0.0 Introduction 

This paper presents a study of the phenomenon of resyllabification in the dialect of a Spanish speaker from the municipality of San Lorenzo, in Puerto Rico. Section 1.0 reviews a well accepted previous description of resyllabification in Spanish and demonstrates the need for a more constrained rule, specifically one which limits resyllabification to specific prosodic constituent boundaries. Section 2.0 is a discussion of the prosodic hierarchy in Spanish and Section 3.0 looks at the behaviour of two syllable final phenomena: /s/ deletion, and the alternation of [r] and [l]. These phenomena are both described in terms of their occurrence in the speech of a native informant. 

1.0 Resyllabification: a previous definition. 

Resyllabification is a common phenomenon in the Romance languages. Concerning the process in Spanish, Harris (1983) wrote the following: 

Data will be presented in Section 3.0 which show that resyllabification cannot occur across all types of word boundaries, but in fact is limited to the domain of the Intonational Phrase. 

2.0 Prosodic domains in Spanish 

It has been propsed by Nespor and Vogel (1986) that there is a hierarchy consisting of seven prosodic constituents. Each constituent in this hierarchy must be made up exclusively of the next lower constituent. This hierarchy is shown in (2). 
 

(2)	phonological utterance (U)  à intonational phrase (I)  à phonological phrase (j)  à clitic group (C)  à phonological word (w)  à foot (S)  à syllable (s)

  
Because resyllabification is a phenomenon that seems to work at the syllabic level, an understanding of Spanish syllable structure is crucial. The basic structure is described in the manual of the Real Academia (1983), and also in some detail in Harris. This structure is shown in (3). 

(3) The syllable in Spanish: 

(N = nucleus; Cd = coda) 

Spanish has a relatively limited group of allowable onsets and codas. The result of resyllabification may never be a branching onset; that is, the following word must begin with a vowel because any onset will block resyllabification. Only possible syllable final consonants are shown in (4), because, as just mentioned, syllable initial consonants will block resyllabification. 

 
 

(4)		consonant		word medial		word final
	/s/		pas-ta		ca-pas
	/r/		bar-co		mi-rar
	/l/		bol-sa		me-tal
	/d/		mad-re		ver-dad
	/n/		man-ta		mon-tón
	/q/		mez-cla		ca-paz
	/p/		cap-ta		*
	/b/		cab-le		*
	/f/		suf-ro		*
	/t/		cuat-ro		*
	/k/		dec-laro		*
	/g/		diag-nosis		*

 
A few syllable final clusters are possible word internally in Spanish, but not word finally. The second segment in these clusters is always /s/ and is always preceded by /b, d, k, r,/ or /n/. These clusters usually occur near the junction of an initial derivational morpheme, as in substancia, adscribir, constancia, instinto, perspicaz, and transformar (Real Academia Española 41-43). Harris does mention a small number of words with consonant clusters in word final position, yet they are all of obvious foreign origion (1983, 14, 28). According to the Real Academia, "La coda compuesta dista mucho de ser en español una forma canónica. Casi todas las palabras en que aparece son extranjerismos, sentidos como tales por la conciencia idiomática" (42). 

Resyllabification can occur in Spanish only if there is a word final consonant, /s, r, l, d, n/ or /q/, followed by a vowel in the next word. 

The foot, the phonological word, and the clitic group all dominate the syllable. However, because resyllabification occurs across all of these domains, they will not be defined here. 

The constituents above the phonological phrase are less rigorously defined in terms of morphology and syntax than those below it on the hierarchy. A single string of phonological phrases can be divided into intonational phrases in a variety of ways, depending on the style of speech employed, and the particular intonational pattern used by the speaker. A very long string may consist of a single intonational phrase if uttered very quickly or in a colloquial style. However, it may also be broken up into a number of smaller phrases. An example of this restructuring in Spanish (5), is given in Nespor and Voegel (212-213). 
 

(5)	(a)	(i)	[Quería invitar a Juan Carlos Carmen y Beatriz]I
		(ii)	[Quería invitar a Juan]I [Carlos]I [Carmen]I [y Beatriz]I
	(b)	(i)	[Eso es el escorpión que espantó al tucán que espantó al faisán que se paseaba en el jardín]I
		(ii)	[Eso es el escorpión]I [que espantó al tucán]I [que espantó al faisán]I [que se paseaba en el jardín]I

 

It can be seen in (5) that the intonational phrase can vary in length from that of the phonological phrase to that of an entire phonological utterance. The boundaries of the intonational phrase do not coincide regularly with the boundaries of any syntactic constituent. 

By the same token, the phonological utterance, as the largest constituent in the phonological hierarchy, need not share its boundaries with the sentence simply because the latter is the largest constituent in the syntactic hierarchy. Vogel (1986) shows that many phonological rules apply across sentence boundaries indicating that the domain of the phonological utterance may include more than one sentence. 

3.0 Two syllable boundary phenomena 

The deletion of /s/ and the alternation between two allophones of /r/, [r] and [l], are two phenomena which, in a traditional segmental framework, have been described as occurring at the syllable boundary. Both of these are common phenomena in dialects throughout the Spanish speaking world. 

Earlier rules presented to account for these phenomena would have looked something like those in (6). 
 

(6)	(a)	Rule for /s/ deletion.  /s/ à ø / [ . . . ___ ]s
	(b)	Rule for [r] and [l] alternation.  /r/ à [l] / [ . . . ___ ] s

The rules presented in (6) do not adequately describe, let alone explain, the vast amount of variation found in the speech of a single dialect of Spanish, or even of a single speaker. Some might claim that the rules are optional and that the phenomena are in free variation. It will be argued here that the differences displayed depend on the style of speech used in a particular discourse. 

3.1 The data 

The data were collected from a native of Puerto Rico, who now lives with her immediate family in Newark, Delaware. All of the data were elicited in approximately one hour using a tape recorder in the informant's home. Three elicitation techniques were used. The first two dealt with /s/ deletion. First, the informant was asked to repeat a series of sentences containing environments in which the phenomenon to be studied occurred. Each sentence was pronounced by the researcher once, and the informant was asked to repeat it five times. If, during the repetitions, she deleted twice, it was counted as a deletion. The second task was a set of value judgements. Given individual words in Spanish, the informant was asked if /s/ could be deleted, or if she could imagine anyone pronouncing it with /s/ deleted. 

Lastly, the researcher and the informant carried on a free conversation, during which, occasionally, other members of her family became involved. This last set of data has been used almost exclusively for the conclusions of this study. This is because the informant differed greatly in her production from one task to the next. Simply put, she was very careful to speak in a way that she considered proper during the first two tasks. It was felt that the the data gathered in these two tasks were not indicative of her normal, everyday pronunciation. 

3.2.1     /s/ deletion in Spanish 

The phonemes /s/ and /n/ have remained in Spanish after centuries of erosion in Romance (Poplack 1980, 371). Their weakening and deletion in word and syllable final position is an ongoing process in many dialects of Spanish. For present purposes, any significant weakening of /s/ will be considered deletion. In most dialects the phenomena varies from heavy aspiration, through weak aspiration, to complete deletion. However, discrepancies in the application of a rule for /s/ deletion, (6)(a), within a single dialect, have been explained somewhat satisfactorily by arguing that as /s/ is deleted in more and more possible environment, the likelihood of ambiguity increases (Poplack; also Terrell 1978). Still there is a great deal of dialectical variation that cannot be explained in this way. Some Spanish speakers occasionally delete an /s/, while others do it most of the time. This phenomenon is particularly prevalent in Puerto Rico, where the plural /s/ is deleted 65% of the time, more that any other Carribean dialect studied (Poplack 383). 

3.2.2     /s/ deletion in this dialect. 

The informant deleted much less during the repetitions and judgements than during the conversation. In the sentences she was asked to repeat there were 21 possible deletions, (that is, 21 occurrences of /s/ in implosive position), yet she deleted only 4, or 20%. Similarly, when asked to make judgements, she claimed that her dialect would only allow 5 out of 11 deletions, or 45%. However, during the conversation, she deleted consistently in all appropriate environments (98%). In fact she only pronounced an /s/ in syllable final position six times, and all of these occurred in either the article los, (four instances), or the quantifier dos, (two instances). Deletion did occur with los on three occasions, these however all preceded consonants, whereas the lack of deletion preceded vowels. Compare (7)(a) where deletion did occur with (7)(b) where it did not. 
 

(7)	(a)	lo(s) pe(s)cado(s)  lo(s) colore(s)
	(b)	los hijo(s)  los otro(s)

 
An easy conclusion would be that deletion is blocked before vowels, however, as shown in (8), deletion always occurred after las and there are many other examples of deletion before vowels. 
 

(8)	la(s) playa(s)  la(s) almeja(s)  ello(s) e(s)tán

  
The word dos also occurs with /s/ deleted before a vowel, as in (9). 
 

(9)	do(s) han ido

  
It can be seen in (10) that /s/ may be deleted within any domain in the prosodic hierarchy. 
 

(10)	(a)	Deletion between syllables.  [pa(s)] s [to] s
	(b)	Deletion between feet.  [bu(s)] S [car] S
	(c)	Deletion between words  [la(s)]w [playa(s)]w
	(d)	Deletion between clitic groups.  [no(s)otro(s)]C [teníamo(s)]C
	(e)	Deletion between phonological phrases.  [no(s)otro(s) teníamo(s)]j [un mapa]j
	(f)	Deletion between intonational phrases.  [Tú lo tiene(s)]I [¿verdad que sí?]I
	(g)	Deletion between phonological utterances.  [la(s) almeja(s) son di(s)tinto(s) colore(s).]U [E(s) bien precioso.]U

 
These data suggest that we can, at least temporarily, leave the rule for /s/ deletion as presented in (6)(a) because /s/ is deleted at the end of all syllables, within and between all prosodic domains. 

3.3.1     [r] and [l] alternation in Puerto Rico. 

The variation in the application of rule (6)(b) is not as easily explained. The rule only applies within a certain prosodic domain, namely, the intonational phrase. 

The alternation of [r] and [l] is a common phonological phenomenon in Puerto Rico. Canfield (1981, 76) tells us that "except for the sections of San Sebastián, Lares, and Las Marías in western Puerto Rico, there is a strong tendency toward acoustic equivalence of /l/ and /r/ syllable final. . . ." Navarro Tomás (1948, 80) states that the [l] allophone is very common in the northeast, whereas the [r], typical of Standard Spanish is heard in the southwest. The informant used in this study is from the municipality of San Lorenzo and speaks a dialect of Puerto Rican Spanish that is somewhere between the two poles. San Lorenzo is about half way between the northeast corner and the center of the island. 

3.3.2     [r] and [l] alternation in this dialect 

In contrast to the way she applied the /s/ deletion rule, the informant did not use the [l] allophone of /r/ in all syllable final environments. Indeed, if this were the case, neutralization would result with no distinction between the [l] allophone of /r/ and the phoneme /l/ in syllable final position. Two distinct conditions affect the application of the rule as it is stated in (6)(b). First, and most obvious is that [r] does not seem to become [l] if the next syllable begins with a vowel. The rule can thus be rewritten as shown in (11). 
 

(11)	Revised rule for [r] and [l] alternarion.  /r/ à [l] / [ . . . ___ ]s [C . . . ]s

  
The rule in (11) loses the similarity that it had with the /s/ deletion rule in (6)(a). This researcher argues that the change is not actually blocked by the vowel in the next syllable, but rather, it is blocked because the /r/ is resyllabified into the next syllable. The resyllabification rule proposed in (1), if applied before the rule in (6)(b), will account for this condition, and will allow both rules to be maintained as presented in (6). Resyllabification must take place after /s/ deletion, because /s/ is deleted in all proper environments, but it must take place before the [r] to [l] change, at least in this particular dialect of Spanish. This rule ordering is shown in (12). 
 

(12)	(a)	Rule for /s/ deletion.  /s/ à ø / [ . . . ___ ]s
	(b)	Rule for resyllabification (modified from (1)).  [ . . . C] à [V . . . ]s Æ [ . . . ]s [CV . . . ]s
	(c)	Rule for [r] and [l] alternation.  /r/ à [l] / [ . . . ___ ]s

  
The second condition that affect the [r] and [l] allophones is that /r/ does not always remain an /r/ before vowels. That is, sometimes the rule in (12)(c) applies where it would be expected that resyllabification should have removed the environment. The data show that this happens only in domains greater than that of the intonational phrase. That this phenomenon does occur in the foot and word domains is shown in (13). 
 

(13)	/r/ becomes [l] between syllables and feet.
	(i)    fuerte à [[[fuel]s [te]s]S]w
	(ii)    verdad à [[[vel]s]S [dad]s]S]w

  
The behaviour of syllable final /r/ at the boundary between words is shown in (14). In the data thare are no examples of resyllabification to a word beginning with a vowel within the domain of the clitic group. However (14)(b) presents what is predicted to happen in that environment. 
 

(14)	/r/ becomes [l] between words.
	(a)    before # consonant          levantarme à [[levantal]w [me]w]C
	(b)    before # vowel          levantaros à [[levantar]w [os]w]C

The behavior of syllable final /r/ within phonological and intonational phrases is shown in (15) and (16), respectively. Notice that within both of these domains /r/ resyllabifies blocking the change to [l]. 
 

(15)	/r/ becomes [l] between clitic groups.
	(a)    before # consonant          el primer día à [[el primel]C [día]C]j
	(b)    before # vowel          por ahora à [[por]C [ahora]C]j

  
 

(16)	/r/ becomes [l] between phonological phrases.
	(a)    before # consonant          (i)     llevar la cabra à [[lleval]j [la cabra]j]I          (ii)     hablar duro à [[hablal]j [duro]j]I
	(b)    before # vowel          (i)     cruzar el puente à [[cruzar]j [el puente]j]I          (ii)     te puedes meter en el agua à [[te puedes meter]j [en el agua]j]I

Finally, (17) shows the boundaries between intonational phrases and utterances. The allophone [r] never occurs at this juncture because resyllabification will never occur here. 
 

(17)	/r/ becomes [l] between intonational phrases and utterances.
	(a)    before # consonant
	(i)    Tenía que venir, darme un baño y... à [[Tenía que venil]I [dalme un baño y...]I]U
	(ii)    En español tiene que ser. ¿Qué tú dijiste? à [[En español tiene que sel]U [¿Qué tú dijiste?]U
	(b)    before # vowel
	(i)    Carlos no quiere aprender, y no quiere atender à [[Carlos no quiere aprendel]I [y no quiere atendel]I]U

  
4.0 Conclusion 

The data show that the domain of resyllabification in this dialect of Spanish is the intonational phrase. A new resyllabification rule is presented in (18). 
 

(18)	Final resyllabification rule:  [ . . . C]s [V . . . ]s à [ . . . ]s [CV . . . ]s / [ . . . ___ . . . ]I

  
Of the two implosive phenomena studied, the first, /s/ deletion, gives us few clues for the domain of resyllabification. The /s/, if indeed it exists in the underlying representation of this particular dialect, must be deleted before any resyllabification occurs. The second phenomenon must, however, occur after resyllabification. The environment for /r/ to [l] alternation is bled by the rule for resyllabification, that is, many syllable final /r/s resyllabify to the next word, and are therefore not available for the change to [l]. This bleeding order is desplayed in (19). 
 

(19)	(a)	Rule for /s/ deletion.  /s/ à ø / [ . . . ___ ]?
	(b)	Rule for resyllabification (modified from (1)).  [ . . . C]s [V . . . ]s à [ . . . ]s [CV . . . ]s / [ . . . ___ . . . ]I
	(c)	Rule for [r] and [l] alternation.  /r/ à [l] / [ . . . ___ ]s

  

WORKS CITED 
Canfield, D. Lincoln. Spanish Pronunciation in the Americas. Chicago: University of Chicago Press, 1981. 

Harris, James W. Spanish Phonology. Cambridge, MA: MIT Press, 1969. 

__________. Syllable Structure and Stress in Spanish: A Nonlinear Analysis. Cambridge, MA: MIT Press, 1983. 

Navarro Tomás, Tomás. El español en Puerto Rico. Río Piedras, PR: Editoral de la Universidad de Puerto Rico, 1948. 

__________. Manual de pronunciación española. Madrid: Consejo Superior de Investigaciones Científicas, 1982. 

Nespor, Marina. "The Phonological Word in Italian." Advances in Non-Linear Phonology. Ed. N. Smith. Dordrecht, Holland: Foris, 1984. 193-204. 

Nespor, Marina, and Irene Vogel. Prosodic Phonollogy. Dordrecht, Holland: Foris, 1986. 

Poplack, Shana. "Deletion and Disambiguation in Puerto Rican Spanish." Language 56 (1980): 371-385. 

Real Academia Española. Esbozo de una nueva gramática de la lengua española. Madrid: Espasa-Calpe, 1983. 

Resnick, Melvyn C. Phonological Variants and Dialect Identification in Latin American Spanish. The Hague: Mouton, 1975. 

Selkirk, E. "The French Foot: On the status of 'mute' d." Studies in French Linguistics 1 (1978): 141-150. 

Terrell, Tracy D. "Functional Constraints on the Deletion of Word-final /s/ in Cuban Spanish." Proceedings of the First Annual Meeting of the Berkley Linguistics Society (1975): 431-437. 

__________. "Sobre la aspiración y elisión de /s/ en el español de Puerto Rico." Nueva revista de filología hispánica 27 (1978): 24-38. 

Vogel, Irene. "External Sandhi Rules operating between sentences." Sandhi Phenomena in the Languages of Europe. Ed. Henning Andersen. Berlin: Mouton de Gruyter, 1986.