 1m I wools
0
- - . - A" A L - a .Brasington(l971) and St. Clair(1971) posit a stem final vowel 15
which is elided after 1, I,:, z, and n. This vowel is to - - -- -
This underlying vowel is also present "in the diphthongs formed in the plural forms such as -caes and-capitais. The derivations & these forms are:
kanes kapitales kEnes 11 nasal
The loss of the stem final-vowel after -1, -r, 2, and -n, but retention when not final is also the philological solution to
i a these alternations:- 1 !,'1f e was the final letter of the word in
------A- P*
final e. ,The 'difference inmaterial as St, Claik also uses a r& which raies e to i when unstressed. Herein, the underlying vowel is taken to Ee -e Zs this reflects the historical fact. .vowel. - -- -s or c. or the -nroup *- t+i preceded by a it feil: so:lem > sol, ...venit > vem, ... amo:rem > amor, ...-facit > -6az ... If -e was not the final letter. it did not fall: although it was preceded by -1, -n, r, 2, and -c, or the group t+i: so:les > sois, vie-es .,, -16 vezes, ,;.amo:res > amores, ...-canes > -caes" . The ruld for the elision of stem final -e is
-- . where -r refers to the the Rho value of the preceding consonant.
The problem is now to determine how the diphthong in, for rn
caes, and that e elides after -n indicates that the diphthong
cannot be formed in the same manner as that of caes.
Brasington proposes the rule
That is, nasal vowels followed by a word final -n diphthongize. 17 Williams appears to posit the same rule. However, this , solution maintains that unstressd vowels diphthongize, for it is
also used to explain the diphthongs of the third plural forms
*. r where the final vowel is unstressed: Cf. levam [lgv&], movem
[m~v~~],Since unstressed vowels do not normally diphthongize,
diphthongization of final nasal vowels appears to be a
1 L '"~illiams(1962: pg. 47). 7~illiams(1962 : pg 34) C 18 . Theoretically suspect solution . ------r * The form capital [kapitaw] indicates that the rule for 'the -. vocaiization of -1 also applies in word final posit'ion. Moreover,. -s .paJatalizes in word final position: Cf . pais [paigl, paise6 I [paizig]. 'It is-concldded that the process strengthening s and 1 * - - has further generalized in environment to include a final word , boundary. Since -s vocalizes in word final position, it is
a predicted by the IDP that -n also vocalizes in this position. The Portuguese process of vocalization is consequently:
The glide e'lement is apparently determined by the preceding vowel. In word final position, the nasal elide@ producing a nasal diphthong. The derivations of -cao and movem illustrate the
- vocalization of -n:
kane moven kan It e-->$/VCr -#
b, b, , " + k3n . moven nasal
------18~otethat the hypothesis that sufficiently strong nasal vowels 7Ephthon~,which was used in the explanatlon of tendes < tene:tis,'does not imply that unstressed vowels dlphthongize. I * 9------. This analy-Sis -does not claim tkt unstr?ed vowelspp /
I b cftptrrl-rmgkre, a c+&rkmh-i+wMVen 6 / 7 'J .. , , - ~ffeted.Rather, at reveals nasal diphthongs- to be-the product, , . of the same process which is responsible for the epenthecic sl
+A- nasals, the palataliktion bf s, and the vocalization of 1. d -7 ' - * - - 1 The vocaliza,tion of nasals is not restricted to Portuguese. .
The eff,acement of a nasal in Polish also produces a nasal
diphthong: Cf. kaski Lk5wsk8iIL'pieces', konslci [k?fysk'i],
0 'equine' 19. This is to be. expected. Theoretical Phonology claims -.
that there are no language particular processes. *
p------p ------Finally, thispana5ysis reveals that the non-palatalizat-ion
of -s in Sao Paulo and the ~raziliandialects described by
8' ~a11(1943$and ~eldman(1967)~~is nots enbmalous. It is the i7
result, of the failure of the Latin vocalization rule to 1
generalize as extensively as it has in ~io'deJaneiro and
Portugual. i
------p -- -- p------
r=2 in Sao Paulo r=3 in Portugal and Rio de Janeiro
In consonance with the IDP, this process applies to -1 and -n, or
I -1, -n, and -s, but does not apply to -s unlgs6 it also applies to L ,
------L----- 19~ightner(1970). *O~eitherHall nor Feldman explicitly state which dialect their - are desc r rbliigf Fromthe l i€tflecfata wh-ich-th-ey-ptavfdeeno~--- [menus] 'less') it can only be determined that palatalization ly t - - apply to preconsonantal -s. and -n.
Sdry
There are a constellation of facts which argue for the generalization of the Latin rule
In previous analyses, these facts have been thought Yo be unrelated and consequently have been accounted for by a set 'Lf ------.- unrelated language particular rules. In the analysis offered here, it is argueh that the following are all the result of the . - application of the same pro-cps:
1, The apparent retention of nasals before consonants; , 2. The raising af vowels before nasals;
- - 5The format ion-of -nasa*s-dtphth ongswhen word--f-ina-37 nasals are effaced;
4. The vocalization of preconsonantal and word final -1' 5. The.palatalization of preconsonatal and word final -6.
The process which is responsible for these phenomena is one which strengthens sufficiently strong consonants: In Latin,
Parochial Condition: n=4
In Romance, this process has generalized so that the morpheme - boundary is no longer required:
In Portuguese, the environment of this process has further
\ position:
This process has alga to apply to weaker elements. 7
majority of Portuguese dialects, it applies to 1,2, and s. r -
Universal Condition: r=>n Parochial Conditions: n=3 in Sao Paulo n=2 in Rio de Janeiro and Portugal
The advantages of this analysis of nasalization over-those Brasington or Williams are: . a - - - -- p------p- - -- *2 1. It does not require rules which delete nasals only $
2 6 -Y -
- -- a ------* 2. It reveals the underlying regularity of disparate.
phenomena using rules which are manifested in other
Romance languages, whereas the rules of Brasington and
- Williams are language particular; ii-
3. It reveals these phenomena to be the result of a ,
coherent linguistic- change, the generalization of a ,
Latin rule\n a manner which is predicted by the I theory, whereas the rules of Brasington and Williams 111. Radical Vowel Alternations in the Portuguese Verb ' % EJ.
X. The Portuguese verbs exhibit alternations among s/e/i and p/o/u in the present tense1... ~reviousanalyses have all assumed -* that the process of assimilation is in part responsible for
P@'-- - these alternations. In this chapter, two analyses which make - ---
this claim are examined. It is argued that neither can explain
this phenomena ih a Theoretically adequate manner nor reveal the
------relation between Portuguta se and its historical precursors. As > data, the singular form% n the present tense of the following - xerbs .are considered: +
tocar 'to touch' * * tpco tgque t9cas twues t9ca tgQue
------+z- - he symbols p and p represent open 5 and 2 respectively. rhe symbols -e and -o represent closed 5 and 2 respectively. * - * mover 'to move' movo , mova I ~ves mova's
w m~ve mova A-
2 dormir 'to sleep du rmo du rma
dgrme du rma
7 As the data illustrates, in the present
. first per.son singular present-indicative the radical vowels 3 - and 2 raise to -i and -u respectively in the third (i-themeY conjugation, to -e and -o in the second (e-theme) conjugation, but rellrain in the first (a-theme) con jugation. 2 Williams proposes the following rules to explaik the , .
ams(1962). Atkinson(l954) also supports the analysis of ' 3 Although Williams maintains that these are assimilation rules b it.is not clear why a final -o should have a,different effect
depending on the "assimilating" vowel. However, from underlying
*mpvo and-~~rv~~he apparently derives -movo and servo as ---
These rules do not account for the subjunctive forms, which f not end in -o. Indeed, elsewhere ~illiarns'proposes the rule
------pp- -
"S -- Cf. me: tam > sda, fo:rma > fgrma. This rule incorrectly
predicts the subjunctive forms.
Williams explains the subjunctive forms as a product of
analogy. The vowels in the subjunctive raise by analogy with the
first singular prestht indicative forms. He does not explain why
vowels should raise bi analogy with the first singular rather ------7- L
3 than remain open by analogy with the statistically mpre frequent
3Williams(1962: pg. 97). 4 Latin moveo: and servio: respectively. ,- invoke analogy whenever convenient renders it vacuous as an
explanatory concept. ~cientific concepts are typified by the d *. fact. that they prohibit certain events from occuring. Analogy * prohibits nothing. .Indeed, it is frequently invoked, as 'in this
case, when a prohibited event, the raising not lowering of the - - rgdikal vowel in -the subjunclive, is found to actually occur. * This inadequacy is especially evident in Williams'
explanation for why the radical vowel in the first person
*levo. Since these forms contain word final o, Williams' rule -3 - k-' k-'
predicts that the vowel will raise. Williams claims that the voOwelremainedopen by anal-ogv with the second and third persons. b singular. He provides no rea'son why the radical vowel in the
first person singular should remain by analogy with the second
and third personsosingular in the first conjugation put not in
the other conjugations. / It is concluded that Williams' analysis is,inadequate. It
is forded to rely an the rron-scientific concept of analogy and F
it uses. rules which are inexplicably sensitive to morhpbological ff The analys>s of ~arris~comprises an argu~ntfor the
"Elsewhere Condition", a hypothesis about the ordering of rules ><* in Transformational Phonology. Consequently, many of the details
e-. e-. of his argument are not relevant to a ~heogeticalibnsideration'
.of this problem. Of concern is his pla&ib$e suggestion that the
radical vowels assimilate to the theme vowel of the verb.
Harris-proposes a rule- by which the radical vowel - -
*_ _ ". A assimil'ates to the height of 'the theme vowel when the theme
vowel is followed by another vowel. He argues that the
The theme vowel is always followed by another vowel in the
/I subjunctive as, according to Harris, the subjunctive morpheme is -e in the.first conjugation and -a in the second and third
------conjugation. Moreover, as the first singular morpheme is -o in
the present inddcative these forms also meet the environment of * his rule. The first singular forms of mover in the present indicative and present subjunctive are underlyingly [mgv+e]+@to. . and fmgv+e]+a+0. ~~fulegoverning the alternations expands as Tperficia this appears to answer the questions posed
by he data. There ishno raising in the first conjugation because the assimilation rule onlyI assimilates following -e or -i. Vowels do not raise in second-and third person singular forms in the present $ndicative because the theme vowel
is not followed by another vowel in these forms,
- - There are-, however, several problems with this solution,
* h not the least of which is the fact that it uses(language
particular rules, rules which are not* found in'any other Romance
------language. As a consequence, it is difficult to relate the
alternations of Modern Portuguese with the rules of Latin,
Rather, it must be nraintained that the assimilation rules were
added to the grammar of Portuguese, contrary to the claim of
Theoretical Phonology.
This rule was not addedto the grammar of Portu~ueseas it ------
appears in Modern Portuguese. In Old Portuguese, the radicel vowel in verbs with -e-themes raises: --sejo [se!lu] < sedeo:6. This follows the rule of Harris. Howevet,_it does not raise until 7 later in verbs with -i-the~s: Hube_r lists the following alternate Old Portuguese first singular forms of servir and sirvo
d'ormi r dormio
- 4 - -u Although Tt i's pqssible t6-argue that the rule generalized
C
from Old 8brtuguese tp Modern Portuguese, the problem of the k
lack of a source for the old Portugaese rule remains. Moreover; ' ------the Old Portuguese data reveals that the radical vowel does not
raise unless the theme vowel is lost. The analysis of Harris
does not aknowledge this. There is, however, no evidence to deny
it; that is, there is no form such as *sirvio which would
confirm that the loss of tk theme vowel is unrelated to the
rdY&of +he-F d e&veueL+rff-%~rgmd4ate~
fact is crucial toF understanding the Po~tugueseradical vowel
alternations. 1
on ale must be restricted to
verbs, wdh 8aks in fact do&; thereby limiting its ability
/ to explain these alternations as phonological phenomena.
Notice for example thaQ in LatI/" n terminurn, Old Portuguese termio, Modern Portuguese -termo [termul the radical vowel does ------not raise to -1 as predicted by Harris' assimilation & rather. to closed 2. The restriction that this rulekipplies only r Though Har~Js' assimilation rule appears to answer the
relevantC" quest-ions. about the behaviour of the radical vowels in .f verbs, it is. nonetheless inadequate for-it uses a rule which is
t unattested in a&y other language and makes incorrect predictions -
about noyns which contain the environment in which the rule
- _ should apply but.does not. 'I
The analysis presented with two observatiods.
First, as Harris correctly obber- the theme vowel plays a
vowel will raise, The Old Portuguese forms above appear to bear
this out, for it is only when the theme vowel is lost ;hat the
radical vowel raises. This suggests that the process by which - the theme vowel is lost plays an important role in the raising of the radical vowel, Second, in general vowel raising is a ------,------manifestation of a weakening process. The raising of open -e ad
-CA to -i and -u is weakening on the Eta parameter, which is represented below as the vertical a~isof the Eta/hega 8. parameter ------2%' FoFey( 1977 :" pg, 47). Foley frequently represents the Eta/Omega parameter as ieu oa 123 45 -a rguingthXtF3sTtFZinger- thZiiallvXls-(C fTtatiKamicamY- French amie, but Latin amicum > French -ad) and that the -ofech ofirelativetoeachne representation of Eta/Omega as a two dimensional matrix is used here because processes often range over only one of Eta or Omega. For example, in French a nasalized -i strengthens to -e To determine the relative strengths of open and closed -e
+ and -o, the application of diphthongization in French, Spanish
LATIN FRENCH SPANISH PORTUGUESE f lo: rem f leur . f lor f lor ho :ram heur hora hora f amo :sum f ameux f amos o f amoso
. tre:s t rois t res t res me :nsem mois mes mes ------cre:dit croit cree cre
0 (cont'd)(fi:nem > fin [fF]), nasal 'e strengthens to a (tempus > temps [t~])(~ope(1934: pp. 171, 236; 240)). Each -is a sttrengthening-by 1 on the Eta parameter. In Latin medial -a weakens to e in closed syllables (facio:, perfectus) and to,i- in open syllabies (facio:, perf icio: ) indicating succesive ------weaenings on the Eta parameter. The greater strength of a is
represented 'on the two dimensional matrix as the highest- Eta / - -"-- -value. me ami u 1s represente e equallty or- f heir combined%ta/Onrega strengths (3). - - - - novum neuf nu evo novo motre t . ~ut mueve , move probarn preuve prueba proba
lepore lievre - liebre ' lebre < petra pierre pietra pet ra tenet tient t iene tern
. . ~i~thon~izatidnapplies preferentially to strong elqpents.
Romance, it exhibits the following configurations: ------
FRENCH 9-->uo-->ue-->eu 9-->ie o-->ou->eu e->ei->oi (->we-->wa )
SPANISH 9-->uo-->ue -- -- >=- e->idem o-->idem
PORTUGUESE 9->idem
g-->idem , o-->idem e -->idem
As diphtkongization applies preferentially to p and p, it is concluded that they are stronger than -e and -o. This and the - -- fact that -e and -o a&tronger than -i and -u establishes that -- following parameter : From this parameter, it is evident that the raising of the
radical vowel in the ~ortu~ueseverb is a manifestation of a F weakening process, In particular, the alternations & and 9/0
represent a weakening of 1; the alternations e/i and p/u a
weakening of 2. This appears initially anomalous for the radical 9 ~owelswhich-raise----- a res t ressed-, and stresSsed VOW~~&-not ------
typically weaken. I
A similar phenomena is found in the ati in weakening of
#' medial vowels. Examples of weakening of stressed vowels include:
facio: . perfsctus ; teneo : attfneo: legGntur
~raditionall~'~,the weakening of stressed vowels is .J * explained by positing an initial stress accent. After weakening
of, posttonic vowels, the stress is ashif ted to' the penultimate
syllable ff it is long, else to the antepenultimate syllable.
------'~he radical vowel raises in the secpnd and thiid persons in the -ppseFit is. oy-hw. See Buck(1933: pp. 100-102), Kent(1966: p$f$%9-101) . and # Niedemnn(l953: pp. 18-28). pGrf actus hiteneo : p6rf ectus sttineo: weakening perf 6ctus attheo: stress shift
. C
Within Theoretical,Phondogy, the assumption of an initial
st2ess is unnecessary. The theory maintains that
4 "morphological units such as wokds and - syllables have a certain inherent, rather constant strength. The more phonological elements this strength must be distributed over, the weaker each p?lonological element. 111 1
/
------The addition of an affix d,istributcs the strength of the word
-over more ele~ts.~~his strength attenuation is manifested as
the weakening of the-radical vowel. Thus, the derivation of 1 attineo : contains the following stages.:
a t+t -eneo: atZe neo: strength attenuation ------.attineo: e -->i
~his~analysisis not directly applicable to Portuguese
where vowels weaken without the addition of elements to the
word. It does however indicate that vowels may weaken even under
stress.
As indicated above, this weakening is apparently related to
------the theme vowel. This relationship between radical and theme?
11 Feley(19ff: pg. 86). - vowels is not unusual; ~6ri~~~~explains- diphthongization. in d French tient (Latin tenet) but not in leve (Latin levat) as the C result of strengthening of theme 5 but not theme e (in
cons6nance.with the 1DP)"with concommitant weakening of the
L radical vowel in leve but not in tient (in consonance with the - principle of strength conservation13). The weakening of the
vowel in 3eve blocks diphthongization:
tenet leva$ 11 le-va t strength fluxion
------Liex-t---- taila- &phthonnFzation ---=------= A tient 11 e-->6 - 9 I1 leva t->@la-# II leve a-->e (-->@)
The influence of the theme vowel over the radical vowel in
Portuguese appears to depend crucially on the loss of the theme
------A ------7 vowel: Cf. Old Portuguese alternates- servio, sirvo. Consequently it is necessary tb consider the process respofrsible for the loss
of the theme vowel.
ihe Transformational analysis of the Portuguese verb by
Harris(1974) contain's a "truncation" rule
12~oley( 1979 : pg; 204). 13~oley(1979 : pg. 198). The principle of strength conservatiorr
- - - is that- alluded------to above: Each word-- maintains - -a constant unit of -- strength, Consequently, the weakening of an element releases strength which may be used to strengthen other elements; the -- strengthening of an element removes strength from other - elements. This flow of strength throughout the word is-appelated "strength fluxiont1. -
to account for the &OF of the theme vowel in the present
subjunctive and the first person present indicative.
--> . INDICATIVE \ compra r -'to buy' comp i-o compramos compre comp remos comp ras comprais comp re s compreis comp ra comp ram comp re comprem
------
vender 'to sell' vendo vende'mos venda vendamos vendes vendeis vendas vende vendem venda vendam
p-artir - - -- 'to0departL parto part imos parta partamos partes partis partas parta-is
parte partem - partam
'r
Harris' rule deletes the final vowel of verb stem when it is
I followed by 'another vowel. The second vowel is either the first
person singular morpheme -o or the sujunctive morpheme, 5 in - e In verbs w3th theme -e-or -i. For example, -- - +eemgrie{+c3 - f~eump&&+c comp rwo comp rwe Truncation compro * comp re
Since -a is stronger than = which is stronger than -i, an f Wnof the sort'sh ich Harris proposes has the following
expansion : a r
Given only the synchronic data, no decision can be made I about whether elisfon is the correct process, for all three
vowels elide. To determine whether a process is correctly Y- formulated, it is necessary to have first, the relative
strengths of the participating 9lements (X, Y), and second, a
If X is stronger than Y then the process is incorrectly
gL 5 formulated as it violates ths IDP. To decide onc this question, , the behavior of the theme vowel in Latin'and Old Portuguese is
considered, -- moneo < *moneo inf, monered audio < *audio inf. audire
If elision is the correct,interpretation thefi the only rule required is
ow ever, as -a is the strongest vowel its elision implies (by the
IDP )
As these rules are contradicted by the data, elision cannot be the correct interpretation.
The Theoretical interpretation is contraction. Foley(1977)
VnVm-->Vo Universal Condition: In-m 1 Sufficiently similar vowels contract. Similarity is measured by combining the Eta and Omega strengths of the participating 14 vowels . Herein, the relevant contraction process is Right to Left contraction;. that is, the right-most element dominates. The process d*fom&&e&s - -- - 36 above for the Eta and Omega parameters. The following parameter reeults from the addition of the -L Eta and Omega values of each vowel: When the Eta/Omega strengths of the vowels are compared, it is contract : a+o-->o la-o l=0 ' e+o->idem ( e-o The value of D for Latin is set at 0, -- * fin Old Portuguese, both themes -a and -e are lost in the 15 st person present indicative. Huber lists as examples: d P cantar 'to sing' canto 'I sing' cantas 'you (sg.) sing' ? vender 'to sell' vendo 'I sell' However, in the third conjugation the loss of the .theme i- does not ochr until later. ~uber'~lists the following first person / singular alternaws of servir and dormir: f , servio servho sirvo dormho durmo the time at which servio coincides with canto and -9 vendo the rule conf igurati-on is the weakest vowel, is maintained in an environment in which stronger vowels elide. However, if the relevant process is contraction then the ,--' rule configuration is ---- 7------16~uber(1933: pp. 200-202). This rule configuration is in consonance with the condition on contraction that only sufficiently similar elements, contract.' Moreover, it is evident thatthe loss of the theme 2 in Old t * b .Portuguese is the result of the generalization of contraction. That is, the linguistic change from Latin to Old Porruguese is P Returning to ~bdernPortuguese, - if contraction is the here then the rule configuration is ------e+o-->o i+o-->o . , As neither this configuration nor that of the "truncation" rule of Harris violate any of the principles of the theory there can - be ho choice between them, based.on Modern Portuguese alone. 4 'However, if one wishes to relate the loss of the theme vowel in - Modern Portuguese with the loss of the theme vowel in Old ------ 3 4 Portuguese and Latin then contraction must be selected as the A ------1 -3 , relevant process..Whereas the elision analysis must ma 4 X3 ------3 4 6 lost and replaced by a general elisisrule, tk analysis proposed here repcesents-'the loss of "the theme vowels in Modern Portugu~seas theelt of a continuing, coherent t generalization of contraction: - D-f& ==> -1 ==> D=2 - - - A I - * It is now possible to explain the different manifestations of the radical vowel. Contraction is an endotherm$c process ------requiring strength1'. According to the principle of strength " consemation,- when an element strengthens, other elements in the 3 w0r.d correspondingly weaken; when an element weakens, other --- 18 elements in the word correspondingly strengthen. Fole'y utilizes this concept to explain the assibilation of -d in French I-. croie from underlying *credeya as the resuLt of contraction of -a-u in the former which utilizes the strength necessary for the contraition of 2 to nheremajning 2 cluster is the etymon of -s in close. The- strength necessary for contraction is 19 provided by 'the elision of the theme vowel -e . . Y ------I------* 17 ? L8E~ley-(HZ9:-pg,_LUL------Foley( 1979: pg. 202) he symbol 'S' is us& represent this unit of strength. The notation represent$ elihon as the release of strength concomitant with the foss of an element, and contraction as the / absorgtion of this str@ngth. ------A -- - - cred~ya claudsya - cred ya claud ya e-->0+S I1 clodya au+S-->o . creya I1 dpt S-->y It /' I cloza dy-->dzy-->z croia It 8 ey-->oi L croie, ' cloze a-->e (-->@) The relative amount of strengthprequired for contraction iK - A --- - . - - LA - determined by the difference between the relative strengths of repeated here from page 38: I (/, The following table illustrates the relation between contraction anG the weakening of the radical vowel: CONTRACTION STRENGTH RAD I CAL VOWEL \ a+o->o lgvo \ 9--> i dem tQC0 if 9-> idem e+o->o le-01-1 rego 9->e movo 9->o i&-->o ) i-o (-2 sirvo g-->i du rmo 9-->u 6 - - -~ -- The raising of the,vowel is explaaed as a weakening concomitant with the contraction of dissimilar vowels. When I vowels which differ by 1 contract, the ra'dical vowel weakens By 1. When vowels which differ by 2 contract, the radical vowel weakens by 2. The relationship between contraction and-vowel weakening is 'l illustrated in 'the derivations -of -'levo -movo and sirvo . Note that this aqalysis need not be restricted to verbs, limpidum (with regular lowering of Latin -i to Romance -e) and term0 [termu] < termio < termi;@m: A limpidu tgrminu 1 . lempidu 11 - i-->e limpiu I1 1e-mpu limpu e- -->i : * 11 tern 9 -->e ,, vowel assimilates to the,height of ,the following .vowel if it is % termio. The analysis offered here, which maintains that the * difference in the pPonological strengths between contracting * vowels determines the extent to which the preceding vowel weakens, makes the.correct prediction, As the difference between -i and -u is 1, the preceding vowel weakens by 1, - - - - If this relation-is not spurious, then it should-also be > A - -- - . A - responsible for the weakening of the radical vowel in the subjunctive. As before, the analysis begins with a comparison of In ~att(, the subjunctive morpheme appears as a in verbs -, with -e and themes, but as -e in verbs with -a theme, Ignoring the first conjugation for the moment, the following forms bear out the prediction that only vowels whose phonological difference is -0 contract in Latin: moneam first sf ngluar monea :s second singular audiam first singular audia :s second singular The rule configurarion is I. e+a->idem 1 e-a 1 =l i+a->idem ( i-a (=2 - --- Huber2*-lists Old Portuguege forms such as first singular subjunctive venda 'and, aduga with e-f hemes but first singular -5 - subjunctive servia, servha, and dormha with -i-theme. The rule -. configuration is e+a->a 1 e-a l=l i+a->idem (i-a)=2 This follows the prediction that vowels which differ by no more In Modern Portuguese, the theme vowel is lost in verbs with both -e and -i themes. Note the first singular forms rega inf. reger mova inf. mover sir-va-inf,--servA= - duma inf. dormir The rule configuration is 1 e-a l=l ) i-a 1x2 The claim that contraction has generalized from Latin to - - Partugueae _tofiorcon~act~f-increasingldissimilar vowels is upheld. Moreover, the difference between the - - -- radical vowels in the subjunctive. This is illustrated in the derivations of rega and sirva: sgrvia --11 . ea+S-->a sg rva - ia+SSt->a-- sirva 9- -->i Returning to the first conjugation, it appears initially counterexample to the claim that on$y vowels which are equal,- in phonological strength contract in Latin. Moreover, from the contraction of -ae in Portuguese lgve < lgvae, it is predicted that the vowel should weaken by 1. Thus, lgve appears to be a a counterexample to the claim that weakening is caused by the ------ ~oktr~ctionof dissimilar vowels. L , 2 1 Williams argues that contraction applies preferentially ? when the contracting vowel is more open than the dominant vowel. He interprets contraction as assimilation of the lower vowel to the higher, followcd by contraction of identical vowels: e.g calentum > caente > gueente > quente, veni:re > -veir > -vdir > -vir. nf which thp first C (i,e stronger) than the second explains why -ae contracts in ! Latin amem < *arnaem but ea does not contract in Latin mneam. -9 -9 - Second, whereas ,the SDP does not explicitly govern contraction, - it does state chat assimilation applies preferentially to , 22 similar elements- . Consequently, by interpreting ccinfcaction as b two stages beginning with assi.milation, this process is brought under the domain of the IDP while simultaneously retaining the analysis of the lass of the theme vowel in Portuguese as the result of generalization of a Latin rule: The similarity --- - -p--p-p---p------requirement on assimilation has been relaxed. Finally, the i assimi1ation.analysis reveals in greater detail why the contraction of -ea weakens the radical vowei by 1 in Portuguese, but the contraction of -ae doe6 not. The rule represents a strengthening of -e by 1 on the Eta parameter. - - According to the principle of strength conservation, the strengthening of -e by 1 weakens the word by 1, This weakenifig is- manifested by the weakening of the radical vowel by 1. Similarly, the rule . . ------ppp-p------represents a strengthening of 2 by 1 on the Omega parameter, and the------rules '. 22~oley(1977:pg. 143). ------.- - - f represent a strengthening of by 2. In each case the vowel is "eakened accordingly. However, the rule represents a weakening of a.- Consequently, the contraction of ae to -e does not remove strength from the wordand the radical vowel does not weaken. ----p---p------p------The process of Right to Left contraction is reformulated as .. V nVm-->Vm Universal Condition: X<=n-m<=Y -- The condition still stipulates that the difference between the relative strengths of the participating vowels must be ------sufficiently small. Howeyer, it is now possible to stipulate that the left vowel must be stronger than the right. The rule - configuration for Latin is Xhe complete Latin rule is I Universal Condition: Xn-mY Parochial Condition: X=fl Y=l This permits the contraction of E, but not the contraction of -ea. The linguistic change from Latin through Old Portuguese to - - Modern- ~ortuguesei-s khe change in- X: ------Note Portuguese -~air < sali: re where though the vowels differ by 2 contraction does not apply because the firsr is stronger than f- the second by 2, not 1 as the condition stipulate;. . a It is concluded that the radical vowel alterations of the 3 I Portuguese verb are caused by the gbneralization of the Latin L rule of Right to Left contraction. This-generd&atFo+rrnFm------the contraction of increasingly dissimilar elements. The contraction of di~imilar~elementsr.emoves strength from the - word. This weakening is manifested'by the weakening of the radical vowel. t sunrmary u f . - ------, 1t is argued that the raising of radical vowels is the product of a weakening process ranging1 over the parameter This weakening is caused by the contraction dissimil r vowels i & Y which remove strength frqm the word. The weakening of the word \L -I is manifested as ,-the -weakening of the radical vowel. The amaunt - A - - -- - of strength required for cmtraction determines. the ddgree of 1. It does language particular rules, but in Latin; 2. It reveals the vowel alternations as the product of n a linguistic change, the generalization of the Latin contraction rule, whereas the rules o4 Williams and 3. It explains rather than describes why the radical - vowels do not weaken in the first conjugation where contraction does not use sufficient strength to cause vowel weakening; 4, It explains why the vowel in Old Portuguese forms such as servir does not weaken until the theme vowel is lost, whereas the analyses of Williams and Harris 0 1 treat vowel raising and loss of the theme vowel as f unrelated; ------5. It does not require morphologically sensitive rules , . to restrict their application to verbs, but also 1 explains the weakening of the vowel in nouns such as tenno < termio. IV. Conclusion P Previous analyses of Portuguese nasalization and radical b vowel alternations have represented the& phenomena as the L products of rules peculiar to ~ortu~uese.This is, in many ------ cases, a result af analysis of Modern Portuguese as an-isolated . - entity. As a consequence, the rules postulated for Portuguese . bear no relation to those found in Latin. This necessitates ------perception of linguistic change from Latin to Portuguese as the t arbitrary addition of rules to thek rammar of Portuguese. This perception of linguistic change fails to address tk problem 7 raised by Meillet, the continuity of linguistic change. . Theoretical Phonology uaintains that linguistic change - - - -- - ?isnatcam&h~eaddFtio~~fa1e6-~ - grammar, but the repetition or modific tion of , rules already existing in the language f .I, This claim follows directly from the claim that the.rules of a a particular language are 'instantiations of universaLphonological rc 1 \ processes. Moreover, it also follows from the ,1atht~al , Development Principle that the "modification" of r&ksi is itself coherent. As the-IDP predicts that no language contains a rule L weakening a strong elerhent unless- it also weakens all weaker ------p-p---p------elements appearing in the same environment, or a rule ------'~ole~(1975 : pg. 54). # ------strengthening a weak elemeat unless it also strengthens all i - C- stronger e,lements appearing-in the same environment, no rule .will generalize in a manner which will create such a situation. Rather, a weakening process generalizes to include the next strongest element; a strengthening process generalizes to d inegude the next weakebt ele+nt.- By-denying-the sj%hron"ic/diachronic distinction, ' Theoretical Phonology permits explanation of facts perceived- in C the sxnchronic grammars of languages, as well ah explanation of ------, ------the linguistic changes which produced the rules responsible for .-d theie facts. Moreover, the analyses which result represent the linguistic change as a continuing coherent generalization of processes. - This is perhaps most evident in the case of the Latin rule f This rule is, in fact, an instantiation of a general process strengthening consonants in a particular environment: consonant. It is not the case tbat contiguity to morpheme M ------3 L boundary is a strong environment. RatherYsthemorpheme boundary -- -- serves as a catalyst, facilitating the application of all a L processes . The environment for this.* rule has generalized in Romance so that the morpheme bou6dary is no longer necessary: . * ,' - - Spanish -otro, French autre, Portuguese 060. ------.- . LLatin a11-2 In Portuguese, though in no other Romance'language, the environment of this process has generalized further to include a ------&fi=E&n&qE+ ==-a: -==------/ . , Cf. Finally, the process of vocalization has generalized in * Portuguese,th~ugh_again_nnt-inany stherommce lane include nasals and -s: This generalization has particular importance in the explanation , C 4s. of aspects of Por'tuguese nasalization. Traditional analyses have postulated rules of spontaneous < ------, diphthongization to account fo guese nasal diphthongs. L -- L, -IFhese~es*~ect+ecaus&%$papply to unsiressed vowels, 4 -PI 'r wfiich typically do not diphthongize. The nasal diphthongs were , * explained here as the-result of vocalization of the nasal. Subsequent to vocalization, the nasal effaces leaving a glide to fohn a diphthong with the preceding-vowel. *-. Although it is recognized that nasalization 'typically applies wbn the nasal is tautosyllablic with the preceding vowel, the apparent failure of.preconsonanta1 nasals to efface has not been previously perceived as a problem, perhaps because ------nor other theory claims. tha phonological rules used to explain --- phenomena of a particular language are+universal. As Theoretical Phonology predicts that if intervocalic nasals efface then preconsonantal nasals efface also, their apparent retention presents a problem. It is argued that the phonetic nasal is again the result of ------. vocalization of the nasal. The nasal does in fact efface, I However, the remaining-glide consonantalizes when followed by a stop. u Whereas the ana ses of Brasington, St. Clair, and Williams represent these two as unrelated, the Theoretical 4 analysis reveals a which relates both nasal - , diphthongs and th&art$t retention of nasals, as well the a Latin process. * - ---- z------ Q. N 6 1 It is possible to isolate four stages in the';histon of -- this process. The-first is that of Latin wherein it applied to the strongest el$ment to a restricted number of forms. The second is the Romance process wherein the environment has generalized to -exclude the requirement of a morphpme boundary, The third is the process which applies in those Brazilian - Portuguese dialects described by %~a11(1943) and Feldman( 1967) - -. . - wherein the environment has generalized to include word final s boundary and the process applies to weaker -n though not -s. In pp- -- - -4kPMrn9aqpF=t-+as--- a further generalized further to apply to -s. Universal Condition : r=>r Panochis1 konditions: r=3 for Sao Paulo n=2 for Portugual and Rio de Janeiro ------This is the manner of generalization which is predicted by the Inertial Developmpt ~rknci~le. L, - Consideratioh bf Portuguese nasalization illustrates two t/ f types of generalization: generallization of environment and \Fneralization of rule. The thirdtypt of generalization is F relaxation of the similarity requirement permitting assi lation \ J and subsequent contraction of increasingly dissimilar vowels."n, ------, VnVm->Vm , :Universal Condition: Xn-mY ~a,rochialconditions: X=b Y=1 for Latin X-1 for Old Portuguese X=2 for Modern Portugese 4 As the parochial conditions indicate, this rule has generalized in contin&hg coherent manner. - A! is the case invocalization, the generalization of "> .- . A . contraction has ramifications which are not superficially apparent. In the case of contraction, the contraction of this strength attenuation which explains why the radical vowel raises only when contr*ction has occurred. In Latin, ,where 1nly phonologically identical- vowels contract, the radical vowels are not weakened by contraction. In Old Portuguese, only identical vowels or those which differ by 1 -- - - p------contract. Correspondingly, the radical vowels in the second conjugation, where eo and ea contract, weaken by one unit of - - - - ,! , strength when contraction has applied: E.g. sejo [sefu]- < sedeo:. In Modern Portuguese, contraction applies-to vowels * /' which differ by no mre than 2. Consequently, the-radical vowels * in the second conjugation weaken by 1 unit of strength and the -? vowels in the third conjugation weaken by 2 units of strength, ------WhPreaa_p~e~al~trave~~~-P-~~~~es+~~----~------t isolated entity, the analyses provided here exp-w set of data using rules which are manifested in other languages, and ------continuing linguistic change, Bibliography Almeida, A. "The Portuguese nasal vowels: Phonetics and phonemics", in J. Schmidt-Radef eldt (ed. ) , Readings in Portuguese Linguistics. Amsterdam: North ~olGndPublishing Company, 1976. c Atkinson, D. "A re-examination of the Hispanic radical-changing verbs", in Estudios Dedicados a Menendez Pidal-: - - - --Tomo V. Madrid: Conse jo superior de Invegtigaciones Cientificas, 1954. , Boyd-Bowman, P, --From Latin to Romance --in Sound Charts. Kalamazoo College, 1954T in Journal -of Linguistics, vol. 7, 1971, pp. 151-177. Buck, C.D. Comparative Gramar ----of Greek and Latin . Chicago: University of Chicago Press, 1933. Feldman, J. "A comparision of the segmental phonemes in Brazilian Portuguese and American Spanish!', in Linguistics, vol, 29, pp, 44-57, 1967. \ - L - Toley; J. "Ph-onologicaT~ange%~RT~pitition', in Papers --from the Seventh Regional Meeting Chicago Linguistic Society. Chicago: Chicago Linguistic Society, 1971. Foley, J. "An interpretation of Lachmann's Law", in Proceedings ---from the Tenth International Congress -of Linguists, 1970. Foley, 3. "Rule presursors and phonological change by laeta-rule", in R. Stockwell and R. Macauley (eds.), Linguistic Change -and Generative Theory. Bloomington: Indiana University Press, 1972. Foley, J. "Latin origin of Romance rules", in M. Saltarelli and - --- - W.-Uan~er<~ed6,) , u iachrOniS tK&EFiK - Romance- Linguistics. The Hague: Mouton, 1975, Foley, J. Foundations of Theoretical Phonology. Cambridge: ~arnbrid~runiversityPress, 1977. Foley , J. Theoretical Morphology_ --~L-thePrench-AmfitPrArun: - John Benjamins B.V., 1979. Hall, R. "The ;nit phonemes of Brazilian Portuguese", in Studies -in Linguistics, vol. 1, pp. 1-6, 1943. Harris, J. "The 'elsewhere condition' in phonology: Evidence ' from Portuguese", in Linguistic Inquiry, 1974. Huber, J. Altportugi'esisches Elementabuch. Beidelberg: Carl Winters Universitatsbuchhandlung, 1933. - Kent, R.G. The Sounds of Latin: A Descriptive and Historical - - - > "Phonolijgy. New ~ork:Kraus Reprint Corporation, 1966. Koutsoudas, A. Verb Morphology of Modern Greek: A Descriptive Analysis. ~loominzon:Indiana ~niyersity Research ------t.em er - i-n-?ii&m-gyjf k lore , XGFL~uisSt-ic8,- 1962. Lightner, T. Why and how nasalization takes place", in Papers -in Linguistics, vol. 2, pp. 179-226, 1970. Meillet, A. "L'Etat actuel des etudes de linguistique generaler1Cl906); reprinted in his Linguistique historique et linguistique generale. Paris: Librairie ~zoreChampion, Editeur, 1965. Niedermann, M. Precis de phonetique historique --du latin. Paris: ~ibrairiyc.Klincksieck, 1953. Pope, M.K. ---From Latin to Modern French -with Especial Consideration of Anglo-Norman Manchester: Manchester uniZrsity Press, 1834. St. Clair, R.N. "The Portugu~~enoun pluralization", in Linguistics, vol. 68, 1971, pp. 90-102. Williams, E.B. ---From Latin to Portuguese. Philadelphia: T University of Pennsylvania Pre~,1962. ------Willis, R.C. ----An Essential Course in Modern Po guese . London : - - -- - George C, Harrap and Co., Ltd