|
Article Excerpt
ABSTRACT. This is a longitudinal study on the emergence of complex onsets in the speech of one monolingual Spanish-speaking female child complemented with data from two additional female children. It shows that they tend to reduce onset clusters to stops, a widely-attested crosslinguistic process. This reduction responds to a hierarchy of segmental constraints, which is informed by the Sonority Sequencing Principle. For the first time in the Spanish acquisition literature, this study reports on vowel epenthesis in complex onsets, thereby providing evidence against the U-shape hypothesis and further showing that it is not unique to European Portuguese. I argue that the reason for inserting a vowel is due to the propensity to conform to the universal CV structure. The paper also includes a discussion on the relative ranking of the acquisition of liquids, where *RHOTIC-ONS outranks *LATERAL-ONS as the data show that /l/ is acquired before /r/. *
1. INTRODUCTION. Some earlier and current studies on children's phonological development have centered their attention on the perception and production of individual segmental features. This feature-based approach has been extended to examine the acquisition of phonological features within the context of a larger prosodic domain, namely, the syllable, particularly with respect to the subsyllabic constituents, the onset and coda. Generally, many studies have argued that the acquisition of a single consonant or grouped consonants in these constituents appears to be driven by several phonological or morphological factors, such as the preservation of their segmental features in stressed syllables (Revithiadou anal Tzakosta 2004), their unmarked status (Jakobson 1968, Beers 1996, Grijzenhout and Joppen 2002), headedness (Jongstra 2003, Goad and Rose 2004), morphological markedness (Lleo 2003, Freitas 2003), frequency of production (Stites, Demuth, and Kirk 2004), relative sonority (Fikkert 1994, Lleo and Prinz 1996, Ohala 1999, Barlow 2003, Pater and Barlow 2003, Gnanadesikan 2004, Barlow 2007), perceptibility (Cote 2004, Howe and Pulleyblank 2004, Vanderweide 2005), or a combination of some of these (Stites, Demuth, and Kirk 2004, Cote 2004). Regardless of their theoretical approach, most researchers have argued that when acquiring consonants, children tend to produce the universal CV syllabic prototype. Many of the researchers cited above have also maintained that in the development of complex onsets the least sonorous consonants are first to emerge. To illustrate, when children hear a stop+liquid cluster and attempt to articulate it, they tend to produce the stop rather than the sonorant consonant. In codas, the opposite has been observed; that is, children resolve the perception of a final liquid+stop sequence by favoring the more sonorous liquid while omitting the stop.
In this study, I investigate the emergence of complex onsets in the speech of Seihla, a monolingual Dominican Spanish-speaking girl, and for cross-dialectal comparisons I complement it with additional data coming from two female Mexican children. (1) I show that both she and the Mexican children are inclined to reduce onset clusters, favoring the survival of the least sonorous consonant, as has been attested in the development of onsets in various Spanish dialects as well as in other languages. (2) Thus, this study constitutes a dialectal addition to the growing body of works which hypothesizes that consonantal onset reductions are guided by the Sonority Sequencing Principle (SSP). This principle essentially states that an onset cluster must increase in sonority toward the syllable peak or nucleus and a coda cluster must decrease in sonority from the nucleus.
Additionally, the collected data show a first documented occurrence of an epenthetic vowel in Spanish as Seihla begins to master complex onsets. Following Freitas (2003), I argue that vowel insertion to resolve complex onsets occurs because of children's propensity to conform to the universal CV structure. As is the case with the Portuguese children, the emerging speech of the children analyzed in the present study provides many instances of syllables without onsets, which, in OT terminology, calls for a low ranking of the ONSET CONSTRAINT which sanctions onset-less syllables. Contrary to the children in Freitas' study (2003), I demonstrate that Seihla presents no evidence of the U-shape development strategy exhibited by European Portuguese-speaking children. This strategy refers to the development of stages that begins by reducing clusters to a singleton, passing next through intermediate stages of full cluster production and vowel epenthesis, and then returning to full cluster production. In other words, rather than producing the U-shape development/tinbre/ [right arrow] ['timbe] [right arrow] ['timbre]/['timb[??]re] [right arrow] ['timbre] 'door bell', section 7 will show that Seihla just evinces the progressive acquisition/tinbre/ [right arrow] ['timbe] [right arrow] ['timb[??]re] [right arrow] ['timbre].
The most recent OT approach of Correspondence Theory (McCarthy and Prince 1995), which expands on earlier OT work by Prince and Smolensky (1993), can be usefully exploited to elucidate the range of consonant cluster reductions and epenthesis effected in the acquisition of the three children's speech. In OT the linguistic competence of a speaker corresponds to a set of hierarchically ranked, violable constraints. For any underlying input there is a universal function referred to as GENERATOR that supplies an infinite number of potential outputs. The hierarchy of constraints chooses the candidate that best satisfies the constraints. There are two basic types of ordered constraints I will be considering here. The first is the markedness constraint, which is a universal restriction that acts against marked structures, especially for syllable structure in this study. The second type is the faithfulness constraint, which requires a correspondence of identity between the lexical input and its surface output. This means that segmentai deletions and epenthesis must be avoided, whenever possible. Since these constraints are universal and their rankings reflect possible languages, it is generally presumed in acquisition research that the learner must also acquire the specific rankings provided by her own language. An underlying assumption of language acquisition in OT is that children's initial language contains a grammar in which structural constraints rank over faithfulness constraints (Smolensky 1996, Tesar and Smolesnky 2000, Davidson, Jusczyk and Smolensky 2004). The grammar, however, does not remain at this primordial state; it evolves, possibly going through an intermediate stage (Levelt and van de Wijver 2004), until reaching a final adult grammar in which faithfulness is promoted over structural constraints. During their onset acquisition, the data obtained appear to support this initial state of ordered rankings.
To frame the above discussion, I have organized the paper as follows. Section 2 presents an overview of the Spanish syllable structure and a brief introduction of the SSP. Section 3 provides an overview of the methodology employed. Section 4 gives a snapshot of the single onsets available to Seihla as she develops complex onsets. Section 5 presents the bulk of complex onsets produced by Seihla and the other two children. Section 6 outlines all data presented in previous sections and begins to analyze them in terms of the framework of Optimality Theory. In section 7 I look at the development of complex onsets, the order of liquid acquisition, and the insertion of a schwa as an onset simplification strategy. Section 8 provides a conclusion.
2. SPANISH SYLLABLE STRUCTURE. Let us look first at the basic structure of the syllable that I will follow (Harris 1983). Essentially, I am going to assume the normal speech of Latin American adult speakers in which coda consonants are pronounced. Bearing in mind the phonological inventory given in (1), a basic Spanish syllable is structured as in (2).
(1) /p t k b d g f s m n [??] l rt[??] j x/h w j/
(2) ILLUSTRATION OMITTED]
The syllable may be headed by an optional Ons(et) and an obligatory Rhyme, as shown in (2a). Since the focus of this study is the onset, I will disregard the structure of the rhyme and will focus instead on the onset. In (2b), the X represents a skeletal slot to which the C(onsonant) is attached; any one of the consonants in (1) may occupy that C, thus being a viable simple onset. This consonant assignment is essentially guided by the universal tendency to privilege the basic, non-branching CV syllabic prototype. The natural assumption here is that when confronted with the unstructured segmental sequence VCV, languages are likely to construct their syllable by providing it with an onset first, thus syllabifying this string as V.CV, rather that VC.V due to the Maximal Onset Principle (Selkirk 1982, 1984, Steriade 1982). Similarly, upon encountering an unstructured syllabic string like CVC, Spanish first syllabifies it as CV.C, and then incorporates the final C as a coda in a second pass, yielding CVC. This CVC syllabic type can be seen in the first and final syllables of the word ful.mi.nar. The tendency to accept the CV canonical shape in Spanish is not always respected, as shown by the first syllables of a.plicar, a.presar, om.bligo, a.tlas and a.fronta, a fact seen by the representation (2a). An onset may be more complex as in (2c), consisting of the maximal sequence [C.sub.1][C.sub.2] where [C.sub.1] must be narrowly interpreted as an obstruent stop or the fricative/f/, and [C.sub.2] as a liquid, except for the */dl/ sequence which never surfaces. (3) No other kinds of [C.sub.1][C.sub.2] onsets exist in Spanish. (4) (3) and (4) provide samples of word initial and internal onset realizations, shown in bold types and marked by the syllable boundary symbol [.]:
(3) Non-branching onsets (5) petaca [pe.'ta.ka] 'tobacco pouch' dato ['da.to] 'datum' fulminar [ful.'mi.nar] 'fulminate' loro ['lo. ro] 'parrot' rosado [ro.'sa.[??]o] 'pink' gentio [x/hen.'ti.o] 'crowd' bodega [bo.'[??]e.'[??]a] 'grocery store' galeno [ga.'le.no] 'physician' sena, ['se.na] 'signal' mucho ['mut[??]o] 'a lot' carreta [ka.'re.ta] 'cart' cabulla [ka'[??]u.ja] 'hemp' (4) Possible complex onsets playa ['pla.ja] 'beach' premio ['pre.mjo] 'award' blanco ['blan.ko] 'white' bruja ['bru.ha] 'witch' Nestle [nes.'tle] 'Nestle' drama ['dra.ma] 'drama' gloton [glo.'ton] 'greedy' crema ['kre.mal 'cream' gremio ['gre.mjo] 'syndicate' flota ['flo.tal 'fleet' fracaso [fra'ka.so] 'failure' aplicar [a.pli,'kar] 'aplicar' apresar [a.pre.'sar] 'to arrest' ombligo [om.'bli.[??]o] 'navel' cumbre ['kum.bre] 'summit' atlas ['a.tlas] 'atlas' tundra ['tun.dra] 'tundra' mangle ['man.gle] 'mangrove tree' suscribe [sus.'kri.[??]e] 'subscribe" mugre ['mu.[??]re] 'dirt' conflicto [kom.'flik.to] 'conflict' afrontar [a.fron.'tar] 'to face'
There are no complex onsets headed by palatals or coronal fricatives. Some dialects admit the combination [hl] as in the sumame Majluta [ma'.hluta] in Dominican Spanish. Contrary to other dialects, Mexican and Dominican Spanish tend to group [tl] in onsets.
To account for the complex onsets formation in (4), I will presume that the SSP in (5) is at work in Spanish. This principle has been variously captured by Foley (1970), Hooper (1976), Kiparsky (1979), Steriade (1982), Selkirk (1984), and Clements (1990), among others.
(5) Sonority Sequencing Principle (SSP, adapted from Kenstowicz 1994:254) Onsets must rise in sonority toward the nucleus and codas must fall in sonority from the nucleus
The SSP (5) dates back to Sievers (1881), Jespersen (1913), Saussure (1916), and Grammont (1933). According to these authors, languages tend to display a very common pattern in the manner in which consonants order themselves with respect to a syllable-peak. Given that the SONORITY concept has been difficult to define both phonetically and phonologically, here I intuit it to mean the acoustic properties of the loudness of a sound in relation to another with equal length, stress, and pitch, which interface with major class features, the latter in the sense of Clements (1990:284). Assuming that these...
|
