Paper presented at the 3rd UKLVC-University of York England - July 2001 Page 1 Lateral Vocalization in Brazilian Portuguese Thaïs Cristófaro-Silva (UFMG) Marco Antônio de Oliveira (UFMG) Abstract One of the differences between Brazilian Portuguese (henceforth BP) and European Portuguese (henceforth EP) is that post-vocalic laterals are vocalized in the former. This fact has been traditionally analysed as a phonological phenomenon in which a lateral is realized as a back rounded glide when in syllable final position: /sal/ [sáw] sal ''salt'' (cf. Câmara (1970), Callou & Leite (1990)). According to this process forms which have a distinct pronunciation in EP become homophonous in BP. Thus, EP has the forms ''bad'' mau [máw]/ ''badly'' mal [mál] and ''tail'' cauda [káwda]/ ''syrup'' calda [kálda]. On the other hand, BP forms ''bad'' mau/ ''badly'' mal are pronounced as[máw] and forms such as ''tail'' cauda/ ''syrup'' calda are pronounced as [káwda]. In this paper we intend to evaluate the adequacy of assuming a phonological input such as /sál/ for sal ''salt'' [sáw] in BP. We argue that assuming an input such as /sáw/ is more adequate to a synchronic analysis of BP. Our proposal is based on variation involving forms which present postvocalic /l/ in wordinternal position. In order to discuss the so-claimed postvocalic /l/ in word internal position as in calda ''syrup'' [káwda] one has to consider the behaviour of ''r'' sounds in BP. There are two types of ''r'' sounds in BP which we will refer to as weak and strong ''r's'' (cf. Oliveira (1983, 1999)). The weak ''r'' occurs intervocalically as a tap: caro ''expensive'' [káru]. The strong ''r'' - which we will represent as R - may be manifested as a number of different sounds and occurs in syllable initial position: carro ''car'' [káru], rato ''rat'' [Rátu], desrespeito ''disrespect'' [desrespejtu]. When an ''r'' follows a syllable that ends in a vowel either the tap or the strong R may occur: caro ''expensive'' [káru], carro ''carro'' [káru]. When ''r'' follows a syllable that ends in a diphthong it is typically manifested as a tap as in Laura ''Laura'' [láwra] and feira ''street market'' [féjra] and occasionally as the strong R as in bairro ''neighborhood'' [bájru]. When ''r'' follows a syllable that ends in a consonant as in desrespeito ''disrespect'' the strong R occurs systematically: [desrespéjtu]. When considering BP words which are claimed to have a word-internal postvocalic /l/ followed by an ''r'' sound such as bilro ''lace bobbin'' or guelra ''gill'' we get either the tap ''bi[wr]o,gue[wr]a'' or the strong R ''bi[wr]o,gue[wr]a''. Different speakers may vary the choice for the ''r'' sound depending on the word in question. The analysis is based on data from Brazilian university students. We intend to evaluate the best approach to analyse the variation concerning ''r'' sounds followed by a postvocalic /l/. We will consider a perspective which assumes it reflects a sound change in progress and also a proposal which assumes that it reflects a case of lexical diffusion. One of the differences between Brazilian Portuguese (henceforth BP) and European Portuguese (henceforth EP) 1 is that post-vocalic laterals are vocalized in the former. This fact has been traditionally analysed as a phonological phenomenon in which a lateral is realized as a back rounded glide when in syllable final position (cf. Câmara (1970), Callou & Leite (1990)). The data in (1) illustrates this proposal.
Paper presented at the 3rd UKLVC-University of York England - July 2001 Page 2 (1) a. /sal/ [sáw] sal ''salt'' b. /salto/ [sáwtu] salto ''jump'' Note that some forms which have a distinct pronunciation in EP become homophonous in BP. This is due to the process of lateral vocalization. Forms in (2) indicate major features of word-final /l/ in BP and EP. (2) European Port Brazilian Port [mál] [máw] ''mau'' bad [máw] [máw] ''mal'' badly [kálda] [káwda] ''calda'' syrup [káwda] [káwda] ''cauda'' tail In this paper we intend to evaluate the adequacy of assuming a phonological input such as /mál/ for mal ''badly'' [máw] in BP. This is the traditional view assumed by all works in BP phonology. We will provide evidence which shows that such a view is in fact inadequate. We argue that assuming an input such as /máw/ mal ''badly'' is more adequate to a synchronic analysis of BP. The data we analyse was collected from 60 subjects (18 male and 42 female) all of whom are undergratuate students at the Federal University of Minas Gerais. The data is divided into three sets. The first set comprises a group of words which present a syllablefinal /l/ (eg. ''bilro'' pronounced as ''bi[w]ro'') and some words have a syllable-final u- diphthong (eg. ''besouro'' pronounced as ''beso[w]ro''). The second set of data is inteded to investigate plural forms of final-/l/ (eg. ''anel'') and final-/u/ (eg. ''troféu'') words. The final 1 For the matter addressed in this paper EP and Southern Brazilian dialects display the same behaviour, i.e /l/ vocalization does not take place. When reference is made to EP in this paper it also includes the Southern Brazilian dialects.
Paper presented at the 3rd UKLVC-University of York England - July 2001 Page 3 set of data is aimed at investigating derived forms of final-/l/ and final-/u/ words. By analysing these data we intend to show that the alleged syllable final-/l/ words are better analysed as presenting a high back rounded vowel in word-final position. To defend our proposal we will consider formerly those forms which have an alleged syllable final /l/ in word internal position (calda ''syrup''). Further evidence will follow from cases where it is assumed there is an /l/ in word-final position (mal ''badly'') and in some derived forms (saleiro ''salt shaker'', salgado ''salty''). In order to discuss the so-claimed postvocalic /l/ in word internal position as in calda ''syrup'' [káwda] one has to consider the behavior of ''r'' sounds in BP. There are two types of ''r'' sounds in BP which we will refer to as weak and strong ''r's'' (cf. Oliveira (1983, 1999)). The weak ''r'' occurs intervocalically and it is manifested as a tap and the strong ''r'' - which we will represent as R - may be manifested as a number of different sounds and occurs in syllable initial position: 2 (3) Weak 'r' 3 a. caro ''expensive'' [káru] Strong 'R' b. 1. carro ''car'' [káru] 2. rato ''rat'' [Rátu] 3. desrespeito ''disrespect'' [desrespejtu] When an ''r'' follows a syllable that ends in a vowel either the tap or the strong R may occur: caro ''expensive'' [káru], carro ''carro'' [káru]. When ''r'' follows a syllable that 2 The strong R may be phonetically manifested in BP as [h,, x,, r ].
Paper presented at the 3rd UKLVC-University of York England - July 2001 Page 4 ends in a diphthong it is typically manifested as a tap as in Laura ''Laura'' [láwra] and feira ''street market'' [féjra] and occasionally as the strong R as in bairro ''neighborhood'' [bájru]. When ''r'' follows a syllable that ends in a consonant as in desrespeito ''disrespect'' the strong R occurs systematically: [desrespéjtu]. Let us then consider the output for BP words which are claimed to have a wordinternal postvocalic /l/ followed by an ''r'' sound: (4) a. bilro ''lace bobbin'' b. guelra ''gill'' If we consider the dialects which pronounce an /l/ in syllable final position as EP and Southern Brazilian ones we observe that the strong R always occurs after the syllable final /l/. This is illustrated in (5a). In those dialects where /l/-vocalization occurs we find the data shown in (5b). (5) a. bi[lr]o ''lace bobbin'' gue[lr]a ''gill'' b. bi[wr]o or bi[wr]o ''lace bobbin'' gue[wr]a or gue[wr]a ''gill'' 3 The weak 'r' also occurs in branching onsets: prato [prátu] ''plate''. This environment is not relevant to the
Paper presented at the 3rd UKLVC-University of York England - July 2001 Page 5 The first syllable in the words in (5b) ends in [w] for BP speakers. Concerning the output for the ''r'' sound which follows [w] we get either the tap ''bi[wr]o,gue[wr]a'' or the strong R ''bi[wr]o,gue[wr]a''. Different speakers may vary the choice for the ''r'' sound depending on the word in question. That is, a speaker may have ''bi[wr]o'' and ''chi[wr]ear''. This shows that the same speaker does not have a single pattern as expected if the sound change from /l/ to [w] applies systematically the same way to any word which displays the structural environment requested. The fact that either the tap or the strong R may occur in the words in (5b) show a similar behavior of ''r'' sounds when followed by a vowel (ca[r]o, ca[r]o) or a diphthong (Lau[r]a, bai[r]o). If the glide [w] behaved like a consonant in ''bilro, guelra'' the ''r'' sound should systematically be the strong R (cf. des[r]espeito) and this is not the case in BP. Note, thus, that in those cases where the /l/ is pronounced as in EP and South Brazil (cf. (5a) the strong R always occur. Our proposal is that ''bilro, guelra'' in (5a) dialects have inputs such as /bíwro, bíwro/ and /g wra, g wra/. We assume that these competing forms show that these words have not yet completed their history in the language. Thus, the input for the alleged word-internal postvocalic /l/ is in fact /w/: /káwda/ for both calda ''syrup'' and cauda ''tail''. issue under discussion and thus will not be addressed in this paper.
Paper presented at the 3rd UKLVC-University of York England - July 2001 Page 6 In search of further supporting evidence we will consider the otehr two set of data which were pointed out earlier in this paper. Let us now consider the behavior of the so-claimed postvocalic /l/ in word final position. Analysis which propose inputs such as /sál/ for sal ''salt'' intend to account not only for the occurence of [l] in some derived forms such as [saléjru] saleiro ''salt shaker'' but they also aim at accounting for plural forms of the so-claimed final /l/ forms. It is important to mention that forms which end in a /w/-diphthong make their plural by adding /s/ to the singular form: degrau-degraus ''steps''. On the other hand forms which end in a so-claimed word-final /l/ make their plural by deleting the final /l/ in the singular form and adding /is/ to it: sal-sais ''salts''. Notice that the arguments for assuming an input with word-final /l/ in order to account for plural forms is actually ad hoc. In fact when using the plural forms of the so-claimed final-/l/ words (sal-sais) or words ending in a /w/-diphthong (degrau-degraus) BP speakers typically have three allternatives. Following the most current pattern for synchronic BP they only use the plural mark in the article: ''os sal'' and ''os degrau''. They may also use the plural ending inadequately: ''os sa[ws]'' and ''os degra[is]''. Of course the third alternative is to use the standard plural forms ''os sais'' and ''os degraus''. We argue that this behavior shows that the so-claimed final-/l/ words and the words ending in a /w/-diphhtong display the same behavior as other words which have irregular plural forms in BP: ovos, pães, irmãos, leões, etc. So, they should be treated as exceptional individually learned words. Thus, there is no need to assume a word-final /l/ to account for differences involving plural forms of syllable final /l/ and final /w/-
Paper presented at the 3rd UKLVC-University of York England - July 2001 Page 7 diphthongs words in BP. The output for both cases is a syllable-final /w/: /degráw/ degrau ''step'' and /sáw/ sal ''salt''. Let us finally consider the relationship between forms such as ''sal, saleiro, salgado'' for ''salt, salt shaker, salty''. We ought to account for the fact that an [l] appears in derived forms such as [saléjru] saleiro ''salt shaker'' being that we assume a representation for the underived form as /sáw/ sal ''salt''. We argue that the BP sound structure does not typically allow for intervocalic glide /w/ in pretonic position. Thus, a form like [sawéjru] may be interpreted as not fitting the sound pattern of the language. Considering that [sawéjru] is not adequate speakers find the possibility of an output such as [saléjru]. It remains for us to address how words such as ''sal, saleiro, salgado'' are related in the language. We claim that they may have a semantic relationship but not a phonological one. Thus, these words are lexically listed as individual items: /sáw, saléjro, sawgádo/. In this paper we argue that there is no syllable final-/l/ in BP. We analysed the so-claimed syllable final-/l/ words in medial position (calda), in word-final position (sal) and the related forms (saleiro, salgado). Our proposal is that the so-called process of lateral vocalization actually reflects a sound change which has been completed for most words which actually have a glide /w/ in syllable final position. Some words such as ''bilro, guelra'' - still display some competing forms as a consequence of an undergoing sound change.
Paper presented at the 3rd UKLVC-University of York England - July 2001 Page 8 Bibliography BYBEE, Joan (1995). Regular Morphology and the Lexicon. Language and Cognitive Process, v. 10, N. 5, pp.. 425-455 CÂMARA JR., J. M. (1970). Estrutura da Língua Portuguesa. Editora Vozes. Rio de Janeiro. CALLOU, D. & Y. LEITE (1990). Iniciação à Fonética e a Fonologia. Editira Zahar. Rio de Janeiro. CHEN, M. Y. & WANG, W. S-Y. (1975). Sound Change: Actuation and Implementation. Language, V.51, N. 2, pp..255-81 CRISTÓFARO-SILVA, Thaïs. (2001). Fonética e Fonologia do Português Roteiro de estudos e guia de exercícios. Editora Contexto. São Paulo. (2000). Sobre a quebra de encontros consonantais no português brasileiro. Estudos Lingüísticos XXIX. Unesp. São Paulo. FIDELHOLTZ, J. (1975). Word Frequency and Vowel Reduction in English. Papers from the 11 th Regional Meeting Chicago Linguistic Society pp..2000-213 HINSKENS, F. et al. (2000) Um balanço de dados e teoria no Estudo da Variação e da Mudança Fonológica. Porto Alegre: Letras de Hoje. V.35, n.1, pp.. 7-46 KIPARSKY, Paul. (1995). The phonological basis of sound change. pp.. 167-197. In: Goldsmith J. (ed). KIPARSKY, Paul (1968) Linguistics Universals and Linguistic Change. IN: BACH & KAMS (Ed) Universals in Linguistic Theory, New York pp.. 171-202 LABOV, W.(1994) Principles of linguistic Change. A proposed Resolution of the regularity question. Vo1: Internal Factors. Basil Blackwell, in press. pp. 143. LORD, R. & ZUNG, C.T. (1992) How does the lexicon work? In: Journal of the International Linguistic Association. Vol 43. pp.349-373. MOLLICA, Maria Cecilia.(1999) O início da Variação e o papel dos parâmetros controladores. OHALA. John J. (1993). The phonectics of sound change. Reprinted From: Historical Linguistics: Problems and Perspectives. London/New York: Longman. Pp. 237-278. OLIVEIRA, M. A (1983). Phonological Variation and Change in Brazilian Portuguese: The case of the liquids. PhD. University of Pensylvannia. (1991) The neogrammarian controversy revisited.in: International Journal of the sociology of Language :Berlin. Pp. 93-105. (1999). Reanalisando o processo de cancelamento do (r) em final de sílaba. Revista de Estudos da Linguagem. V. 6. Number 2. Belo Horizonte. PHILLIPS, Betty (1998) Lexical diffusion is not lexical analogy. Word, V.49, N.3 (1984) S. Open syllable lengthening and the ormulum. Language, V.60, N.2 pp. 320341. TASCA, M. (2000) A preservação da lateral alveolar na Coda: uma explicação possível. Letras de Hoje, V. 35, N.1, p.331-354. WANG, W. S-Y. & LIEN, C. (1999) Bidirectional diffusion in sound change. In: Historical Linguistics, London: Longman, pp.345-400 WANG, W. S-Y (1969) Competing Changes as a cause of Residue. Language, N.45, pp.9-25. WANG, W. S-Y. (1990) Theoretical Issues in studying Chinese Dialects. V.25, N.1 pp.1-34