This report of possessive structures in Latino1
The term Latino is used here to refer to children of
Puerto Rican, Mexican, and Mexican-American descent.
children
relies on the sociolinguistic study of spontaneous speech and looks at the
effects of multiple social and linguistic factors on the production of
several linguistic variables. Three types of English language possessives
were considered: third person possessive pronouns; periphrastic
of possessives; and attributive
-s possessives. It was
found that the Latino children confused
his for
her and
her for
his, as in (1); used periphrastic
of
constructions more than native English speakers, as in (2); and omitted
the attributive
-s marker, as in (3).
(1) … my mother's not going to go
work—tomorrow his day off. (= her day off)
(2) … and the friend of my brother brought it
back. (= my brother's friend)
(3) … like when I go to my cousin house. (=
cousin's house)
The deviations from other English dialects seen in these examples are
consistent with contrastive analyses of English and Spanish. Specifically,
the pattern evident in (1) is generally considered a feature common to
nonnative English speakers from Spanish language backgrounds. In such
constructions, it is believed that the pronoun is selected to agree with
the possession, or noun that follows, instead of the possessor (Hill & Bradford, 2000:111). As Whitley stated,
“in Spanish one rarely uses a noun without marking its gender,
thanks to the required agreement of modifiers” (Whitley, 2002:148). The periphrastic of
construction in (2) is common to Spanish speakers learning English, as the
syntax of such constructions mirrors the most common form in Spanish for
expressing possession (Whitley, 2002:153–154), shown in example (4).
(4) el libro de Juan (the book of Juan/Juan's
book)
What follows from this idea about Spanish-speaking English language
learners is that they should show a greater dependence on the periphrastic
form and, as a consequence, the attributive possessive construction in
example (3) should be less common (Fernández
Domínguez, 2000a:134).
A contrastive analysis also tells us that when the attributive
construction in (3) is used, there is no reason to expect a high rate of
absence of the -s, which is a feature generally seen in African
American Vernacular English (AAVE) (Baugh, 1983:94–97; Labov et al.,
1968:169) and English first language acquisition (Brown, 1973:335–337). Furthermore, it actually has been
proposed that possessive -s should be easily acquired by
Spanish-speaking English language learners because Spanish syntax enclitic
formation is similar. Fernández Domínguez (2000b:91) stated, “The status of the English
genitive as a clitic should certainly cause no major problems to the
native speakers of Spanish, since they should be used to the frequent use
of clitic elements in their own language, where, for instance, a
syntactically independent direct object pronoun can enclitically merge
with an intransitive verb, other than the verb that governs it, and
behave, from an orthographic and phonological point of view, as a suffix:
Podrían estarlo construyendo sin permiso (They can beit building
without permission).”
PARTICIPANTS
In the 2001–2002 school year, spontaneous speech samples were
collected from 630 elementary school children in Atlanta, Philadelphia,
and several cities in California. The present study included speech
samples from 61 Latino children who learned to read in Spanish (referred
to here as Latino Spanish) and 65 who learned to read in English (Latino
English),2
Although all of the Latino
children spoke English, the two Latino groups represent different types of
bilingualism, in which the Latino Spanish group learned to read in Spanish
and the Latino English group was exposed to English instruction in the
earlier grades. Details about their early academic experiences were not,
however, solicited from the children, who were just asked which language
they learned to read in.
and a comparison group of 28 African
American and 28 white children. All study participants were enrolled in
elementary schools where 65% or more of the student body qualified for the
federal low or reduced lunch program. All of the children were below
reading grade level, and many of them participated in bilingual education
programs. The children were evenly distributed between the language
groups, boys and girls, and among the second, third, and fourth grades.
The number of children coming from the three cities was not evenly
distributed, however, with more children in the study cohort coming from
California overall, and within each subgroup. A detailed distribution of
the children is shown in
Table 1.
Distribution of Latino study participants
DATA COLLECTION
Part-time tutors collected the speech samples. They received minimal
training in the methods of sociolinguistic interviewing and were provided
with a question guide to assist them in the interview process. The types
of questions used to elicit speech from the children included items
like:
‘Did you ever do something that your mom told you not
to do?’
‘Have you ever gotten blamed for something that you
didn't do?’
‘Do you know any place that's really
scary?’
The interviews lasted 30 minutes to 1 hour and were recorded on Sony
mini-disk recorders. The quality and quantity of the speech samples were
variable, as would be expected with 8 to 10 year-old children. Some
children were not really forthcoming, while others continued to talk
whether prompted or not. Approximately 170,000 words were transcribed, for
an average of about 1,300 words per child. Despite an equal number of boys
and girls, the number of words for each was 75,000 and 95,000,
respectively, with girls producing, on average, 300 more words per
interview than boys.
THIRD PERSON POSSESSIVE PRONOUN
CONFUSION
The variables considered for the analysis of third person possessive
pronouns were those used to mark possession in pronoun + noun
constructions involving the third person pronouns his and
her. All instances of his + noun and her + noun
were identified. The distribution of third person possessive pronoun
tokens by speaker sex and pronoun choice is shown in Table 2. The rate for pronoun confusion is 12%
overall, which is sufficiently common to warrant further investigation.
Also, while the number of boys and girls in the cohort is almost equal,
the girls used twice as many third person possessive pronouns as the boys.
This is not surprising as it has been shown that female discourse is
characterized by more talk about people than male discourse, which is more
focused on things (Macaulay, 2005:129–138).3
In his study of adults and adolescents
in Glasgow, Macaulay found that the proportion of talk about people was
64% for the girls compared to 24% for the boys. The adolescent girls also
showed a much higher rate of use for the personal pronoun
she,
and adult women used a higher rate of both
he and
she in
their conversations.
Expected and other pronoun by sex for Latino children
To explore the possible social and linguistic influences on pronoun
confusion, a multivariate analysis using goldvarb 2.0 on the
Macintosh was conducted. The dependent variable was use of incorrect
pronoun (his when her was expected, or vice versa). The
factor groups shown in Table 3 were selected as
contributing significantly (p = .035) to pronoun confusion. These
findings confirm that pronoun confusion is more likely when the pronoun
agrees with the Spanish translation of the noun, as shown in example (5),
where name in Spanish is a masculine noun and the pronoun
his is selected instead of her. Out of all 81 cases of
pronoun confusion, 75% involved this type of pronoun and noun
agreement.
(5) His name is Jacqueline.
Third person pronoun confusion: varbrul results I (binomial
up and down)
There are also some interesting group differences shown here. First,
pronoun confusion is more likely to occur in the Latino Spanish children,
who learned to read in Spanish first. Furthermore, there is a
developmental effect, with second graders more likely to confuse pronouns.
In terms of geographical distribution, Atlanta favors pronoun confusion,
California is neutral, and Philadelphia favors correct pronoun use. This
distribution suggested a difference between Mexican origin and Puerto
Rican origin children, which was explored by recoding the data to include
Spanish language origin plus sex as a factor group. Factors were created
for Mexican boys and girls by combining Atlanta and California data, and
for Puerto Rican boys and girls based on the Philadelphia data. The new
varbrul results (Table 4) show that it
is actually the Mexican origin females who are responsible for the greater
part of gender confusion. The effects of learning to read in Spanish and
pronoun and noun agreement continued to be favoring factors, as did the
developmental effect. Of particular interest, however, is the sharp split
between Mexican origin and Puerto Rican origin females, with the first
group favoring pronoun confusion and the latter group disfavoring it. The
factor weights for both male groups are in the neutral range and the
difference between the two is not significant. The reorganization of the
data uncovered interaction between speaker sex and language origin, which
is quite striking, though not easily accounted for.
Third person pronoun confusion: varbrul results II (binomial
up and down)
USE OF PERIPHRASTIC of VERSUS
ATTRIBUTIVE -s
The variables used in the analysis of periphrastic of were
noun + of + noun constructions that could be expressed as
noun + -s + noun (or noun + noun when the -s is
not expressed), as in example (6).
(6) So my dad, he got the gun of the bad guy. / So my
dad, he got the bad guy's gun.
There are many factors that may affect the choice of one or the other
form, including topicality of the possessor [± topical],
animacy of the possessor [± animate], and the nature of
the possession relation [± prototypical]. In a recent
study of U.S. and British English, Rosenbach (2002) found that the most influential factors
affecting the choice of periphrastic of over attributive
-s were the presence of an [− animate] and
[− topical] possessor, and a [−
prototypical] possessive relation, as in the fumes of a car
compared to a car's fumes. Accordingly, the preferred
environment for the attributive possessive involved a [+
animate] and [+ topical] possessor, and a [+
prototypical] possessive relation, as in the girl's arm
compared to the arm of the girl.
The possessive categories included in the present study are based on
three main types identified by Rosenbach (2002):
real possessives, subjective possessives, and objective possessives. Real
possessives describe an intimate relationship, often involving parts of a
whole, as in, John's arm or the arm of John. Real
possessives also express actual possession, as in, the girl's
dog or the dog of the girl. Subjective possessives imply a
transitory or abstract relationship, as in, the pool's
clarity or the clarity of the pool. And finally, objective
possessives do not signal actual possession, but the two nouns are
nonetheless closely related, as in, Mary's murder or the
murder of Mary.
The rates for periphrastic of compared to attributive
-s for all groups are shown in Table 5.
Both the Latino Spanish and Latino English groups had a significantly
higher rate of periphrastic of than the African American group
(chi-square, p < .001), which virtually does not use this
form. The Latino Spanish group had a significantly higher rate of
periphrastic of than the white group (chi-square, p =
.0028), but was not significantly different from the Latino English
group.
Use of periphrastic of compared to attributive -s
for all groups
An initial varbrul run was done to evaluate the selection of
of compared to -s for all four language groups. The
dependent variable was periphrastic of as opposed to attributive
-s. The attributive -s factor group included all
noun + noun possessive constructions whether or not the
-s was expressed. The factor groups that were selected as
contributing significantly (p = .009) to periphrastic of
use are shown in Table 6. The result for the
third factor group is in line with our current understanding of the
selection of the periphrastic form over the attributive form, which is
preferred when the possessor is [− animate], as mentioned
earlier. Furthermore, it was the Latino Spanish children who favored
periphrastic of, as did the Latino English children.
Periphrastic of compared to attributive -s for all
groups: varbrul results I (binomial up and down)
In an initial exploration of the preference for periphrasis by the
Latino children, the author conducted a review of the constructions
produced by them. As mentioned earlier, many factors are involved in the
selection of periphrastic of versus attributive -s and
it is difficult to determine what is actually different about the Latino
children's constructions. Accordingly, it was concluded that a first
step was to have the constructions rated as ‘native’,
‘nativelike’, or ‘nonnative’ in a pilot
grammaticality study involving native English speakers. Preliminary
results show that the Latino children's utterances were considered
‘nonnative’ more often than those of the other speakers. More
details on the grammaticality of the utterances will be reported at a
later time.
In the initial varbrul run for only the Latino children, the
factor groups that were selected as contributing significantly (p
= .04) to of use are shown in Table 7.
These data demonstrate more clearly that the children from California and
Atlanta are equally likely to favor the periphrastic construction compared
to the children from Philadelphia, as was the case with pronoun confusion.
Accordingly, the data were again reorganized to include a Spanish language
origin plus sex factor group. In the second varbrul run for the
Latino groups (Table 8), two factor groups were
selected as significant (p = .009). The factor weights for these
factors confirm that periphrastic of is a feature preferred by
Mexican origin girls as compared to the Puerto Rican children in general.
Mexican origin boys are neutral and Puerto Rican boys did not use the
periphrastic of construction at all, hence the knockout category.
Furthermore, learning to read in Spanish clearly contributes to a
preference for this form.
Periphrastic of vs. attributive -s for Latino
groups: varbrul results II (binomial up and down)
Periphrastic of vs. attributive -s for Latino
groups: varbrul results III (binomial up and down)
What might have been considered a Spanish-language feature has proven
to be quite variable among the Spanish speakers in the cohort. This is not
to say that the parallels between Spanish and English are not related to
an increased reliance on the periphrastic of possessive, but
being Latino by no means guarantees an overgeneralization of this
construction. It may be that the different histories of bilingualism and
English contact of Mexicans and Puerto Ricans has contributed to differing
patterns in Spanish influence. For example, the use of English as the
official language in Puerto Rico has afforded Puerto Ricans more
opportunity for English contact (Zentella, 2000:137–138). Furthermore, in areas with high
concentrations of Mexican origin Spanish speakers (in particular in the
Southwest United States), Spanish continues to be used in the home more
than for most other groups for whom English is a second language (Alba et al., 2002). There are also different patterns
of -s treatment in the Spanish of the two groups, with Puerto
Ricans allowing -s elision freely (Poplack,
1980), which is not generally associated with Mexican Spanish. The
preference on the part of Puerto Rican origin males for the attributive
construction will be considered now by looking more closely at the use of
attributive -s.
ATTRIBUTIVE -s ABSENCE
With the attributive -s construction, absence of the
possessive -s marker also varies among children from different
racial and ethnic backgrounds. Absence of attributive -s is well
documented in the African American community (Baugh, 1983:94–97; Labov et al., 1968:169; Labov & Harris, 1986:11–12; Rickford, 1992:185–186), and occurs at a rate of over 70%
for the African American children in the present study (Table 9). Interestingly, it also seems to be a feature
common to the Latino children. As compared to the white children, there
were significantly higher rates of -s absence among the Latino
Spanish (chi-square, p = .0001) and Latino English (chi-square,
p = .002) groups.
Absence of attributive -s for all groups
An initial varbrul run to assess these differences included
five factor groups, with -s absence as the dependent variable.
The factor groups selected as contributing significantly (p =
.001) to -s absence are shown in Table
10. The findings confirm what was already evident, which is that
African Americans favor the absence of -s. We also see that there
are regional differences, with Philadelphia being a favoring factor.
Following segment did not significantly affect the model however, which
suggests that we are dealing with a morphological rather than a
phonological variable, as has been noted in other studies of the absence
of -s in AAVE (Labov, 1984:147–150; Labov et al., 1968:170; Rickford,
1992:186).
Absence of attributive -s for all groups: varbrul
results I (binomial up and down)
In the initial varbrul analysis for the Latino children
alone, two factor groups were selected as contributing significantly
(p = .04) to -s absence (Table
11). Once again, we see a bigger difference between Philadelphia,
on the one hand, and Atlanta and California on the other, suggesting a
Spanish language origin effect. But this time, boys favor the form over
girls, and language group is not included. Data from a second Latino group
varbrul analysis including the language origin group are shown in
Table 12. Spanish language origin plus sex is
the only factor group selected as significant (p < .001); the
absence of attributive -s is clearly favored by the Puerto Rican
boys, and to a lesser extent by the girls, but not among the Mexican
girls.
Absence of attributive -s for Latino groups: varbrul
results II (binomial up and down)
Absence of attributive -s for Latino groups: varbrul
results III (binomial up and down)
DISCUSSION
The present findings support a substrate effect that is clearly
mediated by social and cognitive factors. While variation in the
expression of possession could be attributed to incomplete acquisition,
the different sociolinguistic patterns for the different variables
suggests otherwise. A relationship between the increased frequency of
pronoun confusion and agreement in Spanish between the possessor pronoun
and the possession noun was shown. Such agreement was found in 75% of the
cases of pronoun confusion. There is also a developmental effect evident
in pronoun confusion, with an increased frequency found in the second
graders. This can be viewed in combination with the high level of salience
of this feature, where the use of the incorrect pronoun might impede
communication and cause misunderstandings, which underscores the
difference between this and the other variables being studied. In the two
cases where Spanish language influence was expected—pronoun
confusion and periphrastic of—there was an increased
occurrence of the nonstandard form among the children who learned to read
in Spanish first compared to those who learned to read in English. With
possessive -s absence, there was no difference between these two
groups, which contradicts the idea that attributive -s
acquisition may be facilitated by Spanish structural similarities.
The difference between boys and girls for two of the analyses was
surprising, as was the interaction between speaker sex and language origin
for all of the analyses. Most striking here is that a group of girls
favored the use of periphrastic of and a group of boys favored
absence of the attributive -s. An explanation for the preference
of periphrastic of by the Mexican girls may lie in the fact that
girls have closer ties to their homes, and less outside contact, which
could limit their exposure to English. However, the same does not hold
true for Puerto Rican girls, which could be due to the sociocultural
differences between the Latino groups. At the same time, the increased
absence of attributive -s among the Puerto Rican boys, and to a
lesser extent girls, suggests that there is a link between them and black
street culture. Poplack (1978) first reported on
Puerto Rican bilingual boys use of AAVE features in Philadelphia in the
late 1970s. In this study, Poplack looked at the realization of six
phonological variables in the speech of Puerto Rican bilingual children
attending a Catholic high school in North Philadelphia that was 51% Puerto
Rican, 46% white, and 3% black. The purpose of the study was to examine
the acquisition of the local Philadelphia dialect compared to AAVE dialect
features by the Puerto Rican bilinguals. The boys in the study not only
showed a higher percentage of the AAVE variants than the girls, but also
had significantly more AAVE use in casual compared to careful speech. In
their discussion of segregation in Philadelphia, Labov and Harris (1986:12) showed that Philadelphia African American
speech patterns also extend to Puerto Rican males in that city.
Specifically, it was the Puerto Rican males linked to the African American
community through marriage who showed higher rates of both verbal and
possessive -s deletion. An earlier study on assimilation by
Wolfram (1974) found similar evidence of AAVE
influence on Puerto Rican English in New York. Current, ongoing
ethnographic and linguistic studies being conducted in Philadelphia
confirm that Puerto Ricans have a close relationship with local African
Americans and, as a consequence, certainly have the opportunity to absorb
their grammatical features and are doing so (Wolford,
Evans, & Cakiades, 2005).
Finally, the differences between the Mexican and Puerto Rican children
are striking, beyond the differences between boys and girls noted earlier.
The Mexican and Puerto Rican origin children are clearly divided in their
use of both periphrastic of and absence of possessive
-s. These findings point to a possible differentiation in Mexican
or Mexican American and Puerto Rican English, with the latter being more
and more characterized by features of AAVE, with less apparent influence
from Spanish. At the same time, the former group may continue to be
subject to a broad range of influences, as well, including Spanish
language structures and forms.
