Professional Documents
Culture Documents
Gilda M. Socarrás
www.continuumbooks.com
EISBN: 978-1-4411-90307
Preface ix
Acknowledgments xiii
Notes 207
Bibliography 214
Index 221
Preface
Example 1
(a) La casa linda (Referent ‘one house’)
‘The (fem/sg) house (fem/sg) pretty (fem/sg)’
(b) *La casa linda (Referent ‘two houses’)
x Preface
Example 2
(a) El niño exitoso
‘The (masc/sg) boy (masc/sg) successful (masc/sg)’
(b) Las niñas exitosas
‘The (fem/pl) girls (fem/pl) successful (fem/pl)’
(c) El/La estudiante exitoso/a
‘The (masc/sg)/fem/sg) student (unm/sg) successful(masc/sg)/
(fem/sg)’
Many beings have supported this monograph from its inception to the
present form. First, a heartfelt thanks to the Puerto Rican children who
participated in this research and who were always ready to “play” with me in
those incredibly hot afternoons in the Island. You continue to inspire me!
Second, I will always be indebted to my dissertation committee members.
To Dr Lardiere, my advisor, many thanks for her professional guidance and
patience. My gratitude goes to Dr Thornton, who taught me the beauty of
child language research. Thanks! Many thanks to Dr Crain, who believed
in the linguist in me and helped me grasp the big picture of language. To
Dr Campos, the eternal learner and the example of overall professional
excellence. ¡Gracias profesor!
Many thanks to all the parents for allowing their children to participate
in this research and to the University of Puerto Rico and the Fajardo
Montesori School for opening their facilities to conduct my experiment.
Special thanks go to my parents, Efraín and Nieves, for their love, support
and lifelong example. Los quiero y admiro. And to Yiannis, for his love, faith
in me and unconditional support. Ευχαριστώ. To Adriana Merced, who
made me believe in miracles.
Chapter 1
Theoretical Foundations
Example 1.1
[DP D la [FP F casa [AgrP bonita AGR casa [NP casa]]]]
In Example 1.1, the nominal head casa raises to the head of the func-
tional projection AgrP to check the agreement features of the adjectival
head bonita in a Spec-head configuration.5 In the determiner case, I assume
that the agreement features of the N-head are copied (or inherited) by the
D-head. Then, the agreement features of the Determiner head are checked
by raising the nominal head to the functional category FP, as shown in the
example. Notice that in the case of the agreement features of the Determiner
head, the checking mechanism does not involve a Spec-head configuration
but is checked via percolation, that is, the agreement features of the N-head
percolate up from N to D (Radford 1997, among others). For the present
monograph, I adopt Cinque’s (1994) and Bosque and Picallo’s (1996)
4 First Language Acquisition in Spanish
analyses for adjectives and assume that they are generated as specifiers
of functional projections, in this case, an Agreement Phrase, as shown in
Example 1.1. Furthermore, I assume the availability of a functional projec-
tion between the DP and the Nominal Phrase, where Spanish nominals
raise to check strong features.6
Notice that N-raising in the example above is motivated within the MP
by an additional characteristic of the features, namely, strength. In this
framework, a formal feature such as gender or number may or may not be
strong, forcing the operation Move (Chomsky 1995). Spanish, among other
Romance languages, has been characterized as a language with strong
nominal features, hence movement to the appropriate functional head(s)
within DP is required for the derivation to converge (Bernstein 1991; Brugè
2002; Cinque 1994; Longobardi 1994; among others). In contrast, languages
such as English are presumed to have weak N-features, and therefore no
N-movement is required. The notion of feature strength in the MP, although
difficult to define in concrete terms, accounts for some cross-linguistic
differences between Spanish and English. For example, the attested attribu-
tive adjectival placement: in pre-nominal position in English, for example,
pretty house, and in post-nominal position in Spanish, for example casa bonita
‘house pretty.’7 That is, in Spanish the noun casa raises to check its strong
features, hence appearing before the adjective bonita, whereas in English
the noun house does not because it has weak features.8
The overview of agreement within the MP above reveals a puzzling aspect
about the agreement process: feature content is added to unspecified con-
stituents and then the same content is deleted later on. Miyagawa (2010)
addresses this issue, pointing out the redundant, asymmetric, and appar-
ently irrelevant nature of agreement. For example, in the DP la casa bonita,
the relation is asymmetric because only one element, the nominal casa,
holds the values for the features that are “copied” to the other elements.
Agreement is also redundant because the same information, in this case,
the features feminine and singular, is repeated on three constituents: the
nominal, the determiner, and the adjective. Furthermore, once the uninter-
pretable features receive a value specification through agreement, they
must be deleted before LF so that they will not receive a semantic inter-
pretation. Therefore, the actual content of agreement seems irrelevant.
Miyagawa sheds some light on this matter, proposing that the purpose
of Agree is to create functional relations between a functional head and
an XP. These relations are critical because they “substantially enhance the
expressive power of human language” (Miyagawa 2010, 8). In the example
above, the purpose of N-raising or Move is to keep record of the functional
Theoretical Foundations 5
relation created in the syntax, that is, agreement, so that it can be used by
the semantic interpretation or LF.
input given the fact that they have no principles or constraints guiding the
acquisition process. Second, this hypothesis is very language specific because
it relies solely on the input to which a child is exposed in a particular lan-
guage. Hence, it fails to explain cross-linguistic generalizations, for exam-
ple, the attested phenomenon of determiner omission in child language
and how do children convey to the target language. On the other hand,
under the continuity approach, children are guided by UG, and one possi-
ble explanation for determiner omission is the Nominal Mapping Parameter
(Chierchia 1998a, 1998b; Chierchia, Guasti, and Gualmini 2000). This
hypothesis accounts for determiner omission cross-linguistically, asserting
that children start the acquisition process with a mass-like interpretation
of nominals and the primary linguistic data guides them to choose the
target-like setting of the parameter. Moreover, this parameter unifies other
seemingly unrelated phenomenon present in child language, namely, the
production of unmarked plurals. In addition, data-driven accounts of
acquisition fail to explain how children acquire knowledge in the absence
of experience, for example, the interpretation of anaphoric relations.
Continuous approaches on the other hand rely on the availability of linguis-
tic constraints, in this case Binding Theory Principle C.12 For the present
monograph, a data-driven analysis predicts that children will be conserva-
tive learners and will omit determiners but when they do start producing
them, their utterances should match the primary linguistic data, the only
source of information available, according to this approach. In contrast,
a continuous approach will predict that children’s production should not
violate the principles of UG even though it might not match the target
language.
Chapter 2
Experimental Methodology
2.1 Data
feminine, cas–a ‘house,’ and toys with names that are not marked overtly
for gender, for example, ending on –e like lech–e ‘milk,’ or on a consonant,
like –n in león ‘lion.’ In this manner, the experimenter can explore the exis-
tence of initial values in child grammar for nominals that do not provide
any morphophonological clues for gender, that is, the gender value assigned
to these nominals is not overtly marked for gender. Moreover, the inclusion
of nominals that ended in a consonant sound, for example, león, provided
the opportunity to assess whether children marked plurality or not with the
allomorph –es. This was crucial given the fact that one of the characteristics
of the Puerto Rican dialect is the aspiration of the final –s. Notice that this
experimental control might have not been possible with the naturalistic
approach to data collection, as pointed out by Snyder (2007). He studied
the acquisition of DPs, using the data of Juan Linaza from CHILDES
(MacWhinney & Snow 1985), finding out that at age 1;7, Juan produced
only one instance of a DP. Hence, no conclusions were made regarding this
issue (Snyder 1995).
The second experimental approach used in the present study was a com-
prehension task. This task was designed to investigate children’s compre-
hension of number, that is, singular versus plural, separating the cognitive
ability to comprehend the difference between one and more than one
from their linguistic ability to produce the plural marker –(e)s. In this task,
the child did not have to respond verbally to a request but instead had
to select among a number of available objects, for example, one horse, two
dogs, or all the cats. This task points to a third advantage of the experimen-
tal approach: the evaluation of a particular scientific hypothesis, for exam-
ple, whether children distinguish between mass and count nominals in
their grammars. Nonetheless, experimental tasks have some drawbacks, as
stated by Crain and Thornton (1998). As with all experimental techniques,
the success of an experimental task relies on its careful design and develop-
ment, that is what might seem appropriate for the experimenter at the
designing stage might result in a task that is boring or uninteresting for
children. Second, this type of task is labor intensive, especially with children
around the ages of 2;0 years old. It relies on the active participation of the
children involved, hence a good rapport between the experimenter and
the child is vital to its success, as well as an interesting “game” so that chil-
dren are compelled to participate in the task. In this respect, the naturalis-
tic approach to data collection is easier to administer because it does not
require the subjects to attend to a particular task. However, the output in
terms of the robustness of the data is not comparable, as discussed above.
The rest of the chapter is presented in three Sections. Section 2.2 pro-
vides a description of the children who participated in the present research.
Experimental Methodology 11
Section 2.3 discusses the experimental tasks conducted to gather the acqui-
sition data discussed in this monograph. Finally, Section 2.4 describes how
the data were transcribed and coded and presents a summary of the data
coded by constituents under study.
2.2 Subjects
The data for the present study were extracted from the transcripts of
three monolingual Spanish-speaking children under the age of three: two
boys, Elián (2;1,18) and Alonso (2;2,29), and one girl, Londa (2;9,4); pseud-
onyms were given to all participants.1 In addition, the data of two other
children over the age of three—a girl, Diana (3;6,3) and a boy, Pepe
(4;3,10)—were introduced in discussion when interesting. These two older
children were included in the experiment for two reasons. First, their data
were included to assess the validity of the experimental tasks conducted for
this study, that is, to test if these older children were able to complete the
tasks. The three experimental tasks posed no problems for them, produc-
ing target-like responses in almost all cases. Hence, their data are not
analyzed in detail in the present monograph, but they are introduced into
the discussion when relevant. Second, their data were included to explore
whether children over the age of 3;0 already have acquired the agreement
system between determiners and nominals attested to in the literature
(Hernández-Pina 1984; Snyder 2007; Pizzutto & Caselli 1992 for Italian,
among others).
Subjects were recruited informally through family and friends both inside
and outside the metropolitan area of San Juan, Puerto Rico. Children were
chosen after a first visit or telephone contact with one of the parents. The
selection criteria were: (1) the linguistic development of the child, that is,
the children selected were at least at the two-word stage, because the main
focus of the present study is nominal agreement; and (2) that the children
were raised in a monolingual Spanish environment, that is, the childcare
provider and/or the home. Of the five children, only Pepe was enrolled in
some formal schooling, that being pre-Kindergarten. Table 2.1 summarizes
the information for the five subjects involved in the study.
Children completed the tasks in two sessions with the exception of Alonso,
whose first visit was part of the piloting phase of the study, and Pepe, a
4-year-old who was able to complete all the tasks in one session. Table 2.1
also shows that the mean length utterance (MLUw was calculated in words
instead of morphemes).2 First, there is no agreement in the literature
regarding what constitutes a morpheme in a highly inflected language
12 First Language Acquisition in Spanish
like Spanish, as Schnell de Acedo (1994) points out. Second, MLU in words
seems to be the preferred method for Romance languages (Bottari, Cipriani,
& Chilosi 1993/1994; Caselli, Leonard, Volterra, & Campagnoli 1993;
Pizzutto & Caselli 1992; Schnell de Acedo 1994; among others). By using
the same measurement, we can compare the results of the present study
with the findings of previous studies in Romance.
The children were recorded either at home or in the childcare facility they
attended. Video and audio were recorded during two 1-hour sessions per
child, with the exception of Pepe, who successfully completed all the tasks
in the first hour. Video recordings were made using a JVC-GR-DVM90 digi-
tal video camera, and audio was recorded with a Sony-ECM-ZS90 external
microphone placed in a floor stand. The camera was placed on a tripod on
the side of the room where the tasks were conducted. Audio recordings
were also made with a Sony-WM-D6C tape recorder, with the exception of
Elián; in his case, the childcare facility did not have more than one electri-
cal outlet available.
Example 2.1
Protocol: “The Animal House”
Story: Monkey, the puppet, wants to introduce some of his friends,
other animals, to the child. (The animals are inside a bag.)
Monkey: ¿Quieres conocer a mis amiguitos? ‘Would you like to meet my
little friends?’
(The child responds by approaching Monkey or nodding. The
experimenter takes out the first set of animals, for example,
three dogs.)
Monkey: ¿Qué es eso? ‘What’s that (neuter)?’
Child: Un perro or perro. ‘A dog, or dog.’
Monkey: ¿Me prestas uno? or ¿Puedes encontrar un perro? ‘Would you lend
me one?’ or ‘Can you find a dog?’
(The experimenter makes sure that the three dogs are together
and visible to the child. In this fashion, the child is given the
option of selecting the appropriate number of animals accord-
ing to the puppet’s request. The task was repeated with the
other sets of animals.)
Once all the animals were taken out of the bag and Monkey’s requests
were made but not necessarily answered, especially by the younger subjects,
14 First Language Acquisition in Spanish
the experimenter took out a toy house big enough for the child to get
inside. This served as a break from the experimental situation and allowed
the children to play freely, getting in and out of the house. After a few min-
utes of play, the Comprehension Task was continued with the variation in
Example 2.2.
Example 2.2
Protocol: “The Animal House:” Follow-up
Monkey: Creo que los animales están cansados. Busca dos gatos, tienen que ir
a dormir a la casita. ‘I think the animals are tired. Find two cats;
they need to go to the little house to sleep.’
(As mentioned earlier, before making a request, the experi-
menter made sure that the animals, for example, cats, were
grouped together by type and in full view of the child.)
Child: (Finds the requested animal, for example, the two cats, and
takes them to the house to sleep.)
through the additional toys presented to the children, for example, llave
‘key,’ leche ‘milk,’ and luz ‘light.’ For a detailed listing of the additional
props used in the experiment, refer to the section entitled Other Props.
Table 2.2 also shows that adjectives were added to the requests, such as in
the example of la vaca sucia ‘the dirty cow.’3 This was done for the purpose
of measuring the understanding of the adjectives in this first task because
they were going to be elicited in the task: “The Race.” This first task primed
the children regarding the differences among objects. The addition of
adjectives did not affect children’s response to the request, that is, when
children responded, they did it correctly in the majority of cases. In addi-
tion to the nominals or animals listed in Example 2.2, other less common
animals were presented to the children, given the fact that the study was
dealing with children of different age groups. For example, two giraffes
‘jirafa in Spanish,’ were shown to the children. Finally, I would like to point
out that even though some of the objects had two characteristics, for exam-
ple, small blue butterflies, only one adjective at the time was used in the
requests, in this case, the request was as follows: Could you find the small
butterflies or the blue butterflies?
Masculine nominals used in this task are summarized in Table 2.3.
In contrast to the feminine nouns presented in Table 2.2, it was easier to
find masculine nouns unmarked for gender that were recognizable to the
children, for example, pez ‘fish’ and león ‘lion.’
16 First Language Acquisition in Spanish
Example 2.3
Protocol: “Time to Eat”
Story: Puppet tells the child that it is time to eat, to find an
animal that she thinks might be hungry.
Monkey: Es hora de comer. ¿Tú sabes quién tiene hambre? ‘It is time to
eat. Do you know who is hungry?’
Experimental Methodology 17
In this task, as the experimenter took out a food item, children instinctively
tried to identify it and name it. If they did not say what the item was, the
experimenter would take one of the items in her hand and label it for the
child, for example, esto es arroz ‘this [neuter] is rice,’ with one of the two
piles of rice in her hand. Once the child had identified the food choices,
she would make a food selection for the “hungry animal,” for example, rice.
On the one hand, if the child has a default value in her grammar of “mass”
she is expected to produce all the nouns in the singular number because
mass nominals are not pluralized. On the other hand, if the child’s gram-
mar distinguishes between mass nominals (not pluralizable) versus count
nominals (pluralizable), the child should produce singular mass nominals
and plural count nominals. Crucially, mass nouns were presented in two
sets, for example, two piles of rice. In this way, the experiment gave the
children the option of pluralizing mass nominals if the distinction between
mass and count nominals was not present in their grammar.
The task proceeded with the animal (manipulated by the experimenter)
eating from the other food group, for example, flowers, and complaining
that the food tasted bad, as shown in Example 2.4.
Example 2.4
Protocol: “Time to eat” (Continuation)
Hungry animal: Ah eso sabe mal. ¿Qué comí? ‘Oh that tastes bad! What did
I eat?’
Child: Flores. ‘Flowers’ (four flowers).
Hungry animal: ¿Flores? ¿Qué como? ¡Tengo hambre! ‘Flowers? What should
I eat? I’m hungry!’
Child: Arroz or arroces. ‘Rice’ or ‘*Rices’ (two piles of rice).
18 First Language Acquisition in Spanish
The child was expected to produce the count nominal in the plural form,
for example, flores ‘flowers’ to mark the distinction that in her grammar
count nominals are pluralized. Finally, the animal would eat the rice and
comment on how good it tasted: Eso sabe bien, ¿qué comí? ‘This tastes good,
what did I eat?’ The child was expected to say arroz ‘rice’ and not arroces
‘rices’ if there was a difference between the mass and count nominals in her
grammar. If the child did not respond, the experimenter asked the child
what did the animal eat, ¿qué comió? The same routine was repeated with
several food items since some of the children would identify one item of the
mass/count set but not the other. The task would end after all the food
items were “eaten” and responses were obtained from the child or when the
child was no longer interested in this particular game. The food sets used
in this task are listed in Example 2.5.
Example 2.5
Mass and Count Nominals
In this task what was important was the pairing of two different kinds of
nouns, for example, mass versus count, so no strict order was followed
regarding the pairing of two specific items, for example, hair was paired
with bananas or flowers, for instance. The items in Example 2.5 marked
with an asterisk (*) were introduced at the end of the task. These items
were presented to the child in the following situations: (1) when the child
did not respond to some of the items listed in Example 2.5; or (2) when the
child had completed the task but was still interested in the feeding game. In
the case of the two bottles of milk, the procedure followed was to make the
animal drink first from one bottle and then immediately from the second
bottle. Then the child was asked: ¿Qué tomó? ‘What did the animal drink?’
The expected answer was leche ‘milk’ or la leche ‘the milk’ even though there
were two bottles given. In the case of the cups, the children immediately
pretended to be drinking something. In this context, the experimenter
told the child to drink from the two cups and then asked her: ¿qué tomaste?
Experimental Methodology 19
‘What did you drink?’ Children’s responses were agua ‘water,’ café ‘coffee.’
None of the children pluralized these particular nouns, even though the
possibility is available in the target grammar, that is, two cups of water/cof-
fee, two bottles of milk. Finally, another variation to the main protocol of
this task was added to make it more interesting for the children, “Time to
Sit Down” (see Example 2.6). Like in the other two versions of the task, two
choices were given to the child as the seating place for the doll, for exam-
ple, a pair of chairs or two bunches of grass.
Example 2.6
Protocol: “Time to sit down”
Story: A human doll (or the child’s favorite animal) indicated to
the child that he was tired and needed to sit down.
Doll: Estoy cansado. ¿Dónde me siento? ‘I am tired, where can
I sit?’
Child: En la silla or silla. ‘On the chair or chair.’
(Doll sits on the grass and complains.)
Doll: ¡Au! ¡Esto pica! ¿Dónde me senté? ‘Ouch! This is itchy! Where did
I sit?’
Child: En la hierba/*hierbas ‘On the grass/*grasses’
Doll: Oh, me equivoqué. ¿Dónde me siento? ‘Oh I made a mistake. Where
should I sit?’
Child: En la silla. ‘On the chair.’
(Doll would sit first on one of the two chairs and would fall
down because it was too small for him; then the experimenter
would put the two chairs together indicating to the child that
two will hold the doll.)
Doll: Esto es más cómodo. ‘This is more comfortable.’
told the child that it would be fun to have a race to the house. Usually the
children got very excited about participating in a competition. Before the
race took place, the child was asked to choose which vehicles she would like
to drive, as shown in Example 2.7.
Example 2.7
Protocol: “The Race”
Story: Monkey likes races and he likes to win.
Experimenter: ¿Quieres hacer una carrera para ver quién gana? ‘Do you want
to have a race with Monkey to see who wins?’
Child: Sí, yo voy a ganar. ‘Yes, I am going to win.’
Experimenter: Sí, pero primero tienes que escoger qué carros vas a manejar.
‘Yes, but before you have to choose which cars you are
going to drive.’
(Experimenter points at the two different sets of cars, for
example, big vs small.)
Child (expected response): Los grandes/los carros grandes or los chiquitos/
los carros chiquitos. ‘The big ones/the big cars or the small
ones/the small cars.’
Monkey: En su marca, listos. . .¡fuera! ‘On your mark, get set, go!’
(The race takes place, and then the experimenter asked
the child who won.)
Experimenter: ¿Quién ganó? ‘Who won?’
Child (expected response): Los carros grandes or los grandes. ‘The big cars
or the big ones.’
(Then the task was repeated with other vehicles, for
example, motorcycles and trains.)
The production of attributive adjectives from these children was very lim-
ited, especially in the case of the younger ones, who preferred to point at
the vehicles when asked to choose between the two sets. When the younger
children did not produce the adjective after several trials, the experimenter
would establish a contrast between the two sets of objects as follows: Mira, yo
tengo los grandes. ‘Look, I have the big [ones].’ In some cases this phrase was
enough to trigger a response from the child like: Y yo tengo los chiquitos. ‘And
I have the small [ones].’ However, if the child did not respond, the experi-
menter followed up with the question: ¿y tú qué tienes? ‘And what do you
have?’ As in this example, the experimenter would always try to choose the
unmarked adjective, for example, grand–e instead of chiquit–o, to limit the
information given to the child in terms of the agreement needed. As stated
Experimental Methodology 21
earlier for the other tasks, the nominals selected for this task had to be
objects recognizable to the children and they have to provide a variety of
endings, that is, canonically marked and non-canonically marked for gen-
der, as its goal was to explore the nature of the agreement system in chil-
dren’s grammar. In particular, I want to explore whether children acquiring
nominal agreement Spanish are creating agreement relations based on the
clues overtly on the morphology, or there is evidence of an underlying
agreement system. As a result, an attempt was made to select a variety of
noun types and modifying adjectives. Masculine nominals included in this
task are listed in Table 2.4.
As illustrated in Table 2.4 only one overtly marked masculine nominal
was used as a prop for this task, for example, carros ‘cars’ since there were
issue, for example, llave ‘key,’ as can be seen in Table 2.5. As with the mas-
culine nominals, the vehicle choice made by the child would determine the
agreement pattern required. For example, if the child chose to drive la
motora roja ‘the red (fem/sg) motorcycle,’ then she would have to create an
agreement relation between the nominal motora and an overtly marked
adjective, roja.
included in the analysis for comparison purposes, that is, to examine if the
agreement patterns children created with these adjectives were similar to
those established with attributive ones. Finally, the data also include the
analysis of the agreement patterns (i.e., number and gender features) of
demonstrative pronouns and third person clitics. The structures under
study were entered into an Excel spreadsheet to be coded, as illustrated in
Example 2.8.
Example 2.8
Coding System
a. Utterance: actual child production
b. Context
c. Child behavior: what the child is doing at the moment of the
utterance
d. Structure
e. Agreement: Target-like/Non-Target-like
f. Noun marking: Canonical/Non-canonical
All utterances pertaining to this study were coded, along with the context
in which they occurred and what the child was doing when she uttered a
particular structure. This last piece of information was vital for the analysis
of the acquisition of the number feature, especially when it came to deter-
mining if there was agreement between the referent and the nominal
produced by the child. Structures were coded as: (1) Bare Noun Phrases;
(2) Determiner Phrase (DP), including structures with full determiners
and structures with a Monosyllabic Place Holder; (3) DPs with an adjective,
including structures with an overt nominal and those in which the nominal
was non-overt; (4) Adjective Phrase; (5) Determiner, referring to demon-
strative pronouns produced in isolation; (6) Other, including third person
clitics. Furthermore, the agreement relations were coded as target-like or
non-target-like in reference to number and gender. Finally, nominal end-
ings were coded as canonical, that is, –o and –a, and non-canonical, that
is, –e and final consonant sounds, to examine if there was any relation
between the information encoded in the nominal endings and the agree-
ment relations children created.
The following were excluded: repetitions of immediately preceding
utterances (as well as the child’s immediate exact repetitions of her own
utterances, for example, caballo, caballo ‘horse, horse,’ (Alonso); songs and
formulaic expressions. Furthermore, ambiguous cases were excluded from
the analysis. Ambiguity was found with regard to Monosyllabic Place Holders
Experimental Methodology 25
(i.e., vowels that appeared in front of the nouns; Bottari et al. 1993/1994),
in particular a and e as shown in Example 2.9.
Example 2.9
(a) A cashita ‘to house’/‘a house or the house’ (Alonso)
(b) ee mío ‘(It) is mine’ / ‘the mine’ (the one that belongs to me)
(Elián)
Bare noun phrases 76 (55%) 105 (41%) 213 (43%) 394 (44)
Gender Agreement
3.1 Agreement
Spanish nouns possess at least three features: gender, number (the focus of
this monograph), and definiteness. Gender and number are marked mor-
phologically on the noun in the form of affixes, for example, –a for femi-
nine as in cas–a ‘house’ and –o for masculine as in castill–o ‘castle,’ whereas
definiteness is marked by a prenominal determiner, for example, la casa
‘the house’ or el castillo ‘the castle,’ that also encodes the gender and num-
ber features. In addition, gender and number agreement is also marked
overtly on constituents such as attributive and predicative adjectives and
demonstrative and object pronouns. Koehn (1994) points out the complexity
learners face in the acquisition of these two features, stating that it involves
at least four tasks: (1) the child has to develop the underlying semantic
concept for number, that is, the distinction between one and more than
one; (2) the child has to recognize that gender and number are systemati-
cally encoded on specific syntactic categories, that is, the corresponding
grammatical features have to arise; (3) the appropriate morphophonologi-
cal realizations of these features have to be acquired; and finally, (4) the
appropriate agreement paradigms have to be learned. This chapter exam-
ines the acquisition of gender agreement, and it is organized in three major
sections. First, Section 3.2 presents an overview of agreement in Spanish,
with special focus on gender. Section 3.3 discusses some of the previous
studies on the acquisition of gender in Spanish. Finally, Section 3.4 focuses
on experimental findings for the present research.
not and can be expressed (or not expressed) with the same phonetic
ending or suffix (Martínez 1999). One of the most salient characteristics
of the Spanish language is the redundancy present in terms of its agree-
ment morphology, that is, all nominal modifiers and determiners must
agree in gender and number with the noun with which they enter into
an agreement relationship. For example, in la casa amarilla ‘the (fem/sg)
house (fem/sg) yellow (fem/sg)’ (the yellow house), the nominal casa
shares the same word marker –a as the adjective amarill–a and the deter-
miner la.
Agreeing constituents within Spanish Determiner Phrases (DPs) include
determiners and attributive adjectives. In Spanish, the determiner class
encompasses: (a) the definite articles, for example el ‘the (masc/sg)’ or la
‘the (fem/sg)’ and their corresponding plural forms for masculine and
feminine respectively: los and las; (b) the indefinites articles, for example,
un ‘a (masc/sg)’ and una ‘a (fem/sg);’ and (c) quantifiers such as poco/a
‘few (masc/fem/sg),’ and their plural forms pocos/pocas and todo/a ‘all
(masc/fem/sg)’ and the corresponding plurals todos/todas. In addition,
demonstrative adjectives are included in the determiner class for the
purpose of this study: este/esta ‘this (masc/fem/sg),’ ese/esa ‘that (masc/
fem/sg),’ and aquel/aquella ‘that over there (masc/fem/sg)’ and their cor-
responding plural forms: estos/estas ‘these,’ esos/esas ‘those,’ and aquellos/
aquellas ‘those over there.’ In general, determiners precede the noun in
Spanish, and they play a crucial role in terms of gender agreement when
they occur in DPs such as la noche ‘the (fem/sg) night (unm/sg).’ In this
example, the nominal noche is not marked for gender overtly; therefore the
only overt evidence of the feminine feature of the nominal is encoded in
the determiner la. In addition, the monograph will also discuss agreement
patterns that children established in the production of demonstrative
pronouns. These pronouns, like their adjectival counterparts, have to
agree in gender and number with the noun they represent.
Table 3.1 shows how each set of demonstrative pronouns, that is, este ‘this
(one),’ ese ‘that (one),’ aquel ‘that (one) over there,’ has a unique morpho-
logical representation according to the values it encodes for the gender
and number features. For example, the demonstrative esta ‘this (one),’
encodes the feminine/singular features whereas the demonstrative ese ‘that
(one),’ encodes the masculine/singular features. Moreover, this paradigm
also includes one of a few examples of the “neuter” gender in the Spanish
language, for example, esto ‘this (abstract idea or situation)’ (see discussion
in Section 3.3).
Gender Agreement 29
Singular Plural
Neuter Esto
Eso
Aquello
These pronominals were included in the analysis because they also serve
to assess the agreement system available to the children under study, as
shown in Example 3.1.
Example 3.1
Quiero esta ‘(I) want this [one] (fem/sg).’
In Example 3.1, the demonstrative pronoun esta agrees with the feminine
singular features of the nominal it represents. In the agreement analysis of
these pronouns, the context of the utterance becomes of utmost impor-
tance in determining the grammaticality of a particular token.
Another class of agreeing constituents within DP is the qualifying adjec-
tives. Adjectives in Spanish, like determiners, obligatorily agree with the
nouns they modify in gender and number, for example, la casa roja ‘the
(fem/sg) house (fem/sg) red (fem/sg)’ (the red house). Most adjectives
are overtly marked for gender, expressed morphologically with the –o end-
ing in the masculine and the –a ending in the feminine (Harris 1991). But
there are also adjectives that lack a word marker, and they are invariable in
both genders, for example, el gato inteligente ‘the cat intelligent (masc/sg)’
(the intelligent cat) versus la gata inteligente ‘the cat intelligent (fem/sg).’
In this example, the adjective inteligente has the same morphology for the
male and female cat. Spanish qualifying adjectives in attributive position
usually occurred in postnominal position, for example, la manzana roja ‘the
apple red.’ The prenominal position, although possible, entails a change in
30 First Language Acquisition in Spanish
the meaning of the DP, for example, la roja manzana ‘the red apple,’ in
which the redness of the apple is emphasized (Demonte 1999).1 The pres-
ent research focuses on postnominal adjectives, the unmarked position for
Spanish.2
In addition to the determiners and attributive adjectives, the present
monograph will also include a discussion of other constituents produced by
the children under study, which are also marked overtly for agreement but
are not necessarily within DP, such as predicative adjectives. Predicative
adjectives in Spanish, as in English, occur with the copula to be. However, in
Spanish the copula has two representations, ser and estar, with contrasting
semantic interpretations, as illustrated in Example 3.2.
Example 3.2
(a) Pedro es gordo ‘Peter is fat’
(b) Pedro está gordo ‘Peter is fat’
Singular
1st person Me ‘me’ Me ‘me’
2nd person Te ‘you’ Te ‘you’
3rd person Lo ‘it (masc)’ Le ‘him/her’
La ‘it (fem)’
Plural
1st person Nos ‘us’ Nos ‘us’
2nd person Os ‘you’ Os ‘you’
3rd person Los ‘them (masc)’ Les ‘them’ (masc/fem)’
Las ‘them (fem)’
Gender Agreement 31
Example 3.3
Marker Gender Example
(a) –o masc. only muchacho ‘boy’
fem only mano ‘hand’
masc or fem testigo ‘witness’
none dentro ‘inside’
(b) –a fem only muchacha ‘girl’
masc only día ‘day’
masc or fem turista ‘tourist’
none fuera ‘outside’
Example 3.3 illustrates Harris’ point that these two markers (–o and –a) can
be attached to nominals of either gender, such as muchacho/muchacha, as
well as to other constituents, such as the adverbs dentr–o ‘inside’ and fuer–a
‘outside.’ In addition, not all Spanish nominals have a WM; such nominals
typically end in –e or in a single coronal consonant and can be of either
gender, as illustrated in Example 3.4.
Example 3.4
Gender Nominal
(a) masc. only padre ‘father’
sol ‘sun’
(b) fem. only madre ‘mother’
col ‘cabbage’
(c) masc. or fem. amante ‘lover’
mártir ‘martyr’
(d) none delante ‘ahead’
atrás ‘behind’
In Example 3.4, the sequences padr–, madr– in 3.4(a) and 3.4(b), and
the –nt– in amant– in 3.4(c) are not permissible codas in Spanish, therefore
an epenthetic –e appears after otherwise unsyllabifiable segments (Harris
1991; Klein 1989; among others). On the other hand, words such as col, sol,
and atrás are syllabifiable without the addition of a final –e. Nonetheless,
Gender Agreement 33
Example 3.5
Gender Ending %
(a) Feminine –a 96.30
–d 97.57
(b) Masculine –o 99.87
–l 97.85
–r 98.55
–i 93.13
–u 95.10
(c) Either –n 51.61 (Fem)/48.39 (Masc)
–z 61.63 (Fem)/38.37 (Masc)
–s 42.68 (Fem)/57.32 (Masc)
In this monograph, I address the issue of an initial default value for gen-
der from a different perspective, proposing that children’s production of
feminine nominals with masculine determiners or masculine pronominals
to refer to feminine referents is in fact part of an acquisition strategy in
order to converge. Similar proposals on the availability of default values
in child language have been advanced in the literature by Phillips (1996)
and Borer and Rohrbacher (1997) to account for root infinitives (RIs) in
child language. In particular, these researchers propose that the produc-
tion of RIs is target-like in children’s grammar as a default verbal form
(Phillips 1996) or as an avoidance of non-target-like agreement (Borer &
Rohrbacher 1997).
Phillips argues that RIs produced by 2-year-olds should not be interpreted
as evidence of a deficit or lack of functional projections in initial grammars
(cf. Clahsen, Eisenbeiss, & Vainikka 1994; Radford 1990, 1994; Wexler
1994). Instead, he states that children have all the syntactic components
of an adult structure but they are missing the derivational step, which
combines the verb with inflection; that is, children have difficulties in
accessing morphological knowledge. Crucially, Phillips distinguishes in his
analysis two types of constraints: (a) rigid constraints, or “constraints that
must be satisfied or else the sentence is ruled out automatically” (1996,
589); for example, in Germanic V2 languages, wh-questions and topicaliza-
tions are Complementizer Phrases (CPs) involving V-raising to C; another
example is Case-licensing in V-raising languages, and (b) violable constraints,
“which can only be violated if nothing better can be done” (p. 589); for
example, V-raising to inflection is not an absolute requirement.
Phillips’ first argument for the availability of functional projections
(FPs) in initial grammars is the contrast in the production of RIs in V2
languages in declaratives versus their disappearance in wh-questions and
topicalizations; that is, analysis of Dutch, German, and Swedish child data
showed that although children were producing RIs in declarative struc-
tures, RIs disappeared almost completely in wh-questions and topicaliza-
tions, structures that require V-raising to C. Phillips argues that if children
are lacking FPs (e.g., inflection or C), then we should expect random pro-
duction of RIs regardless of the structure involved. This prediction was
not borne out by the Germanic V2 child data. Moreover, the author points
out that when these children produced verbal inflection, their production
was target-like.
Phillips’ second argument in favor of the availability of FPs in initial gram-
mars is based on the correlation found in the data between the production
of RIs and null subjects.5 He found the interaction of finiteness with subject
overtness was not uniform across languages, as analyses based on syntactic
36 First Language Acquisition in Spanish
In both the Greek and the Spanish nominal systems, it is not morphologi-
cally possible to produce a bare verbal stem, and there are no infinitival
forms in this language. Varlokosta et al. found that Greek children initially
prefer the verbal forms ending in the third person singular (3sg) suffix –i,
for example, anitsi ‘open (3sg.Perf.Subj).’ The authors point out that this
form is ambiguous with the participial form, for example, έχω ανοίξει (eho
anixi) ‘(I) have opened.’ The children produced the i-form in non-third-
person contexts, but only when the forms were ambiguous between the
third person interpretation and the non-agreeing participial reading. The
production of this form can be interpreted as the production of a default
verbal form, similar to the default masculine gender argued for in Spanish
in the present study.
Several studies have addressed the topic of the acquisition of gender agree-
ment in Spanish as a first language. Most of the studies reviewed agree on
the gradual nature of the acquisition of gender agreement morphology,
which starts with the production of mostly masculine singular DPs and
progressively expands to the inclusion of feminine ones (Aguirre 1995;
Hernández Pina 1984; López Ornat 2003; Schnell de Acedo 1994; among
others). However, these studies as a whole do not provide a coherent pic-
ture regarding the nature of gender agreement in Spanish child language.
Discrepancies in research findings arise from several factors, such as the
different methodological approaches used (naturalistic vs experimental);
the subject’s linguistic development (chronological age vs Mean Length
Utterance); and the type of data analysis conducted (qualitative vs quantita-
tive). Researchers such as Hernández Pina (1984) and Maez (1982), for
example, present purely descriptive analyses of the acquisition of gender
agreement using naturalistic data, whereas Schnell de Acedo (1994) and
Snyder (1995, 2007) provide quantitative analyses of the naturalistic data used.
Furthermore, the studies reviewed present opposite views on the nature of
gender agreement in initial grammars, one of the focuses of the present
monograph. On the one hand, some researchers portray an image of the
acquisition process with an apparent abundance of non-target-like produc-
tion until acquisition is achieved (Hernández Pina 1984; López Ornat 2003;
Maez 1982), whereas others present a picture of a virtually errorless process
(Aguirre 1995; Schnell de Acedo 1994). This section presents a discussion
of these opposite views on the nature of gender agreement.
Gender Agreement 39
Example 3.6
Rafael’s production Target-like
(a) mota rota moto rota ‘motorcycle (fem) broken (fem)’
(b) tierra azula tierra azul ‘earth(fem) blue (fem)’
(c) un llave una llave ‘a (masc) key (fem)’
A quick review of the DPs in Example 3.6 indicates that Rafael may have
been aware of the Spanish canonical feminine morphological marker,
namely a final –a. In particular, he tried to regularize the morphological
endings of the nominal mot–o (short for motora) to mot–a ‘motorcycle’ in
Example 3.6(a) and the invariable adjective azul ‘blue,’ to azula.6 Finally,
Example 3.6(c) illustrates the use of the masculine determiner un ‘a/one’
with the unmarked feminine noun llave ‘key.’ This could be an indication
that masculine is the unmarked value in Spanish initial grammars. Unfortu-
nately, the researcher does not provide any quantitative data to support her
assessment of Rafael’s acquisition of gender agreement.
Maez (1982) presents a similar view of the acquisition of gender agree-
ment, namely, a process characterized by great variability and the produc-
tion of non-target-like utterances. He studied the acquisition of nominal
morphology for a period of six months, collecting spontaneous production
data every two weeks from three children: Karina, Celena, and Ana, ages 1;6
40 First Language Acquisition in Spanish
Example 3.7
Age Utterance Error type
1;7 *un (una) boca (Celena) ‘a/one mouth’ Masc. Det/Fem. Noun
1;8 *un (una) cosita (Karina) ‘a/one little thing’ Masc. Det/Fem. Noun
*la (el) pan (Karina) ‘the bread’ Fem. Det/ Masc.Noun
1;10 *un (una) leche (Ana) ‘a/one milk’ Masc. Det/Fem. Noun
1;11 *un (una) uva (Ana) ‘a/one grape’ Masc. Det/Fem. Noun
2;0 *el (la) televisión (Celena) ‘the television’ Masc. Det/Fem. Noun
*un (una) calabaza (Celena)‘a/one pumpkin’ Masc. Det/Fem. Noun
*una (un) jabón (Ana) ‘a/one soap’ Fem. Det/Masc. Noun
*lo (la) silla (Celena) ‘the chair’ Neuter Det/Fem. Noun
Example 3.7 shows that the most common type of non-target-like agree-
ment utterance with respect to gender (six out of nine) is the production
of a masculine determiner with a feminine noun, for example, un calabaza
‘a/one (masc) pumpkin (fem)’; this might be an indication of the avail-
ability of an initial value for gender. Example 3.7 also includes two examples
of the use of a feminine determiner, la pan ‘the (fem) bread (masc)’ and
una jabón ‘a/one (fem) soap (masc).’ Interestingly, Hernández Pina cites
similar examples in Rafael’s production; these examples point to the pos-
sibility that these children are not merely matching the input heard but
working out their own internal agreement system. Notice that these two
nominals, jabón and pan, are readily available in the input to these children.
The last DP listed in Example 3.7, lo silla ‘the (neuter) chair (fem),’ illus-
trates the use of the neuter pronoun lo with a feminine noun. This case is
interesting because it might reflect an attempt by the child at regularizing
the definite article paradigm from el/la to lo/la. I should point out, this
neuter pronoun is used only with abstract reference in the target language,
for example, lo importante ‘the important (matter).’ As a result, this use
might not be part of the primary linguistic data. As in the study by Hernán-
dez Pina, Maez’s findings point to an acquisition process in which children
are producing non-target-like utterances together with target-like utter-
ances; however, no major conclusions can be drawn given the chosen meth-
odological approach.
In contrast with the studies reviewed, Schnell de Acedo (1994) and
Aguirre (1995) present a view of a virtually errorless gender acquisition
process in terms of the gender agreement relations established by the chil-
Gender Agreement 41
Definite Indefinite
16 4 0 0 0 4 3 1 0 0 4 8
23 2 5 10 1 18 0 0 0 0 0 18
29 8 13 1 0 22 31 12 2 0 45 67
Total 14 18 11 1 44 34 13 2 0 49
dren under study. First, Schnell de Acedo examines the acquisition of agree-
ment in determiner-noun structures using the longitudinal naturalistic
data of a monolingual Venezuelan girl, Morela. The researcher included in
her study three recordings: 16, 23, and 29 months. Similarly to Hernández
Pina’s findings, Morela’s production of determiners increased in a gradual
fashion, as illustrated in Table 3.3.
Table 3.3 also shows the agreement patterns present in Morela’s Deter-
miner Phrases (DPs). First, at 16 months with a MLUw of 1.35, most of her
DPs are masculine (seven out of eight) and all were target-like with respect
to agreement. Morela produced only one instance of a feminine DP, and it
was non-target-like: una papá ‘a/one (fem) dad (masc).’ This example is
interesting because the nominal has the feminine canonical mark for gen-
der, namely, a final –a. This non-target-like utterance seems to support
Pérez-Pereira’s (1991) findings on gender assignment clues. The researcher
conducted an experimental study with 160 children, ages 4 to 11 years
old, to study the relevance of intralinguistic clues (syntactic and mor-
phophonological) versus extralinguistic ones (male vs female) in children’s
recognition of the gender of 22 created nouns. He found that intralinguis-
tic clues are more relevant than extralinguistic ones in terms of compre-
hension, for example, children better understood items such as una pilín
‘a/one (syntactic clue) pilin (unmarked)’ than items such as dos borales ‘two
borales (natural gender clue).’
At 23 months (MLUw = 2.15), Morela produced a total of 18 DPs (12 mas-
culine and 6 feminine), all of them target-like with respect to agreement.
This marks a dramatic change in her production, from 8 to 18 and from mostly
all masculine to one-third feminine. Finally, at 29 months (MLUw = 2.26), the
child produced a total of 67 DPs: 42 masculine and 25 feminine. Schnell de
Acedo states that Morela has a complete command of the full inventory of
42 First Language Acquisition in Spanish
forms at this age. However, this claim does not seem to be supported by
Morela’s data, shown in Table 3.3. For example, Morela did not produce
the feminine articles in the plural form. Aguirre (1995) brings support for
Schnell de Acedo’s findings of an errorless acquisition of gender. In her
study on the acquisition of the Spanish determiner, Aguirre finds that the
children under study present an almost errorless production of gender,
noting only one example of a non-target utterance with respect to gender,
for example, un niña ‘a/one (masc/sg) girl.’ Interestingly, the author also
reports that children were also producing nominals with the invariant
vocalic element a, for example, a niños ‘a children’ and a pollo ‘a chicken.’
This points to an acquisition strategy of using a when they are not able to
produce the target determiner, as noted by the researcher.
Schnell de Acedo’s findings are at odds with Maez’s and Hernández Pina’s
results in terms of the nature of gender agreement in early grammars; that
is, the former two stress the fact that children are indeed producing non-
target-like utterances at the beginning stages, but the latter finds only one
instance of a non-target-like utterance in her analysis.7 This striking differ-
ence between these findings might be the result of a methodological arti-
fact, namely, frequency of recordings.
Snyder (1995) presents a more balanced view of gender acquisition within
DP, using the corpus of Juan Linaza from CHILDES (MacWhinney & Snow
1985) from ages 1;7 to 3;5. His analysis reveals that even though Juan’s
determiners correctly agree with their head nouns in gender in the vast
majority of cases from the earliest stages, the child also produces non-
target-like utterances. Juan’s DP production ranges from 83 to 100 percent
target-like. In particular, at age 2;0, Juan produces examples of contrastive
gender uses of the determiner, for example, el/la ‘the (masc/fem).’
The current review of research on the nature of gender agreement in
Spanish early grammar yields inconclusive results. On the one side, research-
ers agree on the emergence of the masculine DPs, bringing support for the
availability of an initial unmarked value for the gender feature, namely the
masculine. On the other side, studies disagree on the process of conver-
gence to the target grammar. As mentioned earlier, this might be a result of
the different methodological approaches adopted.
BNPs 394 44
(Full) DPs 94 11
MPH/DPs 102 11
Attributive adjectives 18 2
Predicative adjectives 77 9
Target-like 83 (89%) 87 (85%) 140 (85%) 30 (70%) 12 (67%) 61 (79%) 413 (83%)
Total 94 (100%) 102 (100%) 165 (100%) 43 (100%) 18 (100%) 77 (100%) 499 *(100%)
a
(Full) DPs
b
Monosyllabic Phrase Holder DPs
c
Demonstrative pronouns
d
Third person clitic pronouns
e
Attributive adjective
f
Predicative adjective
*
Notice that Table 3.5 totals do not include the 394 BNPs produced by the children under study. This total
was excluded from the table because BNPs occur in isolation, that is, they do not enter into an agreement
concord.
The strong target-like production in Table 3.5 points out the fact that
children acquiring Spanish are able to establish adult-like agreement rela-
tions from the early stages of the acquisition process. Notice that two of the
children are below the two-word stage in their production, that is, Elián,
MLUw = 1.5 and Alonso, MLUw = 1.9. Moreover, children’s target-like pro-
duction brings support to previous acquisition studies that claim that the
percentage of non-target-like production of children is limited (Aguirre
1995, Snyder 1995, and Schnell de Acedo 1994, for Spanish; Pizzuto &
Caselli 1992 for Italian, among others).
The second major generalization is in regard to the timing of the acquisi-
tion of gender and number features. Specifically, my analysis indicates
that gender morphological markings are acquired before number in these
structures, as the three children exhibited problems marking the plural
number feature in almost all the constituent types under study. The issue
of number acquisition is addressed in Chapter 4. These results provide
support for previous cross-linguistic and Spanish acquisition studies that
claimed that plural number posed difficulties in the acquisition process
(Caselli et al. 1993; Koehn 1994; López Ornat 2003; Maez 1982; Marrero &
Aguirre 2003). Moreover, this difference between the acquisition of num-
ber and gender points to a contrast in the acquisition of grammatical versus
semantic feature markings. Children seem to acquire grammatical gender
markings before plural (semantic) number markings. Contrary to the pres-
ent findings, Koehn (1994) argues that Ivar (a child acquiring German and
Gender Agreement 45
French) has no gender distinctions but has acquired the semantic concept
of singularity versus plurality, although the development of the grammatical
notion of number takes place later on. Similarly, Müller (1994) also claims
that grammatical features are unavailable at the beginning of the acquisi-
tion process of the two bilingual children under study (German and
French). Notice that these discrepancies might be based on the nature of
the linguistic systems being acquired: although it is not morphologically
possible in Spanish to produce a nominal (or the agreeing constituents)
without marking their gender, in German and French the learner has this
option, that is, German adjectives can be produced bare and French is not
a very morphologically rich language in terms of number and gender.
In terms of the overall non-target-like production, the percentages were
fairly low in general. However, the nature of the non-target-like utterances
produced in each structure was not the same. In particular, the three chil-
dren seem to have acquired gender agreement between determiners and
nouns and (apparently) between nominals and attributive adjectives,8 but
they had problems with number agreement in these structures. In contrast,
non-target-like utterances in the production of third-person clitics and
demonstrative pronouns mostly involved the application of the masculine
“default” gender;9 whereas non-target-like predicative adjectives utterances
involved mismatches of both gender and number features.
Mismatch Default
(Full) DPs 62 1 – 1
MPH/DPs 33 4 – 4
Attributive adjectives 18 – – –
Predicative Adjectives 65 – 10 10
Demonstratives 118 6 13 19
Total 333 12 31 43
Mismatch Default
(Full) DPs 62 1 – 1
MPH/DPs 33 4 – 4
Attributive adjectives 18 – – –
Predicative Adjectives 65 – 10 10
Demonstratives 118 6 13 19
Total 333 12 31 43
Target-like 4 19 38 61
Non-target-like 1 0 0 1
Total 5 19 38 62
48 First Language Acquisition in Spanish
Example 3.8
Child utterance Target
*la yuguete el juguete
‘the (fem/sg) toy (unm/masc/sg)’ ‘the (masc/sg) toy (unm/masc/sg)’
In Example 3.8, Elián utters the feminine determiner la with the mascu-
line nominal juguete, creating what I have called a feature mismatch, that is,
the production of a masculine nominal with a feminine determiner. This is
an interesting mismatch because the noun juguete is not overtly marked for
gender, ending in the vowel –e. As a result, this noun does not offer any
overt clues to the language learner regarding the value of its gender feature.
In the Spanish language, nominals ending in this vowel can belong to either
the masculine or the feminine gender. Although it is tempting to suggest
that he is using the feminine gender as the “default” value, given the limited
production of DPs by this particular child, no conclusions can be reached
regarding this matter.
In contrast to Elián’s all feminine production, Alonso exhibited an almost
all-masculine determiner production (17 out of 19 tokens), as shown in
Table 3.9.
Demonstrative 2 – – – 2 (11%)
Other 2 – – – 2 (11%)
Total 12 5 2 – 19 (100%)
Gender Agreement 49
Crucially, the two feminine DPs this child produced were also target-like
with respect to gender agreement. Moreover, Alonso’s data reflects a wider
variety of DPs in comparison to Elián’s (almost) all definite production,
as seen in Table 3.9. The following types of determiners were present in
Alonso’s data: (a) definite articles, for example, el caballo ‘the (masc/sg)
horse;’ (b) indefinite articles, for example, un carro ‘a/one (masc/sg) car;’
(c) demonstrative adjectives, for example, ese guauguau ‘that (masc/sg)
dog;’ and (d) the indefinite otro ‘another,’ for example, otro caballo ‘another
horse.’ Table 3.9 illustrates the distribution of determiner types produced
by Alonso (see Appendix A for a complete list of Alonso’s (full) DP tokens).
Table 3.9 also shows that the definite article was the determiner with the
highest production frequency: 10 tokens out of a total of 19. However, the
difference between the production of definite versus indefinite articles was
not found to be statistically significant (χ2 = 1.7, p < 0.20). Moreover, the
definite determiner was the only determiner produced also in the feminine
form, for example, la comida ‘the (fem/sg) food (fem/sg)’ and la lu ‘the
(fem/sg) light (un/fem/sg)’ for la luz. Regarding the indefinite article, it
was only produced in the singular masculine form un, for example, un nene
‘a/one (masc/sg) kid (masc/sg).’ Alonso uttered two demonstrative deter-
miner tokens, both in the singular masculine form, for example, ese guau-
guau ‘that (masc/sg) dog (masc/sg)’ and este caballo ‘this (masc/sg) horse
(masc/sg).’ Finally, he also produced two tokens of the indefinite otro
‘another,’ in the masculine singular, for example, otro caballo ‘another
(masc/sg) horse (masc/sg).’
Similar to the almost all target-like production of Elián and Alonso with
respect to gender agreement, Londa’s (full) DP utterances reflect an adult-
like agreement production, as illustrated in Table 3.10.
Londa produced 29 instances of masculine nouns (24 singular and 5
plural) and 27 instances of feminine nouns (24 singular and 3 plural), as
illustrated in Table 3.10. Notice that in Table 3.10 there is an instance of
Singular 24 – 24 1*
Plural 3 2 3 –
Total 27 2 26 1
50 First Language Acquisition in Spanish
Example 3.9
Child Target
*Mira él, pielna Mira las piernas de él/ sus piernas
‘Look he/the leg’ ‘Look (at) the legs of him/his legs’
The child uttered Example 3.9 as she showed the legs of her toy dog to the
experimenter. This example is ambiguous because it can be interpreted as
the child saying mira el pielna ‘look the (masc) leg (fem)’ in which case
there would be an instance of a non-adult-like gender agreement utter-
ance, or what I call default, involving the use of masculine definite deter-
miner el ‘the’ with the feminine nominal pielna ‘leg.’ On the other hand,
Example 3.9 can be interpreted as mira él, pielna ‘look at him, leg.’ In this
second possible interpretation, el ‘the’ is read as the object pronoun él
‘him.’ Notice that these two constituents, the definite determiner el and the
object pronoun él, are homophonous in Spanish, adding to the ambiguity
of this utterance. Nonetheless, factors such as the context of this utterance,
the pause between él and the nominal pierna and this child’s overall produc-
tion favor the second interpretation as the intended one. In this interpreta-
tion, the child might have omitted the possessive adjective su ‘his’ or simply
the definite article la ‘the.’ Moreover, there is evidence in Londa’s data
that seem to indicate that she has not acquired yet the (prenominal) pos-
sessive markers (e.g., mi ‘my,’ tu ‘your,’ su ‘his/her,’ etc.) and uses the (post-
nominal) emphatic possessive instead, as illustrated in the dialogue in
Example 3.10.
Example 3.10
Experimenter: ¿Con quién hablas? ‘With whom are you speaking?’
Londa: Papá mío ‘Dad (of) mine’
The exchange in Example 3.10 occurs as the experimenter and the child
pretend to talk on the phone. When the experimenter asks the child with
whom she is speaking, Londa responds with the phrase Papá mío ‘Dad (of)
mine’ instead of the adult-like mi papá ‘my dad.’ Hence Example 3.9 above
might indicate the difficulty the child is having in marking the possessive
structures, namely, producing mira su pierna ‘look at his leg’ or mira la pierna
de él ‘look at the leg of his.’ Finally, in Table 3.10 we can see that Londa
Gender Agreement 51
Demonstrative 4 – 2 1 7 (18%)
Other – – 1 – 1 (3%)
Total 15 4 16 3 38 (100%)
Example 3.11
Experimenter: ¿Qué es esto? ‘What is this (neuter)?’
(Experimenter asks the child as she is taking
out a toy dog from the bag.)
Child’ expected response: Un perro ‘a dog’ or Es un perro ‘[It] is a dog.’
In Example 3.11, the felicitous answer is one that includes the indefinite
determiner un perro ‘a/one dog.’ However, as will be discussed in Chapter 5,
the analysis of the data reveals a significant rate of determiner omissions in
the children under study.
Nonetheless, the contrast regarding the production of definite versus
indefinite determiner type might point to the fact that definite determiners
are acquired before indefinite ones. Previous acquisition research on this
issue is inconclusive, with some studies stating children are producing only
definite determiners at the earliest stages of acquisition (e.g., Mariscal 2008;
Schnell de Acedo 1994), and others claiming both types of determiners
are present in children’s production from the start of the two-word stage
(e.g., Aguirre 1995).
Overall, the previous discussion on (full) DPs reveals that the production
by the three children under study was almost completely target-like with
respect to gender agreement, even though they are at different stages of
acquisition, as illustrated in Table 3.12.
In particular, Table 3.12 indicates that of a total of 62 full DP tokens pro-
duced by the three children, only one token was non-target-like with respect
to gender agreement. This finding is very important because it supports
the hypothesis on the availability of an adult-like feature-checking mecha-
nism in early grammars. Interestingly, this conclusion finds support in the
L2 literature (e.g., Bruhn de Garavito & White 2000). Bruhn de Garavito
and White conducted a study on the L2 acquisition of Spanish DPs using as
subjects two groups of French speakers. Group 1 consisted of 30 high school
students who were finishing their first year of Spanish, and Group 2 con-
sisted of 12 students who were at the end of their second year of Spanish.
The researchers found that the two groups had a high accuracy rate on the
Total 36 1 25 – 62
Gender Agreement 53
studies (e.g., Aguirre 1995; Bottari et al. 1993/1994; Hernández Pina 1984;
Lleó 1997, 2001; López Ornat 1997; Pizzuto & Caselli 1992), but there is
disagreement regarding what they really represent in child grammar. On
the one side, researchers like Hernández Pina 1984 and López Ornat
(1997) interpret the production of these vocalic elements in phrases such
as e caballo ‘a horse’ as a reflection of a pre-grammatical stage, whereas
researchers such as Lleó (1997, 2001) interpret these vocalic elements as
grammatical but unspecified morphophonologically.
To determine the nature of these vocalic elements (MPHs) in children’s
grammar, an analysis was conducted on the agreement patterns established
between MPHs and the nominals they precede. The null hypothesis is that
MPHs function as pre-syntactic elements, that is, pre-determiners that hold
no features in the earlier stages of acquisition, as claimed by Bottari et al.
(1993/1994). The alternative hypothesis is that these vocalic elements
might contain some of the same features as their full determiner counter-
parts. The present analysis of MPHs is based on several assumptions. First,
vowel a can be a representation of the feminine determiners la ‘the’ or una
‘a’ and, as such, these pre-determiners might contain some of the same
features as the full determiner forms. Second, vowels e and o/os can func-
tion as phonetically shortened forms of the masculine determiners el ‘the’
(masculine/singular) and los ‘the’ (masculine/plural). Under the previous
assumptions, a target-like utterance is defined as one in which the assumed
number and gender features contained in the MPH match those of the
accompanying nominal.
The analysis of data on the production of MPH/DPs by the three chil-
dren under study yields a total of 102 utterances containing a Monosyllabic
Place Holder (MPH) with a noun, shown in Table 3.13.
Overall, Table 3.13 reveals the production of a majority of target-like
MPH/DPs by the three children, with the production of 87 target-like (84
percent) MPH/DPs out of a total of 102 utterances and 15 (16 percent)
non-target-like. Moreover, Table 3.13 indicates that the production of MPH/
DPs among the three children is not even, with the two younger children,
Masculine Feminine
Elián Target-like 3 4 – – 2 2 – – 11
Non-target-like – – – – – – – 1 1
Londa Target-like – 2 – – 2 1 – – 5
Non-target-like – – – – – – – – –
Alonso Target-like 3 3 1 1 2 1 – – 11
Non-target-like 2 1 – 2 – – – – 5
Total 8 10 1 3 6 4 – 1 33
a
–oth includes the unmarked nominal endings, for example –e, –consonant.
with both marked and unmarked nominals: a bola ‘a/the (fem/sg) ball
(fem/sg)’ for una/la bola; a uu (luz) ‘a/the (fem/sg) light (fem/sg)’ for
una/la luz. Elián’s MPH/DP production provides more information regard-
ing his gender agreement system. That is, although all of Elián’s full DP
production was feminine, his MPH/DP production in both the masculine
and feminine genders (all target-like) indicates that he has an operating
gender agreement system in place.
Londa’s MPH/DP production was very limited in comparison with Elián’s
(12) and Alonso’s (16), with only five tokens produced, all in the singular
and all target-like, as shown in Table 3.14 above. Her tokens were distrib-
uted as follows: two tokens involving the production of the masculine MPH
e with masculine nominals not overtly marked for gender (e.g., e guauguau
for el guauguau ‘the (masc/sg) dog (masc/sg),’ e lón for el león ‘the (masc/
sg) lion (masc/sg)’); and three tokens involving the production of the fem-
inine MPH a: two with (overtly) gender marked nominals (e.g., a pielna
for la pierna ‘the (fem/sg) leg (fem/sg),’ a ñama for la grama ‘the (fem/sg)
grass (fem/sg);’ and one with a non-overtly marked nominal: a lu for la luz
‘the(fem/sg) light (fem/sg).’ Londa’s limited production of MPHs seems
to indicate that she is already moving away from the production of MPHs
instead of full DPs; that is, Londa is at a later stage of development of the
determiner system than the other two children. Evidence for this conclu-
sion is demonstrated by the fact that she produced a significantly higher
proportion of full DPs in comparison to MPH/DPs (χ2 = 31, p < 0.01).
In contrast with the near-perfect MPH/DP production of Elián and
Londa, Alonso’s data present a significantly higher number of non-target-
like tokens, that is, 5 out of 16. First, Alonso’s target-like tokens are
examined. This child produced a total of 11 target-like MPH/DP tokens,
distributed as follows: five singular tokens with the masculine MPH e,
divided between canonically marked nominals (e.g., e caballo for el caballo
‘the (masc/sg) horse (masc/sg)’) and non-canonically marked ones (e.g.,
e bibí for el bibí or biberón ‘the (masc/sg) baby bottle (masc/sg)’); one singu-
lar token with the MPH u (e.g., u caballo ‘a/one horse’); and two tokens
in the plural. In addition, this child produced three feminine singular
tokens with the MPH a: two with canonically marked nominals (e.g., a bola
for una/la bola ‘a/the (fem/sg) ball (fem/sg)’); a vaca for una/la vaca ‘a/
the (fem/sg) cow (fem/sg)’) and one with a non-canonically marked nomi-
nal (e.g., a mano ‘the (fem/sg) hand (fem/sg).’
Now we turn the discussion to Alonso’s non-target-like production. The
non-target-like tokens produced by Alonso can be divided in two types:
the first type relates to an apparent mismatch of the gender feature; and
Gender Agreement 57
the second one involves a mismatch of the number feature. Example 3.12
illustrates Alonso’s gender mismatches; plural mismatches are presented in
Chapter 4.
Example 3.12
Child’s utterance Target
(a) *a caballo el caballo
‘the (fem/sg) horse (masc/sg)’ ‘the (masc/sg) horse (masc/sg)’
(b) *a pelo el pelo
‘the (fem/sg) hair(masc/sg)’ ‘the (masc/sg) hair (masc/sg)’
(c) *a guauguau el guauguau
‘the (fem/sg) dog (masc/sg)’ ‘the (masc/sg) dog (masc/sg)’
The first three utterances in Example 3.12 involve the production of MPH
a with masculine nominals in an apparent gender mismatch. Specifically, in
Example 3.12(a) Alonso utters the masculine nominal caballo with the MPH
a, which in the present analysis is assumed to carry the feminine gender
feature. There were five instances of this token in the data. One of them
involved the following alternation: a caballo, e caballo. This immediate alter-
nation between the MPHs a and e seems to be an attempt by the child at
self-correction. A second instance of this token occurred in the context
illustrated in Example 3.13.
Example 3.13
Child: ¿Vao a juar? for ¿Vamos a jugar? ‘Are we going to play?’
Experimenter: Sí, vamos a... ‘Yes, we are going to...’
Child: ¿A juar? for ¿A jugar ‘To play?’
Experimenter: Jugar ‘Play’
Child: A caballo ‘to horse [horses]’
In the dialogue above, the child asks the experimenter if the two of them
are going to play together as the experimenter takes out the experimental
props. In this particular context, the vowel a seems to be acting not as a
determiner but as the preposition in the phrase vamos a jugar a los caballos
‘We are going to play horses.’ The other three instances of a caballo seem to
involve the use of the vocalic element a as a determiner. In two cases, the
child utters this token in the form of a question as he searches for the
horses, for example, ¿Dostá a caballo? ‘Where is the horse?’ for ¿Dónde está el
caballo? ‘Where is the (masc/sg) horse (masc/sg)?’ in what seems to be an
58 First Language Acquisition in Spanish
appropriate context for a determiner. Finally, the context for the last
instance involving this token was as the child offers the horse to an adult,
for example, pati, a caballo ‘for you, a/the horse’ instead of para ti, un/el
caballo ‘for you, a/the (masc/sg) horse (masc/sg).’
To determine the preponderance of these five instances of the token a
caballo in the MPH/DP data, the number of instances in which Alonso
uttered the nominal caballo ‘horse’ with an MPH was calculated. A total of
30 instances were found, out of which only three were clear cases of non-
target-like uses as we discussed above. Moreover, the child produced the
appropriate masculine MPH e with the noun caballo in the majority of
instances (25 out of 30 instances).
The context of Example 3.12(b) a pelo ‘the hair’ instead of el pelo ‘the
(masc/sg) hair (masc/sg)’ is as Alonso looks at himself on the screen of the
camera while touching his hair. In this token, the child seems to be using
MPH a as a determiner. In Example 3.12(c), we see the production of the
MPH a with a non-canonically marked (i.e., ending in –u) masculine noun,
for example, guaugua–u ‘dog.’ Alonso produces four instances of this
particular token. As in the case of the token a caballo, the vowel a is not
used (unambiguously) as a determiner in all of the instances of this token.
The context for the first two instances is the child showing toys to other
people, for example, mía vaca, a guauguau ‘look cow, the/to dog’ instead
of mira una vaca, un guauguau ‘look (at) a (fem/sg) cow, a (masc/sg) dog.’
In this context, the a could be interpreted as the personal a and not the
MPH a. Spanish grammar marks the accusative noun object with an a if it is
a definite person or personified thing, for example, mira a Margo ‘look at
[personal a marker] Margo.’ Alonso might be marking the dog toy with the
personal a. The child utters the third instance of this token as he passes
one of the dogs to an adult, for example, ¡mía guauguau, pati a guauguau!
‘look dog, for you the dog.’ This utterance seems to involve a mismatch
between the MPH a and the masculine nominal. The last instance of this
token occurs while the child pretends to talk on the telephone with one
dog on each ear, hence the meaning intended is not clear. Alonso also pro-
duces two target-like instances of this token with the MPH e, for example,
Mira, e guauguau ‘look, the (masc/sg) dog (masc/sg).’
Several hypotheses can be invoked to account for Alonso’s non-target-
like MPH/DP utterances. The most obvious one would be to conclude that
MPH a is acting as a default determiner in this child’s grammar. This con-
clusion, however, is not borne out by Alonso’s overall production. First, as
discussed previously, this child’s (full) DP production is all target-like with
regard to gender and is almost all masculine, that is, 17 out of 19 full DPs
Gender Agreement 59
produced are masculine. Notice that the two feminine DPs that he pro-
duced were also target-like. Second, regarding his MPH/DP production,
Alonso utters three target-like feminine MPH/DPs, including a mano ‘the
hand.’ As explained in the case of Elián, this token is unique within the
Spanish nominal paradigm because it involves the only case of a feminine
nominal marked with the canonical masculine word marker –o (in com-
mon use), as pointed out by Harris (1991). Nonetheless, Alonso produces
this token with the target feminine MPH, an indication that this child has
some awareness of the feminine feature. Third, Alonso produces the mas-
culine singular form of demonstrative pronouns, third person clitics, and
adjectives to refer to both feminine and masculine nominals, as will be pre-
sented in the following sections, that is, evidence of a masculine default
value for gender. In light of this evidence, it would be an ad hoc solution to
conclude that MPH a is the default value for MPHs. Furthermore, Spanish
nominal agreement would be rendered unlearnable for this particular
child under the hypothesis that the masculine gender is the default value
for all the structures under study except in the case of MPHs.11
Another possible hypothesis is the one proposed by Bottari et al.
(1993/1994) for Italian. Interestingly, the authors found that the interpre-
tation of MPH a was ambiguous because it is difficult to discern between a
true MPH (void of morphological content) and a phonetic approximation
of the feminine determiner la ‘the.’ Notice that Bottari et al. distinguish in
their analysis between true MPHs and phonetic approximations. True
MPHs are defined as vocalic elements void of morphological value and, as
such, display a nearly free distribution, for example, MPH variants [a] and
[e] would occur with both feminine and masculine nominals. In contrast,
phonetic approximations would occur only in contexts in which full deter-
miners with the same vocalic elements would occur, for example, e in the
context of the masculine determiner el ‘the.’ In their data analysis, they
found that Italian children produced MPH a with feminine nominals
that ended in –a, but also in other contexts that could not be interpreted as
the determiner la. Bottari et al. concluded that the utterances involving the
MPH a with feminine nominal ending in –a were not cases of target-like
agreement, but rather they involved a strategy of phonological matching
between the MPH a and the final vocalic element in the nominals, –a. This
hypothesis predicts that if a is acting in Alonso’s grammar as a true MPH
void of features, the data should exhibit examples in which this MPH occurs
in contexts in which functional elements without the vocalic segment a
would occur. An analysis of Alonso’s overall production yielded evidence in
support of this prediction, as illustrated in Example 3.14.
60 First Language Acquisition in Spanish
Example 3.14
Child Target
(a) A Wili Wili
‘[personal a] Wili’ ‘Willy’
(b) A fío tiene frío
‘(a) cold’ ‘[he] is cold’
(c) A gusta te gusta
‘(A) (it) is pleasing’ ‘To you [it] is pleasing’
off a doll’s clothes because then the doll will be cold: ¡Ay qué frío, tiene frío!
‘Oh how cold, [he] is cold!’ First the child indicates that the doll was not
cold, saying: no fío ‘not cold.’ Then, as the experimenter introduces a new
toy to the task, the child shows the doll without clothes to the experimenter,
saying: A fío, a fío for perhaps tiene frío, tiene frío ‘(he) is cold.’ Even though
it is not certain that the child substituted a for the verb tiene, he uses the
MPH a to fill the structural position of another constituent. In the last utter-
ance listed in Example 3.14, a gusta, the child substitutes the objective pro-
noun for the MPH a. This token occurs as a response to the experimenter’s
question regarding a flower: ¿Te gusta? ‘Do you like (it)?’ The child answers
repeating the verb gusta, but substitutes the objective pronoun me for the
MPH a. Utterances listed in Example 3.14 lend support to the hypothesis
that in this child’s grammar, MPH a acts in some contexts as a true MPH,
void of feature content. Nonetheless, to conclude that in all instances this
MPH is void of featural content would be too strong a conclusion, given the
fact that the interpretation of this MPH is confounded by the homophony
present in the Spanish language, that is, it could stand for determiners una
‘a’ or la ‘the,’ for the preposition a as in a jugar ‘to play,’ or for the marker
of objective case a. Alonso’s overall production seems to indicate that he is
on the process of eliminating the “generic” uses of MPH a for MPHs that
have feature content, at least in the case of MPHs used as determiners.
The overall analysis of the MPH/DP production data indicates that these
vocalic elements are more than pre-grammatical fillers, as claimed by some
researchers (Bottari et al. 1993/1994; López Ornat 1997; Mariscal 2008),
but rather shortened approximations of adult-like determiners. Support
for this conclusion is found in the fact that out of the 33 tokens produced
by the three children, 27 were target-like in terms of agreement, that is,
these elements reveal the presence of an agreement relation at work.
Moreover, I found a strong correlation between the production of MPH
e for el ‘the (masc/sg)’ with masculine nouns and MPH a for una ‘a/one’ or
la ‘the’ with feminine nouns. Specifically, out of the 14 tokens of MPH e
produced by the three children, all of them were with masculine singular
nominals, whereas out of the 15 tokens of MPH a produced by the three
children, 11 were with feminine singular nominals.
Children’s vast production of target-like MPH/DPs could be taken as
evidence of the availability of an underlying checking mechanism in early
grammars, responsible for this strong target-like production. However, one
should be careful not to rush to premature conclusions before taking a
careful look at the nature of the agreement patterns these children are
establishing. In order to achieve this goal, I conducted a further data analysis
62 First Language Acquisition in Spanish
to address what Bottari et al. (1993/1994) call the linear agreement hypothesis,
that is, whether children are establishing agreement relations between
MPHs and nominals by matching the phonological nominal endings with
the phonological features of the MPH selected. Notice that this hypothesis,
if proven, supports discontinuous analyses of language acquisition such as
the usage based, in which children’s linguistic development is solely based
on the extraction of input regularities.
The linear agreement hypothesis predicts that if there is any agreement
present in the MPH/DP production data it would be purely phonological,
that is, children would produce MPH o with nouns ending in –o and MHP
a with nouns ending in –a. This hypothesis also predicts that children
will fail to produce target-like agreement with nominals not overtly marked
for gender, as there are no overt clues to guide the matching strategy.
In contrast, the continuity hypothesis would predict that children would be
successful regardless of the presence of overt morphological markings, as
they are guided by an underlying system.
In order to test these hypotheses, the 33 MPH/DP tokens were analyzed
in terms of the relation between the MPH used and the ending of the
nominal, as shown in Table 3.14 and repeated here as Table 3.15.
Crucially, the analysis in Table 3.15 shows that children produced MPHs
a and e with nominals canonically marked for gender (i.e., masculine nouns
ending in –o and feminine nouns ending in –a) as well as with nominals not
overly marked for gender (i.e., nouns ending in the vowels e, í, u or conso-
nants s, z, n, m). Moreover, the majority of nouns produced with MPH e
involved nominals not overtly marked for gender (i.e., 9 out of a total of 15),
whereas the distribution of MPH a was equally divided between nouns
Masculine Feminine
Elián Target-like 3 4 – – 2 2 – – 11
Non-target-like – – – – – – – 1 1
Londa Target-like – 2 – – 2 1 – – 5
Non-target-like – – – – – – – – –
Alonso Target 3 3 1 1 2 1 – – 11
Non-target-like 2 1 – 2 – – – – 5
Total 8 11 1 3 6 3 – 1 33
Gender Agreement 63
canonically marked for gender (7) and non-canonically marked for gender
(8). In particular, the data analysis points to the conclusion that overt mor-
phological clues do not play a significant role in the agreement relations
these children establish, with Londa and Elián uttering an all-target-
like production of MPH/DPs regardless of the nominal ending involved.
Interestingly, Alonso’s production of ungrammatical tokens reveals that
he was as likely to produce non-target-like MPH/DPs with nominals canoni-
cally marked for gender, for example, a caball–o ‘the/a (fem./sg.) horse
(masc./sg.),’ as with nominals unmarked for gender, for example, a guau-
guau ‘the/a fem./sg.) dog (masc./sg.).’ These results support Pizzutto and
Caselli’s (1992), Lleó´s (1997, 2001) and Valian’s (2009b) claim that these
bare vowels produced with nominals carry some of the adult-like feature
information, in this case the gender feature.
Moreover, these findings lend support to the hypothesis that children
acquiring Spanish exhibit target-like agreement relations that go beyond a
phonological matching strategy, for example, a cas–a ‘the house,’ but that
they are building their agreement systems on the basis of feature checking
mechanism, for example, e guaugua–u ‘the dog,’ a man–o ‘the hand.’
Namely, these children are checking the gender feature of the determiners
with that of the nominals they precede. This conclusion brings support to
the Weak Continuity Hypothesis adopted in the present research (Borer &
Rohrbacher 1997; Crain & Thornton 1998; Crain & Wexler 1999; Phillips
1996; Pinker 1984) in that children start the acquisition process with
a grammar governed by the principles of UG, that is, the availability of a
checking mechanism and the corresponding functional categories, such as
the DP. Support for this conclusion is found in the analysis of the agree-
ment patterns these three children establish in (full) DPs, that is, the three
children produced a majority of target-like full DPs with regard to gender
agreement regardless of the nominal endings.
In the next section, I discuss findings on gender agreement in attributive
adjectives, that is, adjectives within the DP, for example, la casa roja ‘the
house red’ (the red house). In addition, the discussion will also include
children’s production of adjectives in predicative structures. Even though
these adjectives are not part of the DP, the agreement patterns children
establish in structures outside the DP serve to create a better picture of
gender agreement as a whole in Spanish early grammar.
Mismatch Default
(Full) DPs 62 1 – 1
MPH/DPs 33 4 – 4
Attributive adjectives 18 – – –
Predicative adjectives 65 – 10 10
Demonstratives 118 6 13 19
Third person clitics 37 1 8 9
Total 333 12 31 43
Example 3.15
Child utterance Target utterance
(a) *Mira gande (Elián) Mira una/la grande
‘Look (at) big (one)’ ‘Look (at) a/the (fem/sg)
big (one)’
(b) Si cayó, ecito (Elián) Se cayó, pobrecito
‘(he) fell down, poor (masc/sg)’ ‘(he) fell down, poor (masc/sg)’
(c) ¡Mira carito bonito! (Alonso) Mira un carrito bonito
‘Look car (dim/masc/sg) pretty ‘Look a (masc/sg) car (dim/
(masc./sg.)!’ masc/sg) pretty (masc/sg)!’
Example 3.16
Child Target Referent
(a) No, el quito No, el chiquito [león]
‘No, the (masc/sg) ‘No, the (masc/sg) small
small (masc/sg) (masc/sg) (one)’
(one)’
(b) Yo engo gande Yo tengo el grande [teléfono]
‘I have big (unm/sg) ‘I have the (masc/sg) big
(one)’ (masc/sg) (one)’
(c) Tistre ya anó El triste ya ganó [tren]
‘Sad (unm/sg) (one) ‘The (masc/sg) sad (masc/
already won’ sg)(one) already won’
(d) Feli[h] El feliz [tren]
‘Happy (unm/sg) ‘The (masc/sg) happy
(one)’ (masc/sg) (one)’
(e) Azul El azul [pez]
‘Blue (unm/sg) (one)’ ‘The (masc/sg) blue (masc/
sg) (one)’
Example 3.16(b) illustrates one of these cases. In this example, Londa pro-
duces the adjective grande ‘big (umn/sg)’ to refer to the masculine nominal
teléfono ‘telephone.’ The context of this example is the experimenter’s state-
ment: Mira, yo tengo el chiquito ‘Look, I have the (masc/sg) small (masc/sg)
[one],’ pointing at a small telephone. The child answered by using the
same structure but dropping the determiner, that is, Yo engo gande ‘I have
big (unm/sg) [one].’ These results confirm that the child is neither imitat-
ing the input heard nor learning the nominal and the determiner as a unit.
The analysis of the attributive adjective production reveals several aspects
of these structures. First, they are not very common in children’s produc-
tion, as has been reported in the acquisition literature (e.g., Mariscal 2008;
Snyder, Senghas, & Inman 2001); even when they were elicited by a parti-
cular task, that is, “The race.” As discussed earlier, the two smaller children
barely produced examples of this particular structure, and the older child’s
production was higher, although it does not seem very rich. Second, the
issue of determiner omission seems to be pervasive in this structure. Specifi-
cally, the two younger children produced a total of three utterances and all
three involved the omission of the determiner. Regarding Londa, 14 out of
15 instances produced were examples of determiner omissions. These
results are consistent with the results obtained for [Det-N] structures in
which the three children omitted determiners in 64 percent of the obliga-
tory contexts (see Chapter 5).
These general findings on attributive adjectives were compared with the
production of the two older children, Diana (3;5,27) and Pepe (4;3,10).
Recall the data of these two children are introduced in the discussion when
interesting. In this case, they are introduced to assess the attributive adjec-
tive production of children above the acquisition age of 3;0. Table 3.18
illustrates their production in terms of agreement.
Table 3.18 demonstrates that these two children have already mastered
the adjectival agreement system, displaying perfect target-like production.
Examples of their target-like production are illustrated in Example 3.17.
Example 3.17
(a) El chiquito ‘The (masc/sg) small (masc/sg) (one)’ (Diana)
(b) La motora roja ‘The (fem/sg) motorcycle (fem/sg)
red (fem/pl)’ (Diana)
(c) Un león chiquito ‘A (masc/sg) lion (unm/sg)
small (masc/sg)’ (Pepe)
(d) La[h] azule[h] ‘The (fem/pl) blue (unm/pl) (ones)’ (Pepe)
Example 3.17 shows that the production of these two older children con-
trasts with that of the three smaller children on several aspects. First, regard-
ing agreement, the examples above show target-like attributive adjective
structures both in the masculine, as in Examples 3.17(a) and 3.17(c), as well
in the feminine, as in Examples 3.17(b) and 3.17(d). As discussed earlier,
Elián, Alonso, and Londa produced attributive structures (overtly marked)
only in the masculine singular. In addition, Diana and Pepe produced struc-
tures involving N-drop, but with the obligatory determiner, for example, el
chiquito ‘the small [one],’ las azules ‘the blue [ones].’ Moreover, in Example
3.17d Pepe produces a target-like utterance in the plural las azules ‘the blue
[ones]’ and with the required determiner las. Finally, Diana’s and Pepe’s
attributive adjective production (17) and (20) respectively, reveals that
these two children produced similar proportions to Londa (15).
Regarding the issue of determiner omissions, Diana omitted the obliga-
tory determiner in two instances:
Example 3.18
Child Target
of the experimenter’s phrase: yo tengo los trenes tristes ‘I have the (masc/pl)
trains (unm/pl) sad (masc/pl).’ The child repeats the phrase by dropping
both the nominal trenes ‘trains’ (allowed by the grammar) and the deter-
miner los ‘the (masc/pl),’ which is not allowed in the Spanish grammar.
I should point out that this example presents a certain level of ambiguity.
One could argue for a predicative reading of this particular structure, for
example, Sí son/están tristes ‘Yes, [they] are sad (unm/pl),’ an interpretation
that will make this example adult-like without the determiner.
As we have seen, a comparison between the three children under study
with the two older children (over the age of 3;0) yielded several contrasts.
First, the production of attributive adjectives is, in general, limited, with the
two younger children, Elián and Alonso, barely producing any instances
and the other three children, Londa, Diana, and Pepe, producing a maxi-
mum of 20 examples each. Second, regarding gender agreement, the data
of Diana and Pepe show that these two children have already mastered the
nominal agreement system. This finding is consistent with previous studies
that have claimed that children acquire agreement around the age of 3;0
(Hernández Pina 1984). Third, the issue of determiner omissions is perva-
sive in Londa’s production whereas in the production of the older children
it is almost non-existent, although there are still some cases in Diana’s
production.
The next section focuses on the analysis of agreement in predicative
adjectival structures, that is, adjectives that occur with the copula ‘to be,’ for
example, el caballo es verde ‘the horse is green.’
Example 3.19
(a) María es feliz ‘Mary is happy’
(b) María está feliz ‘Mary is happy’
In general terms, these two forms of the copula serve to mark the contrast
between a characteristic that is perceived as permanent (ser) and one that
is perceived as temporary or changeable (estar), that is, ser refers individual-
level predicates whereas estar refers to stage-level predicates (Fernández
Gender Agreement 71
Example 3.20
Wili está [SC pro (masc/sg) sucio (masc/sg)]
Example 3.21
Child Target
Ii cayente está caliente
‘(It) is (permanent) hot (un./sg.)’ ‘(It) is (changeable) hot (unm/sg)’
72 First Language Acquisition in Spanish
The context of Example 3.21 is that the child touched the video camera
and ran back to the experimenter complaining that it was hot. In this utter-
ance the long vowel ii could be interpreted as the copula es as in es caliente
‘it is hot.’ The target-like utterance should have been está caliente because
the child was referring to a changeable state of the camera, that is, the tem-
perature of the camera. Elián produced two additional instances of this
token, but with the adjective in isolation, for example, cayente ‘hot.’ One of
them occurred the first time the child touched the camera, that is, it had
exactly the same context as the previous example. The second instance
occurred in the context of a game: the child told the puppet to sit on a
(turned off) light. When the puppet sat on it, the child turned the light on
and the puppet screamed. The child produced cayente after the puppet
refused to sit on the light saying: ¡No, me quemo! ‘No I will burn (myself)!’
These two adjectives produced in isolation seem to be predicative within
the contexts described. However, because this adjective is not marked
overtly for gender (i.e., it ends on the vowel –e, which could refer to either
gender), it provides no information in terms of the gender agreement.
Alonso produced a total of ten predicative adjective utterances, as shown
in Table 3.19 above, of which five were target-like and five non-target-like in
terms of agreement. His target-like adjectives included three different
adjectives: sucio, duro, and caliente, as illustrated in Example 3.22.
Example 3.22
¿Está limpio? ‘Is [it] clean (masc/sg)?’ with ee sucio for es sucio ‘[it] is (per-
manent) dirty (masc/sg)’ instead of está sucio.12 In this context, the target-
like form of the copula is estar because the child is referring to a state of
dirtiness that can change. Examples 3.22(b) and 3.22(c) were produced
as responses to the experimenter’s follow-up questions: ¿Y Wili? ‘And
Wili (is he dirty)?’ and ¿Y Alonso? ‘And Alonso (is he dirty)?’ respectively. In
these two cases, the child responded in a target-like fashion regarding
agreement and the use of the appropriate form of the copula, the verb estar.
In the three examples discussed above, the masculine singular features of
the adjective sucio matched the masculine singular features of the modified
nominals: muñeco ‘male doll,’ Wili ‘male whale,’ and Alonso ‘the child him-
self.’ In Example 3.22(d) Alonso uttered the adjective caliente ‘hot (unm/
sg)’ to refer to the video camera. As pointed out in the discussion of Elián’s
production, this adjective is not marked overtly for gender, so it provides no
agreement information with regard to gender. Finally, the child produced
the adjective duro ‘hard (masc/sg)’ in 3.22(e) to refer to the fact that it was
hard to unstick a sticker. Notice that all these examples referred to mascu-
line nominals. This is consistent with Alonso’s overall production, in which
the vast majority of his data is in the masculine singular form.
Alonso produced five non-target-like examples of predicative adjectives,
illustrated in Example 3.23.
Example 3.23
namely the child used this adjective marked as masculine singular to modify
feminine nominals. In Example 3.23(a) the experimenter asked Alonso for
the ball: Préstamelo ‘Lend it (neuter/sg) to me,’ and the child refused by
saying tá sucio ‘(it) is (changeable) dirty (masc/sg).’ In this particular exam-
ple, the experimenter used the neuter form of the clitic pronoun lo ‘it’
(instead of the feminine form la ‘it (fem/sg)’) not to mark the nominal
bola ‘ball (fem/sg)’ as feminine. In the following Example 3.23(b), the
experimenter marked the feminine gender of the nominal bola ‘ball’ in the
demonstrative pronoun: ¿Y esa? ‘And that (fem/sg) [one]?’ pointing at
another ball. Nonetheless, the child gave exactly the same non-target-like
gender response, tá sucio ‘[it] is (changeable) dirty (masc/sg).’ Similarly, in
Example 3.23(c), the experimenter asked the child if the sock was clean,
marking the feminine gender of the nominal media ‘sock (fem/sg)’ in both
the adjective and the clitic: ¿Está limpia? Mírala ‘Is [it] clean (fem/sg)? Look
[at] it (fem/sg)’ as the experimenter showed him a dirty sock. Alonso
answered no, ucio ‘No, dirty (masc/sg).’ Notice that the nominals involved
in these three non-target-like utterances were all canonically marked for the
feminine gender, that is, they all ended in the feminine word marker –a:
bol–a, medi–a. This finding brings support to the conclusion that at this stage
of acquisition the morphophonological clues present in the input are not
vital in the process of establishing agreement, as we saw earlier in the case
of (full) DPs and MPH/DPs. Moreover, this result contradicts Pérez-Pereira’s
(1991) claim that children pay attention to intralinguistic clues such as
morphological endings and syntactic agreement morphemes, in gender
assignment. The difference in the results might be due to developmental
differences between the participants of the two studies. In particular, the
children who participated in the Pérez-Pereira’s study ranged in age from
4 to 11 years old, whereas the children in the present study are under the
age of 3. It is possible that at this earlier stage of acquisition they are not
making use yet of the clues present in the input.
Example 3.23(d) illustrates a non-target-like utterance involving a differ-
ent adjective, that is, the masculine gender feature of the adjective despierto
‘awake’ does not match the feminine feature of the nominal vaca ‘cow
(fem/sg).’ In the context of this utterance, the experimenter instructed
the child to be quiet because the cow was sleeping: Shh, no lo despiertes ‘Ssh,
do not wake him/it up.’ In this utterance, the experimenter used the
masculine (also the neuter form) clitic lo ‘him/it’ not to provide the target
gender cues to the child. The child responded by picking up the cow and
saying: Dame, dame tá despierto ‘Give me, give me [it] is awake (masc/sg).’
The discussion of Alonso’s predicative adjective production brings sup-
port for the hypothesis of a masculine default value for the gender feature
Gender Agreement 75
Example 3.24
(a) chiquito/chiquita ‘small (masc/fem)’
Este ee ito for Este es chiquito ‘This (masc/sg) (one) is small
(masc/sg)’
No, ita for No, chiquita ‘No, small (fem/sg) (one)’
(b) dormido/dormida ‘asleep (masc/fem)’
Tá domío for Está dormido ‘[He] is asleep (masc/sg)’
Mira tá domida for está dormida ‘Look [she] is asleep (fem/sg)’
(c) sucio/sucia ‘dirty (masc/fem)’
Tá sucio for Está sucio ‘(He) is dirty (masc/sg)’
Tá sucia for Está sucia ‘(She) is dirty (fem/sg)’
Example 3.25
Child Target Referent
(a) *Etá abielto Está abierta boca ‘mouth (fem/sg)’
‘(It) is opened (masc/sg)’ ‘(It) is opened (fem/sg)’
(b) *Tá jerrado Está cerrada boca ‘mouth (fem/sg)’
‘[It] is closed (masc/sg)’ ‘[It] is closed (fem/sg)’
76 First Language Acquisition in Spanish
a result she marks the adjectives with the masculine default value. This find-
ing is supported by Londa’s utterances with demonstrative pronouns in
which she assigned the wrong gender to the horses, as discussed in the next
section.
Example 3.26
Child Target
(a) ése es mamá ésa es la mamá
‘that (masc/sg) (one) is mom (fem/sg)’ ‘that (fem/sg)(one) is the
mom’
(b) ésa es papá ése es el papá
‘that (fem/sg) (one) is dad (masc/sg)’ ‘that (masc/sg) is the dad’
Example 3.27
Experimenter: Pero ¿qué motora ganó?
‘But what motorcycle (fem/sg) won?’
78 First Language Acquisition in Spanish
Child: Esta.
‘This (fem/sg) (one).’
Experimenter: ¿Y de qué color es esa?
‘And what color is that (fem/sg) (one)?’
Child: *Jojo for rojo
‘Red (masc/sg)’
In the dialogue in Example 3.27 the child knew the nominal motora
‘motorcycle (fem/sg)’ was feminine because she has used the feminine
form of the demonstrative pronoun ésta ‘this (one)’ to refer to it. Mariscal
(2008) reports that the children in her study exhibited similar gender
agreement issues with color adjectives. I would like to propose that gender
mismatches with the adjective rojo ‘red’ may be due to the child’s lack of
understanding of the structure used in the question or a structure misinter-
pretation (Socarrás 2003). That is, the child does not know the difference
between two related structures, or has a parsing preference for one of the
possible structures (Crain & Thornton 1998) as in Example 3.28.
Example 3.28
(a) La motora es [SC pro (fem/sg) roja (fem/sg)]
The motorcycle is [SC pro (fem/sg) red (fem/sg)]
(b) La motora es [PP de color (masc/sg) rojo (masc/sg) ]
The motorcycle is [PP of color (masc/sg) red (masc/sg)]
In Example 3.28(a) the adjective roja ‘red’ has to agree with the feminine
noun motora, for example, roja ‘red (fem/sg),’ and in 3.28(b) the adjective
agrees with the masculine nominal color ‘color,’ for example, rojo ‘red
(masc/sg).’ It seems that Londa is establishing agreement with the nominal
color instead of with the nominal motora.
In order to explore the validity of the structure misinterpretation hypo-
thesis, Londa’s results were compared with the predicative adjective pro-
duction of the two older children, Diana and Pepe.
Table 3.20 shows that Diana produced a total of 27 tokens (once repeti-
tions were excluded) of predicative adjectives, out of which 24 were target-
like and 3 non-target-like in terms of agreement. In the case of Pepe, Table
3.20 shows that he produced a total of 50 tokens (excluding repetitions) of
predicative adjectives, out of which 47 were target-like and 3 were non-tar-
get-like. Overall, these two children exhibit an almost perfect production of
predicative adjectives in terms of agreement. Hence we can conclude that
they have acquired the agreement system in this particular structure.
Gender Agreement 79
Example 3.29
(a) Child *Estee ... Ella[h] son ‘Um..They (fem/pl) are yellow
amarillo[h] (masc/pl)’
Target Ellas son amarillas ‘They (fem/pl) are yellow
(fem/pl)’
Referent: mariposas ‘butterflies’
(b) Child *De amarilla ‘Of [the color] yellow
(fem/sg)’
Target (Son) amarillas ‘[They] (are) yellow
(fem/pl)’
Referent: bolas ‘balls’
(c) Child *Son azul ‘(They) are blue (unm/sg)’
Target (Son) azules ‘(They) (are) blue (unm/pl)’
Referent: bolas ‘balls’
Example 3.30
(a) Child: *Esto[h] son flore[h] ‘These (masc/pl)are flowers
(unm/fem/pl)’
Target: Estas son flores ‘These (fem/pl) are flowers (unm/
fem/pl)’
(b) Child: *Esto[h] son uita[h] ‘These (masc/pl) are grapes (dim/
fem/pl)’
Target: Estas son uvitas ‘These (fem/pl) are grapes (dim/
fem/pl)’
Referent: [uvas ‘grapes’]
Example 3.31
*Ella[h] son amarillo[h] ‘They (fem/pl) are yellow (masc/pl).’
First, one could argue that this example is the result of the application
of the masculine gender as the default value. However, this hypothesis
cannot account for the fact that this default value was only triggered
with color adjectives, that is, these were the only non-target-like examples
produced by Diana with respect to gender, as discussed earlier. Also notice
that in this example, the child marked the feminine gender in a target-like
fashion on the pronominal ellas ‘they (fem/pl).’ Another possibility is to
argue that the child misinterpreted the structure used in the question to
elicit the color adjective, that is, she was interpreting the structure of the
question as the one in Example 3.32(a) instead of the one in Example
3.32(b).
Gender Agreement 81
Example 3.32
(a) Las mariposas son [PP de color amarillo]
Butterflies are [PP of the color [mas/sg] yellow [mas/sg]]
(b) Las mariposas son [SC pro amarillas]
Butterflies are [PP pro [fem/pl] yellow [fem/pl]]
Example 3.33
Experimenter: ¿Bolas qué? ¿de qué color?
‘Balls (fem/pl) what? ¿what color (balls)?’
Child: Este... bola[h]
‘Um... balls (fem/pl)’
Experimenter: ¿Bolas de qué color?
‘Balls (fem/pl) (of) what color?’
Child: *De amarilla
‘Of yellow (fem/sg)’
number is possible, cf. De color amarillo ‘Of color yellow (masc/sg)’ versus
*De color amarillos ‘Of color yellow (masc/pl).’ The child produced this type
of response with other color adjectives as shown in Example 3.34.
Example 3.34
Experimenter: ¿De qué color es? ‘(Of) what color is (it)?’
Child: *De azul ‘Of blue (unm/sg)
Experimenter: ¿De azul? ‘Of blue?’
Child: De azulito ‘Of blue (diminutive/masc/sg)’
In the Example 3.34 above, Diana uttered again the preposition de ‘of’ with
a color adjective. When the experimenter repeated her response, she con-
firmed her answer by repeating the adjective azul ‘blue’ this time in the
diminutive form azulito ‘blue (dim/masc/sg).’
In her last non-target-like token shown in Example 3.29(c), *son azul
‘[they] are blue (unm/sg),’ Diana uttered the copula in the plural but
the adjective azul in the singular to refer to the plural nominal bolas ‘balls
(fem/pl).’ This token was produced as an answer to the experimenter’s
question: Pero, ¿de qué color? ‘But, (of) what color?’ If our structural misinter-
pretation hypothesis is correct, this answer can be paraphrased as las bolas
son de color azul ‘The balls are of color blue (unm/sg).’ As we have seen,
Diana’s production confirms the prediction made by our hypothesis, that
children would have problems with color adjectives due to the possible
ambiguity in the interpretation of this particular structure, even those that
have mastered agreement.
Pepe’s non-target-like production is discussed next. The analysis of his
production reveals three non-target-like examples with respect to agree-
ment, but only one pertains to the gender feature.
Example 3.35
(a) Child: *Y este veide, veide, y amarillo.
‘And this (masc/sg) [one)] green (unm/sg), green
(unm/sg) and yellow (masc/sg)’
Target: Y esta (es) verde, verde y amarilla
‘And this (fem/sg) [one] green (um/sg), green (unm/sg)
and yellow (fem/sg)’
(b) Child: *Igual a esto[h]
‘Similar (unm/sg) to these (masc/pl)’
Target: Iguales a estos
‘Similar (unm/pl) to these (masc/pl)’
Gender Agreement 83
(c) *Desenmotao
‘[They] (are) disassembled (masc/sg)’
Target: [Están] desmontados
‘[They] (are) disassembled (masc/pl)’
default value for gender. In particular, Alonso and Londa produced mascu-
line singular predicative adjectives to refer to feminine singular nominals,
for example, *Tá sucio (Alonso) ‘[It] is dirty (masc/sg)’ instead of Está sucia
‘[It] is dirty (fem/sg)’ to refer to the feminine nominal media ‘sock.’ Third,
the data of the two older children revealed that the vast majority of their
utterances were target-like, for example, Diana (3 non-target-like out of 27)
and Pepe (3 non-target-like out of 50). This was not true in the data of the
younger children; specifically Alonso produced 5 non-target-like utterances
out of a total of 10 (50 percent) and Londa produced 11 non-target-like
utterances out of 54 tokens (20 percent). In Alonso’s case, the difference
between target and non-target-like production is not significant. On the
other side, Londa produced a significantly higher number of target-like
predicative adjectives (χ2 = 19, p < 0.01). In addition, the analysis showed
that non-target-like utterances in this structure pertained to both gender
and number mismatches (see Chapter 4). In the former, it involved the
application of the masculine default value to both feminine and masculine
nominals and the apparent difficulty in the interpretation of the structure
used to elicit color adjectives. Notice that this particular result pertaining to
a non-target-like gender production was not found in the attributive data,
that is, all the non-target-like production involving adjectives in attributive
structures involved a number mismatch. This contrast in targetness points
to a difference in the acquisition of these two structures, with attributive
adjective acquisition preceding predicative one. Moreover, notice that the
production data on attributive adjectives patterns with that of full DPs, that
is, in both structures non-target-like production was circumscribed to num-
ber mismatches. In the case of predicative adjectives, the acquisition data of
this structure parallels the acquisition of demonstratives, in which both
gender and number mismatches were found.
In the next section gender agreement in pronominals is examined; in
particular in demonstrative and clitic pronouns produced by children
under study.
default values in Spanish child language. As in the case of the other con-
stituents included in the present analysis, these pronominals also mark
overtly the gender and number features, as illustrated in Example 3.36.
Example 3.36
(a) esta tiene mucha[h] manchita[h] (Pepe)
‘this (fem/sg) (one) has many spots’
(b) Y este mira (Alonso)
‘And this (masc/sg) (one) look’
(c) A comer eso (Elián)
‘To eat that (neuter)’
Example 3.37
(a) ¿Tú lo viste? (Diana) ‘Did you see him (masc/sg)’
(b) Sácala (Alonso) ‘Take it (fem/sg) off’
(c) Coelo[h] (Alonso) ‘Take them (masc/pl)’
86 First Language Acquisition in Spanish
Third person clitics (direct object pronouns) mark both the number and
gender features overtly, as illustrated in Example 3.37. In Example 3.37(a)
the masculine singular clitic lo ‘it’ agrees with the masculine singular fea-
tures of the nominal it refers to, that is, nene ‘boy (masc/sg),’ whereas in
3.37(b) the clitic pronoun la ‘it’ marks overtly the feminine singular fea-
tures of the noun it refers to, that is, ropa ‘clothing.’ On the other hand,
Example 3.37(c) illustrates how the masculine plural feature of the clitic
pronoun los ‘them’ agrees with the corresponding features of the nominal
caballos ‘horses (masculine/plural).’
Elián Target-like 4 – – – 3 7
Non-target-like 1 – – – 1 2
Alonso Target-like 4 1 – – 13 18
Non-target-like 3 – – – 1 4
Londa Target-like 29 8 4 1 33 75
Non-target-like 7 2 2 – 2 13
Total (%) 48 (40) 11 (9) 6 (5) 1 (1) 53 (45) 119 (100)
contexts, given the fact that there are no nominals marked as neuter in this
language, for example, esto es un perro ‘this (neut) is a (masc/sg) dog (masc/
sg);’ esto es una silla ‘this (neut) is a (fem/sg) chair (fem/sg).’ The majority of
the masculine and neuter pronouns produced were target-like as can be seen
in Table 3.22. Specifically, 13 (4.5 percent) out of a total of 59 masculine
demonstrative tokens were non-target-like and 4 (13.3 percent) out of 53
neuter demonstrative tokens were non-target-like with respect to agreement.
We will turn our attention to the 19 non-target-like demonstrative
pronoun tokens produced by the three children. Notice that all the non-
target-like demonstrative tokens involved the gender feature, the focus
of the present chapter. In particular, the most common type of non-adult
like production involved the use of a masculine demonstrative to refer to a
feminine nominal, accounting for 13 out of a total of 19 non-target-like
demonstratives produced. First in Elián’s production, two non-target-like
demonstratives were found.
Example 3.38
Child Target
(a) *Eso Ese
‘That (neut)’ ‘This (masc/sg) (one)’
(b) *Ese e una Esa es una
‘That (masc/sg) (one) is one (fem/sg)’ ‘That (fem/sg) (one) is
one (fem/sg)’
The context of Example 3.38 was the experimenter asking the child: ¿Cuál
quieres? ‘Which one [of the cats] do you want?’ The child responded with
88 First Language Acquisition in Spanish
the demonstrative in the neuter gender *eso ‘that’ as he took one of the cats.
The target-like answer should have been a demonstrative in the masculine
gender, that is, este ‘this [one]’ or ese ‘that [one].’ A possible explanation for
this non-target-like example could be that the child does not understand
yet the difference between qué ‘what’ versus cuál ‘which [one].’ Notice that
if the question had been ¿qué quieres? ‘what do you want?’ the child’s answer
would have been target-like, that is, eso ‘that (neuter-stuff).’ Another possi-
bility is that Elián was using eso instead of the masculine determiner ese in an
attempt to regularize the paradigm, that is, the child used the canonical
masculine gender mark –o.13 However, no conclusions can be reached given
this child’s limited production. The second non-target-like utterance pro-
duced by Elián involves a mismatch of the gender feature, as shown in
Example 3.38(b). Elián uttered the masculine singular demonstrative *este
‘this [one]’ to refer to mariposa ‘butterfly’ a feminine singular nominal.
Notice that in this same example the child produced the feminine pronoun
una ‘one’ in agreement with the nominal mariposa. This example might
point towards the presence of a default gender value for the demonstrative
pronouns, that is, the masculine gender.
In the case of Alonso, his four non-target-like utterances were of two kinds
as illustrated in Example 3.39.
Example 3.39
Child Target
(a) *So[h] Wili Ese es Wili
‘This (neut) is Willy’ ‘This (masc/sg) [one] is Willy’
(b) *Toa, ese[h] tuyo Toma, esa es tuya
‘Take, that (masc/sg) ‘Take, that (fem/sg) [one] is yours
[one] is yours (masc/sg)’ (fem/sg)’
(c) *Te[h] tuyo Esta es tuya
‘This (masc/sg) [one] is ‘This (fem/sg)[one] is yours (fem/sg)’
yours (masc/sg)’
(d) *Ese es tuyo Esa es tuya
‘That (masc/sg)[one] is ‘That (fem/sg)[one] is yours (fem/sg)’
yours (masc/sg)’
Example 3.40
Child Target
(a) *Este ona Esta funciona [luz ‘light’]
‘This (masc/sg)[one] works’ ‘This (fem/sg)[one] works’
(b) *Este e sita Esta es [una] casita [casa ‘house’]
‘This (masc/sg)[one] is house’ ‘This is (a) house (dim/fem/sg)’
(c) *Estos son uitas Estas son uvitas [uvas ‘grapes’]
‘These (masc/pl) are grapes ‘These (fem/pl) are grapes (dim/
(dim/fem/pl)’ fem/pl)’
(d) *Esa[h] papá Ese es papá [caballo ‘horse’]
‘This (fem/sg) [one] is dad’ ‘This (masc/sg)[one] is the dad’
(e) *Esto Este [pez ‘fish’]
‘This (neuter)’ ‘This (masc/sg) [one]’
(f) *Esto lo senta Este lo siento
‘This (neuter) (he/she) sits him’ ‘This (masc) [one](I) seat him’
90 First Language Acquisition in Spanish
small house.’ This contrast might point to the fact that the child’s problem
in establishing target-like agreement relations resides on the nature of the
agreement with the pronominal demonstratives.
The discussion of the production data on demonstrative pronouns con-
sistently supports the masculine value as default for the gender feature.
Moreover, these findings are in contrast with those of the full DPs and
MPH/DPs; although non-target-like demonstrative pronouns involved a
mismatch of the gender feature, non-target-like DPs and MPHs involved a
mismatch of the number feature. Regarding reference, no mismatches
were found between number feature encoded in the demonstratives and
those of the referents.
Target-like (%) 2 1 15 5 5 2 30
Non-target-like (%) – 1 9 1 2 – 13
Total (100%) 2 2 24 6 7 2 43
92 First Language Acquisition in Spanish
type of non-target-like utterance (10 out of 13) was the use of the masculine
clitic pronoun lo ‘it’ to refer to a feminine noun, as illustrated in the
Example 3.41.
Example 3.41
Child Target
(a) Londa: A abilo [Vamos] a abrirla [casa ‘house]
‘To open it (masc/sg)’] ‘[We are going] to open it (fem/sg)’
(b) Londa: Yo lo encendo Yo la enciendo [luz ‘light’]
‘I turn it (masc/sg) on’ ‘I turn it (fem/sg) on’
(c) Alonso: Sácalo Sácala [ropa ‘cloths’]
‘Take it (masc/sg) [off]’ ‘Take it (fem/sg) [off]’
(d) Alonso: Se lo come Se la come [serpiente]
‘[It] eats it (masc/sg)[up]’ ‘(It) eats it (fem/sg) [up]’
In the examples above, Alonso and Londa produced the masculine clitic lo
‘it’ to refer to feminine nominals, both canonically marked for gender as in
Examples 3.41(a) and 3.41(c), as well as non-canonically marked nominals,
as in Examples 3.41(b) and 3.41(d). This points to the fact that these chil-
dren are not making use of the available linguistic clues in the input. Notice
that these children produced target-like utterances with these nominals in
other structures, for example, una sita ‘a/one (fem/sg) small house (fem.
/sg.),’ (Londa), or in the same type of structure, for example, sácala ‘take
it (fem/sg)(out),’ (Alonso).
In addition to the non-target-like utterances discussed above, one other
instance of non-target-like production was found in the data.
Example 3.42
Alonso: Vo gadarla ‘(I) am going to put it (fem/sg)[away]’
Target: Voy a guardarlos ‘(I) am going to put them (masc/pl) [away]’
Referent: *[bloques ‘blocks’]
Number Agreement
The number feature receives two values in Spanish, singular or plural, with
plurality being marked formally by the allomorphs (–s, –es) and singularity
carrying the zero mark or left unspecified (Alcina & Blecua 1983). This
points out a contrast between number and gender morphology in terms of
acquisition, namely, children acquiring Spanish need to mark nominals,
and their agreeing constituents, with gender morphology, while they have
the “option” (although not grammatical) of leaving the number feature
unspecified, that is, producing nominals in the singular. Therefore, the ref-
erent of the utterance becomes of extremely important in the assessment
of the acquisition of number, that is, while singular utterances such as una
casa ‘a house’ might be grammatical in terms of morphological agreement,
they might be ungrammatical if used to refer to ‘many houses.’ This aspect
would be addressed in this chapter.
In general the selection between the plural marker allomorphs –s and –es
is mostly based on phonological stress patterns of the language. Most nomi-
nals in Spanish follow these basic rules of pluralization: (1) words that end
in an unstressed vowel take –s as the mark of plural, for example, libro/libro–s
‘book/books;’ (2) words ending in consonants (except –s) or a stressed
vowel (except –e) form their plural by adding –es, for example, jamón/
jamon–es ‘ham/hams,’ tabú/tabú–es ‘tabu/tabus;’ (3) nouns ending in a
stressed –é form their plural by adding only –s, for example, bebé/bebé–s ‘baby/
babies’ (see Ambadiang 1999 for a complete discussion of number morpho-
logy). As a result, a child acquiring Spanish needs to be aware of all these
morphophonological restrictions involved in the pluralization process.
In addition to the phonological restrictions described, plural formation is a
very complex process because it involves the semantics of pluralization, which
goes beyond the distinction between ‘one’ and ‘more than one’ depending
on the noun type involved, as pointed out by Prado (1989). For example, an
Number Agreement 95
abstract noun like cólera ‘anger,’ becomes concrete and countable when it is
pluralized, acquiring the meaning of ‘acts or instances’ of the abstraction
cóleras; a concrete nominal like esposa ‘wife’ suffers a complete change in
meaning when pluralized: esposas ‘handcuffs’ or ‘wives.’
Bosque’s (1999) examination of a four-way classification for common
nouns illustrates the complexity involved in the pluralization process: (1)
count/mass nouns, for example, casa ‘house’ versus aire ‘air;’ (2) nouns
that can be numbered/pluralia tantum, for example, libro/dos libros ‘book/
two books’ versus celos ‘jealousy;’ (3) single/collective, for example, árbol
‘tree,’ versus ejército ‘army;’ (4) abstract/concrete, for example, verdad
‘truth’ versus flor ‘flower.’ As Bosque points out, this noun classification is
not strict, that is, nominals can belong to more than one classification.
Of particular interest for the present research is the distinction between
mass and count nominals, that is, whether children distinguish between
mass and count nominals by pluralizing the latter but not the former. Accord-
ing to Bosque, the contrast between mass and count nominals has the most
syntactic consequences. First, the nature of the nominal determines the
presence or absence of a determiner, as shown in Example 4.1.
Example 4.1
(a) Esto es *libro *‘This is book’ (cf. Esto es un libro ‘This is a book’)
(b) Esto es pan ‘This is bread’
(c) Quiero un pan ‘(I) want a bread’
(d) Quiero pan ‘(I) want bread’
(e) *Cae niño *‘Falls child’ (cf. Cae un/el niño ‘A/the child falls’)
(f) Cae agua ‘Water falls’
(g) *Entra mujer ‘Woman enters’ (cf. Una/la mujer entra ‘A/the woman
enters’)
(h) Entra frío ‘Cold enters’
enter as: (1) kind denoting; (2) predicate denoting; or (3) they might be
free to start as either. Each of these three choices corresponds to a major
language type. Specifically, in languages like Chinese and Japanese, all
nouns appear to be mass-like, exhibiting the morphosyntactic properties
of mass nouns, that is, they: (a) do not combine with plural morphology
(e.g., grass/*grasses); (b) do not combine with numerals (e.g.,*one grass/two
grasses); and (c) need classifier/measure phrases to be quantified (e.g., three
plies of grass). Chierchia, Guasti, and Gualmini (2000) provide as an exam-
ple the noun phrase ‘three tables’ in Chinese, liang zhang zhuozi which liter-
ally means ‘three pieces of table.’ Notice that another characteristic of this
type of language is that nouns can always occur bare even in argument posi-
tion. In contrast, languages like English present a mixed system, as they
mark the singular/plural contrast, distinguish mass versus count nouns,
and allow bare (count) plurals and bare mass nouns in argument position.
Finally, languages belonging to the Romance languages type mark the
singular/plural and the mass/count distinctions like in English, but bare
nominals are either not allowed (e.g., French) or restricted (e.g., Italian).
Chierchia et al. (2000) formalize this cross-linguistic classification by pro-
posing the Nominal Mapping Parameter (NMP):
Example 4.2
Nominal Mapping Parameter
(a) N [+ arg – pred] Classifier languages, for example, Chinese
(b) N [+ arg + pred] Germanic languages, for example, English
(c) N [– arg + pred] Romance languages, for example, Italian
Example 4.2 illustrates the three possible values for the NMP. In languages
like Chinese, as in Example 4.2(a), NPs are arguments; therefore, they can
occur bare in argument position. Chierchia et al. (2000) argue that in this
type of language, when nominals are turned into predicates, they exhibit
the same characteristics as mass nouns; for example, nouns show neither
plural morphology nor the obligatory use of classifiers with numerals (e.g.,
‘three pounds of rice’). On the other hand, Germanic languages like
English, present a mixed system, as shown in Example 4.2(b). In these lan-
guages, nouns may behave like arguments (e.g., they can occur bare) or
predicates (e.g., they require an overt determiner). Finally, in Romance
languages, NPs are predicates, as shown in Example 4.2(c). As a result, in
Romance languages, bare Ns cannot be arguments; that is, Ns need an overt
determiner to turn into an argument in these languages. This proposal
accounts for the cross-linguistic distribution of bare nominals and it is
Number Agreement 99
supports the singular value as the unmarked one. The three non-target-like
tokens with respect to number involved the production of singular clitics
used to refer to plural referents, for example, Elián: *la tiele ‘(she/he) has
it (fem/sg)’ instead of las tienes ‘[you] have them (fem/pl).’
In sum, the discussion above brings strong support for the availability of
singular as the unmarked value in children’s grammar. This was evidenced
in almost all the constituents under study (with the exception of demon-
strative pronouns) in which the children consistently uttered singular ele-
ments to refer to both singular and plural referents. Recall that both
hypotheses explored in this research (singular vs mass) are similar in that
they make the same prediction regarding number acquisition, that is, the
children under study will produce singular DPs to refer to plural referents.
In that sense both hypotheses are supported by these results.
Example 4.3
Flowers and Rice
Experimenter: ¿Qué va a comer papá?
‘What is dad going to eat?
(Doll eats from the two piles of rice)
Child: Ayián a come arro, oye.
‘Elián to eat (inf) rice, hey’
(Child complains he wants to eat rice himself)
Experimenter: ¿Qué comió papá ?
‘What did dad eat?’
[Doll eats from two piles of rice again]
Child: La[h] chiore
‘the flowers’
Example 4.4
Bananas and hair
Experimenter: ¿Qué come el caballo?
‘What is the horse eating?’
[The horse eats from two bananas]
Child: (Repeats experimenter’s question)
Experimenter: ¿Qué come el caballo?
‘What is the horse eating?’
[The horse eats from two bananas]
Child: (Child does not answer; makes eating sounds)
Experimenter: ¿Qué come el caballo?
‘What is the horse eating?’
(The horse eats from two piles of hair)
Child: No pelo, no come
‘No hair, [it] does not eat’
Experimenter: Mira lo que come el caballo
‘Look what the horse is eating’ (hair)
Child: Mira pelo
‘Look hair’
In Example 4.4 the child consistently gave the target-like response for the
mass nominal pelo ‘hair,’ that is, he did not pluralize the noun even though
there were two piles hair to trigger a plural response if available. Notice that
in Alonso’s case there is evidence that he knows how to mark plurality on
Number Agreement 105
Example 4.5
(a) Experimenter: ¿Qué comió el guauguau?
‘What did the dog eat?’ [Four flowers]
Child: Flole[h]
‘Flowers’
(b) Dog: ¿Qué como?
‘What should I eat?’
Child: Esto.
‘This (neuter).’ (Pointing at the two chicken legs)
Dog: ¿Qué es esto?
‘What is this (neuter)?’ (Points at the two cups)
Child: Tazo[h]
‘Cups’
In the utterances in Example 4.5, the child marked the plural number in
the count nominals produced. In particular, in Example 4.5(a) Londa
uttered flole[h] ‘flowers’ to refer to the four flowers the dog ate. In Example
4.5(b) she produced the plural noun tazo[h] for tazas ‘cups’ to refer to the
two cups present in the activity.
Regarding Londa’s mass nominal production, she produced the follow-
ing nominals: pelo ‘hair,’ cheche for leche ‘milk,’ ama for grama ‘grass,’ and ito
for pollito ‘chicken,’ as shown in Example 4.6.
106 First Language Acquisition in Spanish
Example 4.6
Milk and Flowers
Horse: ¿Qué como?
‘What should I eat?!’
Child: Cheche.
‘Milk.’
Esto e[h] una cheche.
‘This is a milk’ (She explains to the horse with one bottle).
(Horse drinks from the two bottles of milk)
Experimenter: ¿Qué tomó el caballo?
‘What did the horse drink?’
Child: Esto
‘This (neuter).’ [bottle of milk]
Experimenter: Sí, pero ¿qué tomó el caballo?
‘Yes, but what did the horse drink?’
Child: Cheche.
‘Milk.’
Example 4.6 presents an interesting case. On the one hand, the child
responded with a singular nominal cheche for leche ‘milk’ even though the
horse drank from the two bottles of milk. On the other hand, she produced
an indefinite determiner una ‘a/one’ with the mass noun, for example, esto es
una leche ‘this is a milk.’ This phrase seems to indicate that Londa is treating
this particular mass nominal as a count, that is, it could be quantified with the
indefinite determiner ‘a.’ Another possible explanation is that the child
quantified the nominal as in “a bottle of milk.” However, no other examples
were found of this nature in this child’s data to confirm this hypothesis.
Overall the previous discussion showed that all three children produced
target-like mass nominals, that is, in the singular. The case of count nomi-
nals is confounded by the fact that these children are not marking the
plural number in general. Nonetheless, Elián and Londa produced the
plural count noun flores, providing some evidence that there is a difference
in their grammars between these two types of nominals. Finally, Londa’s
production of una leche raises some questions about the true nature of mass
nominals in child language. One possibility is that she was quantifying the
bottle itself, as mentioned before, and not its contents, that is, the milk.
Similar non-target-like utterances are reported in the literature (Chierchia
et al. 2000; Gordon 1982). In particular, Gordon found that this type of
non-target-like instance accounted for 2 percent of the total production.
In addition, Chierchia et al. found examples of mass nouns treated as count
Number Agreement 107
in the data of the four English children used in their study. They argue that
in a language like English with grammaticized count nouns, children are
open to misanalyses in the lexicalization of mass nominals given the limited
input on collective mass, for example, hair, luggage, and furniture are
examples of lexicalized misanalyses in English.
In order to assess if the two older children Diana and Pepe have acquired
the mass/count distinction, I compared their production with that of the
three children under study. The analysis shows that Diana and Pepe suc-
cessfully produced the mass nominals, as well as the count nominals in the
“Time to eat” task. In particular, Diana produced the following count plural
nominals: flore[h] ‘flowers,’ silla[h] ‘chairs,’ bola[h] ‘balls,’ and taza[h] ‘cups.’
Regarding the mass nominals she uttered the following examples: pasto
‘pasture,’ grama ‘grass,’ arro[h] ‘rice,’ and café ‘coffee.’
Diana produced two interesting examples with mass nominals, as illus-
trated in Example 4.7.
Example 4.7
(a) Mira se cayó un arroz
‘Look a rice fell down’
(b) Doll: ¿Dónde me senté?
‘Where did I sit down?’ [doll sits on the grass]
Child: Aquí.
‘Here’ (She points at the grass)
Doll: Pero, ¿qué es eso?
‘But, what is that (neuter)?’
Child: *Un pasto.
‘A/one pasture’
Example 4.8
Doll: Yo me tengo que sentar.
‘I have to sit down’
Child: En el pasto.
‘On the grass’ (She points at the grass)
Doll: ¡Ay, ay...! ¿Dónde me senté, en la silla?
‘Au, au...! Where did I sit, on the chair?’
(Doll sits on the grass and complains)
Child: No. *Esto[h] son de caballitos.
‘No. These (masc/pl) belong to [the] small horses.’
Example 4.8 presents how Diana alternated between target-like and non-
target-like production of mass nominals. First, the child responded in a
target-like fashion en el pasto ‘on the grass’ using the mass nominal in the
singular. In contrast, she referred to the grass with the plural demonstrative
pronoun esto[h] ‘these (masc/pl)’ as she explained that the grass was for
the small horses. Finally, Diana produced another interesting example per-
taining to the mass nominals, as shown in the dialogue in Example 4.9.
Example 4.9
Experimenter: O.K, deja ver qué más hay aquí.
‘O.K., lets see what else is here’
Child: ¿Qué son estos? ¿qué esto?
‘What are these (masc/pl)? what [is] this (neut)’
Experimenter: ¿Qué es eso?
‘What is that (neut)?’
Child: Pollo.
‘Chicken.’
In the dialogue illustrated in Example 4.9, Diana first refers to the two
pieces of chicken with the plural demonstrative estos ‘these’ perhaps
wondering what the prop actually was. Then once she identified the objects,
she produced the singular target-like mass nominal pollo ‘chicken.’ How-
ever, when the experimenter asked her again what the prop was: ¿Qué es eso?
‘What is that (neuter)?’ the child responded in a non-target-like fashion:
*Un pollo . . . lo[h] pollo[h] ‘A chicken . . . the (masc/pl) chickens (masc/pl).’
In her response, the child produced two non-target-like instances, the
first one involving a quantified mass nominal and the second one the plu-
ralization of the same nominal. The discussion of Diana’s non-target-like
responses points to the production of mass nominals as mass (namely in the
Number Agreement 109
singular) but also as count, in the sense that they can be both pluralized
and quantified. This alternation seems to reflect the difficulty children face
in the interpretation of this type of nominals or what Chierchia et al. call
the recategorization process.
Pepe’s production of nominals in this task was similar to Diana’s produc-
tion, that is, it included both types of nominals. For example, regarding
count nominals, he produced the following: plato[h] ‘plates,’ guineo[h]
‘bananas,’ silla[h] ‘chairs,’ bola[h] ‘balls,’ and taza[h] ‘cups.’ In addition, the
child produced the following mass nominals: grama ‘grass,’ pollo ‘chicken,’
and pelo ‘hair.’
Pepe also uttered two interesting examples involving the mass nominal
grama ‘grass,’ as shown in Example 4.10.
Example 4.10
(a) *Yo la[h] tengo en la pecera pero no son así . . . de larga[h].
‘I have them (fem/pl) in the aquarium but (they) are not that . . .
long (fem/pl)’
(b) ¡Eso[h] son iguale[h]!
‘Those (masc/pl) are equal (masc/pl)’
other hand, and most importantly, it also accounts for how children exposed
to different languages converge to the target grammar, that is, learnability.
In this sense, Chierchia’s hypothesis is preferable over Harris’ (Spanish-
specific) singular parameter.
1;8 1 0 1 (100%)
1;9 11 0 11 (100%)
1;10 15 1 14 (93%)
1;11 9 2 7 (78%)
2;0 42 24 18 (43%)
Example 4.11
Child Target
*Flol (Londa) flores
‘flower’ [four flowers] ‘flowers’
Table 4.4 Non-plural marked BNs and plural marked BNs produced: plural
referent
Non-plural marked Bare Nominals Plural marked Bare Nominals Total
Notice that the phenomenon of unmarked plurality is still very strong in the
case of the oldest child, Londa, accounting for 69 percent of her nominals
produced to refer to a plural entity, as illustrated in Table 4.4. This result
regarding number marking on nominals is consistent with my findings for
the other structures under study, that is, children do not mark the plural
number feature overtly at this stage of the acquisition process.
We turn the discussion now to the findings on number agreement, that is,
agreement relations between nominals and their agreeing constituents.
The presentation of the findings also includes an analysis of reference to
determine whether the structures produced by the children under study
match the features of the referents. Moreover, number agreement results
will be compared to the results obtained for gender agreement to establish
a timing in the acquisition of these two features.
The data analysis revealed that while children established almost perfect
gender agreement relations within Determiner Phrases, this was not the
case with respect to number agreement. In particular, the (full) DP data
showed that the children produced a total of 62 (full) DP tokens as shown
in Table 4.5.
This table indicates that out of a total of 62 (full) DP tokens produced by
the three children, 8 tokens were non-target-like with respect to agreement.
Table 4.5 also shows that the majority of non-target-like utterances (7 out
of 8) involved the production of plural masculine nominals. In particular,
these utterances pertained to mismatches of the number feature, that is,
children either mark the plurality on the determiner or on the nominal,
but not in both constituents as required by Spanish grammar.
Production of (full) DPs was different across the three children, with the
younger children, Elián and Alonso, producing considerably fewer instances
of this structure (5 and 19 respectively) than Londa (38 tokens). Hence
their production will be analyzed individually. In the case of Elián, he
produced only one (full) DP token in the plural, as shown in Example
4.12(a).
Singular 27 1 20 –
Plural 4 6 3 1
Total 31 7 23 1
Number Agreement 117
Example 4.12
Child Target
(a) *es ya flore[h] (Elián) son las flores
‘[it] is the (fem/sg) flowers (un/fem/pl)’ ‘(they) are the (fem/pl)
flowers(un/f/pl)’
(b) Coye la bola coge la bola
‘take (imp) the(fem/sg) ball (fem/sg)’ ‘take (imp) the (fem/sg)
ball (fem/sg)’
Example 4.13
Child Target
(a) oto pece otros peces
‘other (masc/sg) fishes (masc/pl)’ ‘other (masc/pl) fishes
(masc/pl)’
(b) lo juete los juguetes
‘the (masc/pl) toy (masc/sg)’ ‘the (masc/pl) toys (masc/pl)’
(c) lo carro los carros
‘the (masc/pl) car (masc/sg)’ ‘the (masc/pl) cars (masc/pl)’
(d) lo[h] caballo los caballos
‘the (masc/pl) horse (masc/sg)’ ‘the (masc/pl) horses (masc/pl)’
118 First Language Acquisition in Spanish
Example 4.14
(a) ese guauguau ‘that(masc/sg) dog (masc/sg)’
(b) la comida ‘the (fem/sg) food (fem/sg)’
(c) el caballo ‘the (masc/sg) horse (masc/sg)’
The referent of Example 4.14(a) is not clear because Alonso uttered this
DP while pretending to talk on the phone with a dog on each ear, that is, it
is not clear whether the child was referring to both dogs or to only one.
Number Agreement 119
Example 4.14(b) la comida ‘the food’ presents an interesting case for the
analysis of reference. The child uttered this DP when the experimenter
took out four (dinner) plates. In this context it was clear that he was not
referring to the plates, but to the food to be put on them. Since there was
no food present, the referent of the utterance was abstract not concrete. In
the case of Example 4.14(c), Alonso produced this particular token in three
separate instances. In two instances, the referent was one horse; hence
these were target-like uses. In one of the instances, however, the child
uttered this DP, el caballo ‘the horse,’ while holding two horses. This seems
to be a case of a number feature mismatch, that is, using a singular DP to
refer to more than one object.
Regarding the reference examination of Alonso’s six plural (full) DP
tokens produced, the analysis found only one case of a semantic number
mismatch, as illustrated in Example 4.15(a).
Example 4.15
Child Target
(a) Lo[h] caballo[h], el caballo el caballo
‘the horses, the horse’ ‘the horse’
(b) lo pece los peces
‘the fishes’ ‘the fishes’
(c) lo carro los carros
‘the (masc/pl) car (masc/sg)’ ‘the (masc/pl) cars (masc/pl)’
In Example 4.15(a) the child uttered a plural DP, lo[h] caballo[h] ‘the
horses’ to refer to one horse. Interestingly, he immediately produced the
singular DP equivalent el caballo ‘the horse,’ in what appears to be a case of
self-correction. The remaining utterances in Example 4.15 were target-like
with respect to semantic number agreement. Target-like plural (full) DPs
included requests for a particular group of toys, as in Example 4.15(b), that
is, the child requests lo pece ‘the fishes.’ Another target-like plural DP is
shown in example 4.15(c) lo carro ‘the (masc/pl) car (masc/sg).’ In this
example the child uses a definite plural DP to complain about the construc-
tion trucks that have been passing outside the house, making a lot of noise.
In the target grammar this particular DP would receive a generic inter-
pretation, that is, a non-specific set of cars. However, there is no evidence in
Alonso’s data that he has the knowledge of generic interpretations.
The discussion of Alonso’s (full) DP production in terms of number,
indicates that he had difficulties marking the plurality of the nominals and
determiners with the final –s, while he marked successfully segments that
120 First Language Acquisition in Spanish
Definite 2 1 2 1 6 (16%)
Indefinite 9 3 11 1 24 (63%)
Demonstrative 4 – 2 1 7 (18%)
Other – – 1 – 1 (3%)
Total 15 4 16 3 38 (100%)
did not involve this marker exclusively, for example, lo pece. This finding
points to the conclusion that in these examples the nature of the problem is
not grammatical but phonological in nature. This finding seems to be at odds
with previous research stating that the acquisition of plural allomorph –es
poses more problems for children than final –s (Pérez-Pereira 1989).
The analysis of Londa’s (full) DP utterances yielded a total of 38 tokens
produced (excluding all repetitions): 19 masculine and 19 feminine, as
shown in Table 4.6. This table displays the distribution of (full) DPs in terms
of number, with 31 tokens of singular DPs and 7 tokens of plural DPs
produced by Londa. This points to the fact that the vast majority of (full)
DPs was in the singular. The data analysis found that Londa produced
three non-target-like utterances with regard to (grammatical) number
agreement, one pertained to a gender mismatch, and two to number mis-
matches. The two non-target-like utterances involving a number mismatch
are illustrated in Example 4.16:
Example 4.16
Child Target
(a) *¡Uno ece! ¡Unos peces!
‘A (masc/sg/indef)/some ‘Some (masc/pl) fishes (masc/pl)’
(masc/pl/indef) fishes’
(b) *Lo duce Los dulces
‘The (masc/pl) candy (sg/pl)’ ‘The (masc/pl) candies (masc/pl)’
mismatch is between the plural definite article lo ‘the’ and the singular
noun dulce ‘candy.’ As discussed in Alonso’s case, these two examples could
be two instances of the final –s plural marker aspiration, that is, uno for
unos, and dulce for dulces, and in that case there would be no mismatch pres-
ent. The production of uno can be interpreted as a shortened form of the
plural determiner unos ‘some,’ however in this child’s case this inference
cannot be done because she produces all singular indefinite articles as uno
instead of un (cf. Appendix A for a listing of Londa’s DP tokens).
In terms of the analysis of reference, an evaluation of the 38 tokens
yielded 6 cases in which the grammatical features marked on the DPs did
not match those of the referent, as can be seen in Example 4.17.
Example 4.17
Child Target
(a) Mira él, pielna Mira las/sus piernas *(Two legs)
‘Look at him, leg (sg)’ ‘Look the (pl)/his (3rd pers/pl) legs (pl)’
(b) Una flor Unas flores *(Four flowers)
‘A /one (sg) flower (sg)’ ‘Some (pl) flowers (pl)’
(c) Una aita Unas alitas *(Two pairs of wings)
‘A/one (sg) wing (sg)’ ‘Some (pl) wings (pl)’
(d) Un dente Unos dientes *(Teeth)
‘A/one (sg) tooth (sg)’ ‘Some (pl) teeth (pl)’
(e) Es un carro Son unos carros *(Many cars)
‘(It) is a/one (sg) car (sg)’ ‘(They) are some (pl) cars (pl)’
(f) Ese ee uno ijo Estos son unos lagartijos *(Two snakes)
‘this (sg) is a/one lizard’ ‘These are some lizards’
All the utterances listed in Example 4.17 are of the same nature, that is,
they involve the production of a singular nominal to refer to a plural refer-
ent. In Example 4.17(a) the child showed two of the dog’s legs to the exper-
imenter but spontaneously uttered the nominal in the singular, pielna ‘ leg’
instead of piernas ‘legs.’ Examples 4.17(a) to 4.17(e) were produced as
responses to the experimenter’s inquiry, for example, ¿Qué es esto? ‘What is
this (neuter)?’ In particular, in Example 4.17(b) the child answered with
una flor ‘a flower’ as the experimenter showed her four flowers, a clear
(semantic) number mismatch between the singular DP and the plural ref-
erent. The context of Example 4.17(c) was the experimenter pointing out
to the child the wings of two bees drawn in one of the toy houses. Even
though there were two pairs of wings present, the child uttered the nominal
122 First Language Acquisition in Spanish
in the singular, for example, una aita ‘a/one little wing.’ Example 4.17(d)
was uttered as the experimenter showed the child the teeth of a toy snake.
As in the previous example, even though there were many teeth present the
child may have been referring to only one of them. The context of Example
4.17(e) was the experimenter taking out two cars and asking the child: ¿Qué
es esto? ‘What is this (neuter)?’ The child responded with the singular DP:
es un carro ‘(it) is a car.’ Finally, in Example 4.17(f) the experimenter took
out two snakes (one big and one small), then the child identified them as a
lizard saying este ee uno ijo ‘this (sg) is a/one lizard.’
One possible explanation for these semantic mismatches between the DP
features and the objects they referred to, can be that the child at this stage
of acquisition does not pay attention to the number of objects present, but
to the object itself, that is, the child is identifying the object type or labeling
it, and not specifying the amount of objects present. Another explanation
could be related to the acquisition of the feature definiteness, involving the
contrast between definite versus indefinite. In all but one of these examples
(i.e., 4.17(a)), the target-like answer required the plural indefinite unos/
unas ‘some.’ If this hypothesis on the acquisition of determiners is on
the right track, these non-target-like examples would be explained by the
inability of this child to mark the contrast definite versus indefinite, that is,
Londa’s grammar initially may be set initially with the definite value for this
feature. Then, in this analysis singular examples with the determiner uno ‘a’
for un would be explained as instances of the quantifier one and not as the
indefinite determiner ‘a.’
The discussion of number agreement in the (full) DP data points to the
fact that marking plurality is problematic for these children (confounded
by phonological factors) while gender is not. Notice that even though
the majority of DPs produced were in the singular, (48) versus (14) in the
plural, the majority of non-target-like utterances in terms of agreement per-
tained to the number feature. This result seems to indicate an order in the
acquisition of these features, that is, gender precedes the acquisition of
number in Spanish, or grammatical features are acquired before semantic
features in this case.
Further support for a delay in the acquisition of the number feature is
found in the MPH/DPs, that is, vocalic elements produced in the slot a
determiner would occupy in the adult language (Bottari et al. 1993/1994).
An evaluation of the 102 instances of MPH/DPs, produced by the three
children, yielded a total of 33 tokens once repetitions were excluded, as
shown in Table 4.7.
Number Agreement 123
Masculine Feminine
Elián Target-like 7 – 4 – 11
Non-target-like – – – 1 1
Londa Target-like 2 – 3 – 5
Non-target-like – – – – –
Alonso Target-like 6 2 3 – 11
Non-target-like 3 2 – – 5
Total 18 4 10 1 33
Example 4.18
(a) *a fore[h] las flores [Elián]
‘a/the (fem/sg) flowers ‘the (fem/pl) flowers
(un/fem/pl)’ (un/fem/pl)’
(b) *o pie los pies [Alonso]
‘the (masc/pl) foot (masc/sg)’ ‘the (masc/pl) feet
(masc/sg)’
(c) *a pece los peces [Alonso]
‘the (fem/sg) fishes (masc/pl)’ ‘the (masc/pl) fishes
(masc/pl)’
Example 4.19
experimenter showing four cats to Alonso while telling him: ¿Qué habrá
aquí? ‘I wonder what is in here;’ his answer was: mia, e guauguau ‘look, the
(masc/sg) dog.’ Notice that as in Example 4.19(d), the child uses the MPH
e interpreted as the definite article el ‘the’ instead of the indefinite one to
identify a new object, un. There was another instance of this token and it
was target-like in terms of reference because the child was asking for another
dog. The last two utterances in Example 4.19 involve a different type of
mismatch, namely the use of a plural MPH/DP to refer to a single object.
Specifically, in Example 4.19(e), Elián produces a plural MPH/DP even
though he is holding one flower in his hand. One possible explanation for
this token is that the child might only know the plural form of the nominal,
that is, flore[h] ‘flowers.’ Evidence for this hypothesis was found in the analy-
sis of Elián’s DP production, in particular the non-target-like utterance es ya
flore[h] ‘(it) is the (fem/sg) flowers (unm/fem/pl).’ In this utterance all
the constituents are in the singular except the nominal flore[h], perhaps
because it is the only form of this particular noun that the child knows.
Finally, in Example 4.19(f) Alonso produces a plural DP even though he
is presented with only one object, in this particular case a crocodile. The
context of this last utterance was the child requesting the fishes and antici-
pating that the experimenter was going to take them out of the bag at any
moment. In this context, the child might have uttered the plural DP o[h]
pece prior to seeing the actual fishes being taken out. In this child’s data,
there is evidence that he knows both the singular and the plural forms of
this particular nominal, that is, he produced pece ‘fishes’ and pe[h] ‘fish.’
The discussion of the MPH/DP data with respect to number agreement
reveals that children were not having many issues establishing target-like
agreement between the MPHs and the nominals they precede, that is, chil-
dren produce only three non-target-like tokens with regard to number
agreement. In contrast, the analysis of semantic agreement showed that the
three children were having problems matching the number features of the
referent with the corresponding morphological features.
We turn the discussion to the analysis of number agreement in attributive
adjective structures. Children’s production of attributive adjectives was
very limited, with a total of 18 utterances, as seen in Table 4.8. Moreover,
this table reveals a marked contrast between the production of the two
younger children and the production of Londa: Elián (2), Alonso (1), and
Londa (15). Table 4.8 also shows that Elián’s and Alonso’s production was
all target-like.
In contrast, Londa produced six non-target-like attributive adjective
utterances, all of them involving a mismatch of the number feature.
Number Agreement 127
Example 4.20
Child Target
(a) *No, anaron feli[h] (Two times) No, ganaron los felices [trenes]
‘No, happy (unm/sg) [one] won’ ‘No, the (masc/pl) happy
(masc/pl) [ones] won’
(b) *Tistre Los tristes [trenes]
‘Sad (unm/sg) [one]’ ‘The (masc/pl) sad (masc/pl)
[ones]’
(c) *Azul (Three times) Los azules [carros]
‘Blue (unm/sg) [one]’ ‘The (masc/sg) blue (masc/pl)
[ones]’
Example 4.21
Experimenter: ¿Quién ganó? ‘Who won?’
Child: Yo ‘I’
Experimenter: ¿Los tristes? ‘ The (masc/pl) sad (unm/pl)
[ones]?’
¿Ganaron los tristes? ‘The (masc/pl) sad (unm/pl)
[ones] won?’
Child: No, anaron feli[h] ‘No, sad (unm/sg) [one] won.’
128 First Language Acquisition in Spanish
In the dialogue in Example 4.21 the child uttered the singular adjective
feli[h] ‘happy’ to refer to the happy trains that won the race, even though
the experimenter marked the plurality of the nominal in the second ques-
tion, that is, ¿Ganaron los tristes? ‘The (masc/pl) sad (masc/pl) (ones) won
(3rd pers/pl)?’ The child produced a second instance of the adjective feli[h]
‘happy’ after the experimenter explained to her that she had to choose two
trains, either the two happy ones or the two sad ones: Estos dos, tienes que
escoger los dos ‘These (masc/pl) two, (you) have to choose the (masc/pl) two
(trains).’ After listening to the indications of the experimenter, the child
made her selection by saying feli[h] instead of los felices ‘the (masc/pl) happy
(unm/pl) (ones).’ As in the previous instance of this token, Londa pro-
duced a singular adjective even though the experimenter provided her with
plurality cues in the instructions. One could argue that the child inter-
preted the final –s of the singular adjective feli–z ‘happy’ as a plural marker.
However, Londa’s overall production does not support this conclusion, that
is, this child had problems marking the plural number feature in all con-
stituents under study.
Example 4.20(b) was produced as the child completed the experiment-
er’s sentence: Y yo voy a llevar . . . ‘And I am going to carry . . .’ The child
finished the experimenter’s sentence by saying tistre ‘sad (unm/sg)’ instead
of los tristes ‘the (masc/pl) sad (un/pl)(ones)’ to refer to the two trains with
sad faces that the experimenter was going to play with. In this context, the
attributive reading seems more appropriate given the fact that the child was
completing the experimenter’s sentence.
The last non-target-like adjective token produced by Londa was azul ‘blue
(unm/sg) (one).’ The child uttered this token three times as responses to
the experimenter’s questions, as shown in Example 4.22.
Example 4.22
(a) Experimenter: Mira, yo tengo los rojos, ¿cuál tú tienes?
‘Look, I have the (masc/pl) red (masc/pl) [ones],
which [one] do you have?’
Child: Azul, azul.
‘Blue (unm/sg) [one], blue (unm/sg) [one]’
(b) Experimenter: Yo tengo los rojos, ¿cuál tú tienes? ‘
‘I have the (masc/pl) red (masc/pl) [ones], which
[one] do you have?’
Child: Azul
‘Blue (unm/sg) [one]’
(c) Experimenter: ¿Qué carros ganaron?
‘What cars (masc/pl) won (3rd pl)?’
Number Agreement 129
Table 4.9 demonstrates that these two children have already mastered
the adjectival agreement in attributive structures, displaying perfect
target-like production, for example, el chiquito ‘the small (one),’ las azules
‘the blue (ones).’ Interestingly, Diana’s and Pepe’s attributive adjective pro-
duction (17) and (20) respectively, reveal that these two children produced
similar proportions to Londa (15), that is, Diana, Pepe, and Londa pro-
duced less than 20 attributive utterances. This limited production supports
the finding that these structures are limited in child language; hence,
further research should include a larger subject pool to be able to extract
a more robust data set.
We turn our discussion to the analysis of number agreement in the other
constituents under study, namely predicative adjectives, and demonstrative
and third person (direct) object clitic pronouns. These additional constitu-
ents were included in the agreement analysis to have a more global under-
standing of agreement in the grammar of the children under study.
In the case of predicative adjectives, the children under study produced
a total of 77 utterances with a predicate adjective, 61 target-like and 16 non-
target-like in terms of nominal agreement. Table 4.10 illustrates the distri-
bution of the predicative adjectives produced by the three children.
Table 4.10 also shows that Elián’s and Alonso’s production of this struc-
ture was scarce, as in the case of attributive adjective data, in comparison to
Londa’s production, that is, (1), (10), and (66) respectively. Specifically,
Elián only produced one predicative adjective utterance and it was in the
singular, for example, Ii cayente ‘[It] is hot (unm/sg)’ for está caliente to
Number Agreement 131
Example 4.23
Child Target
*Son sucio Están sucios [pies]
‘(They) are (perm) dirty (masc/sg)’ ‘(They) are (chang) dirty (masc/pl)’
Example 4.24
Experimenter: ¿Y eso qué es?
‘And that (neuter) what is (it)?’
Child: ¡Mira oo pie!
‘Look the feet!’
Experimenter: Pies, sí.
‘Feet, yes.’
Child: Son sucio.
‘(They) are dirty (masc/sg)’
Example 4.25
Child Target
(a) *Flole ande (Son) flores grandes
‘Flowers (unm/pl) big (unm/sg)’ ‘[They] are big (unm/pl) flowers
(unm/pl)’
(b) *Ello ya antado Ellos ya están levantados
‘They already awake (masc/sg)’ ‘They are already awake (masc/pl)’
(c) *Tá omido Están dormidos
‘[It] is asleep (masc/sg)’ ‘(They) are asleep (masc/pl)’
(d) *Tá tistre (two times) Están tristes
‘[it]is sad (unm/sg)’ ‘(They) are sad (unm/pl)’
(e) *Feli[h] (Están) felices
‘Happy (unm/sg)’ ‘(They) are happy (unm/pl)’
(f) *No, ita No, son chiquitas
‘No, small (fem/sg)’ ‘No, (they) are small (fem/pl)’
This particular example points to the availability of a default value for the
number feature, expressed morphologically as the singular form. Examples
4.25(d) and 4.25(e) illustrate the use of the singular adjectives triste ‘sad
(unm/sg)’ and feliz ‘happy (unm/sg)’ to refer to a pair of trains. These two
examples are very interesting because Londa produced consistently singu-
lar utterances even though the experimenter (overtly) marked the plurality
of the nominal in the questions. The two examples had the context shown
in Example 4.26.
Example 4.26
(a) Experimenter: ¿Y estos dos?
‘And these (masc/pl) two?’
Child: Tá tiste
‘(it) is sad (unm/sg)’
(b) Experimenter: ¿Y estos, estos dos?
‘And these (masc/pl), these (masc/pl) two?’
Child: Feli[h]
‘Happy (unm/sg)’
that is, the child responded entirely in the singular form even when plural
cues were provided in the input. This issue is addressed in Chapter 5.
The results obtained from the data of the three children under study
were compared with the predicative adjective production of the two older
children, Diana and Pepe.
Table 4.11 displays the almost completely target-like production of Diana
and Pepe, with only six non-target-like utterances produced. In the case of
Diana, two non-target-like utterances were found with respect to number
agreement, as seen in Example 4.27.
Example 4.27
Example 4.28
Experimenter: ¿Bolas qué? ¿de qué color?
‘Balls (fem/pl) what? What color (balls)?’
Child: Este . . . bola[h]
‘Um... balls (fem/pl)’
Experimenter: ¿Bolas de qué color?
‘Balls (fem/pl) (of) what color?’
Child: *De amarilla
‘Of yellow (fem/sg)’
Number Agreement 135
Example 4.29
(a) Las bolas son [PP de color amarillo]
The balls are [PP of the color [mas/sg] yellow [mas/sg]]
(b) Las bolas son [SC pro amarillas]
The balls are [PP pro [fem/pl] yellow [fem/pl]]
In Example 4.29(a) the adjective amarillo ‘yellow’ agrees with the masculine
singular features of the nominal color ‘color,’ whereas in Example 4.29(b),
it agrees with the plural feminine features of the nominal bolas. Example
4.27(a), then seems to indicate that this child’s grammar interprets the
structure in the question as in Example 4.29(a) and as a result, she copies
the preposition, that is, las bolas son de color amarillo ‘the balls are of the color
yellow.’ Notice that within the prepositional structural interpretation, only
the singular number is possible, cf. De color amarillo ‘Of color yellow (masc/
sg)’ versus *De color amarillos ‘Of color yellow (masc/pl).’
In the second non-target-like token shown in Example 4.27(b), *son azul
‘(they) are blue (unm/sg),’ Diana uttered the copula in the plural but the
adjective azul in the singular to refer to the nominal bolas ‘balls (fem/pl).’ This
token was produced as an answer to the experimenter’s question: Pero, ¿de qué
color? ‘But, (of) what color?’ If the structural misinterpretation hypothesis is
correct, this answer can be paraphrased as las bolas son de color azul ‘The balls
are of color blue (unm/sg)’ and as such is target-like in this child’s grammar.
Pepe’s non-target-like production is discussed next. Three non-target-like
examples (regarding agreement) were found in his data, two of them per-
tained to a mismatch of the number feature, as seen in Example 4.30.
Example 4.30
Child Target
(a) *Igual a esto[h] Iguales a estos
‘Similar (unm/sg) to these (masc/pl) ‘Similar (unm/pl) to these
[ones]’ (masc/pl) [ones]’
136 First Language Acquisition in Spanish
The data analysis shows that the three children produced a total of 13
non-target-like clitics, but only three instances involved a number mismatch,
as shown in Example 4.31.
Example 4.31
Child Target Referent
(a) Vo gadarla (Alonso) Voy a guardarlos *[bloques]
‘(I) am going to put it ‘(I) am going to put them
(fem/sg)[away]’ (masc/pl) [away]’
(b) Coelo (Alonso) Cógelos *[zapatos]
‘Take it (masc/sg)’ ‘Take them (masc/pl)’
(c) La tiele (Elián) Las tienes *[mariposas]
‘[He/she] has it (fem/sg)’ ‘[You] have them (fem/pl)’
Example 4.31(a) displays a total mismatch of both the gender and the
number features. Alonso utters the feminine singular clitic, la ‘it’ to refer to
bloques ‘toy blocks’ a masculine plural nominal. Examples 4.31(b) and
4.31(c) involve a mismatch of the number feature. Specifically, in Example
4.31(b) Alonso produced the masculine singular clitic lo to refer to zapatos
‘shoes’ a masculine plural nominal; while in 4.31(c) Elián utters the femi-
nine singular clitic la to refer to the feminine plural nominal mariposas
‘butterflies.’ As discussed previously, these examples may be a product of
phonology not featural mismatch, given the phonological context for
aspiration, namely, final vowel + –s. However, this explanation seems plau-
sible in the case of Alonso because there is evidence in his production of
plurality. In the case of Elián, no conclusions can be drawn given his limited
and mostly singular production.
Grammaticala Semanticb
(Full) DPs 21 7 8 15
MPH/DPs 9 3 5 8
Attributive adjectives 6 1 5 6
Predicative adjectives 6 2 4 6
Demonstratives 11 – – –
Third person clitics 5 1 2 3
Total 58 14 24 38
a
Grammatical non-target-like number refers to a number mismatch between the nominals and the
agreeing constituents.
b
Semantic non-target-like number refers to a mismatch between the number morphology of the DP and
the number of objects it refers to.
on whether the children understood the contrast between one versus more
than one.
The protocol for this task was that one (or more than one) animal was
sleepy and needed to go to the house to sleep. Then the experimenter
would ask the child, using a puppet, for one (or more) animal to take to the
house. Crucially animal props were set up in front of the child for her to
choose from, for example, two cats or three cows. Once a request was made,
children responded by selecting from the set of animals and taking to the
house or giving it to the puppet. Notice that this task separates the cognitive
aspect of number comprehension from the production aspect, that is, chil-
dren were responding by acting not speaking (see Chapter 2 for details).
Children’s responses were coded in two categories: (1) responsive, defined
as the child responding by giving a toy or toys to the experimenter or select-
ing the appropriate number of objects even if she did not pass them to the
experimenter; (2) non-responsive, defined as the child not giving anything at
all. The non-responsive category includes the following behaviors: refusals
(the child overtly refuses to give the requested toy); other responses
(includes behaviors such as hitting the puppet, asking for clarification for
example, ¿ah? ‘uh?,’ or simply pretending not to have heard the request);
and no response (the child is distracted at play). The results for this task are
summarized in Table 4.15.
This table also illustrates that the type of response varies from child to
child, for example, Londa never refused to give a toy or toys away, whereas
Elián and Adolfo frequently responded with an overt refusal. Hence, the
results of each child would be discussed on an individual basis. In the case
of Elián, his responses were distributed as follows: 10 responsive and 26
non-responsive, as illustrated in Table 4.15. Of the ten responsive acts car-
ried out by Elián, eight were target-like, that is, the child gave the number
Responsive 10 15 8 31
Non-responsive refusals 10 15 – 31
Other response 12 14 5 29
No answer 4 12 – 14
Total 36a 56b 13c 103
a
A total of 10 different requests were made, excluding repetitions.
b
A total of 9 different requests were made, excluding repetitions.
c
A total of 8 different requests were made, excluding repetitions.
142 First Language Acquisition in Spanish
of objects requested, and two were non-target-like, that is, the child gave
an incorrect number of objects. Target-like responses were distributed as
follows: two responses were to feminine singular nouns, three to masculine
singular nouns, three to feminine plural nouns, whereas non-target-like
responses involved requests for two masculine plural nouns. This distribu-
tion points to the fact that this child was as likely to respond to requests with
feminine nominals as to requests with masculine ones, that is, five responses
to feminine requests and five to masculine ones.
Regarding Elián’s two non-target-like responses, both of them involved
the experimenter’s request for two objects, as shown in Example 4.32.
Example 4.32
(a) Experimenter: Dame dos, yo quiero dos (three cats available).
‘Give me two, I want two.’
Child: (responds by giving one cat)
(b) Experimenter: Dame dos de esos, vamos a guardarlos, dame dos.
‘Give me two of those (masc/pl), lets put them
(masc/pl) [away], give me two’ (pointing at the
three cats).
Child: (responds by giving one cat).
In Example 4.32(a) the experimenter requested two cats (of the three cats
present) from the child, while the child played with one. Elián responded
by giving the cat he was holding. In 4.32(b) the same request was repeated,
this time the three cats were inside the toy car the child was driving. Again
the child responded by giving only one cat. A possible explanation for the
non-target-like responses is that this child did not understand the numeral
“two” or perhaps he was more interested in playing than responding to the
experimenter’s requests. Another possibility is that the child did not under-
stand the difference between one and more than one. However, there is
evidence that the child understood a different kind of plural request, as
illustrated in Example 4.33.
Example 4.33
(a) Experimenter: ¿Dónde están las grandes? Dame las grandes.
‘Where are the (fem/pl) big (unm/pl) [ones]? Give
me the (fem/pl) big (unm/pl) [ones].’
Child: (responds by giving three butterflies).
(b) Experimenter: Búscalas ‘Find them (fem/pl).’
Child: (responds by picking up two of the batteries).
Number Agreement 143
Example 4.34
(a) Experimenter: Mira las mariposas. Dame las grandes.
‘Look at the (fem/pl) butterflies (fem/pl). Give me
the (fem/pl) big (unm/fem) [ones].’
Child: ¿La[h] grande[h]?
‘The (fem/pl) big (unm/pl) [ones]?’ (repeats the
request; hands the big snake).
In the dialogue in Example 4.34 the experimenter requested the big but-
terflies; there were four butterflies available, two big and two small. Notice
that when the request was made Alonso was not in front of the butterflies.
His immediate reaction was to repeat the request in the form of a question,
¿la[h] grande[h]? ‘the big [ones]?’ and then he proceeded to bring a big
snake instead of the big butterflies. Interestingly enough, the noun serpiente
‘snake’ matches the feminine gender features of the noun ‘butterflies.’
Since the noun ‘butterflies’ was dropped (or omitted) in this request, the
child might have chosen a different referent that matched the gender fea-
tures of the request. However, there was no other big snake to bring to the
experimenter to match the plural number feature of the requested object.
The child never uttered the nominal serpiente or culebra ‘snake’ so in fact,
we have no evidence that he actually knows it. This response could be an
error due to the limited attention span of the child or simply a response
that shows his preference for particular toys. These results seem to point to
the fact that this child has some awareness of the difference between one
and more than one, as illustrated by his responses.
Londa’s behavior in this task contrasted from the behavior of the two
children discussed. First, neither “overt refusals” nor “no answers” to the
experimenter’s requests were found in her data. This might reflect a social
maturation factor, that is, this child was willing to participate in the task
sharing “her” toys, unlike the other two children. Londa’s responses were
distributed as follows: eight responsive answers and five other responses.
Of the eight responsive behaviors, five were target-like and three were
non-target-like. All her target-like responses were to singular requests: two
responses to feminine nouns and three responses to masculine nouns.
Regarding this child’s three non-target responses, all of them involved a
request for more than one object and the child responding by giving one
object, as illustrated in Example 4.35.
Number Agreement 145
Example 4.35
(a) Experimenter: Me puedes dar las grandes. ¿Dónde están las grandes?
‘Can you give me the (fem/pl) big (unm/pl) [ones]?’
Where are the (fem/pl) big (unm/pl) [ones]?
Child: Aquí (pointing at the butterflies). ‘Here.’
Experimenter: Dámelas. ‘Give them (fem/pl) to me.’
Child: Toa (for toma). ‘Take.’
(gave one big butterfly to the experimenter).
Experimenter: ¿Hay más? ‘Are there (any) more?’
Child: Mía for mira. ‘Look’
(gave the other one to the experimenter).
(b) Experimenter: ¿Me das los azules? [Blue elephants]
‘Would you give me the (masc/pl) blue (unm/pl)
[ones]?’
Child: Sí. ‘Yes.’
Experimenter: ¿Dónde están los azules?
‘Where are the (masc/pl) blue (unm/pl) (ones)?’
Child: Aquí. ‘Here’ (gave one blue elephant to the
experimenter).
Experimenter: ¿Hay más?
‘Are there (any) more?’
Child: Aquí. ‘Here’ (gave another blue elephant).
Experimenter: ¿Hay más o no? ‘Are there [any] more or not?’
Child: Sí ira for mira. ‘Yes look’ (gave the last blue
elephant).
the experimenter requested two white cats: ¿Me das dos gatos blancos? ‘Would
you give me two white (masc/pl) cats (masc/pl)?’ Londa gave only one cat
(of the three available) to the experimenter.
The analysis of Londa’s data revealed that she had problems interpreting
the plural requests, that is, out of the three plural requests made all of them
received a singular object as a response. One possibility is to conclude that
this child does not understand the difference between singular and plural.
However, this conclusion would be precipitous in light of her overall pro-
duction. First, Londa produced several target-like plural DPs that referred
to plural referents, as in Example 4.36.
Example 4.36
(a) uno[h] dulce[h] ‘some candies’
(b) una[h] llave[h ‘some keys’
Example 4.37
(a) Child: sita for casita ‘house (diminutive/fem/sg)’
Experimenter: ¿Y quién vive ahí? ‘And who lives there?’
Child: Estos do[h] ‘These (masc/pl) two.’
(b) Child: Mira do[h] pielna[h] ‘Look two legs’
Experimenter: ¿Qué tiene? ‘What does (he) have?’
Child: Do[h] pielna[h]. ‘Two legs.’
Example 4.38
(a) Child: Tá tistre, este tá feli[h], este tá feli[h]
‘[It] is sad, this [one] is happy, this [one] is happy’
Target: Está triste, estos están felices
‘[It] is sad, these [ones] are happy’
(b) Experimenter: ¿Y estos dos?
And these two?’
Child: Este es bebé, bebé
‘This [one] is baby, baby’
Target: Estos son los bebés
‘These two are the babies’
Example 4.39
un cocodrilo, un delfín y una ballena y una ballena
‘a crocodile, a dolphin and a whale and a whale’
148 First Language Acquisition in Spanish
In Example 4.39, Pepe repeats the singular DP una ballena instead of using
the plural form of the DP, dos ballenas ‘two whales’ or unas ballenas ‘some
whales.’
Notice that this type of utterance is available in the target grammar when
the objects to be named are different (e.g., este es el grande y éste es el chiquito
‘this is the big one and this is the small one’) or when the speaker wants to
emphasize the components of a group, for example, este no me gustó, este
no me gustó pero este sí ‘this one I did not like, this one I did not like but this
one I liked.’
Londa’s non-target-like responses might point to a parsing preference in
this child’s early grammar, the distributive interpretation. Miyamoto and
Crain (1991) conducted an experiment the distinction in child grammar
between the collective and the distributive readings of plural NPs, using
24 children between the ages of 3;0 and 6;0. The experiment presented the
children ambiguous situations in two parts to determine if they had access
to the collective and distributive interpretations. In the first situation, two
characters (Ernie and Big Bird) participated in a lifting competition that
ended with each character succeeding in lifting two cans each, that is, the
distributive condition. Then Kermit the Frog described what he thought has
taken place, saying: They are lifting four cans. This statement was false prag-
matically. In the second part, Ernie still wanted to win the competition and
tried to lift four cans by himself but they were too heavy. Then Big Bird
helped him and together they managed to lift the four cans. Once more,
Kermit stated what occurred: They are holding four cans, that is, the collective
condition. Kermit’s statement was pragmatically true. The study found that
children responded affirmatively to the collective interpretation 89 percent
of the time, while they rejected Kermit’s false statement 70 percent of the
time. Miyamoto and Crain concluded that children can access both inter-
pretations. Interestingly, they found that children younger than 5 rejected
Kermit’s false statement in 84 percent of the instances, that is, compared
with 70 percent for all children. They interpret this result as an indication
that younger children might prefer the distributive interpretation over the
collective interpretation. I speculate that this is the case in Londa’s gram-
mar, even in the case of unambiguous plural requests. Another possibility is
that she did not understand her expected response in this particular task.
The discussion of the “Animal House” task findings revealed several
aspects regarding number. First, none of three children had problems com-
prehending the singular number requests. Second, all the non-target-like
responses involved giving a singular object to a plural request. In particular,
Elián seemed to have difficulties understanding the numeral dos ‘two’ but
Number Agreement 149
Responsive 3 3 2 3
Non-responsive – – – –
Total 3 3 2 3
One of the central aims of the present monograph is to define the nature of
nominal in Spanish early grammars, within the framework of the Minimalist
Program (Chomsky 1995, 2001). In this framework, agreement is seen as a
feature checking process, in which functional categories, such as Determiner
Phrases, play a key role. In this chapter I explore the emergence of the
Determiner Phrase in the production of the three children under study
(Elián, Alonso, and Londa) and in doing so, I examine the evidence in sup-
port of the availability of functional projections in Spanish early grammars.
In particular, the analysis will be focused on the data on determiner omis-
sions or Bare Nominals, Monosyllabic Place Holders (Bottari, Cipriani,
& Chilosi 1993/1994) Determiner Phrases, and (full) Determiner Phrases
(DPs), and the interaction among these structures in child language.
In addition, I will address in this chapter the debate about the initial
state, namely, continuous versus discontinuous approaches to language
acquisition. The focus of the discussion will be on the explanatory adequacy
of these two approaches to language acquisition, that is, their ability to
account for the data produced by the children as well as to explain how
these children will converge into the target grammar. Notice that continu-
ous and discontinuous approaches make different predictions regarding
the initial state; and in particular, the availability of functional categories
such as the DP. On the one hand, continuous approaches, specifically the
Weak Continuity Hypothesis, state that children’s grammar is qualitatively the
same as adult’s grammar in that they are both guided by Universal Grammar
(UG). As a result, children’s production may vary from the target grammar
but should not violate any principles of UG (Crain & Pietrosky 2001; Pinker
1984). Crucially, continuous approaches to language acquisition do not
predict instantaneous acquisition because children still have to map their
abstract linguistic knowledge to the primary linguistic data, as pointed by
Valian, Solt, and Stewart (2009). For the present study on the acquisition
of nominal agreement, children are assumed to have the abstract category
152 First Language Acquisition in Spanish
of determiner but they will need to map it to the specification of the Spanish
language, for example, strong agreement features and Case requirements.
Nonetheless, this version of the Weak Continuity Hypothesis predicts
that in the course of the acquisition process children’s non-target-like pro-
duction will be acceptable in a language because they are guided by UG.
Specifically, this hypothesis assumes that children would be as successful
establishing target-like agreement relations with nominals canonically
marked for gender (i.e., masculine nouns ending in the word maker –o,
and feminine nouns ending in the word marker –a), as well as with nomi-
nals non-canonically marked (e.g., nominals ending in the vowel –e or
a consonant), because children are not matching endings but checking
features like the adult grammar does.
On the other hand, discontinuous approaches, such as the usage-based
approach (e.g., Pine & Lieven 1997; Tomasello 2000, 2003) state that chil-
dren start the acquisition process with no language-specific knowledge but
with the general cognitive learning mechanisms. In this approach, children
build or construct their grammar in a piecemeal fashion, based on the reg-
ularities extracted from the input. This hypothesis predicts that children’s
output will match the primary linguistic input because that is the basis of
the grammar they are constructing. Furthermore, children are expected to
extract the regularities available in the input to build their grammar, such
as masculine nominals in Spanish end in the word marker –o and feminine
nominals in the word marker –a. Then, this hypothesis predicts that chil-
dren would be more successful establishing target-like gender agreement
with nominals canonically marked for gender than with nominals non-
canonically marked because of the abundance in the input of examples
exhibiting these agreement patterns, that is, children would be able to
make the connection between these particular endings and their associated
gender in general. We turn the discussion now to the emergence of DP in
the grammar of the children under study.
Studies on the acquisition of Spanish DPs have described the initial stages
of acquisition of this functional category as being characterized by the omis-
sion of the obligatory determiners; the production of a vocalic element in
the slot a determiner would occupy in adult language; and a limited pro-
duction of adult-like DPs (e.g., Aguirre 1995; Hernández Pina 1984; López
Ornat 1997, 2003; among others). Although these acquisition studies mostly
Emergence of DP 153
Example 5.1
Child Target
(a) *Se cayó caballo (Alonso) Se cayeron los caballos
‘Fell (3rd person sg) horse’ ‘Fell (3rd person pl) the (masc/sg)
horses’
(b) *Eso es bebé (Londa) Esos son los bebés
‘That (neut) is baby’ ‘Those (masc/pl) are the (fem/pl)
babies’
(c) *A ve batía (Elián) [Vamos] a ver las baterías
‘To see battery’ ‘(Lets) (to) see the (fem/pl)
batteries’
The Bare Noun Phrases (BNPs) in Example 5.1 are non-target-like for two
reasons. First, the children used a singular noun to refer to a plural refer-
ent, as discussed in Chapter 4. Second, these examples are non-adult-like
because Spanish grammar does not allow BNPs in these particular contexts.
Specifically, in Example 5.1(a) Alonso utters a singular BNP in subject posi-
tion to refer to the two horses that fell down, that is, caballo ‘horse’ instead
of los caballos ‘the horses,’ whereas in Example 5.1(b) Londa identifies the
two small horses that the experimenter is holding as bebé instead of los bebés
‘the (fem/pl) babies (fem/pl).’ Finally, in Example 5.1(c) Elián uses a BNP
to refer to a generic group of objects, batteries. Notice that Spanish marks
generic groups such as batteries, with use of the definite determiner, for
example, las baterías ‘the (fem/pl) batteries (fem/pl).’
In terms of the non-target-like Bare Plurals (BPs) produced, that is, plu-
ral nominals produced without an obligatory determiner, only two exam-
ples were found in the data, as shown in Table 5.4. In these examples Londa
produced a BP in structures that required the production of the definite
determiner, as presented in Example 5.2.
Example 5.2
Child Target
(a) *Flole[h] las flores
‘Flowers (fem/pl)’ ‘the (fem/pl) flowers (fem/pl)’
Emergence of DP 157
Table 5.5 illustrates that the two younger children barely produced any
utterances with an attributive adjective, for a total of three utterances pro-
duced, as seen in Example 5.3.
Example 5.3
The analysis of the utterances in Example 5.3 points to the fact that all of
them involve the omission of an obligatory determiner, as discussed in
Chapter 3. In particular, Example 5.3(a) shows that Elián omitted the
obligatory determiner as he pointed at one orange, for example, mira gande
‘look big’ instead of mira una/la grande ‘look [at] a/the big (one),’ while in
Example 5.3(b), he omitted the determiner el ‘the’ to refer to a doll, el
pobrecito ‘the poor [one].’ Similarly, in Example 5.3(c) Alonso was excited
to see a toy car and produced the attributive adjective mira carito bonito ‘look
pretty car’ instead of mira un carrito bonito ‘look a pretty car’ with the obliga-
tory determiner. In contrast to the limited production of the two younger
children, Londa produced 15 attributive adjective utterances, as shown in
Table 5.5. Interestingly, Londa’s production reflects a pervasive omission of
obligatory determiners, that is, she produced 1 out of the 15 determiners
required. Example 5.4 illustrates some of the utterances produced by this
child.
Emergence of DP 159
Example 5.4
Example 5.5
Nominal Mapping Parameter
(a) N [+ arg – pred] Classifier languages, for example, Chinese
(b) N [+ arg + pred] Germanic languages, for example, English
(c) N [– arg + pred] Romance languages, for example, Italian
Example 5.5 illustrates the three possible values for the NMP. In languages
like Chinese, as shown in Example 5.5(a), NPs are arguments; therefore,
they can occur bare in argument position, whereas Germanic languages
like English, present a mixed system, as shown in 5.5(b). In these languages,
nouns may behave like arguments (e.g., they can occur bare) or predicates
(e.g., they require an overt determiner). Finally, in Romance languages such
as Spanish, NPs are predicates, as shown in 5.5(c). As a result, in Romance
languages, BNs cannot be arguments; that is, nominals need an overt deter-
miner to turn into an argument in these languages (see Chapter 4 for fur-
ther details).
According to this proposal, children would start the acquisition process
with the NMP set to the value of languages like Chinese, hence accounting
for the cross-linguistic phenomenon of determiner omissions in child lan-
guage. In the case of children acquiring a Romance language like Spanish,
they will reset the NMP to the Romance value on the basis of the primary
linguistic input. Notice that this proposal accounts for children’s non-adult
production by acknowledging that it falls within the possibilities allowed by
UG, namely, the value of a parameter set to a different value than the target
language. This proposal has received cross-linguistic support by researchers
Emergence of DP 161
Masculine Feminine
Elián Target-like 3 4 – – 2 2 – – 11
Non-target-like – – – – – – – 1 1
Londa Target-like – 2 – – 2 1 – – 5
Non-target-like – – – – – – – – –
Alonso Target-like 3 3 1 1 2 1 – – 11
Non-target-like 2 1 – 2 – – – – 5
Total 8 10 1 3 6 4 – 1 33
a
–oth includes the unmarked nominal endings, that is, –e and consonants.
Emergence of DP 163
Table 5.8 shows that the children under study produced a majority of
target-like (full) DP utterances with respect to agreement for a total of 89
percent, as discussed in Chapters 3 and 4. These findings bring support to
previous studies claiming that when children acquiring Spanish produce
DPs they do so in a target-like manner (e.g., Aguirre 1995; Schnell de Acedo
1994; Snyder 1995). Moreover, the analysis of the non-target-like utterances
revealed that the majority of non-target-like (full) DP utterances pertained
to a number mismatch involving either the production of a singular DP to
refer to a plural referent, for example, uno dente ‘one/a tooth’ instead
of unos dientes ‘some teeth;’ or marking the plural feature in only one of the
constituents involved in the agreement relation, for example, lo duce ‘the
(pl) candy (sg).’ As discussed in Chapter 4, this second type of non-target-
like production may be a reflection of the phonological factors present in
the linguistic context of the children, namely the aspiration final –s.3 These
results are consistent with those obtained in the analysis of both MPH/DPs
and Bare Nominals, that is, non-target-like production pertained to the
number feature.
Regarding the nature of the initial grammar, the production of (full) DPs
in an almost all target-like fashion, brings support to a continuous approach
to language acquisition, that is, the availability of a checking mechanism in
children’s grammars from the initial stages of acquisition. Furthermore, as
in the case of MPH/DPs, the analysis found that children establish target-like
gender agreement in (full) DP structures regardless of nominal endings.
Specifically, out of a total of 62 (full) DP tokens (once repetitions were
eliminated) produced, 35 involved nominals not overtly marked for gender
(i.e., nouns ending in the vowel –e or a consonant) and all but one were
target-like in terms of gender agreement. This points to the fact that the
three children under study were as likely to create target-like gender agree-
ment with overtly (gender) marked nominals as well as with non-overtly
marked nominals (χ2 = 1.03, p = 0.31). These findings provide additional
evidence in support of the hypothesis that children are not simply match-
ing nominal endings but they are checking the abstract agreement features
of the nominals and their agreeing determiners and modifiers.
Emergence of DP 165
Example 5.6
[DP la [AGRP AGR [FP bonita F [NP casa]]]]
‘the (fem/sg) pretty (fem/sg) house (fem/sg)’
advanced developmental stage than the other two children under study.
Evidence for this conclusion is found in her limited production of MPH/
DPs (8 percent) and her higher production of (full) DPs. Nonetheless,
Londa has the highest rate of BNPs, or determiner omissions in obligatory
contexts. Notice that Londa’s overall production seems contradictory at
first sight. On the one hand, her production seems to indicate that she is in
the process of mastering nominal agreement with an almost all target-like
production of (full) DPs and a variety of determiners used. On the other
hand, her high rate of determiner omissions is at odds with her overall lin-
guistic development. This issue is addressed in Section 5.2.
Note that usage-based approaches will have difficulties explaining the
production distribution in Table 5.9. On the one hand, children are pro-
ducing target-like utterances, but on the other hand, there is a high rate of
omissions. Moreover, a quick review of the data shows that children’s pro-
duction of one nominal can alternate among the three possibilities shown
in Table 5.9, that is, BNP, MPH/DP and (full) DP. Example 5.7 illustrates
Alonso’s production alternation of the nominal caballo.
Example 5.7
(a) el caballo ‘the horse’
(b) e caballo ‘the (MPH) horse’
(c) *caballo ‘horse’
(d) *a caballo ‘a (MPH) horse’
(Full) DPs – 1 – 1
MPH/DPs – (4)c – (4)
Attributive adjectives – – – –
Predicative adjectives 10 – – 10
Demonstrative pronouns 13 2 4 19
Third person clitics 8 1 – 9
Total 31 8 4 43
a
Feminine refers to the use of feminine marked constituents to refer to masculine nominals.
b
Neuter refer to the use of the neuter demonstrative pronoun to refer to masculine nominals.
c
These 4 tokens are between parenthesis to mark that they were not considered gender mismatches as such
but place holders void of any feature content.
170 First Language Acquisition in Spanish
Conclusions
The main goal of this monograph was to explore the nature of nominal
agreement in Spanish early grammars, that is, to examine the gender and
number agreement relations children establish (or fail to establish) in the
Determiner Phrase, within the framework of the Minimalist Program
(Chomsky 1995, 2001). At the heart of this exploration, is the controversy
about the initial state or the continuity versus the discontinuity approaches
to language acquisition, that is, whether there is evidence in support of
the availability of an underlying feature checking mechanism like the one
posited for the adult grammar (Continuity) or there is evidence that chil-
dren are just matching the input heard because they lack any a priori
linguistic specifications (Discontinuity). With these goals in mind, a set
of experimental tasks were administered to three monolingual, Spanish-
speaking children under the age of 3 (Elián, Alonso, and Londa) in San
Juan, Puerto Rico. Utterances involving a noun, a determiner or an adjec-
tive were included in the analysis. In addition, predicative adjectives and
demonstrative and clitic pronouns were also analyzed in the present study
to obtain a more complete picture of the agreement system in early gram-
mars. This inquiry yielded answers to some of the questions posed yet it
raised several questions for future research.
In regard to the nature of nominal agreement, the analysis of the produc-
tion of the three children under study produced three major generaliza-
tions: (1) Children’s production was mostly target-like in terms of agreement;
(2) Non-target-like production followed a principled selection pattern of
initial default values for the gender and number features; and (3) A delay
in the acquisition of number was attested in the data analysis. Regarding
the first generalization, the three children under study produced a signifi-
cantly higher number of target-like utterances (83 percent) with regard to
agreement (χ2 = 43.56, p < 0.0001) than non-target-like, despite the differ-
ences in Mean Length of Utterance in words among them: Elián, MLUw =
1.5; Alonso, MLUw = 1.9; and Londa, MLUw = 2.2. Notice that two of the
Conclusions 177
children, Elián and Alonso, are below the two-word stage in their produc-
tion, pointing to the fact that children acquiring Spanish are able to estab-
lish target-like agreement relations from the early stages of the acquisition
process. In particular, Table 6.1 illustrates that the three children’s overall
production was predominantly target-like across all the structures under
examination, with target-like percentages ranging from a high of 94 per-
cent for the production of attributive adjectives to a low of 73 percent for
the production of third person direct object clitics.
This finding confirms previous acquisition studies stating that the per-
centage of non-target-like production of children acquiring Spanish is lim-
ited (e.g., Aguirre 1995, Schnell de Acedo 1994, Snyder 1995, 2007 for
Spanish; Pizzuto & Caselli 1992 for Italian).
Furthermore, children’s overall target-like production seems to indicate
the availability of a feature-checking mechanism in Spanish early gram-
mars. Support for this conclusion is found in the results of a detailed analy-
sis on the agreement patterns children established with respect to nominal
word markers. Specifically, the data analysis showed that children were able
to establish target-like agreement relations with nominals canonically
marked for gender (i.e., masculine nominals ending in –o and feminine
nominals ending in –a), as well as with nominals non-canonically marked
for gender (e.g., nominals ending in the vowel –e or a consonant).1 These
findings bring support to the Weak Continuity Hypothesis (Crain & Pietrosky
2002; Crain & Thornton 1998, Pinker 1984, among others), indicating the
availability of a feature-checking mechanism in initial grammars, that is, the
children under study are not merely matching the input but they are check-
ing features as hypothesized for the adult grammar. For the present mono-
graph, I adopted Cinque’s (1994) and Brugè’s (2002) N-raising analyses for
Determiner Phrases and proposed that children, like adults, project inter-
mediate functional categories in which the strong nominal agreement
features are checked.
In contrast, these findings do not support a usage-based approach to
acquisition in which children will match the input they hear. In particular,
if children are expected to build categories by extracting regularities from
the input, then this approach predicts that they will be more successful
establishing target-like agreement relations with canonically marked nomi-
nals than with non-canonically marked nominals, given the regular pattern
these nominals exhibit with respect to agreement. This prediction was not
borne out by the data.
Further support for the availability of a checking mechanism in Spanish
early grammars is found in children’s production of MPH (Bottari et al.
1993/1994) DPs, that is, nominals with a vocalic element in the place a
determiner occupies in adult language. The examination of these vocalic
elements showed that they function as reduced forms of the (full) deter-
miners with which the children established target-like agreement relations,
especially in regard to gender. Notice that children were not matching
MPHs with nominal endings, as predicted in a usage-based approach, but
they established target-like agreement relations regardless of the nominal
ending, that is, canonical (ending in –o for masculine and –a for feminine)
versus non-canonically marked (ending in the vowel –e or consonant). This
finding sets the beginning of the acquisition of DP in Spanish at an earlier
stage than the one reported in the literature reviewed (Hernández Pina
1984; Schnell de Acedo 1991; Snyder 1995), as pointed out by Guasti et al.
2008 for the acquisition of Italian. Notice that studies by Hernández Pina
(1984) and López Ornat (1997, 2003) have analyzed these vocalic elements
as amalgams or unanalyzed segments void of any feature content. As a
result, they have not been considered as the starting point in the acquisi-
tion of Spanish nominal agreement.
Moreover, the production of MPHs at the beginning of the acquisition
process instead of (full) DP forms explains why once children start to pro-
duce (full) Determiners, they do it in a target-like fashion; that is, they have
already worked out the morphological agreement markings with the MPHs.
Finally, the production of MPH/DPs and their status as reduced versions of
the full determiners with feature content, brings support to the hypothesis
that children have an awareness of functional categories before they have
learned the full phonological inventory of Determiner forms.
In terms of the second generalization pertaining to the availability of ini-
tial default values for the gender and number features, the analysis revealed
Conclusions 179
(full) DPs – 1 – 1
MPH/DPs – (4)a – (4)
Attributive adjectives – – – –
Predicative adjectives 10 – – 10
Demonstrative pronouns 13 2 4 19
Third person clitics 8 1 – 9
Total 31 (72%) 8 (19%) 4 (9%) 43 (100%)
a
These four instances involved the production of the MPH a with masculine nominals. The analysis
found out that these did not constitute instances of gender mismatches but the production of a
true MPH, as proposed by Bottari et al. (1993/1994).
180 First Language Acquisition in Spanish
(full) DPs 21 7 8 15
MPH/DPs 9 3 5 8
Attributive adjectives 6 1 5 6
Predicative adjectives 6 2 4 6
Demonstratives 11 – – –
Third person clitics 5 1 2 3
Total 58 14 24 38
a
Grammatical non-target-like number refers to a mismatch of the number between the nominals and the
agreeing constituents.
b
Semantic non-target-like number refers to a mismatch between the number feature of the DP and the
number of objects it refers to.
Conclusions 181
particular DP and the number of objects it refers to). Overall, Table 6.3
shows that out of a total of 58 tokens required to be marked as plural,
only 20 (34 percent) were marked for plurality in a target-like fashion; the
remaining 38 (66 percent) tokens involved a number mismatch, semantic
or grammatical. This brings support for the availability of a default value for
the number feature that surfaces as the singular zero mark. In addition, the
present research explored the evidence in support of the availability of
the NMP at work in the grammar of the children under study. The examina-
tion of the non-target-like production with regard to the number feature
provides some support in favor of this parameter. In particular, the avail-
ability of this parameter accounts for children’s production of unmarked
plurals (i.e., singular nominals to refer to plural referents) across all the
structures under study, that is, children are interpreting nominals as
kind-denoting and therefore they cannot be pluralized. The production of
Bare Nominal Phrases by the three children under study brings additional
support for the availability of the NMP. Specifically, the three children
omitted obligatory determiners in 64 percent of the required contexts.
According to this parameter, determiner omissions are expected in the
acquisition process as children set this parameter to a value different than
the language of the linguistic community. In particular, the data analysis
provided evidence that the children under study had set the NMP to the
[+ argument, + predicate] value, as in languages like English in which Bare
Nominals and DPs are allowed, according to their lexical specifications.
As a result, the children were producing target-like DPs, along with Bare
Nominals.2 The availability of this parameter brings support to the Weak
Continuity Hypothesis by accounting for children’s non-adult production
as one of the options provided by UG, namely, the setting of a parameter to
a value different from the linguistic community. Crucially, according to this
analysis, children will converge to the target language once they reset the
parameter to the Romance value [– argument, + predicate] on the basis of
the primary linguistic input.
Notice that usage-based approaches to acquisition will have a difficult
time explaining children’s target-like production alongside with omission
of determiners in obligatory contexts. On the one side, one could state that
children have learned the Det-N pairings in the case of the target-like DP
production. On the other side, it would be difficult for this approach to
explain the variability found in the production data in which one particular
nominal could appear bare (e.g. *Caballo ‘Horse’), in a (full) DP (e.g.
Un caballo ‘A horse’ or in a MPH/DP E caballo ‘E horse.’ In contrast these
alternations are accounted for as options available in UG, or as Crain and
Pietroski (2002) explained it, children are speaking a different language.
182 First Language Acquisition in Spanish
tokens with a plural referent was marked for plurality.3 In addition, the
examination of the attributive adjective production follows the same pat-
tern as the (full) DPs and MPH/DPs. In particular, there were no gender
mismatches; and six out of six non-target-like utterances involved a number
mismatch. Finally, the data on Bare Nominal Phrases (i.e., nominals pro-
duced in isolation) provide further support for a delay in the acquisition
of the number feature. Specifically, the review of the BNPs produced
showed that out of the 87 nominals that refer to a plural referent, only 21
(24 percent) were marked overtly as plural. This indicates that the contrast
in the acquisition of number and gender go beyond grammatical agree-
ment. These results bring support to previous research in Spanish that have
stated that gender is acquired before number (e.g., Hernández Pina 1984;
Marrero & Aguirre 2003; Schnell de Acedo 1994; Snyder 1995).
In the present study I have proposed that part of the reason for this delay
is semantic in nature, that is, the complex process of matching morphologi-
cal features on the DP with the features of the referent, for example, la casa
‘the house’ with “one house.” In order to assess this hypothesis a detailed
analysis of utterance referent was conducted and the results are illustrated
in Table 6.3 above. To my knowledge this is the first attempt to address the
semantic aspect of number acquisition in Spanish early grammars. Table 6.3
shows that the three children under study produced more semantic num-
ber mismatches (63 percent), that is, the use of a singular DP to refer to a
plural entity, than grammatical mismatches (37 percent), a mismatch of the
number feature within the DP. This difference was found significant (χ2 =
5.58, p = 0.018). These results provide strong support for the hypothesis
that the delay on the acquisition of the number feature has its basis on the
semantic or referential aspect of this feature. Furthermore, the adoption of
the NMP entails that children need to reset this parameter to the Romance
value, and in doing so, they have to recategorize nominals from arguments
to predicates on a one by one basis (Chierchia et al. 2000). This process
adds another layer to the complexity involved in the acquisition of number.
The present research inquiry has pointed out several areas of further
research. First, regarding gender, research is needed on the role of phono-
logical cues in gender assignment. For the present study, these cues did not
seem to play a significant role. This raises the question of at what point in
the acquisition process children are ready to make use of this information.
Notice that Pérez-Pereira (1991) found that children between the ages of
4 to 11 were using intralinguistic cues for gender assignment.
Second, further research is needed in the area of acquisition of adjectival
agreement. In particular, the present study had difficulty collecting enough
184 First Language Acquisition in Spanish
Elián
Alonso
Londa
Elián
Alonso
Londa
Elián
Alonso
Londa
Elián
Alonso
Non-Target Tokens
Londa
Non-Target Tokens
Elián
Alonso
Londa
Non-Target Tokens
Elián
Alonso
Londa
(Tokens)
Masculine APs
Feminine APs
26. No, ita No (es) chiquita [mariposa]
‘No, small (fem/sg)’
27. está tistre está triste [luz]
‘(it) is (chang) sad (un/sg)’
28. es tistre está triste [luz]
‘(it) is (perm) sad (un/sg)’
29. tá feli[h] está feliz [luz]
‘(it) is (chang) happy (un/sg)’
30. No, tá sucia No, está sucia [nena]
‘No, (it) is (chang) dirty
(fem/sg)’
31. tá inda está linda [muñeca]
‘(it) is (chang) pretty (fem/sg)’
32. tás fea estás fea [experimenter]
‘(you) are (chang) ugly
(fem/sg)’
Appendices 205
Non-target APs
44. rojo (es) roja *[paleta]
‘red (masc/sg)’ ‘(it) is (perm) red
(fem/sg)’
45. rojo (es) roja *[motora]
‘red (masc/sg)’ ‘(it is (perm)) red
(fem/sg)’
46. etá bielto está abierta *[boca]
‘(it) is (chang) opened ‘(it) is (chang)
(masc/sg)’ opened (fem/sg)’
206 First Language Acquisition in Spanish
Preface
1
Notice that the term ‘default’ is used in this instance to refer to a language-specific
initial feature value, for example, Harris’ (1991) proposal of masculine and singu-
lar as default values for Spanish.
Chapter 1
1
For the present analysis, I have chosen the masculine and singular to label the
nominal features gender and number for expository purposes, because Spanish
only has two values for these features. The same analysis could be applied to lan-
guages with more than one gender by simply labeling these features as [gender,
number].
2
Farkas (1990) proposes a similar analysis based on the underspecification of the
gender feature for neuter gender nominals in Romanian.
3
Some nominals in Spanish are not morphologically marked for a specific gender,
for example, el/la estudiante (the (masc/fem) student). For these cases, I assumed
that they come from the lexicon already specified for the gender feature as either
masculine or feminine and that this specification becomes transparent when they
enter into an agreement relationship in the derivation.
4
See Bruening (2009) for arguments against the DP-Hypothesis.
5
I should point out that structures in this monograph are illustrated as projecting
X-bar levels for expository purposes. However, under the bare phrase structure
(Chomsky 1995), there are no bar levels.
6
As we have seen, functional categories are a vital component in the Minimalist
framework because of their presumed role in feature checking. Consequently,
several intermediate functional projections have been proposed in the literature
to account for gender and number agreement within DP. Among them, Ritter
(1993) proposes a non-unified treatment of gender and number features using a
comparative analysis of Hebrew and three Romance languages, including Spanish.
In her proposal, number projects its own functional phrase that serves as the com-
plement of D, whereas gender is a feature whose position varies cross-linguistically.
Similarly, Bernstein (1991, 1993) finds support for the availability of a Number
Phrase as the intermediate projection between the Determiner Phrase and the
Nominal Phrase in her analysis of prenominal adjectives in Walloon, a Romance
language spoken in Belgium. In contrast, Picallo (1991) states that both gender
and number should project their own functional projections following Pollock’s
208 Notes
(1989) Functional XP Hypothesis. Using data from Catalan DPs, she proposes a
cyclical derivation of nominal agreement, involving N-raising to Gender and
then to Number.
7
Implicit in this analysis is the assumption that the base order of these constituents
is [determiner + adjective + noun], as has been advanced in the literature, among
others by Bernstein (1991), Valois (1991), Picallo (1991), and Cinque (1994).
8
We should point out that word-order relations between the noun and the
adjective are more flexible in Spanish than in English. For example, qualifying
adjectives, such as those referring to colors, can occur also in prenominal position
in Spanish, for example, rojo abrigo ‘red coat,’ as well as in other Romance lan-
guages, such as Italian and French. However, as Demonte (1999) explains, color
adjectives in prenominal position carry a change in the meaning of the nominal
phrase: the use of prenominal color adjectives is literary, for example, in this case
it stresses the redness of the coat.
9
Other interesting variants of the weak continuity variation of the Continuity
Hypothesis are the Structure Building Model (Radford 1995, 2000) and the
Lexical Learning Approach (Clahsen, Parodi, & Penke 1993/1994; Clahsen,
Eisenbeiss, & Penke 1996). These two hypotheses argue that children are guided
by UG principles as they build syntactic structures but the process is a gradual
one based on lexical learning and the interaction of abstract knowledge. These
hypotheses, although interesting, will not be discussed in the present monograph
as the focus is the continuity versus the usage-based approaches controversy.
10
See Bohnacker (1997) for a discussion of Radford’s Maturational Hypothesis.
11
Yang (2004) presents evidence against these analyses, stating that this statistical
mechanism cannot segment sequences of monosyllabic words, which are abun-
dant in the input directed to children acquiring English.
12
See Crain and Thornton (2006) for a complete discussion of the inadequacies of
data-driven hypotheses.
Chapter 2
1
A total 15 subjects participated in the study but not all of them completed all
tasks due to illness. Another cause for data loss was equipment malfunction.
2
The original values of the MLU in words were calculated following Brown’s
(1973) criteria, that is, all repetitions were included in the calculation. The values
found were as follows: Elián (MLUw=1.6); Alonso (MLUw= 2.3); Londa (MLUw=
2.4); Diana (MLUw= 3.2); and Pepe (MLUw= 4.3). However, a second calculation
was carried out eliminating all repetitions from the data resulting in the lower
MLUw values shown in Table 2.1 above. This second calculation was motivated by
the fact that Alonso’s prevalent repetitions altered considerably his MLUw from a
value of 2.3 to a value of 1.9.
3
A revision was made regarding the adjectives included in the tasks. In the design-
ing phase of the study adjectives listed in (i) were selected from the Spanish
transcripts in the CHILDES database (MacWhinney & Snow 1985):
(i) (a) Size:
Chiquito ‘small’ (marked for gender, final –o or –a)
Grande ‘big’ (unmarked for gender, –e)
Notes 209
(b) State:
Mojado ‘wet’ (marked for gender)
Seco ‘dry’ (marked for gender)
Limpio ‘clean’ (marked for gender)
Sucio ‘dirty’ (marked for gender)
Peinado ‘combed’ (marked for gender)
Despeinado ‘uncombed’ (marked for gender)
(c) Emotion:
Feliz ‘happy ‘(unmarked for gender)
Triste ‘sad’ (unmarked for gender)
Only three adjectives unmarked for gender were found to be appropriate to use
for children these ages, grande ‘big,’ feliz ‘happy’ and triste ‘sad.’ Of these three
adjectives, all children recognized the contrast between grande/chiquito ‘big/
small’ consistently in the props, for example, a big phone versus a small phone.
The other two adjectives, feliz and triste, seemed to be obvious for the girls but not
so obvious for the boys. Similarly, we found gender differences regarding the
perception of the cleanliness of a toy, while boys seemed not to worry about
the cleanliness of a particular toy, girls, would immediately voice their concerns
about the dirty toy. The adjectives mojado/seco ‘dry/wet’ were eliminated because
it was not obvious to the children that a particular item was wet.
4
The use of these articles did not seem to have an effect on the children’s produc-
tion, since their answers would vary many times from the information provided,
as see in (i):
(i) Experimenter: Tengo los grandes
‘I have the (masc/pl) big (unm/pl) (ones) (two cars)’
Child: *Y yo chiquito
‘And I small (masc/sg) (one) (two cars)’
Moreover, within the UG framework the underlying assumption is that children
do not acquire grammar by imitating the input but by testing different hypothe-
ses on how the system works. These responses were not cases of immediate
repetitions.
Chapter 3
1
Demonte (1999) states that qualifying adjectives can be classified in several types:
(1) dimension/size, for example, largo/corto ‘long/short,’ (2) velocity or speed,
for example, rápido/lento ‘fast/slow;’ (3) physical characteristic, for example,
redondo/cóncavo ‘round/concave;’ (4) color and shape: for example, blanco/alar-
gado ‘white/elongated;’ (5) age: for example, antiguo/moderno ‘old/modern;’ (6)
valuation or evaluative: for example, excelente/horrible ‘excellent/horrible;’ and
(7) human aptitudes and predispositions: for example, capaz/emotivo ‘capable/
emotional.’
2
Recall from Chapter 1, I am adopting Cinque’s (1993) adjectival analysis, in
which adjectives are generated in prenominal position as specifiers of the func-
tional projection Agreement Phrase. In this analysis, the postnominal position of
the adjective in Spanish is the result of N-raising to check strong features.
210 Notes
3
Spanish gender paradigm assumes that there are two genders (masculine and
feminine); however, there is a third possibility: neuter. This third gender is overtly
expressed in the following five forms: the demonstratives esto ‘this,’ eso ‘that,’
aquello ‘that,’ and the pronouns ello ‘it,’ and lo ‘it.’ These neuter forms are prob-
lematic because they stand in isolation from the rest of the constituents; that is,
there are no other constituents in Spanish marked as such. In addition, when
these “neuter” forms surface with other constituents, such as nouns or adjectives,
they take the agreement markings of the masculine gender, for example, lo buen–o
‘the (neuter) quality of good.’ Researchers seem to agree that these forms do not
constitute a third gender in Spanish (Harris 1991; Klein 1989; Leonetti 1999,
among others). Several proposals have been made to account for these anoma-
lous members of Spanish gender paradigm. Harris affirms that neuter can only
be considered a subgender, because there are no neuter nominals in Spanish.
Klein (1989) proposes to analyze neuter as abstract reference, whereas Ojeda
(1984) proposes a reclassification of these elements as [–count]. Leonetti (1999)
states that the neuter forms do not oppose the rest of the paradigm morphologi-
cally but by a semantic feature, that is, their ability to denote only inanimate
entities or their ability to refer to non-countable referents.
4
Harris’ word classification is motivated by several factors: (1) numerical prepon-
derance: he states that the vast number of nouns, adjectives, and adverbs belong
to the core; (2) productivity: inner and outer cores are productive and accept loan-
words and other types of neologisms; and (3) historical shift: over time residue
constituents have moved to the core.
5
Phillips assumes that case-licensing is a rigid constraint, that is verb-raising must
take place. The author further assumes that case-licensing is regulated by the
Generalized Visibility Condition (Baker 1991; Shlonsky 1987), which states that pho-
netically interpreted NPs are case-licensed by PF but semantically interpreted
NPs are case-licensed by LF.
6
Furthermore, it is not uncommon for children in the beginning stages of acquisi-
tion to shorten words; in this case mota could just be a shortened version of the
motora.
7
Notice that Schnell de Acedo also analyzes in her study the percentage of deter-
miner omissions in obligatory contexts, a variable that was not taken into account
in the studies reviewed. I address the issue of determiner omission in Chapter 5.
8
No major conclusions can be drawn with respect to agreement in attributive adjec-
tival structures given the reduced number of utterances present in the data.
9
I should point out that the use of “default” with respect to the masculine gender
feature represents the most frequent or least marked value of this feature in the
Spanish language (Harris 1991; Teschner & Russell 1984; Trask 1993). In this
sense, this is a language-particular characteristic and as such, it is not specified
within UG.
10
The analysis of the data throughout the discussion is presented in the format of
utterances, which include all instances produced by the children, and tokens,
which reflect children’s production without repetitions.
11
This conclusion is based under the assumption MPH a has encoded the feminine
features.
12
I interpreted the long sound of the e as a shortened form of the copula es ‘is’ and
not está, because Alonso produced the form tá for está consistently in the data.
Notes 211
In addition, Example 3.22(e) provides evidence for the presence of this alterna-
tion between ser and estar, that is, Alonso produces son sucio instead of están sucios.
These two examples seem to indicate that this child has not yet acquired the dis-
tinction between these two representations of the copula ‘to be.’
13
Another possibility is that the child was referring to all the cats available and he did
not pronounce the plural marker, namely the final s, for example, esos ‘these (ones)’
(masculine/plural). In this case the paradigm is regular, with the masculine plural
demonstrative marked canonically with os, for example, esos ‘these (ones).’
Chapter 4
1
Marrero and Aguirre state that children in their study produce singulars to refer
to plural referents at the pre-morphological stage. However, they do not provide
a clear assessment of how pervasive this issue was in the production data.
2
Notice that in this analysis individuals and quantifiers are considered argumental
because of their association with constituents occupying argument positions in
clauses, for example, subject, object.
3
The researchers argue that the initial default value is the Chinese one because it
is the more restrictive in the sense that this type of language lacks determiners,
numerals, and plural morphology.
4
Rigau (1999) states that in Spanish the partitive article formed by [de + definite
article] as in Italian and French, was lost. She argues that Spanish does not pos-
sess a partitive explicit determiner, that is, the partitive determiner appears as an
article without phonological content.
5
The plural las gramas ‘the grasses’ is available in the Spanish grammar but it is
used to refer to different types of grass, for example, Kentucky grass versus blue
grass. In this particular example the child is not referring to different kinds of
grass.
6
Recall that according to the Nominal Mapping Parameter (Chierchia 1998a;
1998b; Chierchia et al. 2000), children start the acquisition process with the
parameter set to the Chinese language value; that is, they would treat all nomi-
nals as mass. Hence, this parameter predicts that at the initial stages of acquisition,
children will not pluralize any nominals (e.g., Chinese), nor will they produce
any determiners since nouns in this setting are arguments. Then, the learner
would reset the NMP to the value of Germanic languages on the basis of positive
evidence; that is, nominals would be treated as both arguments and predicates.
In this stage, the NMP predicts that children would produce nominals with and
without determiners as they start classifying them as count or mass. Finally, chil-
dren acquiring a language like Spanish would converge to the Romance target
setting in which the nominals are predicates; that is, nominals in general need to
have an overt determiner to become arguments.
7
Notice that this conclusion contrasts with our findings on gender, that is, chil-
dren exhibited no problems marking gender inflection on nominals. This brings
support to the hypothesis that the nominals come from the lexicon with their
gender morphology specified.
8
This result does not support Pizzutto and Caselli’s (1992) finding for Italian
regarding bound morphology. In particular they found that the acquisition
212 Notes
Chapter 5
1
Another proposal put forth in the literature to account for the phenomenon of
determiner omission in child language is the Number Underspecification Hypothesis
(Hoekstra & Hyams 1995; Hoekstra, Hyams, & Becker 1997). Proponents of the
Underspecification Hypothesis state that determiner omissions in early grammars
result from the optional underspecification of Number. In this analysis, articles
are the realization of the feature number: when this feature is specified (e.g.,
[+ sg, –pl]), then D is filled by an overt article, but when it is left underspecified,
D would be empty and children would produce Bare Noun Phrases. Crucially
within this hypothesis, D is always projected in child language. This hypothesis
presents the advantage of assuming that children’s grammars always project a DP,
equating them with adults’ grammars. However, it does not explain why children
leave number underspecified, nor does it explain the process of convergence on
the target grammar.
Notes 213
2
Notice though that Marrero and Aguirre’s (2003) study found similar results
even though some of the children in their study spoke a dialect with no final –s
aspiration.
3
The remaining four instances involved Alonso’s production of the MPH (Bottari
et al. 1993/1994) a (considered to mark the feminine gender in the present
analysis) with masculine nouns. However, the analysis presented in Chapter 3
provided evidence that these four cases did not involve gender mismatches;
rather, they were instances of vocalic place holders, void of any featural content.
4
The remaining four instances involved Alonso’s production of the MPH (Bottari
et al. 1993/1994) a (considered to mark the feminine gender in the present
analysis) with masculine nouns. However, the analysis presented in Chapter 3
provided evidence that these four cases did not involve gender mismatches;
rather, they were instances of vocalic place holders, void of any featural content.
Chapter 6
1
This points to the fact that children at the beginning stages of acquisition do not
make use of the morphophonological cues available in the input, contrary to Pérez-
Pereira’s (1991) findings for children between the ages of 4 and 11. The children
in his study paid attention to intralinguistic cues in the assignment of gender.
2
Although Elián’s production can be characterized as singular, he seems to have
set the NMP parameter to the same value as Alonso and Londa because he is
producing target-like MPH/DPs.
3
Notice that I analyzed the four gender mismatches found in the MPH/DP pro-
duction not as the feminine MPH a but as true place holders without a gender
specification. As a result, they do not involve a gender mismatch.
Appendices
1
In the examples above the following abbreviations were used:
masc masculine
N/A Not applicable
fem feminine
un unmarked overtly for gender
sg singular
pl plural
2
The ‘F’ indicates that this nominal has the canonical mark of feminine nouns
even though it is masculine.
3
The ‘M’ indicates that this nominal has the canonical mark of masculine nouns
even though it is feminine.
4
The copula ‘to be’ has two representations in Spanish, ser and estar which serve to
mark the contrast between an attribute perceived as permanent (or individual
predicate) in the case of the former, versus one perceived as changeable (or a
stage level predicate) in the case of the latter. In the text the abbreviations ‘chang.’
and ‘perm.’ refer to changeable attribute and permanent attribute respectively.
Bibliography
Abney, S. P. 1987. The English noun phrase in its sentential aspect. Ph.D. diss.,
Massachusetts Institute of Technology.
Aguirre, C. 1995. La adquisición de las categorías funcionales en el español.
Doctoral diss., Universidad Autónoma de Madrid.
Alcina Franch, J., and J. M. Blecua. 1983. Gramática Española. Barcelona: Editorial
Ariel.
Ambadiang, T. 1999. La flexión nominal. Género y número. In Gramática Descriptiva
de la Lengua Española, eds I. Bosque and V. Demonte, 4843–913. Madrid: Espasa.
Aronoff, M. 1994. Morphology itself: Stems and inflectional classes. Cambridge, MA: MIT
Press.
Baker, C. L., & J. McCarthy. Eds. 1981. The logical problem of language acquisition.
Cambridge, MA: MIT Press.
Baker, M. 1991. On some subject-object non-asymmetries in Mohawk. Natural
Language and Linguistic Theory 9: 537–76.
Behrens, H. 2006. The input-output relationship in first language acquisition.
Language and Cognitive Processes 21 (1): 2–24.
Bernstein, J. 1991. Nominal enclitics in Romance. MIT Working Papers in Linguistics
14: 1–66.
—1993. Topics on the syntax of nominal structures across Romance. Dissertation
Abstracts International 54 (09A): 3418 (UMI No. 9405500).
Bloom, P. 1990. Subjectless sentences in child grammar. Linguistic Inquiry 21:
491–504.
—1993. Grammatical continuity in language development: The case of subjectless
sentences. Linguistic Inquiry 24: 721–34.
Bohnacker, U. 1997. Determiner phrases and the debate in functional categories in
early child language. Language Acquisition 6: 49–90.
Borer, H., & B. Rohrbacher. 1997. Features and projections: Arguments for the full
competence hypothesis. In BUCLD 21st Proceedings, eds. E. Hughes, M. Hughes,
and A. Greenhill, 24–35. Somerville, MA: Cascadilla Press.
Bosque, I. 1999. El nombre común. In Gramática descriptiva de la lengua española, eds
I. Bosque & V. Demonte, 3–75. Madrid: Espasa.
Bosque, I., & C. Picallo. 1996. Postnominal adjectives in Spanish. Journal of Linguis-
tics 32: 349–85.
Bottari, P., P. Cipriani, & A. M. Chilosi. 1993/1994. Protosyntactic devices in the
acquisition of Italian free morphology. Language Acquisition 3: 327–69.
Brown, R. 1973. A first language. Cambridge, MA: Harvard University Press.
Bruening, B. 2009. Selectional asymmetries between CP and DP suggest that the
DP-Hypothesis is wrong. University of Pennsylvania Working Papers in Linguistics
15 (1): 28–34.
Bibliography 215
Crain, S., A. Gualmini, & P. Pietroski. 2005. Brass tacks in linguistic theory: Innate
grammatical principles. In The innate mind: Structure and contents, eds. P. Carruthers,
S. Laurence, & S. Stich, 175–97. New York: Oxford University Press.
Crain, S., & K. Wexler. 1999. Methodology in the study of language acquisition:
A modular approach. In Handbook of child language acquisition, eds. T. Bhatia and
W. C. Ritchie, 387–425. San Diego, CA: Academic Press.
Demonte, V. 1999. El adjetivo: Clases y usos. La posición del adjetivo en el sintagma
nominal. In Gramática descriptiva de la lengua española, eds. I. Bosque and
V. Demonte, 175–213. Madrid: Espasa.
Deprez, V. 1994. Underspecification and functional projections, and parameter
setting. In Syntactic theory and first language acquisition 1, eds. B. Lust, M. Suñer
and J. Whitman, 249–71. Hillsdale, NJ: Lawrence Erlbaum Associates.
Dolitsky, M. 1983. The birth of grammatical morphemes. Journal of Psycholinguistic
Research 12: 353–60.
Farkas, D. 1990. Two cases of underspecification in morphology. Linguisitic Inquiry
21: 539–50.
Fernández Leborans, M. J. 1999. La predicación: Las oraciones copulativas.
In Gramática descriptiva de la lengua española, eds. I. Bosque and V. Demonte,
2357–460. Madrid: Espasa.
Fodor, J. D. 1998. Parsing to learn. Journal of Psycholinguistic Research 27: 339–74.
Goodluck, H. 1991. Language acquisition: A linguistic introduction. Cambridge, MA:
Blackwell.
Gordon, P. 1982. Acquisition of syntactic categories: the case of count/mass distinc-
tion. Ph.D. diss., Massachusetts Institute of Technology.
Gualmini, A., L. Meroni, & S. Crain. 2003. An asymmetric universal in child
language. Proceedings of Sinn und Bedeutung VI, 136–48. Konstanz, Germany:
Arbeitspapiere des Fachbereichs Sprachwissenschaft.
Guasti, M. T., M. Joke de Lange, A. Gavarró, & C. Caprin. 2008. Article omission
across child languages. Language Acquisition 15 (2): 89–119.
Gutiérrez Ordóñez, S. 1999. Los dativos. In Gramática descriptiva de la lengua española,
eds. I. Bosque and V. Demonte, 1855–928. Madrid: Espasa.
Haegeman, L. 1995. Root infinitives, tense and truncated structures. Language
Acquisition 4: 205–55.
Harris, J. W. 1991. The exponence of gender in Spanish. Linguistic Inquiry 22:
27–62.
Harris, T., & K. Wexler. 1996. The optional infinitive stage in child English:
Evidence from negation. In Generative perspectives on language acquisition, ed.
H. Clahsen, 1–42. Amsterdam: John Benjamins.
Hernández-Pina, F. 1984. Teorías psicosociolingüísticas y su aplicación a la adquisición del
español como lengua materna. Madrid: Siglo XXI.
Hoekstra, T. N., & N. Hyams. 1995. The syntax and interpretation of dropped cate-
gories in child language: A unified account. In Proceedings of WCCFL XIV, 293–306.
Stanford, CA: CSIL, Stanford University.
Hoekstra, T. N., N. Hyams, & M. Becker. 1997. The underspecification of number
and the licensing of root Infinitives. In BUCLD 21st Proceedings, eds. E. Hughes,
M. Hughes, and A. Greenhill, 123–36. Somerville, MA: Cascadilla Press.
Bibliography 217
Hyams, N. 1994. V2, null arguments and COMP projections. In Language acquisition
studies in generative grammar, eds. T. Hoekstra and B. Schwartz, 21–56. Amsterdam:
John Benjamins.
—1996. The underspecification of functional categories in early grammar. In
Generative perspectives on language acquisition, ed. H. Clahsen, 91–128. Amsterdam:
John Benjamins.
Klein, P. W. 1989. Spanish “gender” vowels and lexical representation. Hispanic
Linguistics 3: 147–62.
Koehn, C. 1994. The acquisition of gender and number morphology within NP.
In Bilingual first language acquisition: French and German grammatical development,
ed. J. N. Meisel, 29–50. Amsterdam: John Benjamins.
Leonetti, M. 1999. El artículo. In Gramática descriptiva de la lengua española, eds.
I. Bosque and V. Demonte, 787–890. Madrid: Espasa.
Lleó, C. 1997. Filler syllables, proto-articles and early prosodic constraints in
Spanish and German. In Proceedings of GALA 1997: Language Acquisition: Knowl-
edge representation and processing, eds. A. Sorace, C. Heycock, and R. Shillcock,
251–6. Edinburgh: University of Edinburgh.
—2001. Early fillers: Undoubtedly more than phonological stuffing. Journal of Child
Language 28: 262–5.
Longobardi, G. 1994. Reference and proper names: A theory of N-movement in
syntax and logical form. Linguistic Inquiry 25: 609–65.
López Ornat, S. 1997. What lies in between a pre-grammatical and a grammatical
representation? Evidence on nominal and verbal form-function mapping in
Spanish from 1;7 to 2;1. In Contemporary perspectives on the acquisition of Spanish,
eds A. T. Pérez-Leroux and W. Glass, 3–20. Somerville, MA: Cascadilla Press.
—2003. Learning earliest grammar: Evidence of grammar variations in speech before
22 months. In Linguistic theory and language development in Hispanic languages, eds.
S. Montrul and F. Ordoñez, 254–74. Somerville, MA: Cascadilla Press.
Lust, B. 1999. Universal grammar: The strong continuity hypothesis in first lan-
guage acquisition. In Handbook of child language acquisition, eds. W. C. Ritchie and
T. K. Bhatia, 111–55. San Diego, CA: Academic Press.
MacWhinney, B. 2004. A multiple process solution to the logical problem of child
language. Journal of Child Language 31: 883–914.
MacWhinney, B., & C. Snow. 1985. The child language data exchange system.
Journal of Child Language 12: 271–96.
Maez, L. 1982. The acquisition of noun and verb morphology in 18–24 month old
Spanish speaking children. NABE Journal 7: 53–82.
Mariscal, S. 2008. Early acquisition of gender agreement in the Spanish noun
phrase: starting small. Journal of Child Language 35: 1–29.
Marrero, V, & C. Aguirre. 2003. Plural acquisition and development in Spanish.
In Theoretical linguistics and language development in Hispanic languages, eds.
S. Montrul and F. Ordoñez, 275–96. Somerville, MA: Cascadilla Press.
Martínez, J. A. 1999. La concordancia. In Gramática descriptiva de la lengua española,
eds. I. Bosque and V. Demonte, 2695–786. Madrid: Espasa.
Meroni, L., A. Gualmini, & S. Crain. 2006. Everybody knows. In Semantics in acquisi-
tion, ed., V. van Geenhoven, 89–114. Dordrecht, The Netherlands: Springer.
218 Bibliography
Miceli, G., & A. Mazzuchi. 1990. Agrammatism in Italian: Two case studies. In Agram-
matic aphasia: A cross-language narrative source book, eds. L. Menn and L. Obler,
717–816. Amsterdam: John Benjamins.
Miyagawa, S. 2010. Why agree? why move? Unifying agreement-based and discourse-based
configurational languages. Cambridge, MA: MIT Press.
Miyamoto, Y., & S. Crain. 1991. Children’s interpretation of plural pronouns:
Collective vs. distributive interpretation. Paper presented at the 16th Annual Boston
University Conference on Language Development, Boston, Massachusetts.
Montrul, S. A. 2004. The acquisition of Spanish. Morphosyntactic development in monolin-
gual and bilingual L1 acquisition and adult L2 acquisition. Philadelphia, PA: John
Benjamins.
Müller, N. 1994. Gender and number agreement within DP. In Bilingual first
language acquisition: French and German grammatical development, ed. J. N. Meisel,
53–88. Amsterdam: John Benjamins.
Ojeda, A. 1984. A note in the Spanish neuter. Linguistic Inquiry 15: 171–3.
Pérez-Pereira, M. 1989. The acquisition of morphemes: Some evidence from
Spanish. Journal of Psycholinguistic Research 18: 289–311.
—1991. The acquisition of gender: What Spanish children tell us. Journal of Child
Language 18: 571–90.
Peters, A. 2001. Filler syllables: What is their status in emerging grammar? Journal of
Child Language 28: 229–42.
Peters, A., & L. Menn. 1993. False starts and filler syllables: Ways to learn grammati-
cal morphemes. Language 69: 742–77.
Phillips, C. 1996. Root infinitives are finite. In BUCLD 20th Proceedings 2, eds.
A. Stringfellow, D. Cahana-Amitay, E. Hughes, and A. Zukowski, 588–99. Somer-
ville, MA: Cascadilla Press.
Picallo, C. 1991. Nominals and nominalizations in Catalan. Probus 3: 279–316.
Pine, J. M., & E. M. Lieven. 1997. Slot and frame patterns and the development of
the determiner category. Applied Psycholinguistics 18: 123–38.
Pinker, S. 1984. Language learnability and language development. Cambridge, MA:
Harvard University Press.
—1989. Learnability and cognition: the acquisition of argument structure.
Cambridge, MA: MIT Press.
Pizzuto, E., and M.C. Caselli. 1992. The acquisition of Italian morphology: Implica-
tions for models of language development. Journal of Child Language 19:
491–557.
Pollock, J. Y. 1989. Verb movement, universal grammar and the structure of IP.
Linguistic Inquiry 20: 365–424.
Prado, M. 1989. Aspectos semánticos de la pluralización. Hispanic Linguistics 3:
163–83.
Pullum, G., & B. Scholz. 2002. Empirical assessment of the stimulus poverty
argument. The Linguistic Review 19: 9–50.
Radford, A. 1990. Syntactic theory and the acquisition of English syntax. Oxford, U.K.:
Blackwell.
—1994. Tense and agreement variability in child grammars of English. In Syntactic
theory and first language acquisition, eds. B. Lust, M. Suñer, & J. Whitman, 119–34.
Hillsdale, NJ: Lawrence Erlbaum.
Bibliography 219