See discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/282294527 Gender-Mismatch in Monolingual Spanish and Spanish/English Code-Switching RESEARCH · SEPTEMBER 2015 DOI: 10.13140/RG.2.1.1665.5842 READ 1 4 AUTHORS, INCLUDING: William Udziela Rush University Medical Center 3 PUBLICATIONS 0 CITATIONS SEE PROFILE Jeanne Heil University of Illinois at Chicago 5 PUBLICATIONS 3 CITATIONS SEE PROFILE Sergio Ramos University of Illinois at Chicago 4 PUBLICATIONS 0 CITATIONS SEE PROFILE All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately. Available from: William Udziela Retrieved on: 06 December 2015
30
Embed
Gender-Mismatch in Monolingual Spanish and Spanish-English Code-Switching Capstone paper
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Gender-‐Mismatch in Monolingual Spanish and Spanish/English Code-‐Switching William Udziela
University of Illinois at Chicago
Honors College Capstone
May, 2015
Udziela 2
Gender-‐Mismatch in Monolingual Spanish and Spanish/English Code-‐Switching
1. Introduction
The field of Hispanic Linguistics has disagreement in the literature over how
Spanish nouns function when undergoing the linguistic phenomena of ellipsis. The major
theories are: that ellipsis must have gender match in order to be acceptable (Depiante &
Masullo, 2001); that Spanish nouns behave similarly to Greek nouns and fall into categories
that allow either one-‐way, two-‐way, or no ellipsis substitution (Merchant, 2011); or that
Spanish nouns have a partial identity of both syntax and semantics that affect
grammaticality when elided. This set of two experiments probes each of these proposals in
the effort to provide evidence of one proposal that best explains how Spanish nouns
interact with elision.
First ellipsis is the omission of a part of a sentence without a change in meaning. For
example, observe (1), where the part of sentence that is crossed out is the part of the
sentence that is omitted, or elided.
(1) John bought something, but I don’t know what John bought. From (1), it is possible for the person hearing that statement to understand that the
speaker is referring to what John bought. This phenomenon demonstrates that the listener
is able to reconstruct a meaning when there is not explicit information
Yet within the overarching genre of ellipsis, there is a more specific form that can be
called “stripping”. It is possible to learn more about the nature of Spanish NPs (noun
phrases) using ellipsis. This information stems from comparing the grammaticality of
individual items, especially in cases of morphological gender or number mismatches. See
(2) and (3):
Udziela 3
(2) El tío de Juan y los tíos de María. the.MASC.SG of J and the.MASC.PL uncles of M “Juan’s uncle and María’s (uncles).” (3) *El padre de Juan y la madre de María. The.MASC.SG father of J. and the.FEM.SG mother of M “Juan’s father and María’s (mother).”
As can be observed in (2) and (3), the difference of acceptance mismatch in ellipsis is that
number mismatch (2) is allowed, whereas gender mismatch (3) is not allowed (Depiante &
Masullo, 2001) In (2), the noun in the second conjunct (‘uncles’) is not pronounced, yet the
utterance is acceptable with the difference in number between the singular noun in the
first conjunct and the plural one in the second (‘tío’ [MASC.SG] paired with ‘tíos’ [MASC.
PL]). In (3), there is gender mismatch due to the first conjunct containing the masculine
noun [padre], while the second conjunct has an unpronounced feminine noun [madre],
marked by the article ‘la’. The presence of ‘la’ in the second conjunct marks that the noun in
the second conjunct can only be feminine. Not pronouncing the noun in the second
conjunct (‘mother’) results in ungrammaticality, signified by the “*” at the beginning of the
item. It is surprising that this change in gender (‘padre’ vs. ‘madre’) is worse than the
change in number, as seen in (2) with tío [singular] vs. tíos [plural].
In this paper, the status of NPs is tested according to the three leading proposals of
NP-‐ellipsis in Spanish, namely those proposed by Depiante & Masullo (2001), Merchant
(2011), and Saab (2010). This paper analyzes the results of a Mexican dialect of Spanish
monolingual study (Udziela, Ramos, & González-‐Vilbazo, 2013) and Mexican-‐American
dialects of Spanish/English code-‐switched items (Udziela, González-‐Vilbazo, Ramos, & Heil,
2015) involving Spanish NP ellipsis. It is proposed that Saab (2010) gives the best account
Udziela 4
to date of NP ellipsis in Spanish and has strong implications across languages with and
without morphological gender.
2. Literature Review
The first proposal is Depiante & Masullo (2001) in which they state that there must
be complete gender matching in cases of Spanish ellipsis. However, under their proposal,
mismatches would differ by type: a case of number mismatch would be acceptable, unlike a
case of gender mismatch. This was partially shown above in cases (2) and (3). To see the
examples of Depiante & Masullo from their text (2001), see (4) and (5):
(4) Juan visitó a sus tíos y Pedro visitó al tío suyo. John visited his uncles.MASC.PL and Peter visited the (MASC.SG) his.
“John visited his uncles and Peter visited his uncle” (5) *Juan visitó a su tío y Pedro visitó a la tía suya.
John visited his uncle.MASC.SG and Peter visited the (FEM.SG) his. “John visited his uncle and Peter visited his aunt”
(4) is a reversal of (2) in that the noun of the first conjunct is plural, and the second
conjunct’s noun is singular (‘tíos’ [MASC.PL] paired with ‘tío’ [MASC.SG]). (5) is an example
of unacceptable gender mismatch: the first masculine noun in the first conjunct is paired
with a feminine noun in the second conjunct (‘tío’ [MASC.SG] paired with ‘tía’ [FEM.SG]).
Comparing (4) and (5), the differences are differences in number (number-‐mismatch) (4)
versus a difference in gender (gender-‐mismatch) (5). According to Depiante & Masullo
(2001), number-‐mismatch items will have greater acceptability than gender-‐mismatch
items, which is supported by (4) and (5). However, the proposal of required gender-‐match
does not account for all the cases, specifically as regards to some nouns concerning the
office or profession of individuals. See (6) where (6a) has a male antecedent and a female
NP has been elided and (6b) has a female antecedent and a male NP has been elided.
Udziela 5
(6) a. Juan es juez y María es jueza también. J is judge.MASC and M is judge.FEM too.
“Juan is a judge and María too” b. María es jueza y Juan es juez también. M is judge.FEM and J is judge.MASC too.
“María is a judge and John too”
In (6), there is a gender-‐mismatch present and the mismatch has been rated acceptable in
cases where the male noun of the first conjunct is paired with a female noun in the second
conjunct (‘juez’ [MASC.SG.] paired with ‘jueza’ [FEM.SG]), in addition to a female noun in
the first conjunct being paired with a male noun in the second conjunct (‘jueza’ [FEM.SG]
paired with ‘juez’ [MASC.SG]) (Udziela, Ramos, & González-‐Vilbazo 2013). There was no
change in acceptability if the male or female form of the noun (‘juez’) was first in the order.
This presents a problem with the theory in that if gender mismatch is not allowed, then
why would (6) be rated acceptable by native speakers but not (3) or (5)? Alternate
proposals may be able to account for this ability.
Merchant (2011) provides a possible solution to the inconsistencies of Depiante &
Masullo (2001) in acceptability of gender-‐mismatch cases through comparison of Greek to
Spanish. Merchant (2011) cites other works (Brucart 1987, 1999; Ritter 1988; Picallo
1991; Bernstein 1993; among others) as having distinguished “three classes of nouns in
Romance that differ from each other in their behavior under ellipsis”. Merchant (2011)
states that it is possible to see these categorizations of nouns within Greek as well.
Merchant distinguishes three different groups of nouns with regards to their behavior vis-‐
à-‐vis ellipsis. From his findings with studying nouns1 and noun phrases under ellipsis in
Greek, nouns pertained three different categories of N in Greek that were able to vary in
1 In his text, Merchant labels these nouns as “N”, by stating that there are “three categories of ‘N’ that exist”. This also includes not only nouns but noun phrases (NPs).
See (8) for Spanish exemplars of non-‐alternating nouns as would be described by Depiante
& Masullo (2001) and Merchant (2011):
(8) a. *Juan es rey y María es reina también. J is king.MASC and María is king.FEM too.
“Juan is king and María too.” b. *María es reina y Juan es rey también. M is king.FEM and J is king.MASC too.
(8a) is an item where the first conjunct’s noun is masculine (‘rey’) that is paired with a
feminine noun (‘reina’) in the second conjunct. (8b) is the reverse of (8a) in that the noun
of the first conjunct is feminine (‘reina’), and the noun in the second conjunct is masculine
(‘rey’).
Merchant’s second class (“Noun Class 2” or “Two-‐Way Ellipsis”) is a set of nouns in
Greek that is able to vary male to female or female to male. For these nouns, it appears that
the biological sex makes no difference. Here, Merchant (2011) states, “epicene…nouns only
have one form, but their concord and agreement patterns are determined by natural (or
Udziela 8
‘semantic’) gender of their referent.” See (9) for Greek example items of this class taken
from Merchant (2011):
(9) a. O Petros sine jatros, ala i Maria oxi. the.masc Petros is doctor but the.fem Maria not ‘Petros is a doctor, but not Maria.’ b. I Maria ine jatros, ala o Petros oxi. the.fem Maria is doctor but the.masc Petros not ‘Maria is a doctor, but not Petros.’
In (9a) and (9b), while the antecedent noun does not change morphologically, we can know
that their gender is different by looking at the gender of their respective modifiers. In the
same way, the modifiers of the deleted noun show that this unpronounced noun was
different in gender from its antecedent. In both sentences, the ellipsis is allowed.
See Table 2 for a listing of Spanish nouns proposed to be within Noun Class 2 (“Two-‐
way” ellipsis):
Table 2. Nouns of Noun Class 2 (“Two-‐Way Ellipsis”) as proposed by Merchant’s (2011) Greek counterparts (Masculine/Feminine)
Noun Class 2 Spanish English Presidente/Presidenta President P. ministro/P. ministra Prime Minister Doctor/Doctora Doctor Médico/Médica Physician Escritor/Escritora Writer Juez/Jueza Judge Abogado/Abogada Lawyer Secretario/Secretaria Secretary Esposo/Esposa Husband/Wife Jefe/Jefa Boss Niño/Niña Boy/Girl Using these proposed Spanish nouns from Table 2, see (10) for Spanish exemplars of
grammaticality:
Udziela 9
(10) a. Juan es doctor y María es doctora también. J is doctor.MASC and M is doctor.FEM too. “Juan is a doctor and María too.” b. María es doctora y Juan es doctor también. M is doctor.FEM and J is doctor.MASC too. “María is a doctor and Juan too.”
In (10a) the first conjunct is masculine to agree with ‘Juan’ and the second conjunct has an
elided phrase of ‘es doctora’, where ‘doctora’ is feminine marked by “-‐a” to agree with
‘María’. (10b) uses the same sentence but inverts the order of the conjuncts, so the first
conjunct is feminine (‘doctora’ [doctor.FEM]) to coordinate with ‘María’ and the second
conjunct has the elided ‘es doctor [doctor.MASC]’ phrase that would coordinate with ‘Juan’.
(10a) and (10b) both have conjuncts that are assigned a semantic gender to match the
gender of the person within the conjunct, either ‘Juan’ or ‘María’ in these items.
The last class of nouns in Greek (“Noun Class 3” or “One-‐Way Ellipsis”) is only able
to have variance under ellipsis in the direction male-‐to-‐female. That is, ellipsis is only
possible when the antecedent is male and the elided noun female. When the feminine noun
is in the first conjunct and preceding an elided masculine noun, the item is ungrammatical.
See (11) for examples of this in Greek from Merchant (2011):
(11) a. O Petros sine dhaskalos, ala i Maria oxi. the.masc Petros is teacher but the.fem Maria not ‘Petros is a teacher, but not Maria.’
b. * I Maria ine dhaskala, ala o Petros oxi. the.fem Maria is teacher.fem but the.masc Petros not ‘Maria is a teacher, but not Petros.’
Table 3 lists the Spanish nouns selected from their Greek Noun Class 3 counterparts that
Merchant (2011) proposed:
Udziela 10
Table 3. Nouns of Noun Class 3 (“One-‐Way Ellipsis”) as proposed by Merchant’s Greek counterparts
Noun Class 3 Spanish Noun English
noun Maestro/Maestra Teacher Poeta/Poetisa Poet Tío/Tía Uncle/Aunt Soldado Soldier Profesor/Profesora Professor Estudiante Student Enfermero/Enfermera Nurse Alumno/Alumna Pupil Monje/Monja Monk/Nun Testigo Witness Mártir Martyr See (12) for Spanish exemplars of Noun Class 3 (“One-‐way” ellipsis nouns) using ‘tío’/’tía’:
(12) a. Juan es tío y María es tía también. J is uncle.MASC and M is uncle.FEM too.
“Juan is an uncle and María too.” b. *María es tía y Juan es tío también. M is uncle.FEM and J is uncle.MASC too. “María is an aunt and Juan too.”
(12a) uses ‘tió’ [tío.MASC] in the first conjunct and ‘tía’, the female form of ‘tío’, is elided.
(12b) inverts the conjunct order of (12a) to put the feminine form in the first conjunct
preceding the elided masculine form of ‘tío’. Similar to (11), (12) is only rated grammatical
when the masculine noun is in the first conjunct and preceding an elided feminine noun in
the second conjunct.
Merchant refutes Depiante & Masullo’s (2001) proposal by stating that gender
mismatch can be acceptable. In addition, Merchant (2011) provides a schema to explain
why gender-‐mismatch is grammatical in some items but not in others.
Depiante & Masullo (2001) and Merchant (2011), however, do not address the
differences between animate and inanimate objects. For animate organisms, there is the
biological sex feature inherent in their semantics. Saab (2010) proposes a Distributive
Udziela 11
Morphology account for ellipsis in Spanish. Briefly, in Distributive Morphology (DM) words
and sentences are derived by the same mechanism. Further, there are two types of
morphemes: roots, which are language-‐specific combinations of sound and meaning, and
functional morphemes, which combine with roots to give them a category (Embick & Noyer
2007). In Saab’s account, gender is checked in a functional head n. In inanimate nouns,
gender is assigned arbitrarily at n, whereas for animate nouns the morphological gender is
somehow connected to their biological gender found in the root. Number, on the other
hand, heads its own projection, above n, as exemplified in Figure 1. In ellipsis, Saab
contends, what is elided is must be syntactically identical to its antecedent, and in nominal
ellipsis what is elided is nP, which explains why number mismatches are allowed but
gender mismatches are not: the head where number is checked remains outside the ellipsis
site.
Figure 1. Syntactic tree of the Dependent Phrase according to Saab (2010)
Figure 1 shows a phrase tree that breaks down the dependent phrase (DP) into its
components until arriving at the root of the NP. In this tree, the “n” beneath “nP” is the
grammatical or morphological sex signifier to show that the morphological sex is apart
from the root of the word itself. Number mismatch is shown as the separation of “Num”
from “nP” beneath “NumP”. The difference of location of the separations of Num from nP,
Udziela 12
and “n” from √P, supports why number mismatch is more acceptable than gender-‐
mismatch due to being less integral to the semantics of the DP overall.
Saab’s (2010) proposal differs from Merchant’s (2011) proposal in that Saab
provides a more principled way to distinguish the nouns, being if the root is different or a
special suffix for gender is required vs. noun that only change in their noun marker. For
Merchant (2011), the categorization of the different nouns is solely based upon behavior in
ellipsis instead of patterns that the nouns of any particular class (Noun Class 1, 2, or 3) may
share within the category. Also under Distributive Morphology, each and every one of these
roots is specific to each language to which they belong, even if the semantic meaning is the
same. Exemplars of same roots (14) and different roots (15):
(14) a. Abogad-‐o b. Abogad-‐a
(15) a. Duque b. Duquesa
(16) a. El abogado de Juan y la abogada de Pedro The lawyer.MASC of J and the lawyer.FEM of P “Juan’s lawyer and that of Pedro.” b. La abogada de Juan y el abogado de Pedro. The lawyer.FEM of J and the lawyer.MASC of P “Juan’s lawyer and that of Pedro.” c. *El duque de York y la duquesa de Gran Bretaña. The duke.MASC of Y and the duke.FEM of G B “The Duke of York and that of Great Britain.”
(14) demonstrates that if the word only changes in its word marker (‘-‐o’/’-‐a’, etc.)
the mismatch will be slightly tolerated. Example (15a-‐b) is composed of two different roots,
where (15b) ‘duquesa’ has a special suffix to differentiate from ‘duque’ (15a). Case (16a-‐b)
shows the tolerated gender mismatch because the order of the genders within the item did
not change the acceptability. However, in example (16c), the addition of the complex ‘-‐esa’
Udziela 13
to ‘duqu-‐‘, effectively eliminates any acceptability of gender-‐mismatch. This is because,
according to Saab’s analysis, ‘duque’ and ‘duquesa’ are interpreted as different roots,
disallowing mismatch. (Saab, 2010). A code-‐switching perspective can answer questions
not answerable in monolingual data. It is also important, and interesting, to study this
phenomenon in CS for its own sake, without trying to answer a question that may be
unanswerable only using monolingual data, for the simple reason that people have not
previously investigated it.
In light of these three theories, two experiments were proposed: a monolingual
Spanish experiment to determine which of the three proposals better predict how NP
ellipsis occurs in Spanish: Merchant (2011), Saab (2010), or Depiante & Masullo (2001).
The second experiment was a code-‐switching study to test whether the same proposals
found in the monolingual study extended across languages, particularly Spanish and
English.
3. Experiment 1: Stripping in Monolingual Spanish
In Experiment 1, the primary research question was to determine if in a Mexican
dialect of Spanish, there was a difference in acceptability of ellipsis. If there was no gender-‐
mismatch allowed in any of the items, then Depiante & Masullo (2001) would be supported
in that there was no tolerance for gender-‐mismatch. Then, if there was a difference, did it
depend upon root/inflection distinction in nouns (Saab, 2010) or a noun-‐class distinction
(Merchant, 2011)? For this experiment, participants were native speakers of Spanish for
natural intuitions of grammaticality within Spanish. This pilot study included four native
speakers of Spanish, all of whom had learned Spanish before six years of age. All of the
participants were current residents of Aurora, IL, US that had emigrated from Mexico when
Udziela 14
they were greater than 18 years old. Table 4 sums the descriptors of the participant pool
for Experiment 1:
Table 4. Descriptors of participants of Experiment 1 (N=4) Native speakers of Spanish (learned before age 6) Residents of Aurora, IL, US; grew up in Mexico Average age: 31 College educated Age of arrival to US > 18 years
Each of the participants was provided with a sentence judgment task in which there
were 126 total sentences, 66 of which were critical stimuli such as seen in (17), and 60
distractor items. Of the critical stimuli, 11 nouns were from Noun Class 1, 13 nouns were
from Noun Class 2, and 13 nouns were from Noun Class 3. Because the primary focus was
to test Merchant’s categorization, each of the nouns involved were used in 3-‐4 exemplars,
approximately 2 items per gender form (masculine/feminine).
(17) a. Juan es alumno y Teresa también. “Juan is a student and Teresa too.” b. Andrea no es doctora, pero Carlos sí. “Andrea is not a doctor, but Carlos is.”
The participants were asked to read each item and then rate them according to
acceptability on a Likert scale (1=unacceptable, 5=acceptable). After completing the
judgment task, participants were also asked to complete a modified version of the DELE
(Montrul & Slabakova, 2003) to verify the proficiency of each participant in written
Spanish. The results of the Spanish proficiency task indicated that each participant was
indeed a high proficiency speaker of Spanish.
For the results of the judgment task, there was a noted variance among the three
categories proposed by Merchant. The ratings were analyzed according to the noun class to
which they pertained, and then also in accordance to the direction of the elision: male to
Udziela 15
female or female to male. The average ratings are presented in Table 5 in which direction of
elision is entailed by the gender of each noun. Where the listed noun (e.g., ‘escritor’) is
masculine, the elided noun was feminine (e.g., ‘y María es escritora también’ ‘and María is a
writerFEM too’).
Table 5. Statistical analysis of Spanish monolingual NP-‐ellipsis critical items by Noun Class Class Word N M SD Noun Class M 1: Non-‐alternating Rey 11 1.64 0.492 1.50
(18) a. Mesa (‘Table’) root= Mesa, no semantic gender suffix= -‐a, syntactic female gender b. Table root= Table, no semantic gender suffix= ø (null), no syntactic gender (19) a. Padre (‘Father’) root= Padre, male semantic gender (from biological sex) suffix= ø (null), no added syntactic gender b. Father root= Father, male semantic gender suffix= ø (null), no added syntactic gender
As can be seen in (18b) and (19b), English nouns are devoid of a morphological gender;
therefore any gender that is associated with the noun is inherent to the semantic root of
that noun. Here, it is where the identity condition of the roots can be tested.
Although ‘Padre’ and ‘Father’ both are semantically equivalent, they are different
roots under Distributive Morphology (see e.g., (19)). We see now why code-‐switching
provides an interesting ground to test this theory: In Saab’s proposal, no ellipsis should be
allowed between these nouns, because their roots are different. This concept formed the
basis of Experiment 2. We tested, following the categorization of nouns as described in
Experiment 1, gender mismatches in predicate nominal ellipsis in Spanish/English code-‐
switching.
A primary research question for Experiment 2 was to determine if ellipsis was
accepted in code-‐switching cases. If ellipsis was not allowed in code-‐switching items, then
low acceptability ratings would be predicted for the code-‐switched elided items in
comparison to the code-‐switched items with the NP repeated. In the case that ellipsis was
accepted in code-‐switched items, then elided code-‐switched items were expected to have
equal or greater acceptability ratings as the non-‐elided code-‐switch items. The next
question was to determine if there was any difference in the acceptability by NP type
Udziela 19
(English NP, Spanish NP, elided NP), which should be seen regardless of the acceptability of
ellipsis in code-‐switched items and within those differences, was there a difference in the
code-‐switch acceptability by the direction of the code-‐switch (English-‐Spanish, Spanish-‐
English)? Then, within the Mexican-‐American dialect of Spanish that was tested, was there
an effect on licitness of code-‐switched Elided NP by the root type (Same, Different) as
proposed by Saab (2010)?
Participants of Experiment 2 were different than those of Experiment 1 because
bilingual Spanish/English speakers. This second set of participants was a group of six
native, bilingual speakers of Spanish and English who learned both languages before age 6.
All of the participants were female, college educated, residents of the Greater Chicagoland
area, and all spoke a Mexican-‐American dialect. Descriptors of the participants for
Experiment 2 are recorded in Table 6.
Table 6. Descriptors of participants in Experiment 2 Bilingual speakers of Spanish and English (before age 6) Resident of Greater Chicagoland Area Average age= 22.5 College educated Females, N=6 Mexican-‐American dialect of Spanish For Experiment 2, participants were asked to complete an online timed acceptability
judgment task of code-‐switched items containing NP Ellipsis using a Likert scale (1=bad,
5=good). There were 192 total items, of which 144 were critical items [Same-‐Root Spanish
(23), Different-‐Root Spanish NP (24), Different root NP Ellipsis (25)], and 48 were
distractors/fillers (26).
Udziela 20
(20) Pedro is not a good teacher, ni Teresa tampoco es a good teacher P is not a good teacher nor T either.NEG is a good teacher “Pedro is not a good teacher, nor is Teresa a good teacher either.”
(21) Teresa es un buena maestra, but Pedro is not una buena maestra. T is a good teacher.FEM but P is not a good teacher.FEM
“Teresa is a good teacher, but Pedro is not a good teacher.” (22) Teresa is a good teacher, y Pedro también.
T is a good teacher, and P is a good teacher too. “Teresa is a good teacher and Pedro too.” (23) Peter is a brave man, pero Esperanza no es un valiente hombre. P is a brave man.ENG, but E is.NEG a brave man.SPAN “Peter is brave man, but Esperanza is not a brave man.” (24) Esperanza es una buena madre, and Peter is a good mother. E is a good mother.SPAN.FEM and P is a good mother.ENG.FEM “Esperanza is a good mother, and Peter is a good mother.” (25) Peter is a good father, y Esperanza también. P is a good father.ENG.MASC, and E is a good father.ENG.MASC too. “Peter is a good father and Esperanza too.” (26) Pedro toca tres instrumentos, but Mary two. P plays three instruments but Mary two “Peter plays three instruments, but Mary two” All of the critical Spanish nouns (and English counterparts) used within the critical stimuli
items, were proposed by Saab (2010) and were verified to have high frequency ratings in
both Spanish and English to ensure that would be used in every day speech, ideally using
code-‐switching. The critical nouns used in this experiment are listed in Table 7:
Table 7. List of critical nouns by Root Type used in Experiment 2 (Masculine/Feminine) Root Type Spanish English Different Padre/Madre Father/Mother Different Hombre/Mujer Man/Woman Same Maestro/Maestra Teacher Same Abogado/Abogada Lawyer
Udziela 21
The results of the acceptability task appear in Figure 2 and Figure 3.
Figure. 2. Average ratings by Type and Direction
Firstly, the results of average ratings of the Elided NP items are higher overall than either
the English NP or Spanish NP items. This demonstrates that ellipsis in code-‐switched items
is not only acceptable, but also preferred to the English NP or Spanish NP. Between the
English NP and Spanish NP types, English NP was favored over the Spanish NP types, but
neither was rated as well as the elided NP type. This difference in rating by type indicates
that there is an order of preferences among the three types with Spanish NP being the least
preferred, English NP in the middle, and elided NP having the most preference of the three
types. This visual comparison from the graph was also supported by a significant main
effect for Type (F2,10 = 17.749, p=0.001, η2=0.780).
An interaction of Direction*Type (F2,10 = 6.293, p=0.017, η2=0.557) is seen in Figure
2 when comparing three type columns (English NP, Spanish NP, elided NP) among each
other. For two of the three cases, Spanish NP and elided NP, the English to Spanish
direction (blue columns) was rated higher than the Spanish to English direction. When
comparing the English-‐Spanish directionality to the Spanish-‐English direction, while the
Udziela 22
error bars are overlapping, the difference between the overall rating levels suggest that
there was a preference between the two possible directions, while also suggesting that
there is a preference for the direction with each of the types. When visually comparing the
columns, it appears that Spanish NP and elided NP almost differ by the same amount
between the directionalities, just in a different ratio of acceptability. However, statistically,
a post-‐hoc analysis demonstrated that Spanish NP ≠ Elided NP and that there was no effect
for gender either. The occurrence of a visually noted difference between the directionality
preference by type raises the question of why is one direction preferred for Spanish NP and
elided NP, but not the English NP type?
It is thought that English NP items had higher acceptability ratings in comparison to
Spanish NP items due to the employ of a morphologically gender-‐less noun. In comparison
to Spanish, use of a morphologically gender-‐less noun would make it easier for the nouns to
agree when switching between Spanish and English. This agreement of switching from
Spanish to English, or vice-‐versa, can stem from the lack of a morphological gender in
English more easily agreeing with whatever gender is present in the opposing conjunct. See
(27) for a Spanish/English code-‐switched item:
(27) a. Pedro is not a good teacher, ni Teresa tampoco es a good teacher. Pedro is not a good teacher, nor T either.NEG is a good teacher “Pedro is not a good teacher, nor is Teresa a good teacher either.” b. Teresa es una buena maestra, but Pedro is not una buena maestra. T is a good teacher.FEM but Pedro is not a good teacher.FEM “Teresa is a good teacher, but Pedro is not a good teacher.” Example (27a) shows an English-‐Spanish code-‐switch item that employs an English NP
phrase ‘a good teacher’ as part of the first conjunct, but also repeats it as part of the second
conjunct. Similarly, (27b) uses a Spanish NP phrase ‘una buena maestra’ in both the first
Udziela 23
and second conjuncts. Here, the potential difference of preference of English NP or Spanish
NP use could be seen because in (27a), the English NP ‘teacher’ is morphologically gender-‐
less, therefore, potentially easier to tolerate gender-‐mismatch because the English NP itself
does not have a morphological gender that needs to agree with the gender of ‘Pedro’ or
‘Teresa’. It is not the same in example (27b): ‘una buena maestra’, the Spanish NP phrase,
has a morphological gender associated with ‘maestra’ marked feminine through the ‘-‐a’
suffix. This said, Pedro is not presumed to be feminine or female in gender, so while
gender-‐mismatch of this type may be accepted, it may have less acceptance because of the
more obvious mismatch between ‘Pedro’ and the feminine identity of ‘maestra’, than
‘Pedro’ or ‘Teresa’ with ‘teacher’ as the English NP.
While it is interesting to note that ellipsis is allowed in code-‐switching items and
that there was a difference in acceptability by NP type (English NP, Spanish NP, elided NP)
and a difference of acceptability by direction (English-‐Spanish, Spanish-‐English), the
question still unanswered was that critical question of was there a difference by root
(Same/Different) as was predicted by Saab’s (2010) theory? This distinction is necessary to
make inferences whether Saab’s (2010) root proposal is extendable cross-‐linguistically to
code-‐switching items. Figure 3 breaks the different NP types (English NP, Spanish NP,
elided NP) into the distinction of same-‐root versus different-‐root to compare acceptability
levels by root-‐type:
Udziela 24
Figure 3. Average Ratings by Root, NP –type, and Direction
As seen in Figure 3, there was higher overall acceptability for items that had the same root
than items with different roots. Of all the different types, NP ellipsis with same-‐root NPs
had the greatest rating scores, which is significant because it suggests some sort of
accommodation for these types of nouns over different-‐root NPs. The English-‐Spanish
directionality was prevalent in same-‐root Spanish NPs, same-‐root NP Ellipsis, and
different-‐root NP ellipsis. Acceptability for a different-‐root Spanish NP was almost
equivalent in both direction types (English-‐Spanish, Spanish-‐English). There was a
significant main effect for Root (F1,5 = 61.965, p=0.001, η2=0.925). Post-‐hoc analysis for
Type showed that Spanish NP ≠ NP Ellipsis. There was no effect for Gender observed.
There was a strong preference for cases in which the same-‐root was used as
compared to different root nouns. This effect was independent of whether the root was in
English, Spanish, or elided. The finding that type of noun was irrelevant is compatible with
Saab’s proposal that items that use the same root are going to have greater acceptability,
even in non-‐elided cases.
Udziela 25
5. Discussion and Conclusions
In conclusion, the results of Experiment 1 and Experiment 2 combine to create a
coherent picture of NP ellipsis from both monolingual Spanish and code-‐switching
perspectives. Experiment 1 found that the Depiante & Masullo (2001) proposal that
gender-‐mismatch is not allowed in ellipsis was not supported at all. To the contrary of
Depiante & Masullo (2001), the results supported that gender-‐mismatch on the whole was
supported. From Experiment 1, the noun categorization schema proposed by Merchant
(2011) was not as strongly supported as the root partial identity structure proposed by
Saab (2010). Experiment 2 found that the Saab (2010) proposal is further supported by the
code-‐switching data allowing gender-‐mismatch in code-‐switching ellipsis over set of nouns
that had same roots (‘Maestro/Maestra’, ‘Abogado/Abogada’) and was much less
acceptable in ellipsis when nouns had different roots (‘Hombre/Mujer’, ‘Padre/Madre’).
If Merchant’s proposal had been correct, in Experiment 1, the mean ratings of each
Noun Class would have had distinct differences in mean ratings, especially those between
the different gender constructions as well. By this, in Experiment 1, the difference between
the male and female gender directions should have been markedly different from the Noun
Class 2 (Two-‐way ellipsis nouns) and Noun Class 3 (One-‐way ellipsis nouns); this is due to
the importance of the directionality of the item with a masculine noun preceding an elided
feminine noun for Noun Class 3 in comparison to the indifference of noun-‐gender order
within the item for Noun Class 2. For Experiment 2, there would not have been a predicted
effect by root because Merchant did not provide a word structure schema, only a
categorization of Spanish nouns by behavior in ellipsis.
Udziela 26
If Depiante and Mascullo’s proposal had been correct, Experiment 1 would have
indicated that there were no acceptable gender-‐mismatch cases in Spanish in ellipsis. This
prediction would have manifested as very low acceptability scores across all of the Noun
Class types (1, 2, and 3) because the mismatch overall would not be allowed. To extend
this possibility to the code-‐switching of Experiment 2, there would not be mismatch
allowed under circumstances of code-‐switching either; if it is not allowed monolingually in
Spanish, it is presumed that it would not be acceptable in Spanish/English code-‐switching
instances, either.
Saab’s (2010) proposal is the strongest of the three because in Experiment 1, the
participants rated Noun Class 2 (Two-‐way ellipsis) and Noun Class 3 (One-‐way ellipsis)
almost identically. Under Saab’s partial identity of the root proposal (2010), this is
explained by Noun Class 3 nouns as having the same root, which allowed gender mismatch
approximately to the same extent as Noun Class 2 (Two-‐way ellipsis). The root idenitity
proposal is supported, because of the lack of differentiation between One-‐way ellipsis
nouns and Two-‐way ellipsis nouns, which defies Merchant’s (2011) proposal and Depiante
and Masullo’s (2001) proposal. Continuing into Experiment 2, Saab’s proposal is the
strongest of the three due to the results of a significant effect of the root (same root vs.
different root); not only was preference for same root vs. different root documented, but it
was interesting that it was independent of language: English, Spanish, or elided. This
indicates that ellipsis is accepted in code-‐switching cases and that root mismatch may
determine grammaticality, even in cases of code-‐switched items.
Further testing would include additional testing within the same language pair of
Spanish/English in different elliptical constructions such as nominal ellipsis in the subject
Udziela 27
position and in the object positions. Testing these different structures would increase
understanding of ellipsis overall, but is of particular interest as to elliptical behavior cross-‐
linguistically.
Udziela 28
Literature Cited
Bernstein, J. B. (1993). Topics in the syntax of nominal structure across Romance (Doctoral).
City University of New York, New York.
Brucart, J. (1987). La elision sintactica en español. Belaterra: Publicaciones de la Universitat
Autonoma de Barcelona.
Brucart, J. (1999). La elipsis. In I. Bosque & V. Demonte (Eds.), Gramática descriptiva de la
lengua española. Madrid: Espasa Calpe.
Depiante, M. (2001). Ellipsis in Spanish and the stranded affix filter. In M. Kim & U. Strauss
(Eds.), North East Linguistic Society (pp. 215–224). Georgetown University: GLSA.
Depiante, M., & Masullo, P. (2001). Género y número en la elipsis nominal: Consecuencias
para la hipótesis lexicalista. Presented at the Primer Encuentro de Gramática
Generativa., General Roca, Río Negro, Argentina.
Merchant, J. (2011). Not all genders are created equal: Evidence from nominal ellipsis in
Greek. University of Chicago.
Picallo, C. M. (1991). Nominals and nominalizations in Catalan. Probus, (3), 279–316.
Poplack, S. (1980). Sometimes I’ll start a sentence in Spanish y termino en español: toward
a typology of codeswitching. Linguistics, 18(7/8), 581–618.
Ritter, E. (1988). A head-‐movement approach to construct state Noun Phrases. Linguistics,
(26), 909–929.
Saab, A. (2010). (Im)possible deletions in the Spanish DP. Iberia: An International Journal of
Theoretical Linguistics, 2(2), 45–83.
Udziela 29
Toribio, A. J. (2001). On the emergence of bilingual code-‐switching competence.
Bilingualism: Language and Cognition, 4(03), 203–231.
http://doi.org/10.1017/S1366728901000414
Udziela, W., González-‐Vilbazo, K., Ramos, S., & Heil, J. (2015, April). Bare-‐Phrase Ellipsis in
Spanish/English Code-‐Switching. Presented at the University of Illinois at Chicago
Spring Student Research Forum, University of Illinois at Chicago.
Udziela, W., Ramos, S., & González-‐Vilbazo, K. (2013, April). Categorization of Spanish Nouns
through Stripping. Presented at the Spring Student Research Forum, University of