The Acquisition of Spanish Codas - University of Illinois at Chicago

The Acquisition of Spanish Codas: A Frequency/Sonority Approach
Author(s): Rafael Núñez-Cedeño
Source: Hispania, Vol. 90, No. 1 (Mar., 2007), pp. 147-163
Published by: American Association of Teachers of Spanish and Portuguese
Stable URL: http://www.jstor.org/stable/20063476 .
Accessed: 02/12/2013 12:33
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
.
American Association of Teachers of Spanish and Portuguese is collaborating with JSTOR to digitize, preserve
and extend access to Hispania.
http://www.jstor.org
This content downloaded from 128.248.155.225 on Mon, 2 Dec 2013 12:34:00 PM
All use subject to JSTOR Terms and Conditions
Theoretical Linguistics
Preparedby StanleyWhitley
The Acquisition of Spanish Codas:
A Frequency/Sonority Approach1
Rafael N??ez-Cede?o
University of Illinois at Chicago
researchers. They maintain
that its
coda has attracted the attention of many
acquisition
children acquire
responds to considerable variability. Recent studies show that Spanish-speaking
in stressed syllables before producing final ones. They also develop sonorants before obstruents and
acquire the plural I-si before Isl. The present study shows that prosody alone does not determine the codas that
occur in Spanish.
Instead, it claims that sonorant and fricative distributions in a syllable are tied to their
Abstract:
The
development
medial codas
frequency of production and relative prominence
tend to develop at about the same time. It appears
frequent occurrences
final Isl. The
Key
Words:
Frequency
Sonority Hypothesis,
of medial
Hypothesis,
Sonority
nasal
in the universal sonority hierarchy. Medial
and final codas
that there is no distinction between the acquisition of /-s/ and
in codas are explained as being licensed by a following onset.
medial/final
Principle,
Sequencing
codas, Optimality
trochee
Theory,
prosody,
Sonority
Hierarchy,
1. Introduction
1.1General
To
Issues on Acquisitions
ofCodas
speak about the acquisition of syllable structure one must first refer to Jakobson's
(1968) monograph on phonological oppositions, where he discusses the order of feature
acquisition as the first stage of language development. Afterwards, consonants and
vowels follow as thenext step toward thedevelopment of theC V syllable type,which constitutes
theunmarked universal syllable.More recent research on thenature of the syllable has revealed
thatthe syllable typeC VC follows developmentally thecore C V (Levelt, Schiller, and Levelt 2000).
Since Jakobson,many studies have delved into the nature of a single consonant onset, agreeing
thatthefirst to emerge are /ptk/.These are also first in the acquisition of complex initial clusters.
Typically,
while
most
cross-linguistic
studies
show
that in stop+sonorant
onsets,
stops
are
preserved
in fricative+obstruent combinations, children eliminate the fricative and preserve the
following obstruents (Freitas 2003, Gnanadesikan 2004, Ohala 1999, et al.).
Assuming that child phonology is adult-like, recent research on syllable-final consonants
has shown that theiracquisition correlates with the considerable variability in theiroccurrences.
German and Portuguese children produce codas at a very early stage of acquisition, while
English-speaking children favor open syllables (Grijzenhout and Joppen 1999). French children,
on the contrary,produce codas at a later stage but do so by first favoring word-final codas as
opposed tomedial ones (Rose 2000). Spanish, despite itsvariability of acquisition, appears to do
the opposite. Lle? (2003) argues thatSpanish children acquire medial codas triggeredby stress
before producing final ones. She also states that theydevelop sonorant before fricative codas, in
contradistinction to Portuguese childrenwho show a marked preference for acquiring fricative
before sonorant codas (Freitas,Miguel and Hub Faria 2001). Additionally, Lle? contends that
Spanish children show a precedence in acquiring codas of stressed determiners as opposed to
thefinal plural marker /-s/,
whose syllable is unstressed. She furtherclaims that,because of its
morphological status, /-s/is acquired before lexical Isl.
"The Acquisition
of Spanish
Codas:
Rafael
N??ez-Cede?o,
A Frequency/Sonority Approach"
90.1 (2007):
147-163
Hispania
This content downloaded from 128.248.155.225 on Mon, 2 Dec 2013 12:34:00 PM
All use subject to JSTOR Terms and Conditions
148
H?spanla 90March 2007
In this study I first examine the acquisition of codas inMadrid Spanish and look closely at
Lle?'s claims (2003) that surfacing codas relate to the development of prosodie and rhythmic
structures and to their acquisition in functionwords. She hypothesizes that this preference is
bound to syllable development as it interactswith stress.The stressed syllable, being thehead of
a trochaic foot, licenses the acquisition of codas.
Second, my central argument is that stress and rhythmare not the sole determinants in the
acquisition of Spanish. Instead, the distribution of final sonorants and fricatives correlates with
frequency of production and with relative prominence as expressed by the universal sonority
hierarchy (Ohala 1999). Crucial tomy reasoning is the assumption that an early grammar has
moved from a prosodically unmarked structure tomore adult-like syllabic structures.These are
the reasons why nasals, laterals, and fricatives in codas seem tomake their early entry into the
repertoire of sounds acquired early by Spanish children. The facts show thatmedial and final
nasal codas develop at about the same time, contrary toLle?, who asserts that children first ac
quire medial over final codas. It also appears thatno distinction ismade between the occurrence
of Isl and lexical Isl. The occurrences of final nasals are explained away as a result of theirbeing
licensed by the onsets of following obstruents (Ito 1989, Lle? 2003).
We will see that internalobstruent codas are altered during acquisition to thepoint of being
effaced, a process common to adults' speech in all Spanish varieties. This suggests that edu
cators should approach the teaching of pronunciation to English-speaking learners in a more
authentic way, for the latterare expected to pronounce obstruent codas, as their textbooks say
they should,when native Spanish speakers rarely do. The fact thatamong liquids the IV isfirst to
appear and that the tap / (*/
precedes the trill /r/,should help us understand why second-language
learnersmay follow this acquisition order regarding rhotics.
2. Spanish Syllable Structure
To frame the discussion within the context of Spanish, letus look first at thebasic structure
of the syllable that Iwill follow. Essentially, I am assuming the formal speech ofmiddle-class
Dominican speakers, which contains deliberately pronounced codas. Assuming the inventory
given in (1), the Spanish syllable is structuredas in (2).
(1)
/ptkbdgfsmnjilr
(2)
a
tj"jx/hwj/
(O)
Rhyme
Ax
CN
A
(C)
Coda
A (C)
(C)
A syllable consists of a rhymewith an obligatory Nucleus,2 which may be optionally pre
ceded or followed by a single consonant or two consonants. Any of the consonants in (1) above
may head the onset. A more complex onset consists of a stop or the fricative lil followed by a
liquid, except for the */dV cluster, i.e.,playa ['plaja] "beach," bianco ['blanko] "white," trama
['t rama]
"plot,"
drama
['?/rama]
"drama,"
crema
['k rema]
"cream,"gremio
['g remjo]
"syn
dicate," flota [7?ota] "fleet." This study focuses on the rhyme, in particular the optional codas.
Except for palatals, in adult speech all single consonants can close a syllable. Among these,
This content downloaded from 128.248.155.225 on Mon, 2 Dec 2013 12:34:00 PM
All use subject to JSTOR Terms and Conditions
The Acquisition
obstruents
occur
may
in guarded
as
speech,
shown
by
of Spanish
apto
['ap.to]
149
Codas
"apt,"
?tnico
['et.ni.ko]
"ethnic," and acto ['ak.to] "act" ([.] denotes syllable boundary). Isl, however, is the most
frequent obstruent in Spanish rhymes (Harris 1983: 17), as inmes [mes] "month," entonces
in all final
widespread
[al.'tu.
and
"then,"
[en.'ton.ses]
ra]
countless
positions,
delantal
"height,"
as
words
other
in came,
[de.Ian.'tal]
show.
tomar
"meat,"
['kar.ne]
are
There
"apron."
sonorant
The
coronals
l\
"to drink,"
[to.'mar]
coda
complex
clusters
consist of an obstruent or coronal sonorants followed by Isl such as adscrito
"adjunct"
obstruir
[o?z.'trwir]
construir [kons.'trwir]
clusters
in conservative
"to obstruct,"perspectiva
"to build." Adult
speech
style, while
[perz.pek.'ti.?a]
speakers may
sonorant+/s/
1 n/
are
altura
that may
[a?z.'krito]
"perspective,"
and
sparingly produce obstruent+/s/
groups
are more
recurrent,
especially
thenasal+ Isl combination. All obstruent+fricative clustersmay be simplified to a residual [s], i.e,
/obstruir/ is often heard as [os.'trwir].
The coronal In/behaves distinctly depending on the following segment.When followed by
a vowel
across
wa.'ka.tes]
word
"they
it surfaces
boundary,
eat avocados."
However,
as
when
as
coronal,
found
next
in comen
aguacates
to consonants,
['ko.mena,
it assimilates
their
place features: /tronpa/> ['trom.pa]3 "trunk,"/antes/> ['an.tes] "before," /manko/> ['marj.ko]
"lame,"/ko.men
kar
ne/>
['ko.merj
'kar.ne]
"they
eat meat."
Similarly,
assimilation
takes place
with the lateral IV only before coronal consonants within and across word boundary, as in /toldo/
>
['toi .do], "awning," /koltjo/> ['kol.tjb] "cork," /totaldetacfee/> [to.'tal .de. ta. je] "complete
detail."
The various phonetic manifestations of/n/have led phonologists to suggest it lacks point of
articulation and it is thus lexically under- or unspecified for thisfeature (Paradis and Prunet 1989,
1991). The output of/n/ is licensed by the onset of the following consonant (Ito 1988). In
utterance-final position, theplace default specification isfilled by the coronal feature.
3. The Acquisition
of Spanish Codas: Lle?'s Hypothesis
Lle? (2003) examines thedevelopment of codas in the speech of twomonolingual children in
Madrid, with an average age of 2.3. Her data show that the emergence of codas interactswith
prosody, though she also hypothesizes that semantics could potentially be a triggering factor.
Her study reveals thatcodas are produced when the syllable carriesmain stress and are avoided
when unstressed. In otherwords, medial codas4 in trochees surface earlier. In addition, medial
codas develop before final ones, which goes against the tendencies of English (Goad and Bran
nen 2003) and French (Rose 2000) toproduce final consonants beforemedial codas in children's
language. In Lle?'s analysis (2003: 278), unstressed medial codas with unspecified /n/surface,
whose phonetic varieties are due to theirbeing licensed by a following onset, thereforemaking it
easier for children to acquire.
She also claims that stress drives the occurrences of contextualized final IV and Inl, in the
definite and indefinitearticles el "the" and un "a(n)." However, as will be discussed in section 6,
this analysis is complex and must be discarded, because both articles must be restructuredby
switching unfooted prosodie materials into a footed structure.Lle? reasons thatdespite the fact
thatarticles are unstressed proclitics, Spanish carries secondary stress recursively from the left
of themain stressed syllable. As a consequence, these proclitics may carry secondary stress in
some instances, functioning, therefore,as head of a phonological phrase.
While some codas are the reflex of phonetic substance derived lexically, others, such as the
suffixalAn/ inverbs like trabajan "theywork" and theAs/ in casas "houses," denote plurality.
Since children need tomake themselves understood, it follows that these plural morphemes
should surface early. Lle? only examined theplural marker As/ and concluded thatbecause it is
not the head of a foot, itcannot surface.According toLle?, only the head of a foot, the trochee,
will allow codas. However, she predicts thatwhen final consonants begin to appear, those
carryingmorphological weight should appear firstbecause theyconvey semantic information. It
will be shown that this prediction does not hold.
This content downloaded from 128.248.155.225 on Mon, 2 Dec 2013 12:34:00 PM
All use subject to JSTOR Terms and Conditions
150
H?spanla 90March 2007
4. Two Hypotheses: Frequency inCoda Development
and the Sonority Cycle
analysis accurately presents the data of the children she studied, confirming her
stress effects contribute to the occurrence of coda acquisition inyoung children.
that
hypothesis
Other studies have supported her findings fordifferent languages. Klein (1981), Echols (1992),
Echols and Newport (1992) have revealed, for instance, that stress plays a significant role in the
production of codas inmultisyllabic words, while unstressed syllables fail to exhibit them.How
ever, additional studies focusing on syllabic constituencies have refined theirfindings. Schwartz
and Goffman ( 1995) demonstrated, for example, that thepresence of codas did not relate to stress.
Rvachew and Andrews (2002) found that intervocalic consonants preceding stressed or un
stressed syllables were produced as codas by preschoolers with language acquisition delay. Kirk
and Demuth (2003) have argued, on the other hand, thatan integratednumber of factors, among
Lle?'s
themprosodie environment,word-length, and position within theword, affect theoccurrences of
developmental codas. As for thefirst factor, theydetermined thatchildren are less likely to pro
duce codas in unstressed syllables than in stressed ones, thus indirectly validating Lle?'s
results. This may be explained because at a certain age English codas inunstressed positions are
difficult to produce and consequently are hard to perceive (Echols and Newport 1992). Also,
unstressed syllables are produced with less duration,which gives articulators less time to ready
themselves to produce codas (see Kirk and Demuth 2003, and references cited therein).
The key point to note, nevertheless, is thatgiven that children display considerable varia
bility of coda acquisition cross-linguistically, and given thatmy findings support the idea that
these variations are also foundwithin a language, stressmay not be the only factor for explaining
the emergence of Spanish codas but that frequency and sonority play important roles in their
production.
4.1 The Frequency Hypothesis
This studywill test Stites, Demuth, and Kirk's (2004) hypothesis that the development of
consonant types relates to the frequency inwhich they are perceived and produced. Frequency
is taken
to mean
the percentages
of accurately
produced
codas.
These
researchers
have
argued
thatEnglish-learning children acquire codas depending on the rate of frequency towhich they
are exposed to them, and on theirMarkedness in terms of sonority. Their two children showed
considerable variability in coda development. While one child preferred the acquisition ofmost
frequent final stops over sonorants, theirother child showed the opposite preference, that is, the
child privileged Markedness over frequency. According to these researchers, the individual
variation shown by the children seems to suggest that they employ different learning strategies.
Seen within the context ofOptimality Theory (OT), these strategies imply thata child's produc
tion is closer to her ideal forms, ranking higher Faithfulness overMarkedness, while the other
child does the reverse.
These researchers have thus validated other studies showing that frequency pervades the
acquisition of phonological structures, such as stress and minimal word, across languages
( Jusczyk,Cutler and Rendanz 1993, and Kirk and Demuth 2003, among others cited therein).
4.2 The Sonority Hypothesis
The "sonority" concept, both phonetically and phonologically, has been difficult to charac
terize, although various definitions have been offered. Phoneticians have linked the concept to
physical or perceptual parameters, while phonologists have been concerned mostly with issues
of formal explanations (Clements 1990:284). Iwill endorse a hybrid approach by taking"sonority"
tomean the acoustic properties of the loudness of a given sound in relation to anotherwith equal
length, stress, and pitch, all ofwhich interfacewith major class features. The notion of sonority
inphonology has a time-honored precedent thatdates back to Sievers (1881), Jespersen (1913),
This content downloaded from 128.248.155.225 on Mon, 2 Dec 2013 12:34:00 PM
All use subject to JSTOR Terms and Conditions
The Acquisition of Spanish Codas
151
Saussure (1916), and Grammont ( 1933). According to them, languages tend todisplay a common
pattern in themanner inwhich consonants order themselves with respect to a syllable-peak.
Since the latterisnormally a vowel, the segments flanking ittend to show a hierarchical ranking,
with glides closer to the vowel, followed respectively by liquids, nasals, fricatives, and stops.
Such ranking has led to the establishment of the Sonority Hierarchy in (3).
(3) SonorityHierarchy: Optimal Syllable
Syllable-initial
I
I
Stops
Fricatives
Nasals
I Liquids
Glides
Vowels
Syllable-final
(3) has had importantapplications in the construction of complex onsets and codas as guided by
the Sonority Sequencing Principle thatrequires onsets to rise in sonority toward thenucleus and
for codas to fall (Clements 1990, Foley 1970, Hooper 1976, Kiparsky, 1979, Selkirk 1982, and
Steriade 1982). Both principles summarize alternative strategies employed by children inreducing
initialand final consonantal clusters. Still, given that inLle?'s study,as well as inmine, themostly
persistent early syllable output is of the unreduced CV and CVC types, I will closely follow
Clements's principle of sonority cycle (1990) to shed lighton the rate of final codas acquisition
but Iwill maintain the syllable structuregiven in (2).
According toClements, the preferred syllable typehas a specific sonority contour thatbe
gins with consonants thatare lowest on the sonority scale, then rises tomaximal sonority at the
vowel, then slopes down the scale toward the end of the syllable. This means that thepreferred
initial consonants are stops, likepa or tawhile final consonants tend to be sonorants, i.e., tan is
preferredover tat (Stites,Demuth andKirk 2004). He proposes the sonority scale 0<N<L<G<
V (where O = obstruent,N = nasal, L = liquid,G = glide, and V = vowel). In fuller detail, Stites,
Demuth, and Kirk propose the following hierarchy:
(4) SonorityCycle (afterClements 1990)
Stops
< fricative < nasals <
<
<
liquids glides vowels
[t,d] <[s,f]
more marked
less
sonorant
<[m,n]<
/
[l,r] <[j,w]
<[i,a]
lessmarked
more
sonorant
Although Clements views fricatives and stops as being of equal rank, he assumes that seg
mentai sequences thatdiffer in the [continuant] feature are preferred to sequences agreeing in
this feature specification. Therefore, hierarchy (4) captures the spiritof his general hierarchical
formulation.
The sonority cycle (4) helps to make a number of predictions, independent of stress,
concerning the emergence of codas in the speech of Spanish-speaking children, as we will show.
This content downloaded from 128.248.155.225 on Mon, 2 Dec 2013 12:34:00 PM
All use subject to JSTOR Terms and Conditions
152
H?spanla 90March 2007
5. Seihla's Acquisition
of Spanish Codas: A Case-Study Comparison
5.1 General Data Description
and Results
This study is centered on a three-year longitudinal observation of the early development of
the phonology of the author's youngest daughter, Seihla, who, although living in theUnited
States, was constantly exposed tohearing Dominican Spanish spoken in thehome setting.5Initial
exposure toEnglish was confined to a daily half-hour of educational cartoon inEnglish fromPBS,
followed by another half-hour to forty-fiveminutes of similar shows in Spanish transmittedby
Spanish networks. The research sessions began with tape and video-recording, and by
transcribingSeihla's speech from thebeginning of her firstutterances at age 1.2 up to age 3.7. For
the purpose of the present study, Imade a cutoff at age 2.3, which is roughly equivalent to the
ages of Lle?'s two children.Her speech was observed at home while interactingwith members of
a family ofDominican ethnicity and other Spanish-speaking friends. Speech samples were also
obtained by asking the child to name and identifyobjects and animals from children's books, a
taskwhich she embraced as a game and found enjoyable. The author transcribed all productions.
In addition to Seihla's tokens, and to further supportmy hypothesis, I am also relying on
data obtained from two femaleMexican children,whom Iwill name respectively Child I and Child
V. Though not observed longitudinally, these children's ages of 2.0 and 2.4 approximate two
stages of the age span being observed in Seihla's speech.We will see that theyproduce the same
phonological processes noted in Seihla's speech development. Their speech tokens were
collected employing similar methodologies, such as prompting the children to narrate stories
based on children's book theywere asked to read, familiar TV characters, or family life experien
ces. Child I and Child V aremonolingual. Their sample utterances were recorded inmonolingual
settings.
Since the study focuses on the realization of codas, I examined them in stressed and un
stressed contexts. A coda was counted as correct if itapproximated the intended target.Thus, if
[h] and [1]are reflexes of the respective Isl and 111,then theyare counted as realized. Unlike Lle?,
I did not include glide substitutes for consonants because therewere only two instances of oc
currences in Seihla's tokens and bothwere fromEnglish, i.e., birthdaywas rendered as [b?jdej],
and Barbie as [b?jbi]. Notice that thefirstone is obviously an English borrowing because Span
ish tends to reject penultimate stress inwords ending with branching diphthongal rhymes. There
was no record of final codas being replaced by glides in Spanish words.
A coda is considered correctly produced insofaras it is identical to the targetat ormore than
50% of the time (cf. Stites, Demuth, and Kirk 2004). Before age two, Seihla showed the typical
development of theuniversal CV structure, i.e., ['tato] (< /gato/"cat"), followed by occurrences
ofV as in f'awa] (< /agwa/"water"), and of complex nuclei in theformofCVG, where G generally
represents the fronthigh glide, as in ['bejbi] (< /bejbi/) "baby" and [4s8jla] (< /sejla/"Seihla").
Also, at this age, Seihla produced a notable output of bisyllabic structures, thereby seemingly
supporting thedevelopment of theMinimal Word hypothesis, which claims thatchildren's early
prosodie words contain a binary foot (Demuth 1995,Fee 1996,Fikkert 1994).
The onset of age 1.7marks the appearance of thefirst codas. Between this age and 2.3 she
articulates final [nm rj jl 1s h r]. Table 1 displays these segments regardless of stress location:6
This content downloaded from 128.248.155.225 on Mon, 2 Dec 2013 12:34:00 PM
All use subject to JSTOR Terms and Conditions
The Acquisition
TABLE 1
of Spanish
153
Codas
Seihla
Final codas
Medial
target
child
1.7
10
1.8
7
1.9
9
7(70%)
5(71%)
2(22%)
1(16%)
4(40%)
9(75%)
5(71%)
47(62%)
Age
1.10
6
2.0
10
2.1
12
2.2
7
2.3
75
codas
target
2
2
3
5
6
39
2
64
child
1(50%)
1(50%)
0(0%)
0(0%)
1(16%)
15(38%)
0(0%)
36(56%)
This table reveals a heightened propensity to favor occurrences of final codas, as in [sals'al] <
/sereal/ "cereal" and [to'mas] "Thomas," overmedial ones, as in [?mbe] < /anbre/"hungry" and
<
[ta'lente]
/kaljente/"hot" (for clarity, the targeted codas are inbold types). This preference is
sustained even at a laterage, 2.3, as additional data in (5) show.
Figures 1 and 2 illustrate the emergence of type of codas by age and percentages. Figure 1
complements partially Table 1and itsresults,namely, thefinal codas /n 1si, tend to appear earlier,
and simultaneously, inSeihla's speech. Notice, however, there is a valley at about age 1.9which
spikes up at around 2.0. This gap is due to the child not producing sufficientwords with new
codas. The tap / x7makes itsappearance at a rather late age, 2.3. The earlier occurrences of the
former segments, however, validate the fact thatSeihla is fully capable of articulating them.
Figure 1
Types of Final Codas
-/n/
c
-Is/
?0)
?.
-/I/
-/r/
1.071.081.091.101.112.002.012.022.032.04
Age
On the other hand, Figure 2 shows flatness inpercentage of production for liquids and fricatives
inmedial position from the initial stages, privileging only nasals at an earlier stage. Of interest
here is the fact thatbefore age 2.0 Seihla failed to produce coda consonants other than nasals.
This content downloaded from 128.248.155.225 on Mon, 2 Dec 2013 12:34:00 PM
All use subject to JSTOR Terms and Conditions
H?spanla 90March 2007
154
Figure 2
Types of Medial Codas
100
90
80
70
60
50
40
30
20
10
0
1.07
~*r
7n/
7s/
71/
7r/
1.08
1.09
1.10
1.11 2.00
2.01
2.02
2.03
2.04
Age
Both figures show a bias against the hypothesis thatmedial codas tend to surface earlier in
Spanish.
By comparing Table 1 toTable 2 below, we see a striking,proportional similarity in segment
distributions forChild I and Child V incomparison to Seihla's approximate age.
TABLE2
Final
Age
target
Child I
2.0
Child V
2.4
11
78
codas
Medial
child
11(100%)
76(97%)
codas
child
target
15
65
13(86%)
65(100%)
Recall that Seihla's percentages of consonantal segments infinal position are higher than her
medial ones. Notice inTable 2 thatboth children reproduce this tendency,more notably inChild
I,who produced final tokens like [le'on] < /leon/ 'lion' more readily than [la'?ando] < /labando/
"washing."
5.2 Data Description and Results byWord
Position and Stress
In order to grasp fully the relationship holding between stress and codas, it is important to
understand how the Spanish stress system generally works. It is an uncontroversial fact that
most words ending in vowels carry penultimate stress,while those ending in consonants have
ultimate stress.This kind of stress assignment is fairly canonical inSpanish, with less than 10%
ofwords bearing antepenult stress (N??ez-Cede?o andMorales-Front 1999; Lle? 2003:262). To
some phonologists Spanish appears to be a quantity-insensitive language, meaning that the
shape of a syllable, whether closed or not by consonants or diphthongs, does not count for
stress
computation.
However,
others
have
maintained
that, when
closed,
the penultimate
or
ultimate syllable attracts stress.The lexical /karanba/and /fusil/,for instance,would yield respec
*
tively [kar?mba] and [fusil]but not the ill-formed*[k?ramba] nor [f?sil] (Harris 1983,1995). As
This content downloaded from 128.248.155.225 on Mon, 2 Dec 2013 12:34:00 PM
All use subject to JSTOR Terms and Conditions
The Acquisition of Spanish Codas
155
this debate has not been settled yet (Roca 2006), I will follow the canonical pattern sketched
above.
Assuming a non-committal position regarding stress theories, letus move to thedistribution
of final codas in stressed and unstressed positions. Table 3 presents a coda distribution in these
contexts,while thedata in (5) provide corresponding examples.
TABLE 3
Seihla
Final
codas
stressed
Medial
unstressed
codas
stressed
unstressed
target
child
1.7
4
3(75%)
1(50%)
1(25%)
0
0(0%)
1.8
5
4(80%)
0(0%)
2(50%)
0
0(0%)
1.9
4
1(25%)
0(0%)
0(0%)
0
0(0%)
1.10
4
1(25%)
0(0%)
0(0%)
0
0(0%)
2.0
5
3(60%)
0(0%)
1(33%)
0
0(0%)
2.1
10
6(60%)
2(100%)
9(75%)
5
5(100%)
2.2
8
4(50%)
1
1(100%)
8
4(50%)
0
0(0%)
2.3
21
13(61%)
3
1(33%)
30
24(80%)
15
8(53%)
Age
target
child
child
target
target
child
(5) Examples from Seihla's speech in stressed codas
Final
stressed
Target
1.7
1.8
2.0
/dos/
['dos]
/tres/
['tes]
"three"
/sereal/
[sele'al]
"cereal"
['mas]
"more"
/ben/
['men]
"come"
/pan/
['pam]
"bread"
/tomas/
[to'mas]
"Thomas"
[Mus]
['?ol]
"light"
"sun"
[panta'lon]
"pants"
"Halloween"
/sol/
2.2
Gloss
[sale'al]
/lus/
2.1
Child
/sereal/
/mas/
1.9
Medial
/pantalon/
/halowin/
/tukan/
/limon/
[alo'win]
[tu'kan]
[li'mon]
"cereal"
stressed
Target
Child
Gloss
/?nbre/
[?mbe]
"hungry"
/?nbre/
[?mb9]
"hungry"
/linda/
['inda
"beautiful"
/kaljente/
/falta/
[ta'lente]
/este/
['ente]
two
"toucan"
"lemon"
]
['palta]
/sjentate/
['sentate]
/sepi'iandome/
[pic?ndome]
"hot"
"missing"
"this one"
"sit down"
"brushing myself
This content downloaded from 128.248.155.225 on Mon, 2 Dec 2013 12:34:00 PM
All use subject to JSTOR Terms and Conditions
156
.3
Hispania 90March 2007
"armchair"
/elefante/
"nose"
[ele'fante]
/domingo/
[do'mingo]
"Sunday"
['lampana]
"lamp"
/sijon/
/naris/
[si'jon]
/flor/
['flor]
"flower"
/telebisjon/
[bi'sjon]
"television"
[na'dis]
/lanpara/
/lor ando/
"elephant"
"crying"
Qo'dando]
/durmjendo/
[du'mjendo/
"sleeping"
/eskribjendo/
[eski'bendo]
"writing"
According toTable 3, Seihla's speech exhibits a productive pattern pointing to a large number of
final codas in stressed position up to about age 2.0, thus exceeding medial contexts. It is beyond
this age when medial stressed codas improve over final ones. Such a skewed distributional
pattern partially supports Lle?'s hypothesis, since the real difference inpercentages takes place
at ages 2.1 and 2.3, while at 2.2 the percentages are identical to the former.More importantly,
Table 3 illustrates a significant difference of coda realizations inunstressed contexts. Both final
and medial positions display a good percentage inproduction, suggesting that the emergence of
a coda is not necessarily driven by foot-headedness.
Similar conclusions can be reached when comparing Table 3 toTable 4, which corresponds
to the output in (6) produced by the twoMexican children.
TABLE 4
Final
codas
Medial
stressed
unstressed
child
target
Age
codas
target
stressed
child
target
2.0
4
4(100%)
7
7(100%)
2.4
22
21(95%)
56
55(98%)
7
child
unstressed
child
target
6(85%)
8
7(87%)
28 28(100%)
37
37(100%)
(6) Examples of stressed codas fromMexican Child I (2;0;) and Child V (2;4)
Final
stressed
Target
2.0
/asul/
/leon/
2.4
stressed
Medial
Child
[a'sul]
[le'on]
Gloss
Target
"blue"
"lion"
/bejntidos/
/eskobar/
[bejnti'?os]
[esko'?ar]
"twenty-two"
"Escobar"
/kerer/
[ke'rer]
"to want"
/estas/
[es'tas]
"you
/ram?n/
[ra'mon]
/murjel/
/feros/
[mu'rjel]
"Raymond"
"Muriel"
[fe'ros]
"fierce"
are"
Child
Gloss
/elefantes /
[ele'fantes]
"elephants"
/labando/
[la'?ando]
"washing"
/traba'xando/
[tra?a'xando]
/konprame/
['kompame]
"working"
"buy it for me"
/xugando/
[xu'ando]
"playing"
/grande/
/baskes/
['gande]
"big"
['baskes]
"Vasquez"
/subiendo/
[su'?jendo]
/porke/
/mwerto/
['porke]
"climbing"
"because"
['mwerto]
"dead"
/sandwije/
/entonses/
['t/angwitje]
"sandwich"
[en'tonses]
"then"
The rate of final unstressed codas theyproduced is as high as thepattern Seihla displays inTable
3. Also, Child V's medial coda percentages are identical for stressed and unstressed positions,
which supports the idea that stress does seem to be driving the occurrences of medial con
sonants.
On thebasis of these examples, one cannot help but notice thatpenult stress is thepreferred
or perhaps the default solution for stress placement, with the exceptions of antepenultimately
This content downloaded from 128.248.155.225 on Mon, 2 Dec 2013 12:34:00 PM
All use subject to JSTOR Terms and Conditions
The Acquisition of Spanish Codas
157
stressed ['lampana], produced by Seihla, and ['tjarjgwitfe], uttered by Child V. The unusual
marked stressed pattern developed very early in Seihla's speech, as it is evident in /musika/
"music" and /makina/ "machine" in (5), which she rendered respectively as ['mukika] and
['makima]. Most words with closed penult syllables carry stress in thatposition, thus partially
justifying the quantity-sensitive hypothesis. Also, they show the difficulty a child may ex
perience in producing the tap lrl, which is acquired rather late, as additional collected data
demonstrate, or itmay be replaced by IV. The latterprecedes Irl in order of acquisition.
Other examples in (5) further show a good number of infinitivalverbs lacking final con
sonants butwith expected final-stressed syllables. Evidently, Seihla has difficulty inproducing
final rhotics inverbs but not so inother lexical itemsand opts instead for thepreferredC V pattern.
seems to experience a similar process, as shown by [mo'??] which alternates with
[mor'o?r] from /morder/ 'to bite.' Seihla appears to be providing indirect support for ranking
low in the constraint hierarchy theNO-CODA
markedness constraint (Prince and Smolensky
1993) because the facts presented thus far confirm that she can articulate codas elsewhere. In
termsofOT, thiswould require rankingMAX-IO overNO-CODA,
as defined in (7). An example
isprovided in (8) for [to'mas]. For brevity, I am assuming the application of feature identityand
Child V
stress
constraints.
placement
(7) NO CODA: No coda consonants
MAX-IO:
Prince 1995).
Every segment in the Inputmust have a corresponding Output (McCarthy and
(8)MAX-IO? NO CODA
Input: /tomas/
MAX-IO
a. ^[to'mas]
b.
NO-CODA
*t
[to'ma]
While the data in (5) and (6) underscore Lle?'s stress and coda hypothesis, thepercentages
presented in Tables 3 and 4 show opposite effects. That is, she has argued thatmedial codas
surface early in speech development, yet Seihla's final codas, which emerged at the same age as
hermedial ones, seem to challenge Lle?'s hypothesis. The same applies toChild I and Child V.
This is supported furtherby thedata given in (9) for Seihla and in (10) for theMexican children.
(9) Examples of Seihla's unstressed codas at various ages
Final
Medial
Child
Target
1.7
unstressed
/grasjas/
Gloss
unstressed
Target
2.1
[kom'migo]
"with me"
[telmi'no]
"it's
/ratonsito/
[daton'sito]
"little mouse"
/enkontraste/
[enkon'tatle]
"you
[kon'tento]
"happy"
/konmigo/
/termino/
/animales/
2.3
/miki maws/
[anima'des]
[miki
Gloss
"thanks"
['asjas]
2.2
Child
'maws]
over"
"animals"
"Mickey
Mouse"
/kontento/
This content downloaded from 128.248.155.225 on Mon, 2 Dec 2013 12:34:00 PM
All use subject to JSTOR Terms and Conditions
found"
158
H?spanla 90March 2007
(10) Examples of Child I and Child V's unstressed codas at 2.0 and 2.4 respectively
Final
2.3
unstressed
Medial
Child
Target
2.0
unstressed
Gloss
Child
Target
Gloss
/estaban /
[es'ta?an]
"they were"
/estaban/
[es'ta?an]
/elefantes/
[ele'fantes]
"elephants"
[eska'po]
"they were"
"it ran away"
/kotfinitos/
/lamas/
[ko'tf initos]
"piglets"
"call your
/eskapo/
/mansana/
[man'sana]
/berenxena/
"apple"
[beren'xena]
/inkreibles/
[inker'i?es]
"eggplant"
"The
Incredibles"
['jamas]
self
/eskaleras/
[eska'leras]
"stairs"
/eskobar/
[esko'?ar]
"Escobar"
/entonses/
[en'tonses]
"then"
/lexos/
"far"
[bejnti'?os]
['lexos]
/bejntidos/
/ermanito/
"twenty-two"
"little brother"
/trabesuras/
[ta?e'suras]
"mischiefs"
/espanto/
[espan'to]
"he
/inkreibles/
[inke'i?es]
"Incredibles"
/pastel/
/morder/
[pas'tel]
"cake"
/solos/
['solos]
"alone"
/tenian /
[te'nian]
"they
/estaban/
[es'ta?an]
had"
"they were"
[erma'nito]
[mor"?er]
/tfilindrjina/ [tfilin'kina]
"to
scared
it"
bite"
"Chilindrina"
/enkontro/
[enkon'to]
"he
found"
/kontesto/
[kontes'to]
"he
answered"
None of the closed syllables inbold types in these examples, whether infinal ormedial en
vironments, aremanifestly stressed. One could argue that theplural marker I-si, as illustratedby
[ani'mades] and [eska'leras], is exempt from stress because of its special grammatical status.
Also, itmust be assumed, for instance, that thismorpheme isnot computed for stressing,perhaps
because it is extrametrical. Yet, such an argument begs the question in the face of forms like
< /naris/noted in (5), whose final Isl seems to count for stress
[na'dis]
placement. But observe
that the final syllables of /enton.ses/and /le.xos/ forChild V in (10) are unstressed, suggesting
that the final Islmust also be extrametrical, for otherwise theywould attract stress.Whichever
theoretical path is taken to account forfinal Isl, the fact remains stubbornlyclear: thefinal syllable
is unstressed, and this should not have surfaced, following Lle?'s arguments.
6. Codas
inDeterminers
Let us look at the behavior of final codas of the respective definite determiner el and of the
indefiniteun. I looked at Seihla's performance with these variables and found thatat an early age
she shows little evidence of theiruse. At age 1.7 she substitutes the definite article for the re
duced vowel [9], as in [9 le'lo] 'the clock' or fuses thatvowel with a preceding preposition, as in
[en9 'bapo], 'in thebath(room).' The fullformof an indefinitearticle emerges at age 1.9,while the
final IV shows up at age 2.1. Table 5 illustrates the percentages of occurrences of both deter
miners, with a significant burst going on at age 2.3, which distinctivelymakes
has already achieved full command of theirarticulation infinal position.
Table 5
Seihla's
article production
Codas
of definite articles
Age
1.7
1.8
1.9
target
1
0
2
Codas
child
0(0%0
0(0%)
0(0%)
of indefinite articles
target
0
0
1
child
0(0%)
0(0%)
1(100%)
This content downloaded from 128.248.155.225 on Mon, 2 Dec 2013 12:34:00 PM
All use subject to JSTOR Terms and Conditions
itobvious that she
1.10
0(0%)
0(0%)
3(100%)
0(0%)
8(72%)
0
1
2.0
2.1
3
2.2
0
2.3
11
The Acquisition
0
0(0%)
1
1(0%)
0
0(0%)
0
0(0%)
21 21(100%)
of Spanish
Codas
159
Seihla's production of determiners is like thatof the themale child inLle?'s study: he tended
to produce both articles, as opposed toMaria, who preferred the indefinitedeterminer. Seihla
gave proof of providing a good number of allophonic nasals, as shemanaged to tag them before
any kind of following onsets. The definite article, even though more sparsely distributed,
behaves in a similar fashion. The samples (11) and (12) illustratedeterminers' distribution for
Seihla at age 2.3 and for theMexican children, respectively. The latterdid not provide indefinite
determiners
in a consonant.
ending
(11) Seihla's determiners
Definite
[el
[al
[el
[el
[el
Indefinite
"the
car"
[urn 'pedo]
'pa:ke]
"the
[u?
'libro]
"the
park"
book"
'?ipo]
"the boy"
"the lemon"
"the
[un 'sikulo]
"a
[um
"a monkey"
'kajp]
li'mon]
[el 'baj?o]
(12)Mexican
bath"
"a
[un
[um
'baso]
[un daton'sito]
"a glass"
"a little mouse"
[un gi'neo]
"a banana"
'mono]
"a
horse"
circle"
children's determiners
Child I
Definite
[el a'sul]
"the blue
[la man'sana]
"the
[la
"the
apple"
house"
"the
Incredibles"
[los
"a dog"
star"
'teja]
'kabaio]
'kasa]
inke'i?es]
ChildV
one"
Definite
[el kotfi'nito]
[las eska'leras]
[el Uo?o]
[el pas'tel]
[el 'tja?o]
[el pa'pa]
In keeping with her hypothesis, Lle? analyzes the incidence of codas indeterminers on the
basis of their interactionwith stress.That is, since it is the case that indefinitearticles are usually
stressed (Navarro Tom?s 1974), she claims that they project their own prosodie structure, the
Phonological Word, under the aegis of its commanding phonological phrase. Essentially, the
phrase [urn 'baso] would be structurallyrepresented as in (13).
(13)
PrWd
PrWd
F
a
A
u
I
m
F
A
AA
baso
co
This content downloaded from 128.248.155.225 on Mon, 2 Dec 2013 12:34:00 PM
All use subject to JSTOR Terms and Conditions
160
H?spanla 90March 2007
Structure (13) conforms to the prosodie rhythmof Spanish and supports Lle?'s predictions for
medial position. But Iwill argue in section 7 that theonset licenses thepreceding nasal. The story
is differentwhen (13) carries over and applies to definite articles,which are known to function in
a proclitic capacity (seeNavarro Tom?s 1974:193). These determiners rarely, ifever,project them
selves from a prosodie word but directly from a phonological phrase. Lle? proposes that
structure (14), with an unstressed article, is restructured as shown on the right.
PPh
PPh
PrW
PrWd
F
a
n
i
p
A
o el
Ago
AA
n
i
ji
o
The prosodie structure (14) is coherent but factually misguided because Spanish is not
known for accenting definite proclitics. There is no real phonetic substance or rhythmicmotiva
tion that justifies effecting the structural change in (14) other than achieving consistency of
analysis. I am thus submitting that the reason why these codas emerge is due to variability of
production triggered by the frequency of exposure to codas.
7. Discussion
In lightof the analysis presented thus far, letus revisit again Figures 1 and 2 by relating them
to the hypothesis I am advancing here. Recall thatFigure 1 shows an earlier production of codas
/n 1si, at age 1.7 forSeihla, as compared with Lle? (2003), whose subjects produced them later,at
age 2.0 butwith thenasal outnumbering the other two consonants. This can be explained by the
nasal acquiring itsdefault feature specification forplace of articulation,which comes cost-free in
the child's grammar. A similar reasoning applies to stressed and unstressed medial nasals pro
duced by the children in this study, including thenasal for the indefinitedeterminer,whose place
features are licensed by the following onsets, as shown by theirmuch earlier occurrences in
How
Table 2, and by the data in (9).When no consonant follows, the default is the coronal /n/.
ever, inFigures 1 and 2 the fricative I si unexpectedly surpasses inpercentages the occurrences
of the liquid IV, and in fact the former appears more frequently infinal position. This seems to
confirmEmile Benveniste' s early observation, cited by Jakobson ( 1968:57,90), that liquids, being
difficult to produce, are acquired late.We have also argued that I-si surfaces not because of its
semantic property,as claimed forPortuguese and Spanish (Lle? 2003, Freitas et al. 2001 ), but be
cause examples in (5) show thata final, lexical Isi is also part of Seihla's grammar,which indicates
frequency of production.
The importantpoint to notice is that the three codas in question emerge simultaneously,
varying only in frequency of acquisition, suggesting that Seihla is a frequency-based learner
(Stites, Demuth, and Kirk 2004). The variability shown by Seihla in acquiring codas is best
explained as a learning strategy that follows a constraint path with a preference for the occur
rences of the coronal /n 1 si, appearing concurrently but at a different frequency rate. That is,
Seihla's initialperceptions and productions at age 1.7may go against the trendfound in studies
wherein children follow thepath inOT of ranking the constraint ofMarkedness higher, over the
Faithfulness constraints, for codas (Pater 1997, Grijzenhout and Joppen 1999). My findings
support the research on variability by Stites, Demuth, and Kirk, in which they claim that
This content downloaded from 128.248.155.225 on Mon, 2 Dec 2013 12:34:00 PM
All use subject to JSTOR Terms and Conditions
The Acquisition of Spanish Codas
161
Faithfulness outranksMarkedness constraints forfinal codas for theirChild N.
Frequency of coda production works in tandem with sonority. To understand this interac
tion,we must look at Seihla's phonology before age 1.7. The examples in (15) show she produced
mostly open syllables in contexts that called forfinal/medial fricatives or sonorants:
(15) Examples of Seihla's early types of codas
Child
Target
1.2
1.4
/fjorV
[ o]
"Fior"
/papel/
[pe.'pe]
/pan/
/tres/
[pa]
[te]
"paper"
"bread"
/grasjas/
['ta.tja]
/kaljente/
/karta/
Ii c ente/
['ke.te]
"hot"
['ta.tta]
"letter"
"front"
/ratonsito/
['pe.te]
[tato'tito]
(all the above,
/abjon/
1.5
1.6
Gloss
"three"
"thanks"
"little mouse"
plus:)
[bon]
"airplane"
"white"
/blanka/
['ba.ka]
/pwerta/
[p'e.ta]
"door"
/lanpara/
['pa.ta.ba]
/pintar/
[pi.'ta]
"lamp"
"to paint"
(all the above, plus:
)
six
/sejs/
/djes/
[se]
[de]
/karga(me)/
['ka.k9]
/sjentate/
/sinko/
['se.ta.ta]
up"
"sit down"
['si.ko]
"five"
"ten"
"lift me
To account for the above data within OT, at this stage of production one will have to assume that
overMAX-IO, theopposite of the ranking in (8).
Seihla's grammar ranksNO-CODA
But she began early coda acquisition with final Inl, suggesting that the surfacing codas
obey the Sonority Cycle (4 ). The readerwonders how this can be sowhen the child, having been
exposed to both /n 1/,as in avi?n and papel, chooses Inl over IV, thus violating the predictions
made by (4). I have already said that it is due to difficulties in articulating liquids. Furthermore,
Inl, lacking specification forplace, will acquire thisfeature froma following onset or by default in
word-final position. One would also expect the ordered realization of IV and Isl, following the
expected predictions from (4). However, we see in Figure 1 thatboth emerge simultaneously,
although with a differential in frequency. So, in a sense, their realizations are also obeying
which Seihla perceived
hierarchy (4), especially ifwe consider forms such as /akto/"act" /digno/,
but rendered respectively the simplified ['ato] and ['ino] at 1.5. In this early stage, she is
privileging words without codas but begins to demote this tendency by allowing thepresence of
branching syllables with nasals. At a later stage, she is faithful to her underlying coda forms but
eschews obstruents, thus imitatingadult's casual speech, who normally avoid them.
8. Conclusion
To sum up, we have seen thatchildren display considerable variability in theiracquisition of
syllabic structures,particularly with respect to codas. Some are bound by prosodie strategies,
highlighting the interrelationships among stress, syllable structure and rhythm.Others are
guided by consonant types relative to their degree of sonority, and by the frequency of their
exposure in the language. Since sonorants are frequently articulated in Spanish codas, and since
This content downloaded from 128.248.155.225 on Mon, 2 Dec 2013 12:34:00 PM
All use subject to JSTOR Terms and Conditions
162
H?spanla 90March 2007
the sonority cycle predicts they should surface first in the sonority slope, it is expected they
should be realized before stops, a fact thathas been tangentially confirmed in this study.Further
more, itappears thatchildren tend tomake single consonantal morphemes and lexical segments
converge phonetically, without distinguishing them on the basis ofmorphosemantic attributes.
The reportedfindingsmay also have some implications for second-language learning, inasmuch
they shed light in the order inwhich final consonants are acquired and how they affect pronun
ciation.
Finally, given that thepresent longitudinal study is based on the speech of one child, albeit
supported by data from two other children, additional research is needed to achieve more robust
generalizations.
NOTES
!I am grateful to James Compton, Scott Wakely,
and Rosemary Weston
for reading, graphing, and com
reviewers whose suggestions helped
menting on an earlier version of the paper. Thanks go to two of Hispania's
me correct some inaccuracies. None are responsible for potential lapses. My heartfelt gratitude and love go to
informant. I am also most thankful to Antoinette
Seihla, my daughter, for being a playful and cooperative
from Mexico's
Santana Cepero,
Universidad
Aut?noma Metropolitana,
who made
recordings from their archives on children's speech.
2The nucleus in the rhyme may also be preceded by a left-branching structure, consisting of a glide. In this
paper I am concerned only with consonants following the nucleus.
3The brackets "<, >" mean "derives from" or "produces,"
respectively, unless indicated otherwise.
4I will use the term "medial codas" in its broadest meaning as in Lle?'s
2003, which includes both word
Hawayek
available
and Elizabeth
to me
initial and medial
positions.
as attested
5At the time research began, it is the author's impression that Seihla was fully Spanish-dominant
children would say when visiting and
by the fact that she could not understand a word of what English-speaking
playing with her, which caused her much frustration for not being understood when speaking Spanish. Also, her
Dominican
is modeled
after her parents' middle-class
speech, which tends to contain I si or its aspirated
reflex, and non-neutralized
liquids in syllable-final positions. Their pronunciation also lacks final obstruents.
6To facilitate comparisons and presentations, this table, and subsequent ones, will closely resemble those
used by Lle? (2003).
output
WORKS CITED
in Laboratory
N. G. (1990).
of the Sonority Cycle
in Core Syllabification."
"The Role
Papers
1. Between
eds.
the Grammar and Physics of Speech. John Kingston and Mary E. Beckman,
Phonology
283-333.
Cambridge UP.
Cambridge:
to Syntax: Bootstrapping from
Demuth, Katherine.
(1995). "The Prosodie Structure of Early Words."
Signal
to Grammar
in Early Acquisition.
James L. Morgan
and Katherine Demuth,
eds. Hillsdale, NJ:
Speech
Clements,
Lawrence
Erlbaum
Echols, Catharine.
245-96.
Associates.
(1992).
"A
1-17.
Perceptually-Based
Model
of Children's
Earliest
Production."
Cognition
46.
Catharine, and Edward Newport.
(1992). "The Role of Stress in Determining First Words."
Language
2. 189-220.
Acquisition
Fee, E. Jane. (1996). "Syllable Structure and Minimal Word." Proceedings
of the UBC International Conference
on Phonological
Ingram, eds. 85-98. Somerville,
Acquisition. Barbara Bernhardt, John Gilbert and David
MA: Cascadilla
Press.
Echols,
The
Structure. Dissertation.
Paula.
the Acquisition
Leiden,
(1994). On
of Prosodie
University of Leiden.
4: 92-98.
Features." Folia
Distinctive
Ling?istica
"Phonological
Foley, James. (1970).
in European Portuguese." Probus
Jo?o. (2003). "The Acquisition
of Onset Clusters
Freitas, M.
Fikkert,
Netherlands:
15:1.27^17.
"Interaction between Prosody and Morpho
and Isabel Hub Faria. (2001).
Jo?o, Matilde Miguel
to Bootstrapping:
in the Acquisition
of European Portuguese." Approaches
syntax: Plurals within Codas
Lexical, Syntactic and Neurophysiological
Acquisition. Vol. 2.
Aspects of Early Language
Phonological,
45-57.
John Benjamins.
eds. Amsterdam/Philadelphia:
and Barbara Hohle,
J?rgen Weissenborn
Freitas, M.
Gnanadesikan,
Amalia.
Phonological
107.
Goad,
(2004).
Acquisition.
"Markedness
Ren? Kager,
and Faithfulness
Constraints
Joe Pater and Wim
Zonneveld,
in
Constraints
Phonology."
eds. Cambridge: Cambridge UP. 73
in Child
Structure in Syllabifica
"Phonetic Evidence
for Phonological
Heather, and Kathleen Brannen.
(2003).
tion." The Phonological
Spectrum. Vol. 2. Joost van der Weijer, Vincent van Heuven, and Harry van der
3-30.
Publications.
John Benjamins
H?lst eds. Amsterdam:
This content downloaded from 128.248.155.225 on Mon, 2 Dec 2013 12:34:00 PM
All use subject to JSTOR Terms and Conditions
The Acquisition
of Spanish
Paris: Librairie Delagrave.
(1933). Trait? de phon?tique.
Janet, and Sandra Joppen. (1999). "First Steps in the Acquisition
Grijzenhout,
Study." ROA #304. n.p.
Harris, James W. (1983). Syllable Structure and Stress in Spanish: A Nonlinear
163
Codas
Grammont, Maurice.
MIT
?.
Press.
(1995).
of German
Phonology:
A Case
Cambridge, MA:
Analysis.
The
and Edge Marking
in the Computation
of Stress in Spanish."
The Handbook
867-87.
Theory. John A. Goldsmith, ed. Cambridge, MA: Blackwell.
Press.
(1976). An Introduction to Natural Generative Phonology. New York: Academic
"Projection
Phonological
Joan B.
Hooper.
of
Ito, Junko. (1988). Syllable Theory in Prosodie Phonology. New York and London: Garland Publishing.
and Linguistic Theory 7: 217-60.
(1989). "A Prosodie Theory of Epenthesis." Natural Language
and Phonological
The Hague: Mouton.
Universals.
Jakobson, Roman.
(1968). Child Language
Aphasia
Jespersen, Otto. (1913). Lehrbuch der Phonetik. Lepizig und Berlin: B. G. Teubner.
?.
J. Rendanz.
"Infants' Preference for the Predominant Stress
Jusczyk, Peter, Ann Cutler and Nancy
(1993).
64.3: 675-87.
Pattern of English Words."
Child Development
is Cyclic
in English." Linguistic Inquiry 3: 421-^41.
Kiparsky, Paul. (1979). "Metrical Structure Assignment
and Katherine Demuth.
in the Acquisition
of Clusters." Pro
(2003). "Onset/Coda Asymmetries
Barbara Beachley,
ceedings of the 27th Annual Boston University Conference on Language Development.
Amanda Brown, and Frances Conlin, eds. Somerville, MA: Cascadilla
Press. 437^48.
of Consonants
from Polysyllabic
Lexical
Klein, H. (1981).
"Early Perceptual
Strategies for the Replication
Models."
Journal of Speech and Hearing
24: 135-51.
Research
Kirk, Cecilia,
Levelt. (2000). "The Acquisition
of Syllable Types." Language
Clara, Niels O. Schiller, and William
8: 237?64.
Acquisition
of Codas."
15: 255-81.
Probus
Lle?, Conxita.
(2003). "Prosodie Licensing
and Reduplicative
"Faithfulness
in Optimality
John, and Allan Prince. (1995).
McCarthy,
Identity." Papers
Theory. J. Beckman, L. Walsh Dickey and S. Urbanczyk, eds. Amherst, MA: Graduate Linguistic Student
Levelt,
Association.
Navarro
249-384.
Tom?s, Tom?s.
ciones Cient?ficas.
Ohala,
(1974). Manual
de pronunciaci?n
espa?ola.
Diane K. (1999). "The Influence of Sonority on Children's
32.6: 397^22.
tion Disorders
"On Coronal
Paradis, Carole, and Jean Fran?ois Prunet. (1989).
?.
Internal and External
(1991). The Special Status of Coronals:
Press.
Diego: Academic
Madrid:
Consejo
Cluster Reductions."
Superior
Journal
de
Investiga
of Communica
6: 317-48.
Transparency." Phonology
Evidence.
(Phonology and Phonetics 2). San
Joe. (1997).
"Minimal Violation
and Phonological
6: 201-53.
Language
Development."
Acquisition
Prince, Allan, and Paul Smolensky.
(1993). Optimality Theory: Constraint Interaction in Generative Grammar.
MS. Rutgers University, New Brunswick, and U of Colorado, Boulder.
Pater,
22.3: 345-94.
Iggy. (2006). "Saturation of Parameter Settings in Spanish Stress." Phonology
Yvan.
and Prosodie Licensing
in the LI Acquisition
of Phonology. Dissertation.
(2000). Headedness
Canada: McGill
Montreal,
University.
Susan and E. Andrews.
"The Influence of Syllable Position on Children's
Production of
Rvachew,
(2002).
Roca,
Rose,
Consonants."
Clinical Linguistics
Ferdinand. (1916). Cours de
Schwartz, R and L. Goffman.
(1995).
38: 876-88.
Research
Hearing
Saussure,
and Phonetics
16:
183-98.
and Paris: Payot.
linguistique g?n?rale. Lausanne
"Metrical Patterns and Production Accuracy."
Journal
Selkirk, Elizabeth.
(1982). "The Syllable." The Structure of Phonological
H?lst and Norval Smith, eds. Dordrecht: Foris. 337-83.
Representation,
of Speech
and
Part 2. Harry van der
Sievers, Eduard.
(1881). Grundz?ge der Phonetik. Leipzig: Breitkopf and Hartel.
and theNature of Syllabification. Dissertation. Cambridge, MA: MIT.
Steriade, Donca.
(1982). Greek Prosodies
Stites, Jessica, Catherine Demuth, and Cecilia Kirk. (2004). "Markedness vs. Frequency Effects in Coda Acquisi
tion." Alejna Brugos, Linnea Micciulla,
and Christine E. Smith, eds. Proceedings
of the 28th Annual Boston
on Language Development.
Cascadilla
Press. 565-76.
Somerville, MA:
University Conference
This content downloaded from 128.248.155.225 on Mon, 2 Dec 2013 12:34:00 PM
All use subject to JSTOR Terms and Conditions