The Status of Word-Final [E] in Spanish

ABSTRACT. Traditionally the final vowel of Spanish forms like parte 'part', nube 'cloud', etc. has been analyzed as the result of WORD-FINAL EPENTHESIS (Harris 1986a,b, 1991, 1999). The main argument in favor of this proposal is that -e, the epenthetic vowel in Spanish, follows an UNSYLLABIFIABLE CONSONANT or CONSONANT CLUSTER (*nub, *part). This paper argues that status of this generalization is not as clear-cut as generally assumed and that word-final [e] is not always predictable. Furthermore, the benefits obtained by capturing a supposedly systematic aspect of the phonology are offset by the presence of EXCEPTIONLESS INITIAL EPENTHESIS, as in yugoslavo ~ eslavo, deletion of unsyllabifiable word-final consonants, for example club [klu], and by the problems faced by current accounts of pluralization. The analysis put forth argues that the lexical rule of FINAL EPENTHESIS has been dropped from the grammar of Spanish, resulting in grammar simplification and restructuring of word-final /e/ as an underlying terminal element. * INTRODUCTION: THE EPENTHESIS PROPOSAL. Traditionally the final vowel of Spanish forms like parte 'part', nube 'cloud', etc. has been analyzed as the result of word-final epenthesis (FE) (Harris 1986a,b, 1991, 1999). It often has been observed that, whereas terminal elements (TE) -a, -o can be preceded by any consonant/consonant cluster as seen in 1a,b and 2a,b, -e has a somewhat different distribution (Harris 1983, 1986a). Specifically, -e follows consonants and consonant clusters not permitted in word final position as shown in 1c, but it is not common after the set of consonants allowed in word-final position in 2c [d, l, n, r, s, [theta]]. (1) a. -o# b. -a# himno 'hymn' columna 'column' parto 'delivery' carta 'letter' canto 'chant' santa 'saint' pacto 'pact' acta 'minute' tubo 'tube' Cuba 'cuba' pito 'whistle' cita 'date' c. -e# d. -CC# solemne 'solemn' * parte 'part' * guante 'glove' * traste 'fret' (n.) * nube 'cloud' * (2) a. -o# b. -a# dedo 'finger' seda 'silk' polo 'pole' cola 'tail' cono 'cone' luna 'moon' cero 'zero' cera 'wax' caso 'case' casa 'house' cazo 'saucepan' caza 'hunt' c. -e# d. -C# sede 'seat' (n.) sed 'thirst' prole 'progeny' sol 'sun' pene 'penis' atun 'tuna' ere 'r' mujer 'woman' fase 'phase' as 'ace' cruce 'intersection' cruz 'cross' Harris (1986a: 113) points out that 'this highly skewed distribution' of -e must be the result of an epenthesis rule that inserts a mid vowel to parse unsyllabifiable coda consonants and consonant clusters; -e after syllabifiable coda consonants, as in 2c, is considered part of the underlying representation. The same position is adopted in Harris (1991, 1999). The main argument in favor of FE is that a rule of FE captures an important generalization with respect to the syllable structure of Spanish, namely that /CC, c, x, p, t, k, b, g, f/ are not syllabifiable word-finally. An analysis without FE would miss this generalization. Although this is a strong argument that has generally remained unchallenged, I argue that the matter needs to be reexamined. More specifically, I claim that the status of this generalization is not as clear-cut as generally assumed and that word-final -e is not always predictable. Furthermore, the benefits obtained by capturing a supposedly systematic aspect of the phonology are offset by the presence of exceptionless initial epenthesis, as in hemisferio 'hemisphere'~ esfera 'sphere', possible deletion of unsyllabifiable word-final consonants and consonant-clusters, for example club [klu], and by the problems faced by current accounts of pluralization. …