THE ACQUISITION OF GRAMMATICAL GENDER BY ......is no gender agreement between nouns and their...
Transcript of THE ACQUISITION OF GRAMMATICAL GENDER BY ......is no gender agreement between nouns and their...
THE ACQUISITION OF GRAMMATICAL GENDER BY FRENCH AS A SECOND
LANGUAGE LEARNERS ENROLLED IN FRENCH IMMERSION
by
Klaudia Krenca
A thesis submitted in conformity with the requirements
for the degree of Master of Arts
Graduate Department of Applied Psychology and Human Development
University of Toronto
© Copyright by Klaudia Krenca 2015
ii
THE ACQUISITION OF GRAMMATICAL GENDER BY FRENCH AS A SECOND
LANGUAGE LEARNERS ENROLLED IN FRENCH IMMERSION
Master of Arts 2015
Klaudia Krenca
Department of Applied Psychology and Human Development
University of Toronto
Abstract
In Study 1, the relationship between morphophonological cues and gender assignment in
French was examined among a group of 148 French L2 children (M = 114.66 months, SD =
19.69). Study 2 investigated gender-marking ability in French among a subset of 27 children
from the first study (M = 120.74 months, SD = 16.14). In the determiner task, participants
assigned the indefinite article unMASC or uneFEM to pseudowords whose endings were typically
masculine, feminine or neutral. In the picture task, children pointed to the picture of a female or
male Martian that best personified each spoken pseudoword. The results provided evidence that
correct gender assignment improved across grade. More ending-consistent responses were given
to feminine endings than to masculine endings. Children whose L1 marked gender were better at
gender-marking in French than children whose L1 did not mark gender. Finally, there was
evidence of a default subclass for gender-neutral nouns.
iii
Acknowledgments
I am extremely grateful for the hard work and dedication of several people involved in
this journey. First of all, I would like to thank my wonderful supervisor, Dr. Xi Chen, for her
copious feedback while designing the measures and in writing this thesis. Thank you for
providing me with ample opportunities to expand my knowledge and for believing in me along
the way. Second, I am indebted to Dr. Kathleen Hipfner-Boucher for her invaluable insights. I
am incredibly inspired by your passion for research, your witty sense of humour, and the selfless
amount of time that you spend caring for yourself, your family and your students, all while
fighting a perilous battle with your illness. Your strength and courage during this difficult time
are incredible.
I would like to thank the parents, teachers and children for their participation in the
study. Moreover, this study would not have been possible without all of the research assistants,
volunteers, and Research Opportunity Program students who had been involved in some aspect
of the project. Your time, dedication, insights and enthusiasm will not be forgotten.
I am thankful for my faith, which gave me strength throughout this entire journey,
especially during unforeseen circumstances. Finally, I am immensely grateful for the unwavering
support that I received from my family, friends, co-workers and lab mates. Each of them guided,
nourished, laughed and believed in me throughout this one-year master’s program. Thank you!
iv
Table of Contents
Acknowledgments .......................................................................................................................... iii
Table of Contents ........................................................................................................................... iv
List of Tables ................................................................................................................................. vi
List of Figures ............................................................................................................................... vii
List of Appendices ....................................................................................................................... viii
Chapter 1 : Introduction .................................................................................................................. 1
Chapter 2: The Grammatical Gender System ................................................................................. 5
Grammatical Gender ....................................................................................................................... 5
Gender Agreement .......................................................................................................................... 6
Gender Assignment Principles ........................................................................................................ 7
The Gender System in French ......................................................................................................... 8
Processing in Gender Assignment .................................................................................................. 9
The Acquisition of Gender in Monolingual Children ................................................................... 11
Chapter 3: The Acquisition of Gender in Bilingual Children ....................................................... 17
Theoretical Framework ................................................................................................................. 17
Cross-Language Transfer .............................................................................................................. 22
Chapter 4: The Present Research .................................................................................................. 27
Background of French Immersion in Canada ............................................................................... 28
Chapter 5: Study 1 ........................................................................................................................ 31
Method .......................................................................................................................................... 31
Results ........................................................................................................................................... 38
Discussion ..................................................................................................................................... 49
v
Chapter 6: Study 2 ........................................................................................................................ 55
Method .......................................................................................................................................... 55
Results ........................................................................................................................................... 59
Discussion ..................................................................................................................................... 64
Chapter 7: Conclusion ................................................................................................................... 67
Limitations and Future Research .................................................................................................. 67
Educational and Theoretical Implications .................................................................................... 68
References ..................................................................................................................................... 70
Appendix A ................................................................................................................................... 80
Appendix B ................................................................................................................................... 81
Appendix C ................................................................................................................................... 85
Appendix D ................................................................................................................................... 90
Appendix E ................................................................................................................................... 91
vi
List of Tables
Table 1. Frequency Distribution of Language Background .......................................................... 33
Table 2. Frequency Distribution of Maternal Education .............................................................. 33
Table 3. Descriptive Statistics of Measures .................................................................................. 40
Table 4. The median number of gender-neutral responses as a function of grade and task. ........ 41
Table 5. Frequency Distribution of Language Background .......................................................... 58
Table 6. Frequency Distribution of Maternal Education .............................................................. 58
Table 7. Descriptive Statistics of Measures .................................................................................. 61
vii
List of Figures
Figure 1. The proportion of masculine and feminine responses in the picture task for biased
pseudowords as a function of grade. ............................................................................................. 42
Figure 2. The number of gender-unmarked nouns assigned to the feminine subclass in grade 2. 44
Figure 3. The number of gender-unmarked nouns assigned to the feminine subclass in grade 4. 44
Figure 4. The number of gender-unmarked nouns assigned to the feminine subclass in grade 6. 45
Figure 5. The proportion of masculine and feminine responses in the determiner task for biased
pseudowords a function of grade. ................................................................................................. 46
Figure 6. The number of gender-unmarked nouns assigned to the feminine subclass in grade 2. 47
Figure 7. The number of gender-unmarked nouns assigned to the feminine subclass in grade 4. 48
Figure 8. The number of gender-unmarked nouns assigned to the feminine subclass in grade 6. 48
Figure 9. The proportion of masculine and feminine responses (with standard error bars) in the
picture task for biased pseudowords as a function of language. ................................................... 62
Figure 10. The proportion of masculine and feminine responses (with standard errors bars) in the
determiner task for biased pseudowords as a function of language. ............................................. 63
viii
List of Appendices
Appendix A List of pseudowords ................................................................................................. 80
Appendix B The Determiner Task ................................................................................................ 81
Appendix C The Picture Task ....................................................................................................... 85
Appendix D Martians used in the Picture Task ............................................................................ 90
Appendix E Bias Value Calculations Obtained from the Lexique Database ................................ 91
1
Chapter 1 : Introduction
Le problème or La problème? Every individual whose language marks grammatical
gender has encountered the gender assignment problem, at least once. Grammatical gender, also
called noun class, and hereafter referred to simply as gender, may be defined following
Guillelmon & Grosjean (2001): “A subclass within a grammatical class (as noun, pronoun,
adjective, or verb) of a language that is partly arbitrary but also partly based on distinguishable
characteristics (as shape, social rank, manner of existence, or sex) and that determines agreement
with and selection of other words or grammatical forms” (p. 503) (Webster’s Ninth New
Collegiate Dictionary, 1991).
In French, for instance, nouns are either masculine (e.g., bateau, ‘boat’) or feminine (e.g.,
fenêtre, ‘window’). Conversely, English does not assign grammatical gender to nouns and there
is no gender agreement between nouns and their modifiers (e.g., adjectives, articles, and
determiners). Interestingly, to the native French speaker, gender-marking ability appears to be an
innate ability. When asked about this peculiar skill, a typical explanation to account for this
capability is: “through experience, one just knows it, that’s all” (Tucker, 1968, p. 5). In contrast
to the “linguistic myopia” (Tucker, 1968, p. 6) of the native French speaker, second language
learners (L2) of French, especially those whose first language (L1) lacks gender, are quick to
realize that gender-marking may be the most challenging feature to acquire in the French
language. Under this premise, native speakers of English (and other non-gendered languages
such as, Chinese) must learn the French words for things in addition to whether the words are
masculine or feminine (Tucker, 1968).
2
With this in mind, one may inquire, do L2 speakers make use of their L1 that marks
gender when learning another gendered language? In other words, if two languages (e.g.,
Spanish and French) share gender marking as a linguistic feature, are L2 learners able to transfer
the strategies learned from their L1 into their L2? Moreover, gender-marking ability may be
affected more by the structural similarity of gender between the first language and an additional
language than by the mere presence of gender features in the L1. For example, French and
Russian are two languages that share gender-marking traits. Although both languages mark
gender, they do not share the same number of subclasses within their gender system (i.e., French
has a two-tiered masculine-feminine system as opposed to Russian, which has a three-tiered
masculine-feminine-neuter system). Therefore, another question one may ask is whether the
typological distance between the L1 and the L2 influences the accuracy of gender marking
abilities.
Language proficiency is a multidimensional concept. As such, various non-linguistic
skills might relate with linguistic proficiency in different spheres (Slevc & Miyake 2006, as cited
in Serratrice, 2013). According to Surridge and Lesssard (2008), of the 40 pages devoted to the
study of gender, only one or two pages are dedicated to a systematic explanation of how one can
determine the gender of a particular noun in French. Furthermore, since correct gender marking
is rarely essential for comprehension of the message, it carries a low ‘communicative load’
(Harley, 1993; Warden, 1997). For instance, L2 learners can still understand each other without
paying attention to determiners. In fact, teachers may be unwilling to interrupt the flow of the
conversation in order to correct for gender errors, an example of a non-linguistic factor that may
affect grammatical competence in the domain of morphosyntax (Harley, 1993).
3
The rationale for this investigation came from several avenues of observations. First,
native speakers of French attribute gender to nouns unerringly and with celerity, yet are unable to
explain the specific processes that undergird this feature of their language (Tucker, 1968).
Second, French as additional language learners, even those who have acquired near native
proficiency still seem to struggle when it comes to gender assignment (Holmes & Dejean de la
Bâtie, 1999). Third, research concerning the acquisition of gender in monolingual and
simultaneous bilingual children has been inconclusive in regards to the specific factors that
might account for this well-developed skill (Grüter, Lew-Williams, & Fernald, 2012). The study
of gender has been investigated across different disciplines (i.e., applied psycholinguistics,
language and cognition, cognitive development, developmental disabilities etc.) yet the junction
between the theoretical and pedagogical underpinnings remains unclear. Finally, while learning
French as my third language (L3), I believe that I was able to somewhat benefit from the three-
tiered masculine-feminine-neuter system present in my L1. Considering my personal experience
as a sequential trilingual, I was curious about the benefits of additive bilingualism in regards to
gender marking. As a result of increasing international mobility, children are being exposed to
two or more languages simultaneously. Given their diversity of linguistic experience,
multilingual children should not be considered as one homogeneous group. It is important for
researchers, educators and clinicians to consider their various linguistic, cultural and educational
backgrounds in language learning. This thesis consists of two interrelated studies. In Study 1, the
impact of morphophonological cues (carried by the endings of some nouns) on gender
assignment in French was examined in a cross-sectional design. Study 2 investigated the same
relationship as a function of whether the L1 marks gender or does not mark gender among three
groups of French immersion students: children whose home language marks gender (e.g.,
Spanish), children whose home language is a language other than English that does not mark
4
gender (e.g., Chinese) and English-first language children. Hence, French was a third language
for the first two groups and a second language for the final group.
This thesis comprises seven chapters. In Chapter 1, the rationale for the current work is
laid out. An overview of the grammatical gender system (with an emphasis on French) and a
review of the empirical studies concerning the acquisition of gender agreement in monolingual
children are presented in Chapter 2. A review of the relevant theoretical frameworks that are
pertinent to the present research, along with a survey of the empirical studies concerning the
acquisition of gender agreement in bilingual children are discussed in Chapter 3. In Chapter 4, I
will then address the two studies encompassing this thesis along with background information on
French immersion programs in Canada. The studies are then reported in Chapters 5 and 6,
respectively, along with a detailed discussion of the results. To conclude, Chapter 7 presents a
general discussion of the studies, their theoretical and educational implications, and directions
for future research.
5
Chapter 2: The Grammatical Gender System
In this section, I begin by discussing the classification of the grammatical gender system.
Next, the grammatical structure of the French language is described. Finally, I conclude with a
review of the literature concerning the acquisition of gender agreement in monolingual children
(with particular emphasis on French).
Grammatical Gender
In a typological survey, Nichols (1992) found that of 174 languages, over one fourth
marked grammatical gender. The word ‘gender’ derives from Latin genus via Old French
gendre, originally meaning ‘kind’ or ‘sort’(Corbett, 2006, p. 749). In a more concrete sense,
every noun falls into two or more classes, such that noun subclass membership controls the form
of other syntactic categories that modify it or substitute for it (Hockett, 1958; Holmes & Dejean
de la Bâtie, 1999).
There are two components that make up the gender system of a language: gender
assignment (i.e., attributing a particular gender class to each noun) and gender agreement (i.e.,
concord between the noun and its syntactic elements, such as determiners). Grammatical gender
systems vary cross-linguistically on a number of dimensions, including the number of noun
subclasses (Comrie, 1999). There are four main noun subclasses across languages: masculine-
feminine, masculine-feminine-neuter, common-neuter and animate-inanimate, with few or no
nouns occurring in more than one class (Corbett, 1991). Romance languages (French, Italian,
Portuguese, Romanian and Spanish) are examples of a two-tiered masculine-feminine gender
system (e.g., chapeauMASC, ‘hat’; maisonFEM, ‘house’ for French). On the other hand, the
masculine-feminine-neuter subclass is typical of the Slavic languages (e.g., Polish). This system
is similar to the masculine-feminine system, except that there is an additional gender available
6
(neuter) for unspecified sex referents (e.g., stółMASC ‘table’; książkaFEM ‘book’; dzieckoNEUT
‘child’). The common-neuter system is found in languages such as Danish, Dutch, and Swedish.
For example, in Dutch, nouns denoting people are of common gender (e.g., politicusCOM
‘politician’) whereas, other nouns (e.g., voorhoofdNEUT ‘forehead’) are of neuter gender
(Sabourin, Stowe, & de Haan, 2006). Finally, Basque is an example of a language with a two-
tiered gender system that differentiates between animate and inanimate objects (Corbett, 1991).
In contrast, English does not mark grammatical gender (except for semantic gender
reflected by third person pronouns such as his or her and the endings of a few nouns which refer
to male or female entities such as, actor vs. actress) and no gender agreement (Laaha & Gillis,
2007 as cited in, Schwartz et al., 2014; Sera et al., 2002). One of the three groups of children
who participated in Study 2 of this thesis was comprised of speakers of languages characterized
by a two-tiered (masculine-feminine) (e.g., Arabic, Portuguese, Spanish, and Urdu) or a three-
tiered (masculine-feminine-neuter) system (Russian, Polish, and Slovak) for marking gender.
The remaining groups were comprised of speakers of languages such as English, Farsi, and
Tamil. None of these languages mark gender.
Gender Agreement
An important characteristic of gendered languages is gender agreement. Concord, or
agreement is a syntactic property that refers to the gender and number marking (e.g., singular or
plural) between syntactic categories (i.e., adjectives, determiners, numerals, past participles and
pronouns) as a function of the noun they co-occur with (Grüter et al., 2012). The French
language is a ‘head-first language’, meaning that the central element of the phrase (i.e., head
noun) precedes other syntactic categories (Mitchell, Myles, Marsden, 2013). In French, the noun
of the nominal phrase determines the gender and number of its modifiers. For example, leMASC
7
petit garçon means ‘the little boy’, whereas, laFEM petite fille means ‘the little girl’. In this
example, the definite article (le/la) and the adjective (petit/petite) change their form according to
the gender of the noun. Grammatical gender systems vary in terms of the range of syntactic
categories affected by gender agreement (Boloh & Ibernon, 2010). More specifically, in French,
determiners and adjectives are the most frequent syntactic categories that are inflected for gender
(Warden, 1997). There are multiple sources of information that children can use in determining
the gender of a noun namely, semantic cues (whether the noun’s gender is congruent with natural
gender), morphophonological cues (carried by the noun’s ending) and morphosyntactic cues
(carried by modifiers related to the nouns such as, determiners) (Corbett, 1991). The main focus
of this thesis was on gender assignment in French. Specifically, the studies examined the role of
morphophonological cues in gender attribution using two elicited production measures. Elicited
production is an experimental technique designed to tap children’s grammatical ability by having
them produce specific sentence structures (Thornton, 1996).
Gender Assignment Principles
There are two major principles involved in gender assignment: semantic principles and
formal principles (Comrie, 1999). According to the formal principle, nouns are assigned to
gender according to their form, for example, in Spanish, nouns ending in -o are masculine (e.g.,
libro ‘book’), while the majority of nouns ending in –a are feminine (e.g., casa ‘house’)
(Comrie, 1999). However, no language is based purely on formal rules (Corbett, 1991).
According to the semantic principle, nouns denoting male entities are masculine, while those
denoting female entities are feminine. All remaining nouns are neuter, which suggests that
assignment rules in a semantic gender system are transparent. That said, Romance languages do
not feature the neuter subclass, and thus, all inanimate nouns are classified as either masculine or
feminine (Warden, 1997).
8
In some languages such as Dyirbal, an Australian Aboriginal language, gender
assignment is dominated by the semantic gender system, which applies biological cues to
distinguish between nouns (Corbett, 1991; Tsimpli & Hulk, 2013). In other languages, such as
French, Russian, Spanish and Swahili, gender assignment is determined by a combination of
semantic and formal rules (Montrul & Potowski, 2007). For example, in French, the semantic
principle still applies in cases where biological information is a salient feature of the noun (e.g.,
maman ‘mom’ is feminine). Although semantic cues are pertinent to the assignment of nouns
denoting human entities, in French, grammatical gender corresponds to natural gender in only
about 10% of nouns (Corbett, 1991; Séguin, 1969). For example, témoin ‘witness’ is masculine
whether it refers to males or females, and victime ‘victim’ and personne ‘person’ are feminine
whether they refer to males or females (Boloh & Ibernon, 2013). Moreover, nouns whose
referents are not naturally gendered are nonetheless assigned to one or the other gender category
(Grevisse, 1980). Hence, morphophonological cues (based on the noun’s ending) and
morphosyntactic rules (based on the modifier) may be useful when attributing gender to nouns.
The Gender System in French
In French, the target language of the present study, all nouns are assigned to either the
masculine or feminine gender class. For example, leMASC livre, ‘the book’ is a masculine noun
whereas laFEM maison ‘the house’ is a feminine noun. Interestingly, according to Séguin (1969)
approximately 58.4% of French nouns are masculine. Although the French gender attribution
system is opaque and arbitrary (Corbett, 1991), morphophonological cues (carried by the endings
of some nouns) may serve as a useful strategy in noun subclass assignment (Lyster, 2006). For
example, nouns ending with -age are usually masculine (leMASC fromage, ‘the cheese’) but there
are some exceptions, such as uneFEM image, ‘a picture’, laFEM plage, ‘the beach’, and laFEM page,
‘the page’. Similarly, nouns ending with –ion are usually feminine (laFEM station, ‘the station’),
9
yet unMASC avion, ‘a plane’, unMASC lion, ‘a lion’, and unMASC camion, ‘a truck’ are masculine.
Furthermore, the majority of compound words take the gender of the head noun, except for
combinations of nouns with verbs and prepositions, which are generally masculine, such as
porte-monnaie ‘coin purse’ (Andriamamonjy, 2000; Price, 2008). Likewise, there are about 50
nouns that appear in both genders but with different meanings (l’Huillier,1999 as cited in
Presson & Whinney, 2014; Price, 2008). For example, physique is masculine when it refers to a
person’s appearance and feminine when it refers to the science of physics (Presson & Whinney,
2014). Yet despite these apparent regularities, the French gender system has many exceptions,
and thus, prototypical morphophonological cues are not fully reliable in predicting the subclass
of a given noun.
Processing in Gender Assignment
In the literature on the acquisition of grammatical gender, there is a growing body of
research concerning two models for gender assignment: the lexical route and the form-based
route (Gollan & Frost, 2001; Hohlfeld, 2006; Schriefers & Jescheniak 1999). According to the
lexical route, gender for each noun is stored at the lemma (word) level, part of one’s mental
lexicon (dictionary) (Jescheniak & Levelt, 1994; Levelt, Roelofs, & Meyer, 1999; Roelofs, 1992
as cited in, Hohlfeld, 2006) whereas, morphophonological information is obtained at the lexeme
(basic unit of meaning) level. By implication, gender information is retrieved earlier than
phonological information. For example, this process is supported by reports of patients who have
suffered from brain damage, who can often identify the gender of the word that they want to use
without being able to utter the word itself (Holmes & Segui, 2004). Hence, morphophonological
information (carried by word endings) may play a secondary role to lexical items in assigning
gender (Holmes & Dejean de la Bâtie, 1999). On the other hand, the premise of the form-based
route (e.g., the ‘reliable cue hypothesis’ proposed by Gollan & Frost, 2001) is that if nouns are
10
marked by morphophonological information they will be accessed more quickly than nouns that
are not marked by formal regularities. Applying such a framework, Desrochers and colleagues
found that native French speakers retrieved gender information more quickly and accurately for
words with gender-typical endings than for words with unbiased endings (Desrochers & Paivio,
1990; Taft & Meunier, 1998).
At the same time, Hohlfeld (2006) suggested that the lexical route and the form-based
route worked in tandem. She conducted an experiment in which native German speakers
determined the gender of nouns in an online gender assignment task. The stimuli consisted of
transparent masculine nouns (e.g., –ling, Findling ‘foundling’), transparent feminine nouns (e.g.,
-heit, Klugheit, ‘cleverness’), neuter nouns (e.g., -chen, Glöckchen, ‘bell’) and various non-
transparent nouns with unbiased endings (e.g., Atlas, ‘atlas’). The results indicated that there was
no difference in the speed of gender assignment between transparent and non-transparent nouns.
As expected, participants employed the lexical route when determining the gender of ‘real’
words. In a follow-up study, the accessibility of the mental lexicon was blocked by the
introduction of nonwords. The participants were asked to assign gender to transparent (e.g.,
Gremsling*)1 and opaque pseudowords (e.g., Tompeff*). The results revealed that there was no
difference in processing between transparent and opaque pseudowords. However, participants
relied on the form-focused route (i.e., reliable gender cues) for nonwords, given that they did not
have an entry in their mental lexicon for these words. Furthermore, these results support Gollan
and Frost’s (2001) ‘lexical checking’ hypothesis. In other words, by default (in the case of real
words), gender is stored in and activated from the mental lexicon. However,
morphophonological information (based on the noun’s ending), is used as a checking procedure
1 An asterisk refers to an invented noun throughout this thesis.
11
when processing incongruent information (such as pseudowords) from the lexicon
(Mastropavlou & Tsimpli, 2011).
The Acquisition of Gender in Monolingual Children
The gender attribution problem has been at the centre of many studies on gender
acquisition in monolingual children (e.g., Holmes & Dejean de la Bâtie, 1999; Karmiloff-Smith,
1979; Seigneuric, Zagar, Meunier, & Spinelli, 2007; Tucker, Lambert, & Rigault, 1977). A
general trend that can be observed from the literature on French L1 gender acquisition is that by
the age of 3, native speakers make almost no gender attribution errors (Clark, 1985; Karmiloff-
Smith, 1979). In particular, studies in various domains of developmental language research have
shown that gender assignment is based on the ability to determine whether the word ending is
typical for a particular gender on the basis of its nominal endings/suffixes (Holmes & Dejean de
la Bâtie, 1999; Holmes & Segui, 2004; Karmiloff-Smith, 1979; Lyster, 2006; Séguin, 1969;
Tucker et al., 1977).
In the subsequent section, some of the prominent studies that support the form-based
route (gender is assigned via morphophonological information) are surveyed. In a pioneering
study on the development of gender-marking ability, Tucker et al. (1977) hypothesized that
native French speakers use morphophonological rules while assigning gender to nouns on the
basis of their endings. In a series of four studies, the researchers presented French speakers
between the ages of seven and seventeen with a number of pseudonouns whose endings were
typically masculine or typically feminine. The corpus used for this study resulted from a search
of 26,725 nouns in the Petit Larousse dictionary. Noun frequency computations were generated
from Quemada’s (1971) inverse dictionary listing all 31,619 nouns in the Petit Larousse sorted
by noun ending and separated according to gender (Lyster, 2006). Participants were explicitly
12
asked to identify the gender of the noun by circling the indefinite determiner, unMASC or uneFEM.
In one condition, the participants read the words before listening to a recording of them. In the
second condition, the subjects only heard the words, but did not see them. The researchers found
that participants attributed gender to nouns on the basis of their typical endings. For example,
96% of nouns ending in –ette are feminine (e.g., cigarette ‘cigarette’) and 99% of nouns ending
in –ain are masculine (e.g., train ‘train’). This finding served as evidence for a set of rules that
were used to determine the grammatical gender of nouns on the basis of morphophonological
cues. The authors proposed a ‘backward processing’ strategy whereby native speakers actively
process the word from right to left, before assigning the noun to a particular subclass.
In the same vein, Lyster (2006) examined a corpus of 9,961 nouns appearing in the
Robert Junior’s illustrated dictionary designed for 8-12 year old children. An ending was
considered a reliable predictor of gender, if it predicted at least 90 per cent of all nouns in the
corpus with that ending. For example, nouns ending with -age are typically masculine (leMASC
fromage, ‘the cheese’) but there are some exceptions, such as uneFEM image, ‘a picture’ and
laFEM plage, ‘the beach’. Similarly, nouns ending with –ion are typically feminine (laFEM station,
‘the station’), yet unMASC avion, ‘a plane’and unMASC lion, ‘a lion’ are masculine. Aside from
this, some endings are ambiguous, which means that they occur in both genders, for example
nouns ending in –ique are feminine about 60% of the time (Holmes & Segui, 2004). The results
indicated that approximately, 80 per cent of all nouns have endings that systematically predict
their gender. In contrast to noun endings, the beginning of nouns are not as reliable (Holmes &
Dejean de la Bâtie, 1999). For example, looking at nouns with similar beginnings such as,
moucheFEM ‘fly’ and mouchoirMASC ‘tissue’, yields no important information concerning the
noun’s gender (Tucker, 1968).
13
In contrast to the corpus data obtained from dictionaries, elicited production measures
have been designed to assess children’s grammatical ability. In a seminal study, Karmiloff-Smith
(1979) conducted several experiments to evaluate the relative importance of gender cues to
determine the gender of invented nouns. According to the researcher’s hypothesis, in the early
stages of gender acquisition, children assume that grammatical gender is based on natural
gender. However, as children get older and become more experienced with the target language,
they start to pay attention to morphophonological cues. The participants in Karmiloff-Smith’s
(1979) study were 341 French monolingual children between the ages of 3 and 12. To examine
the role of morphophonological cues in gender assignment, Karmiloff-Smith (1979) manipulated
the determiner and the pseudoword’s suffix in two conditions: congruent (e.g., unMASC bicron*
in which the masculine indefinite determiner is consistent with the masculine noun ending –on)
and incongruent (e.g., uneFEM bicron* in which the feminine indefinite determiner is inconsistent
with the typically masculine noun ending –on). In the latter case, the child was presented with
two sources of conflicting information and had to make a decision based on the information
provided by the determiner or the noun’s ending. To investigate whether morphophonological
cues could activate semantic features of the corresponding gender class, children were shown
pictures depicting males or females (with no gender marked determiner) that were either
consistent with the noun’s suffix (e.g., two males referred to as deux bicrons*) or inconsistent
with the noun’s ending (e.g., two females referred to as deux bicrons*). Again, in the
inconsistent condition, the children were required to make a decision between the semantic
component and morphophonological cue. In both tasks, the participant’s gender-marking ability
was measured implicitly.
Taken together, the results from Karmiloff-Smith’s (1979) study revealed that children as
young as 3 years of age assigned gender to invented nouns on the basis of morphophonological
14
cues. In particular, when the participants were presented with conflicting evidence between
indefinite articles and semantic features that contradicted morphophonological cues (word
endings), children opted for the latter to determine the gender of the invented word. Interestingly,
there was “a tendency to attribute masculine gender to all unknown words, despite the potential
feminine phonological clue on the suffix” (Karmiloff-Smith, 1979, p. 159 as cited in Larrañaga
& Guijarro-Fuentes, 2013) for children who were 9 years of age or older. Nevertheless,
Karmiloff-Smith (1979) concluded that morphophonological cues prevailed across all age
groups.
More recently, the role of morphophonological cues and semantic features in gender-
marking ability have been explored by Seigneuric et al. (2007). The participants in this study
consisted of 144 native French children between the ages of 3 and 9. In the ‘determiner
condition’, participants were required to indicate gender class by orally providing the determiner
(unMASC or uneFEM) for invented nouns whose endings were typically masculine, typically
feminine or neutral. In the ‘picture condition’, children assigned gender to invented nouns by
pointing to the picture of a female or male Martian. In terms of the first task, the results
corroborated the findings reported by Karmiloff-Smith (1979) as the ability to assign gender to
pseudowords on the basis of endings increased with age with scores of 60% at age 3 and 82% at
age 9. At the same time, morphophonological cues activated the semantic features of the
corresponding gender class, as performance increased with age and was greater than chance level
at the age of 4.
There is a growing body of research that has looked at gender assignment using real time
processing techniques. In a study by Holmes and Segui (2004), the researchers were interested
in determining whether sublexical cues (i.e., ending typicality) and lexical cues (i.e.,
15
determiners) influence gender assignment. In the first experiment, known as ‘gender
classification’, participants (40 native speakers of French between the ages of 20 to 37 years)
were presented with a single noun and were asked to press a key corresponding to either the
masculine or the feminine. In the second task, known as ‘phrase verification’, the subjects
verified whether the determiner-noun phrase was acceptable or not (by judging the indefinite
article that was employed with the noun), by pressing an affirmative or negative key. Holmes
and Segui (2004) reported longer reaction times when both word endings and the determiner
were inconsistent, compared to shorter reaction times when both were consistent. The results
corroborated the post-lexical checking hypothesis (Gollan & Frost, 2001), which states that
people process both sublexical and lexical cues in tandem, “without any additional boost when
both sources were helpful” (Holmes & Segui, 2004, p. 437).
Finally, more recently, Melançon and Shi (2015) explored grammatical representations in
French monolingual 30-month-olds. In the training phase, toddlers were simultaneously taught
word-object pairs with familiar words and pseudowords (which did not contain phonologically
biased markings of gender) and their gender marked determiners (e.g., unMASC ravole*). In the
gender congruent test phase, ravole* occurred with another masculine determiner (e.g., leMASC
ravole*). In the gender incongruent test phase, ravole* occurred with a feminine determiner
(e.g., laFEM ravole*). Finally, in the neutral trial, the target noun succeeded a gender-unmarked
determiner (e.g., lesPLUR ravoles*). Melançon and Shi (2015) calculated the proportion of looking
time by dividing the total looking time to the target word by the sum of looking time to the target
and the distractor. The results showed that performance was better (i.e., shorter looking times) in
gender-matched trials compared to gender-unmarked trials, and recognition of the target word
was most hindered in gender-mismatched trials. Moreover, there was no difference in processing
time for gender-matched trials across real words and pseudowords. According to Melançon and
16
Shi (2015), “grammatical gender is not tied to familiar words, rather, the feature is represented
abstractly across all relevant words in the lexicon” (p. 13) and can be applied to invented words.
To recapitulate, the literature on gender acquisition in the L1 has demonstrated that the
ability to mark gender is well developed by the age of 3. In addition, the development of gender
appears to be based on the ability to use morphophonological cues (noun endings) as evidenced
by studies that extensively examined words obtained from dictionaries as well as elicited
production measures. Finally, online processing techniques have been designed in order to
examine the underlying processes that underlie gender activation.
17
Chapter 3: The Acquisition of Gender in Bilingual Children
In this section, I will situate the development of grammatical gender within two
prominent theories in the literature, which suggest that second language acquisition might be
contingent on the nature of the L1. Following this, the literature on the acquisition of gender in
bilingual children is framed in terms of two themes. First, from a developmental perspective,
acquisition of gender agreement by second language learners has been shown to mirror the
learning trajectory in L1 speakers, albeit with a delay (Montrul & Potowski, 2007). Second,
research on cross-language transfer is reviewed and its relevance to the study of gender
discussed. In particular, there is some evidence that gender-marking ability acquired in the L1
supports the acquisition of gender in a second language (Sabourin et al., 2006; Schwartz et al.,
2014). Consequently, second language learners whose first language does not mark gender may
develop strategies such as the masculine default, in order to determine the gender of new nouns
(Boloh & Ibernon, 2010; Harris, 1991, Riente, 2003).
Theoretical Framework
According to Noam Chomsky’s theory of Universal Grammar (UG), the ability to learn
grammar is hard-wired in the brain. The issue that emerged, however, is whether second
language learners have full, partial or no access to UG after the sensitive period (during which
the brain is most likely to strengthen important connections and prune irrelevant ones) (Liceras
2010 as cited in, Serratrice, 2013). Following this line of reasoning, second language learners
rely heavily on their L1 in learning an L2 (Serratrice, 2013). However, to what extent does the
L1 influence second language acquisition? To date, two dominant theories that have framed
grammatical gender learning research include: the Failed Functional Features Hypothesis (FFFH)
18
(e.g., Hawkins & Chan, 1997) and the Full Transfer Full Access (FTFA) model (e.g., Schwartz
& Sprouse, 1994; White, Valenzuela, Kozlowska-Macgregor & Leung, 2004).
In contrast to the well-entrenched gender-marking ability of native speakers (by the age
of 3 or 4), the results of several studies on bilingual speakers are somewhat inconsistent as to
whether or not native-like ability can be attained in a second language, even for those at high
levels of proficiency (Holmes & de la Bâtie, 1999; Bartning, 2000; Bruhn de Garavito & White,
2000; Dewaele & Véronique, 2001; Hawkins & Franceschina, 2004). In fact, English L1 learners
of French have been shown to make gender mistakes even after many years of instruction
(Andersen, 1984; Surridge, 1993). The FFFH stipulates that learners of a second language who
passed the critical learning period are unable to acquire grammatical features that were not
represented in their L1 (Hawkins & Franceschina, 2004; Franceschina, 2005; see also Carroll,
1989). As a result, the L2 interlanguage (i.e., evolving system of rules regarding the L2)
(Selinker, 1972) is subject to a representational deficit, given that the L1 does not mark gender
(Unsworth, 2008). In a pivotal study, Franceschina (2005) presented Spanish L2 participants (53
adults who spoke an L1 that featured gender and 15 English-first language learners) with a series
of six offline measures. A group of Spanish L1 speakers was also included in the study. In the
first test known as the ‘guessing game’, the participants heard and read sentences that referred to
a noun indirectly such as Los trajo Martín y dijo que son para usted (‘Martin brought them and
said that they were for you’). In this example, los is a masculine modifier (themMASC) and the
participants were asked to select which object it was referring to from the following choices:
flores ‘flowersFEM’, joyas ‘jewelsFEM’ or chocolates ‘chocolatesMASC’. The correct answer is
chocolates, as the other two options are feminine nouns and do not agree with los, which is
masculine. Results showed a significant difference in gender agreement ability in favour of
native speakers compared to the English-first language group. There was no significant
19
difference between native speakers and the group of learners who spoke an L1 with gender.
According to Franceschina (2005), if one’s first language marks gender, native-like mastery of
an L2 gender system (post-sensitive period) is probable, whereas, if the L1 does not have gender,
native-like attainment is unlikely.
According to the FTFA, grammatical features that are nonexistent in the L1 can be
learned with sufficient exposure to the L2, regardless of the age of acquisition (Schwartz &
Sprouse, 1994). For example, White et al. (2004) investigated the comprehension and
production of L2 Spanish gender and number agreement among a group of 48 French L1
speakers and 68 English-first language learners. The participants were taking Spanish courses as
adults and they were grouped according to their level of proficiency (low, intermediate and
advanced). Four tasks were devised for the study: two elicited production tasks, a vocabulary test
and a picture identification task. For example, in the production task, the participants were
consecutively shown three pictures and asked to describe them. In terms of the vocabulary test,
participants were asked to select the masculine or feminine form of an article (elMASC or laFEM)
that corresponded to the image that was being represented. Finally, the picture identification task
was used to assess gender agreement. The participants were asked to select one of three pictures
that matched a particular sentence on gender and number contrasts. For example, María
contesta: “Sí, claro, va a hacer mucho sol. Ponlas ahí cerca de la roja.” (Maria answers: “Yes,
of course, it is going to be very sunny. Put them over there by the red [one]”). In this example, la
is a feminine determiner and roja is the feminine adjective for red. Therefore, the correct
response is laFEM maleta ‘the suitcase’ which is feminine, instead of elMASC libro ‘the book’,
which is masculine or losPLUR calcetines ‘the socks’, which are masculine and plural. The results
revealed differences that were contingent on proficiency; low proficiency groups differed
significantly from native speakers, but there were no significant differences between advanced
20
and intermediate groups (White et al., 2004). Interestingly, exposure to a L1 with gender (e.g.,
French) did not improve performance on gender in the L2 (e.g., Spanish). In this thesis, then, in
accordance with the FTFA, it is hypothesized that with sufficient input in the target language, L2
learners are comparable to L1 speakers in gender marking. However, unlike White et al. (2004),
experience with a L1 that marks gender should place second (or even third) language learners at
an advantage compared with L1 speakers without gender.
Various theories have been articulated to account for the development of gender-marking
ability among bilingual children. Carroll (1989), in her review of gender attribution to nouns in
French, argued that, “endings do not exist.” (p. 563) In other words, native speakers of French
process nouns and their gender specific determiners as “co-indexed chunks” (e.g., lamaison,
‘thehouse’) or as one word. In contrast, Anglophone learners of French process nouns and their
determiner as separate lexical items (e.g., la maison, ‘the house’). Furthermore, L2 learners do
not recognize nominal endings/suffixes as typical for a particular gender. Thus, second language
learners must learn to develop a strategy that could assist them in assigning gender to new nouns.
Under this premise, Carroll (1989) argued that French native-like mastery in gender assignment
was considered improbable due to the absence of gender in the English language.
Holmes and Dejean de la Bâtie (1999) employed an online processing task using real and
invented French words, with typical or atypical endings to compare performance in gender
attribution of first and second language learners of French. In the first condition, the participants
(50 French L2 learners with English as their L1, M = 21.6 years and 44 French L1 speakers, M =
22.3 years) classified words as masculine or feminine. In the second task, the subjects verified
the accuracy of the determiner-noun combination. Native speakers of French took significantly
longer to classify nonwords than regular words (because they engaged in a lexical search).
21
Hence, the results suggest that for native speakers, knowledge of word endings plays a minor
role in gender assignment (Holmes & Dejean de la Bâtie, 1999). As expected, the gender-
marking ability of the L2 speakers was noticeably inferior to that of native speakers.
Interestingly, it appears that non-native speakers compensated for weak lexical associations with
explicit knowledge of word-ending rules during both tasks.
Among gender-marked languages, it appears that one subclass might act as the default
(i.e., learners would be right without having to learn anything at all) (Roca, 1989 as cited in,
Boloh & Ibernon, 2010). In short, the masculine default implies that a feminine noun has
consistently been used with a masculine determiner (Larrañaga & Guijarro-Fuentes, 2013). For
example, among the Romance languages, the literature reveals that the default gender appears to
be masculine in Spanish (Harris, 1991), in Italian (Riente, 2003), as well as in French (Hulk &
Tellier, 1999 as cited in White et al. 2004). Furthermore, overgeneralizing to one of the
subclasses occurred in Dutch (with the common gender as the default value, Sabourin et al.,
2006). Similarly, Cuza and Pérez-Tattam (2014) found evidence of the masculine as a default in
a picture-naming task among a group of Spanish-English bilingual children.
Applying such a framework to French, the masculine subclass, in particular, the definite
determiner leMASC ‘the’ has been shown to act as the default value standing in for its feminine
counterpart, laFEM ‘the’ (Bartning, 2000). The masculine default is a relatively effective strategy
since approximately 58.4% of French nouns are masculine (Séguin, 1969). Hence, in the event
that French speakers encounter novel nouns, they are more likely than chance to assign the word
to the correct subclass following the masculine as a default principle. Boloh and Ibernon (2010)
examined gender agreement errors in monolingual French speakers grouped into four age
categories (e.g., four-year olds, seven-year olds, ten-year olds and adults). The participants were
22
told that they would send new objects to another person (e.g., un/le kibon* ‘a/the kibon*’ or
une/la bajette* ‘a/the bajette’) and that they had to record a message specifying in which bag
each of the three items would be sent. The results supported the masculine as a default strategy,
as scores for masculine endings reached ceiling at all ages, although agreement on feminine
nouns increased with age and reached ceiling (96.85%) at the age of 10. In summary, the
masculine as a default approach appears to be a robust strategy when compared to the use of
morphophonological cues in speakers in the process of acquiring gender agreement.
Cross-Language Transfer
For the purpose of this work, the term cross-language transfer (or transfer) refers to “the
structural influence of one language on another (see Meisel, 1983), regardless of whether the
influence facilitates language acquisition or results in a grammatical construction.” (Nicoladis,
2006, p. 15) In terms of its relevance to the current thesis, transfer is contextualized within two
theoretical frameworks: the Contrastive Analysis Hypothesis (Lado, 1957) and the Linguistic
Interdependence Hypothesis (Cummins, 1981).
Over the years, the definition of transfer has evolved. Thorndike (1923) concluded that
transfer depends on the existence of ‘identical elements’ in two tasks, and without these identical
elements transfer cannot occur. According to Lado’s Contrastive Analysis Hypothesis (1957), it
is possible to predict which elements of a given L2 will be challenging to learners from a specific
L1 background based on structural (i.e., phonological, morphological, syntactic) similarities and
differences between the L1 and the L2. On the one hand, positive transfer occurs when specific
features in two languages are similar to each other. For instance, gender-marking ability may not
be a difficult skill to master for speakers learning two Romance languages (e.g., laFEM maison
‘the house’, in French, vs. laFEM casa ‘the house’, in Spanish) because the two languages share
23
the same linguistic feature. On the other hand, negative transfer ensues when specific features
differ between two languages such that L2 learners are likely to experience ‘interference’ from
their L1. Hence, gender activation may be difficult for speakers whose first language does not
mark gender (e.g., English). In fact, Carroll (1989) argued that English-speaking children could
not fully acquire the French gender attribution system because English does not have
morphological gender and “there is no trigger for the gender feature of the noun class.” (p. 576)
While the Contrastive Analysis Hypothesis (Lado, 1957) was based on similarities or
differences between specific linguistic features, the Linguistic Interdependence Hypothesis
(Cummins, 1981) stated that common underlying proficiencies are shared across languages,
irrespective of the typological distance between them. Based on this framework, L2 learners may
be able to rely on their L1 as a bootstrapping mechanism that supports learning an additional
language. For example, when two languages share a two-tiered gender system (e.g., French and
Spanish), surface features are transferred from one language to the other (Sabourin et al., 2006).
The transfer of abstract features such as two languages that do not share the same number of
noun subclasses is known as deep transfer. In relation to the current study, one of the three
groups of children who participated in Study 2 was comprised of speakers of languages
characterized by a masculine-feminine (e.g., Arabic, Portuguese, Spanish, and Urdu) or a
masculine-feminine-neuter (e.g., Russian, Polish, and Slovak) gender system. In other words,
gender-marking ability acquired in the L1 should support the acquisition of gender in the L2,
regardless of the typological distance between the languages. The interdependence hypothesis,
however, has been criticized for its lack of specificity in explaining what is actually being
transferred and the conditions under which its effects occur (Odlin, 2003).
24
To date, a small number of studies have found cross-language transfer effects in learners
of a second language grammatical gender system. In their study, Sabourin et al. (2006) explored
the role of transfer in learning the Dutch gender system among adult speakers of L1 German,
English and a Romance language (French, Italian or Spanish). To examine gender assignment,
participants were asked to make a deCOM or hetNEUT ‘the’ judgment for nouns that varied in terms
of word frequency. German speakers scored the highest, followed by the Romance group,
followed by the English group. All three groups showed an effect for assigning a default gender
(common gender) for middle frequency words. To examine gender agreement, participants were
asked to judge whether a sentence (in which the relative pronoun had the appropriate gender
agreement or an inappropriate gender agreement), was grammatically correct (yes/no decision)
and to correct phrases that were ungrammatical. Once again, German speakers performed the
best, followed by the Romance group performing above chance and the English group
performing at chance. Taken together, the hierarchy of performance on both tasks supports
Lado’s Contrastive Analysis Hypothesis (1957). In other words, German L1 speakers obtained
the highest scores because German and Dutch share specific features. Positive transfer still
occurred for the Romance group, although to a smaller degree, given that the former differ from
Dutch in terms of the composition of noun subclasses. Finally, gender activation may have been
difficult for the English-first language group whose L1 does not mark gender.
In the same vein, Schwartz et al. (2014) examined the acquisition of noun-adjective
gender agreement in Russian by comparing four groups of four- to five-year old bilingual
children from diverse L2 backgrounds (English, Finnish, German, and Hebrew) with age-
matched monolingual children. The researchers employed a semi-structured elicitation task in
Russian, with feminine (e.g., ruka, ‘hand’), masculine nouns (e.g., dom, ‘house’) and neuter
nouns (e.g., okno, ‘window’). Again, children whose L2 marked grammatical gender (i.e.,
25
German, Hebrew) outperformed the children whose L2 did not mark grammatical gender (i.e.,
Finnish, English). From a developmental perspective, the bilingual’s performance was shown to
follow the same learning trajectory as that of monolingual children, albeit with a delay (Schwartz
et al., 2014).
Due to a lack of empirical evidence, the process by which speakers of French as an
additional language learn to mark gender remains unclear. Yet Kuo and Anderson (2010)
recently proposed a theory that serves as a useful framework within which to consider this
question. According to the structural sensitivity theory (2010), having access to two or more
languages renders structural similarities and differences between languages more salient and
thus, more readily learned. Based on this theoretical framework, one would expect that gender-
marking ability would be enhanced among learners of an additional language whose L1 also
marked gender.
Taken together, these results suggest that L2 gender marking is affected more by the
structural similarity between L1 and L2 gender marking than by the mere presence of gender
features in the L1. Thus, in accordance with Lado’s Contrastive Analysis Hypothesis (1957),
positive transfer occurred between German and Dutch (Sabourin et al., 2006), and German and
Russian (Schwartz et al., 2014), as the two language pairs are the most similar in terms of their
grammatical landscapes. For example, German feminine and masculine nouns map onto the
common gender in Dutch and German neuter nouns are also neuter in Dutch. At the same time,
both German and Russian share a three-tiered masculine-feminine-neuter gender system.
Furthermore, although gender marking is a shared linguistic feature between Dutch and
Romance languages (Sabourin et al., 2006), and Russian and Hebrew (Schwartz et al., 2014),
Romance languages (and Hebrew) belong to a two-tiered masculine-feminine gender system,
26
which is often transparent (i.e., morphophonological cues are reliable predictors of gender),
whereas, the Dutch (and Russian) gender systems are opaque (Corbett, 1991). Hence, Romance
(and Hebrew) speakers may have experienced some interference from their L1. In contrast,
English (and Finnish) do not have grammatical gender and are the least similar to Dutch (and
Russian). Therefore, in these two studies, L2 learners acquired gender-marking ability to some
extent but, second language learners whose L1 marked grammatical features had a significant
advantage over children whose L1 did not mark gender.
In sum, there seems to be a debate in the L2 acquisition literature with respect to L1 and
L2 gender marking. Despite the delay, the developmental trajectory is similar, albeit, particular
groups of L2 learners may have an advantage given the presence of gender marking in their L1.
As a result, second language learners may develop strategies, such as the masculine default, in
order to determine the gender of new nouns. To gain insight into the development of gender
marking ability in a second (or third) language, the two studies reported in this thesis
investigated the role of morphophonological cues (based on noun endings) in gender assignment
using pseudowords via an elicited production paradigm among a group of multilingual children
enrolled in an early French immersion program.
27
Chapter 4: The Present Research
The focus of this thesis was on the acquisition of gender assignment in French among
children enrolled in French immersion programs. The participants were sequential bilinguals
who began acquiring French between the ages of 4-5, after developing basic command of their
first language (Meisel, 2001).2 This chapter provides a brief overview of the two studies included
in the current study, along with the specific research questions that were addressed. Finally, it
concludes with a description of French immersion programs in Canada, from which the present
thesis sample was drawn.
The objective of Study 1 was to examine gender-marking development in French across
grades two, four and six among a group of second language learners. While the effects of the L1
have been explored on gender acquisition in the L2, relatively less research has been done on L1
speakers whose first language marks gender, while learning an additional language that marks
gender (e.g., Sabourin et al., 2006; Schwartz et al., 2014). In Study 2, I investigated gender-
marking ability in French, as a function of home language status among three groups of French
immersion students: children whose home language marked gender (e.g., Spanish), children
whose home language is a language other than English that does not mark gender (e.g., Tamil)
and English-first language children. Therefore, French was a third language for the first two
groups and a L2 for the final group. Using an elicited production task modified from Seigneuric
et al. (2007), I examined the role of morphophonological cues carried by a noun’s ending (i.e.,
typically masculine or typically feminine) in gender assignment. In the first task, children
indicated their gender classification by pointing to one of two Martian-like figures that were
2 In the present research, some of the participants spoke neither English nor French at home. For
these children, French was a third language (refer to Study 2).
28
distinguished by secondary sexual features. In the second task, children heard a pseudoword and
were asked to verbally indicate which determiner (unMASC or uneFEM) was the most appropriate
for each item. Both measures employed pseudowords that respected the phonological structure
of French. The advantage of using invented words was that it controlled for children’s prior word
knowledge (Melançon & Shi, 2015).
Study 1 addressed three research questions: (a) Does gender-marking ability increase in
French L2 learners? ; (b) Do morphophonological cues play a role in grammatical gender
assignment? In other words, are French L2 learners more likely to respond consistently with
gender-marked pseudowords that refer to male/female entities (picture task) or
masculine/feminine determiners (determiner task); (c) Is one noun subclass more likely to be
overused than the other in cases where gender-endings are ambiguous? Study 2 addressed the
following research question: (a) Do morphophonological cues play a role in grammatical gender
assignment among French as an additional language learners whose L1 marks gender compared
to students whose L1 does not mark gender?
Background of French Immersion in Canada
As mentioned, the participants in the current thesis were children enrolled in French
immersion programs situated in a metropolitan region. This section presents a brief overview of
English-French bilingual programs in Canada.
In 1969, Canada adopted the Official Languages Act, which ensures access to publicly
funded education in both English and French. The aim of French immersion is additive
bilingualism: the development of French fluency at no expense to the majority English-first
language (Swain & Lapkin, 2005; Wesche, 2002). The inception of the first French immersion
program was a result of English-speaking families who were unhappy with the traditional core
29
French programs (Turnbull, Lapkin, Hart, & Swain, 1998). Throughout Canada (with the
exception of Nunavut), French immersion programs are offered in English language school
systems, in which non-native speakers of French are instructed in French. The popularity of
immersion programs has soared and enrollment in French immersion has increased exponentially
from 45,000 students in 1977 to over 350,000 students by 2011 (Canadian Parents for French,
2011). Generally speaking, French immersion students tend to come from families of middle to
high socio-economic status (SES) (Allen, 2004). Moreover, research has shown that Canadians
who speak both English and French have a higher average salary compared to those who speak
only one of the two languages (Jedwab, 2003).
There are three major variants of French immersion programs in Canada: early
immersion that starts in kindergarten or Grade 1; middle immersion that starts in Grade 4 or 5;
and late immersion that starts in Grade 7. These alternative models differ in terms of age of
intake and the proportion of content-based teaching in English and French (Rebuffot, 1993). In
an early immersion program, all academic content material is taught exclusively in French until
grade 4, when English language arts instruction is initiated for one period each day. Over time,
exposure to English instruction increases and by the end of grade 7 or 8, students may receive up
to 50% of their education in English. In middle immersion, at least 50% of the curriculum is
taught in French from Grade 4 or 5, whereas in late immersion, French language instruction
begins at Grade 7 and up (Lyster, 2008). The participants in the current study were enrolled in an
early immersion program that started in Senior Kindergarten.
With the advent of globalization, recent changes in Canadian demographics, especially in
large urban areas, have lead to increased enrollment of linguistically diverse children in early
French immersion (Sinay, 2010). In fact, English Language Learners (ELLs) comprised 32% of
30
the student body in early French immersion in Canada’s largest school board (Sinay, 2010). In
contrast to the non-English speaking population in the United States, which is predominantly
Spanish (Fleischman et al., 2003), in Canada, ELLs are a rather diverse group. Furthermore,
educators and researchers have applied the term ELL to native-born children as well as to new
immigrants (Ontario Ministry of Education, 2008). Although the sample in the current work was
predominantly Anglophone, those who did speak another language were born in Canada (except
two children) and came from families in which languages other than English or French were
primarily spoken in the home.
As a result of the increasing numbers of ELL’s, educators, researchers, test developers
and clinicians are faced with critical issues such as appropriate identification and intervention for
students who are struggling with language acquisition. Furthermore, many parents of ELL
children are concerned that if their children speak a non-official language at home, they will need
to learn English first before they can thrive in a French immersion program. Despite the rise of
ELLs in French immersion, there is a relative dearth of studies investigating outcomes in English
and French among these students. However, some recent studies examining aspects of language
and reading skills (Au-Yeung et al., 2015; Bérubé & Marinova-Todd, 2012) found that ELL
children performed as well as English L1 children on all the French measures and most of the
English measures, providing strong evidence that French immersion education is a viable option
for ELLs.
31
Chapter 5: Study 1
Study 1 was designed to examine gender assignment in French Immersion children. It had
three objectives. The first objective was to determine whether gender-marking ability improves
with age. The second aim was to investigate the effects of morphophonological cues on gender
assignment by L2 children. The third goal was to examine whether second language learners
defaulted to one noun subclass over the other in instances of ambiguity. In terms of the first
research objective, it was expected that gender-marking ability would improve across the three
grades (consistent with Boloh & Ibernon, 2010; Seigneuric et al., 2007). Thus, children in grade
6 were predicted to score the highest, followed by children in grade 4, followed by children in
grade 2. As for the second research goal, it was anticipated that morphophonological cues would
be used for gender attribution when a noun’s termination was typical for its gender in both the
determiner task (e.g., -otteFEM as in uneFEM blaquotte* ‘a blaquotte*’) and the picture task (e.g.,
-our as in lucadour* representing a male Martian). Furthermore, more ending-consistent
responses were expected for biased feminine pseudowords compared to biased masculine
pseudowords (Seigneuric et al., 2007; Karmiloff-Smith 1979). Finally, in terms of the third
research objective, it was expected that children across all grades, would overuse one subclass
over the other when presented with pseudowords with non-typical endings in both tasks.
Method
Participants
A total of 148 children (M = 114.66 months, SD = 19.69, 59 males) were recruited from six
public French immersion elementary schools in working to middle class neighbourhoods. There
were 46 children in grade 2 (M = 89.89 months, SD = 3.77, 16 males), 47 children in grade 4 (M
= 113.09, SD = 3.68, 20 males), and 55 children in grade 6 (M = 136.73, SD = 3.29, 23 males).
32
The children began French language and literacy instruction simultaneously in the fall of Senior
Kindergarten. All academic content material was taught entirely in French, with partial English
instruction beginning in Grade 4. While French was the sole language of instruction in school for
all of the participants, English was the dominant language of the broader community. As a result,
the children did not have much opportunity to hear and speak French outside of the classroom,
which implied that any differences in French were attributed to the children’s language abilities.
Informal observations (e.g. during recess or on the playground) indicated that the children had
acquired conversational proficiency in English.
The parents completed a demographic questionnaire designed by the researchers. Table 1
displays the frequency distribution for language background. The majority of the children
(95.3%) spoke English most often in the home. Of the 148 children, 126 spoke English more
than 75% of the time. Within the group of 148 children, 30 children learned to speak a language
other than English as the first language (L1), of which 7 reported speaking their mother tongue
most often in the home. The languages spoken among these children were Arabic, Mandarin,
Polish, Portuguese, Russian, Tamil and Urdu (all ns = 1). All of the children had at least 10
books at home, with the majority (62%) having more than 100 books. Approximately, 43% of
parents reported that they read with their child almost everyday and about 72% of the children
engaged in daily independent reading. The frequency distribution of maternal education is
reported in Table 2. In this sample, approximately, 90% of mothers had obtained at least a
college/university degree.3 The majority of children were born in Canada (90%) and 10% were
born elsewhere (n = 15).
3 Four participants did not respond to this question.
33
Table 1. Frequency Distribution of Language Background
Language Frequency Percent (%)
Child's most often spoken language in the home
English 141 95.3
Other 7 4.7
Total 148 100
Child's first language he/she learned to speak at home
Arabic 1 .7
English 118 79.7
English/Spanish 1 .7
English/Chinese 2 1.4
English/Urdu 3 2.0
English/Arabic 1 .7
English/Welsh 1 .7
Farsi 1 .7
Filipino 1 .7
French 1 .7
Greek 1 .7
Korean 2 1.4
Mandarin 2 1.4
Polish 2 1.4
Portuguese 1 .7
Russian 1 .7
Russian/English 1 .7
Slovak 1 .7
Spanish 2 1.4
Tamil 3 2.0
Tibetan 1 .7
Urdu 1 .7
Total 148 100
Table 2. Frequency Distribution of Maternal Education
Level of Education Frequency Percent (%)
Some High School 1 .7
Completed High School 4 2.8
Some College/University 10 6.9
Completed College/University 53 36.8
Professional Degree/Certificate 40 27.8
Post-Graduate Degree (Master's, Doctorate, Law) 36 25
Total 144 100
34
Measures
The picture task. In this task, children indicated their gender classification by pointing to
the picture of a Martian-like female or male figure. First, it was ensured that children were able
to differentiate the sex of each object represented in the picture. The instructions were as follows:
Regarde ces deux images. C’est un couple de personnes imaginaires. Montre-moi le garçon et la
fille. Écoute le mot flaison*. C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
flaison*? (Look at these pictures. This is a couple of imaginary beings (Martian-like persons).
Show me the girl and show me the boy. Listen to the invented word flaison*. Who would you
prefer to call flaison* between these two imaginary persons?). In this example, the correct
response is the female Martian because aison is a typically feminine ending. A total of 20
pseudowords were used. Of these, 6 had typically masculine endings (-eau, -ier, -in, -isme, -on, -
our), 6 had typically feminine endings (-aine, -elle, -ette, -ine, -otte, -ure), and 8 had unbiased
endings (-èle, -ide, -ige, -ique, -ode, -one, -oque, -ule). If the picture (male or female Martian)
and the typically marked pseudoword agreed in gender (e.g., female Martian for nasprelle*),
gender assignment on the pseudoword was considered correct and it was assigned 1 point. If the
picture and the typically marked pseudoword disagreed in gender (e.g., male Martian for
nasprelle*), the error was scored as 0. There was no correct or incorrect response for
pseudowords that were not marked for gender. The examiner was asked to circle the student’s
response (i.e., a male or female Martian). The average length of words was equated across
nominal endings: 8 letters for the masculine pseudowords, 9.5 letters for the feminine
pseudowords, and 8.8 letters for the neutral pseudowords.
35
The determiner task. In this task, children were provided with a pseudoword without any
determiner and then they were asked to repeat it. Next, the experimenter asked the children to
verbally provide the determiner unMASC or uneFEM that was the most appropriate for each
pseudoword. The instructions were as follows: J’ai inventé des nouveaux mots. Voici dostral*.
Peux-tu répéter le mot s’il te plaît? Tu préférerais dire un dostral* ou une dostral*? (I have
invented new words. This is dostral*. Can you please repeat the word? Do you prefer to say a
(masculine) dostral* or a (feminine) dostral*?). In this example, the correct response is unMASC
because –al is a typically masculine ending. The experimenter was instructed to emphasize the
distinction between unMASC and une FEM when addressing the children. The order in which the
two possible responses were proposed was counterbalanced across items. A total of 20
pseudowords were used. Of these, 6 had typically masculine endings (-eau, -ier, -in, -isme, -on, -
our), 6 had typically feminine endings (-aine, -elle, -ette, -ine, -otte, -ure), and 8 had unbiased
endings (-èle, -ide, -ige, -ique, -ode, -one, -oque, -ule). If the determiner and the typically marked
pseudoword agreed in gender (e.g., uneFEM fralocture*), gender on the determiner was
considered correct and was assigned 1 point. If the determiner and the typically marked
pseudoword disagreed in gender (e.g., unMASC fralocture*), the error was scored as 0. There was
no correct or incorrect response for pseudowords that were not marked for gender. The examiner
was asked to circle the student’s response (i.e., either unMASC / uneFEM). The average length of
words was equated across nominal endings: 8.5 letters for the masculine pseudowords, 8.7 letters
for the feminine pseudowords, and 8.4 letters for the neutral pseudowords.
All of the stimuli obeyed the phonological structure of French and began with a consonant.
In this analysis, noun endings referred to the last 2-3 phonemes that set the gender-predictive
values at .75 (Spalek, Franck, Schriefers, & Frauenfelder, 2008 as cited in Boloh & Ibernon,
2010). The bias value of each ending was founded on the data provided by Tucker et al. (1977)
36
and was confirmed by further analysis using the Lexique database (New, Pallier, Ferrand &
Matos, 2001). Word frequency estimates were reported for each lemma based on the frequency
of usage per million words (i.e., the number of times each lemma appeared in a corpus of books
per 1 million words of text).4 The bias value was calculated by filtering through real words with
the ending in question. Next, the nouns with the respective endings were separated by gender
(masculine/feminine). The predictive value (extent to which noun endings accurately predict
gender attribution) was obtained by dividing the number of masculine and feminine nouns for
each ending relative to the total number of nouns in that category within the corpus (Lyster,
2006). For example, there were 430 nouns that ended with the typically feminine ending –ette in
the Lexique database. Of these nouns, 412 were feminine (mean frequency of 3.31 per million of
words) and 18 were masculine (mean frequency of .89 per million of words). Therefore, the bias
score indicated that 96% of words that ended in –ette were typically feminine (i.e., 412/430
*100). For nouns with typical endings, bias values towards either masculinity or femininity
ranged from 72% to 100%, with a mean of 93% for masculine nouns and 91% for feminine
nouns. For nouns with neutral endings, bias values towards femininity ranged from 31% to 66%,
with a mean of 51% for masculine nouns and a mean of 49% for feminine nouns.
Control Measures
French receptive vocabulary. French receptive vocabulary was assessed using the
Échelle de Vocabulaire en Images Peabody (EVIP Form A; Dunn, Theriault-Whalen, & Dunn,
1993), which was standardized on Canadian children who were native speakers of French. The
4 The Lexique corpus was based on 14.7 million words obtained from 218 literary texts that were
published between 1950 and 2000 (New, Pallier, Ferrand & Matos, 2001).
37
test was administered according to standardized procedures. In this task, the examiner presented
a target word orally and the child was asked to identify the picture that best represented the word
from a set of four pictures. Children were given three practice items prior to the test items to
ensure understanding of the task. The measure was discontinued when the child made six errors
within a set of eight responses. The items did not elicit grammatical gender because the target
words were presented without determiners. Cronbach’s alpha reliability for this task was .98.
Lecture de mots. ‘French Word Reading’. The word-reading subtest of the Wechsler
Individual Achievement Test II (Wechsler, 2005) was used to assess the following literacy skills
in French: pre-reading skills (e.g., Dis-moi le noms de ces lettres ‘Tell me the name of these
letters’), decoding skills (Quelle lettre fait le son /a/ comme dans avion? ‘What letter makes the
/a/ sound as in ‘avion’?), and reading words from a list. The word card contained 84 words,
which varied in terms of frequency of occurrence, grapheme-phoneme regularity and the number
of syllables. The children were asked to read each word aloud, one by one, to the best of their
ability. This test was standardized on Canadian children who were native speakers of French.
Cronbach’s alpha reliability for this task was .98.
Demographic Measure
Parental demographic questionnaire. The parents completed a demographic
questionnaire designed by the researcher. Parents indicated the languages that the child spoke in
the home and the approximate percentage of time each language was spoken on a 5-point Likert
scale (never (0%), rarely (0-24%), sometimes (25-49%), frequently (50-74%), always (75-
100%)). Parents also specified whether they spoke English, French, or another language to the
child and how often they spoke the language(s) (most often, occasionally, and least often). Data
on parental education was obtained by asking the parents to indicate the highest level of
38
education attained on a 6-point scale (some high school (1), completed high school, some
college/university, completed college/university, professional degree/certificate, and post-
graduate degree (6)).
Procedure
Each child was presented with the two tasks. However, in contrast to Seigneuric et al.’s
(2007) approach, all of the participants were given the picture task first, which prevented
disclosing the purpose of the measure (i.e., no determiners were shown first). In each task, three
practice items (one for each noun subclass) were used to ensure that participants understood the
procedure. However, no feedback was provided on any of the practice and test items, to avoid
priming the children for grammatical gender. Testing for all students took place at the child’s
school during regular classroom hours in the fall term (i.e., November – December 2014) and
took approximately 45-60 minutes (over 1-2 sessions). Children were withdrawn individually
from their classes by carefully trained research assistants and taken to a familiar room in the
school to be tested. All of the measures were administered in French. However, instructions were
given in English to ensure that the children understood the purpose of the task.
Results
For each child, proportions of correctly marked masculine and feminine pseudowords
were calculated per grade by dividing the number of correctly marked pseudowords in each
category by the total number of noun endings in that category (i.e., masculine proportion = total
correct / 6, feminine proportion = total correct / 6). Next, I calculated the proportion of unmarked
pseudowords that were assigned to the masculine or feminine by dividing the number of neutral
endings in each category by the total number of pseudowords in that category (in this case, a
total score of 8, per task). Means and standard deviations of ending-consistent responses
39
according to grade, task (determiner or picture), and type of ending bias (masculine or feminine)
are shown in Table 3. The median and range of gender-neutral responses according to grade and
task are given in Table 4.
There was one univariate outlier (picture task) in the sample based on standard scores
with a value greater than 2.2 standard deviation units away from the mean (Hoaglin & Iglewicz,
1987). This case was excluded from all analyses. The data were examined for normality by
inspecting skewness and kurtosis values for each variable following the guidelines proposed in
Kirby et al. (2012) (i.e. statistic/SE < ± 3.09). Both of the experimental measures were normally
distributed. A one-way multivariate analysis of variance (MANOVA) found a significant
difference between grades 2, 4, and 6 with respect to age F(2, 145) = 2165.94, p < .001, French
receptive vocabulary F(2, 145) = 55.08, p < .001 and French word reading F(2, 145) = 71.42, p <
.001. Post hoc comparisons using the Bonferroni assumption revealed significant differences
between all three grades, p < .001, except on French word reading (p = .551) between grade 4
and 6. As shown in Table 3, French receptive vocabulary and French word reading scores
increased across grade. Overall, French L2 children lagged behind native French speakers as
indicated by the standard scores. A one-way analysis of variance (ANOVA) found no significant
difference between grades 2, 4, and 6 with respect to the distribution of maternal education, F(2,
141) = 2.79, p = .065. The average level of maternal education for all three grades was a
university/college degree.
40
Table 3. Descriptive Statistics of Measures
Grade 2 (n = 46) Grade 4 (n = 47) Grade 6 (n = 55) Combined (n = 148)
Variable M SD M SD M SD M SD
Agea 89.89 3.77 113.09 3.68 136.73 3.29 114.66 19.69
Maternal Educationb 4.74 .95 4.84 .99 4.37 1.17 4.63 1.06
French Receptive Vocabularyc 45.93 25.53 80.13 28.58 99.38 22.93 76.66 33.77
French Receptive Vocabulary SS 72.78 22.07 84.89 23.26 86.65 16.90 81.79 21.45
French Word Readingd 65.78 22.09 105.13 20.03 110.44 18.09 94.87 28.00
French Word Reading SS 84.71 16.56 95.39 13.31 90.28 14.31 90.17 15.25
Picture Feminine Correcte .64* .15 .71* .20 .69* .15 .68 .17
Picture Masculine Correcte .55 .21 .62* .18 .65* .20 .61 .20
Determiner Feminine Correcte .57 .24 .67* .22 .66* .23 .63 .23
Determiner Masculine Correcte .56* .19 .59* .20 .62* .20 .59 .20
Note. SS = Standard Scores. aIn months. b Mother’s highest level of education based on a 6-point scale (some high school (1),
completed high school, some college/university, completed college/university, professional degree/certificate, and post-graduate
degree (6)). Maximum score = 32. cNumber of known words. Maximum score = 170. dSum of the number of correctly identified
sounds/letters, and the number of words read correctly. Maximum score = 131. eProportion of correct trials (chance = .50).
*Performance significantly above chance.
41
Table 4. The median number of gender-neutral responses as a function of grade and task.
Task Gender-neutral
Grade 2
(n = 46)
Grade 4
(n = 47)
Grade 6
(n = 55)
endings Mdn Mdn Mdn
Determiner Masculine 4 (7) 4 (6) 3 (5)
Feminine 4 (7) 4 (6) 5 (5)
Picture Masculine 5 (6) 5 (5) 5 (6)
Feminine 3 (6) 3 (5) 3 (6)
Note. Mdn = Median. The range is provided in parentheses.
Picture Task
Typically Masculine and Feminine Endings. A mixed-design, two-factor ANOVA with
ending bias as the within-subjects effect and grade as a between-subjects effect was conducted to
evaluate the difference in the proportions of correctly marked masculine and feminine
pseudowords for children in grades 2, 4 and 6. The ANOVA yielded a significant main effect for
age, F(1, 144) = 3.28, p = .041, 𝜂𝑝2 = .04, suggesting gender-marking ability increased across
the three groups for both masculine and feminine words. Post hoc comparisons using the
Bonferroni adjustment revealed no significant differences in scores between grade 2 and grade 6,
p = .057, 95% CI [-.14, .001], grade 2 and grade 4, p = .124, 95% CI [-.14, .01] and grade 4 and
grade 6, p = 1.00, 95% CI [-.08, .07] in the proportion of correctly marked masculine and
feminine pseudowords. There was also a significant main effect for ending bias, F(1, 144) =
18.18, p < .001, 𝜂𝑝2 = .11. As shown in Figure 1, French L2 learners made more consistent
responses with gender-biased pseudowords that referred to female entities than to male entities in
each grade. The interaction between ending bias and grade was not significant, F(1, 144) = .62, p
= .541, 𝜂𝑝2 = .01. In addition, t tests performed on these values revealed that performance was
42
greater than chance for the feminine pseudowords (p < .001 at grade 2, 4 and 6) and for
masculine pseudowords (p < .001 at grade 4 and 6, but not significant for grade 2, p = .087).5
Figure 1. The proportion of masculine and feminine responses in the picture task for biased
pseudowords as a function of grade.
5 The homogeneity of variance assuption as assessed by Levene's Test was not satisfied (p < .05) for
feminine words in the picture task. Because the variance assumption was violated for one of the variables,
a Wilcoxon Signed-Ranks Test showed that children performed significantly better on feminine words as
opposed to masculine words in grade 2 (Z = -2.35, p = .019), grade 4 (Z = -2.39, p = .017), and grade 6 (Z
= -2.16, p = .031). A Kruskal-Wallis test evaluated the differences among the three grades on the median
change in number of correctly marked pseudowords. The test, which was corrected for tied ranks, was not
significant for either masculine, 𝝌2 (2, n = 147) = 5.14, p = .077 or feminine pseudowords 𝝌2 (2, n = 147)
= 4.67, p = .097.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Grade 2 Grade 4 Grade 6
Pro
po
rtio
n o
f en
din
g-c
on
sist
ent
resp
on
ses
Grade
Masculine
Feminine
43
Neutral Endings. In each grade, a one-sample Wilcoxon Signed-Ranks test investigated
whether the number of gender-neutral pseudowords that were assigned to the masculine or
feminine differed from a hypothetical median of 4.5 (i.e., there were 8 gender ambiguous
pseudowords, thus, a value of 4.5 would indicate no default in terms of assignment). The median
and range of gender-neutral words according to grade and task are shown in Table 4. The results
indicated that the children assigned significantly more gender-neutral endings to the masculine
subclass than the feminine in grade 2, W = 126.00 p < .001, in grade 4, W = 81.50 p < .001, and
in grade 6, W = 201.00 p < .001. As shown in Figures 2-4, the observed median (Mdn = 3) was
significantly different than the hypothesized median (Mdn = 4.5) in each grade, p < .001. In each
figure, the observed median (represented by the solid line) falls to the left of the hypothesized
median (represented by the dotted line). In grade 2 and 4, the majority of children assigned three
gender-unmarked nouns to the feminine subclass, which implied that the remaining pseudowords
(i.e., five) were assigned to the masculine subclass. In grade 6, the participants assigned six
gender-neutral nouns to the masculine subclass. Taken together, the results suggest that the
masculine gender class was more likely to be overused than the feminine in cases where gender-
endings were neutral (e.g., spadique*).
44
Figure 2. The number of gender-unmarked nouns assigned to the feminine subclass in grade 2.
Figure 3. The number of gender-unmarked nouns assigned to the feminine subclass in grade 4.
0
2
4
6
8
10
12
14
16
18
20
1 2 3 4 5 6 7
Fre
qu
ency
Number of gender-unmarked nouns assigned to the feminine subclass
0
2
4
6
8
10
12
14
16
18
20
1 2 3 4 5 6
Fre
qu
ency
Number of gender-unmarked nouns assigned to the feminine subclass
45
Figure 4. The number of gender-unmarked nouns assigned to the feminine subclass in grade 6.
Determiner Task
Typically Masculine and Feminine Endings. A mixed-design, two-factor ANOVA with
ending bias as the within-subjects effect and grade as a between-subjects effect was conducted to
evaluate the difference in the proportions of correctly marked masculine and feminine
pseudowords for children in grades 2, 4 and 6. The ANOVA yielded a significant main effect for
age, F(1, 145) = 3.49, p = .033, 𝜂𝑝2 = .05. As shown in Figure 5, gender-marking ability
increased across the three groups for both masculine and feminine words. Post hoc comparisons
using the Bonferroni adjustment revealed a significant difference between grade 2 and grade 6, p
= .045, 95% CI [-.15, -.001] but, no significant differences between grade 2 and grade 4, p =
.108, 95% CI [-.15, .01] and grade 4 and grade 6, p = 1.00, 95% CI [-.08, .07] in the proportion
of correctly marked masculine and feminine pseudowords. The main effect of ending bias was
not significant, F(1, 145) = 3.21, p = .075, 𝜂𝑝2 = .02 meaning that there was no difference
between the proportion of ending-consistent responses given to masculine and feminine endings.
0
2
4
6
8
10
12
14
16
18
20
1 2 3 4 5 6 7
Fre
qu
ency
Number of gender-unmarked nouns assigned to the feminine subclass
46
The interaction between ending bias and grade was also not significant, F(1, 145) = .70, p = .498,
𝜂𝑝2 = .01. In addition, t tests performed on these values revealed that performance was greater
than chance for the feminine pseudowords (p < .001 at grade 4 and 6, but not significant for
grade 2, p= .071) and for masculine pseudowords (p = .048 at grade 2, p = .003 at grade 4, and p
< .001 at grade 6).
Figure 5. The proportion of masculine and feminine responses in the determiner task for biased
pseudowords a function of grade.
Neutral Endings. In each grade, a one-sample Wilcoxon Signed-Ranks test investigated
whether the number of gender-neutral pseudowords that were assigned to the masculine or
feminine differed from a hypothetical median of 4.5 (i.e., there were 8 gender ambiguous
pseudowords, thus, a value of 4.5 would indicate no default in terms of assignment). The median
and range of gender-neutral words according to grade and task are shown in Table 4. The results
indicated that the children did not default to one subclass over the other in grade 2, W = 466.50 p
= .410, in grade 4, W = 503.50 p = .511, and in grade 6, W = 782.00 p = .918. As shown in
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Grade 2 Grade 4 Grade 6
Pro
po
rtio
n o
f en
din
g-c
on
sist
ent
resp
on
ses
Grade
Masculine
Feminine
47
Figure 6 -8, the distance between the observed (Mdn = 4 in grade 2 and 4, Mdn = 5 in grade 6)
and the hypothesized median (Mdn = 4.5) is relatively close to the midpoint, indicating that there
was no default subclass. In grade 2 and 4, children were equally likely to assign gender-
unmarked nouns to the masculine as well as the feminine subclass. In grade 6, the observed
median fell to the right of the hypothesized median as children assigned slightly more gender-
neutral nouns to the feminine, as opposed to the masculine subclass. Taken together, the results
suggest that neither subclass appeared to act as the default value in cases where gender-endings
were ambiguous (e.g., flaumide*).
Figure 6. The number of gender-unmarked nouns assigned to the feminine subclass in grade 2.
0
2
4
6
8
10
12
14
16
0 1 2 3 4 5 6 7
Fre
qu
ency
Number of gender-unmarked nouns assigned to the feminine subclass
48
Figure 7. The number of gender-unmarked nouns assigned to the feminine subclass in grade 4.
Figure 8. The number of gender-unmarked nouns assigned to the feminine subclass in grade 6.
0
2
4
6
8
10
12
14
16
2 3 4 5 6 7 8
Fre
qu
ency
Number of gender-unmarked nouns assigned to the feminine subclass
0
2
4
6
8
10
12
14
16
2 3 4 5 6 7
Fre
qu
ency
Number of gender-unmarked nouns assigned to the feminine subclass
49
Discussion
Study 1 investigated grammatical gender assignment ability among three groups of
bilingual children who were enrolled in an early French immersion program. In terms of the first
research objective, gender-marking ability was examined across grades 2, 4 and 6. As expected,
the results revealed that in both the picture and determiner tasks, the ability to assign gender to
typically biased pseudowords improved with grade. However, performance in both tasks lagged
behind that of native French speakers of a comparable age (e.g., Seigneuric et al., 2007). The
second aim of the current study was to investigate the effect of morphophonological cues on
gender assignment. The results indicated that more ending-consistent responses were given to
feminine endings than to masculine endings in both tasks at each time point, although this
difference only reached significance in the picture task. The third query of the present study
concerned the extent to which a particular noun subclass acted as the default gender value. The
results showed that in the picture task, children assigned gender-neutral endings to the masculine
subclass significantly more often than to the feminine subclass. This trend was not observed in
the determiner task.
Although gender assignment increased across the three grades for French L2 learners, its
development appears to be slower than that of native French speakers. Based on research
conducted with monolingual French speakers (e.g., Karmiloff-Smith, 1979), grammatical gender
is well developed by the ages of 3 or 4. Likewise, in Seigneuric et al.’s (2007) study, there was
an increase in the effect of morphosyntactic cues on gender endings in both tasks, such that, by
age 9 monolingual French speaking children correctly identified feminine pseudowords 78% of
the time, and masculine words 87% of the time in the determiner task. In the picture task,
children correctly assigned feminine words 90% of the time and masculine words 95% of the
time. In the current study, the gender-marking ability of the L2 speakers was noticeably inferior
50
to that of native speakers as performance in grade 2 was only slightly above chance for
masculine nouns in the determiner task (56%) and feminine nouns in the picture task (64%), and
was below chance for feminine nouns in the determiner task and masculine nouns in the picture
task. However, in grade 4, L2 learners were significantly above chance in both tasks, and
performance improved relative to children in grade 2, in both the picture task (62% for masculine
nouns and 71% for feminine nouns) and the determiner task (59% for masculine nouns and 67%
for feminine nouns). In grade 6, French L2 learners correctly assigned masculine pseudowords
65% of the time and feminine pseudowords 69% of the time to their appropriate subclass in the
picture task. In the determiner task, children correctly assigned masculine words 62% of the time
and feminine words 66% of the time. In summary, like native French speakers, gender
assignment ability improved across the three groups. However, in the current research, the
bilingual children demonstrated awareness of grammatical gender much later (around age 9)
when compared to the existing literature on French L1 children (Karmiloff-Smith, 1979) and
they never achieved equivalent levels of performance.
According to the Failed Functional Features Hypothesis (FFFH) (e.g., Hawkins & Chan,
1997), the participants in this study missed the sensitive learning period for gender acquisition
(as exposure to French began around 5 years of age) and would thus face numerous challenges
with gender attribution, especially for those children whose L1 did not share the linguistic feature
(e.g., Franceschina, 2005). That being said, from a developmental perspective, the current
findings do not support the assumptions of the FFFH. Given their above chance performance in
both tasks (except at Grade 2), the results suggest that English L1 learners6 have developed at
6 The majority of the children in the current study (95.3%) spoke English in the home.
51
least some knowledge about gender assignment in French, similar to native speakers, although
this ability was still not fully acquired. Instead, these findings can be situated within the Full
Transfer Full Access (FTFA) model (e.g., Schwartz & Sprouse, 1994) as grammatical features
that are absent from the L1 can be acquired with sufficient input to the L2, regardless of the age
of acquisition. Nevertheless, some researchers have argued that despite many years of formal
education, English L1 learners of French continue to struggle with gender assignment (Andersen,
1984; Surridge, 1993). Future research should examine gender assignment ability in an older
group of L2 learners of French enrolled in an immersion program.
In terms of the second important finding in this study, the results showed that French L2
learners made more consistent responses with gender biased pseudowords that referred to
feminine words than to masculine words in both tasks at each time point, although this difference
only reached significance in the picture task. Although, grammatical gender corresponds to
natural gender in about 10% of nouns (Séguin, 1969), in the picture task, it was shown that
children correctly associated morphophonological cues carried by the noun’s ending with
semantic entities (e.g., -ette as in ploumiette* representing a female Martian). In the determiner
task, the participants correctly associated morphophonological cues carried by the noun’s ending
with the corresponding linguistic article (e.g., unMASC bilacron* ‘a bilacron*’). Similar to the
picture task, the results corroborated studies on French L1 speakers, as more consistent responses
were given to feminine endings than to masculine endings (Karmiloff-Smith, 1979; Seigneuric et
al., 2007), although the main effect for ending bias only approached significance. Moreover, the
results concurred with the form-focused route (Gollan and Frost, 2001), as nouns were assigned
to a gender subclass based on their ending. It appears then, that in this study, French L2 children
relied on noun endings as a cue to assigning gender to nouns, given that they did not have an
52
entry in their mental lexicon for pseudowords (consistent with French L2 adult speakers in
Holmes & Dejean de la Bâtie, 1999).
Interestingly, in Seigneuric et al.’s (2007) study, in both tasks, the effect of
morphosyntactic cues emerged earlier for feminine endings than for masculine endings among a
group of native French speakers (between the ages of 3 and 9). However, older children (5, 6, 7,
and 9 year olds in the determiner task, and 6 and 9 year olds in the picture task) made more
ending-consistent responses to masculine pseudowords. According to Karmiloff-Smith (1979),
during the initial stages of gender acquisition, younger children assume that grammatical gender
is determined by natural gender. Over time, they begin to attend to morphophonological cues
given sufficient input in the target language. Likewise, the results of the current study revealed
that more ending consistent responses were given to feminine biased pseudowords in each grade.
Since there is a delay in gender-assignment ability relative to native French speakers, future
research ought to investigate whether an increase in ending consistent responses to masculine
biased words in addition to feminine biased pseudowords holds true in an older group of French
immersion students. Before addressing gender-neutral endings, the analysis investigated whether
the default strategy was employed for typically biased pseudowords. The results indicated that in
contrast to Boloh and Ibernon’s (2010) theory, the default strategy was not observed for typically
masculine or typically feminine endings. In other words, performance on masculine endings was
not at ceiling at all ages, and scores for feminine endings were not significantly below those of
masculine endings, but were also above chance level. Hence, as expected, children relied on
noun endings as an effective cue during gender attribution.
The question that remained though was whether one subclass was more likely to be
overused than the other in cases where the gender-endings were unbiased (e.g., -ide as in
53
flaumide*). Overall, the masculine as a default appeared to be a robust strategy in the picture
task, as children attributed gender-neutral endings to male entities significantly more often than
to female ones in each grade (e.g., pointing to the picture representing a male Martian for
noplide*). The observed median was found to differ significantly from the hypothesized median,
indicating that children overused the masculine subclass compared to the feminine. It appears
that for gender-neutral pseudowords, French L2 children were unable to rely on
morphophonological cues (i.e., the form focused route) for non-words, given that unbiased
endings are equally dispersed across both genders. Conversely, this tendency was not observed in
the determiner task, as roughly half of the pseudowords were assigned to the masculine subclass,
while the other half were assigned to the feminine. One explanation for the discrepancy in the
results lies in the nature of the tasks employed for this study. For example, in the determiner task,
French L2 children might have relied on the ‘backward processing’ strategy (Tucker et al., 1997)
by processing the noun’s ending from right to left in conjunction with the linguistic article
(determiner) before assigning the gender-neutral pseudoword to a particular subclass. It appears
that semantic information (in the picture task) did not have the same effect. Overall, the results in
Study 1 showed that morphophonological rules played an important role in gender attribution for
typically biased words. However, it seems that words syntactically related to the noun such as
determiners may be used in conjunction with morphophonological cues when assigning gender
to pseudowords with gender-neutral endings.
In summary, the objective of Study 1 was to examine the impact of morphophonological
cues on gender attribution across three groups of French immersion children. Taken together, in
both tasks, more ending-consistent responses were given to feminine biased pseudowords.
Within the statistical learning framework, it may be argued that feminine regularities were more
frequent and thus, were more salient to children in the early stages of gender acquisition.
54
Conversely, an ending such as –oque, is considered to be typically neutral. Consequently,
children were more likely to default to one subclass (e.g., the masculine) in instances where they
were uncertain. As such, future studies should employ lexical and sublexical choice tasks to
examine the relationship between orthographic regularities and grammatical gender-marking
ability. Furthermore, another avenue of research should consider the relationship between
spelling patterns and gender acquisition by analyzing the results obtained from spelling tests.
55
Chapter 6: Study 2
Given the paucity of research on minority language children in French immersion, and
motivated by a gap in L3 research, the objective of Study 2 was to investigate gender-marking in
French, as a function of home language status among three groups of French immersion students:
children whose home language marks gender (e.g., Spanish), children whose home language is a
language other than English that does not mark gender (e.g., Tamil) and English-first language
children. Therefore, French was a third language for the first two groups and a L2 for the final
group. The methodology employed in the current study was the same as the one reported in
Study 1. In both tasks (i.e., picture and determiner), children whose L1 marks gender were
expected to outperform children whose L1 does not mark gender (consistent with Sabourin et al.,
2006; Schwartz et al., 2014).
Method
Participants
A subgroup from the 148 children who participated in Study 1 was identified using the
parental demographic questionnaire data. A total of 27 children (M = 120.74 months, SD =
16.14, 10 males) took part in the study. Table 5 displays the frequency distribution for each
language background. Using the parental demographic questionnaire, I first identified 9 children
(M = 121.56 months, SD = 17.26, 2 males) whose L1 marks gender, which included Polish,
Spanish (ns = 2), Arabic, Greek, Portuguese, Russian and Slovak (ns = 1).7 For these
participants, French was a third language. Next, I identified 9 children (M = 119.56, SD = 17.80,
7 In total, there were 11 children whose L1 marks gender. However, two grade 2 students were
excluded from the analysis because there was no match available in the L3_NMG group.
56
3 males), who had been exposed to a L1 other than English that does not mark gender. The
languages represented in this group included: Tamil (n = 3), Mandarin (n = 2), Farsi, Filipino,
Korean and Tibetan (ns = 1). For these participants, French was a third language as it was for the
first group. Finally, the third group was comprised of 9 English-first language children (M =
121.11, SD = 15.13, 5 males) for whom French was a second language. From hereon in, the
group whose L1 marks gender, the group whose L1 other than English that does not mark
gender, and English-first language children are labeled as L3_MG group, the L3_NMG group
and the L2_NMG group respectively. As the pool of English L1 participants was very large, I
matched the L2_NMG group with the L3_MG and the L3_NMG groups on French receptive
vocabulary, French word reading, place of birth, grade and, when possible, for gender. There
were 2 students in grade 2, 4 students in grade 4 and 4 students in grade 6 in each language
group. Due to the small sample size, the data from the three grades were collapsed for all
statistical analyses.
As described in Study 1, the parents completed a demographic questionnaire about language,
demographic and educational information. Within the group of 27 children, 18 learned to speak a
language other than English as their first language (i.e., the L3_MG and L3_NMG groups). Of
the 27 children, 22.2% of the children in the L3_MG group and 55.6% of the L3_NMG group
spoke English more than 75% of the time. Approximately, 44.4% of the children in the L3_MG
group and 33.3% of the children in the L3_NMG group reported speaking their L1 more than
50% of the time. Furthermore, 87.5% of children in the L3_MG group, 44.4% of children in the
L3_NMG group and 66.7% of children in the L2_NMG group had more than 50 books at home.8
Moreover, 29% of parents in the L3_MG group, 25% in the L3_NMG group and 33.3% of
8 One participant from the L3_MG group did not respond to this question.
57
parents in the L2_NMG group reported that they read with their child almost everyday.9
Approximately, 55.6% of the children in the L3_MG group, 33.3% of the children in the
L3_NMG group and 77.8% of children in the L2_NMG group engaged in daily independent
reading.10 The frequency distribution of maternal education is reported in Table 6. In this sample,
100% of the mothers in the L3_MG group, 75% of the mother in the L3_NMG group and 66.7%
of the mothers in the L2_NMG group completed at least a college/university degree.11 Finally,
88.9% of the children in the L3_MG group, 87.5% of the children in the L3_NMG group and
100% of the children in the L2_NMG group were born in Canada. For those who were born
outside of Canada, the average age of immigration was 24 months.12
9 Two participants from the L3_MG group and 1 participant from the L3_NMG group did not
respond to this question.
10 Two participants from the L3_MG group and two participants from the L3_NMG group did
not respond to this question.
11 One participant from the L3_MG group and one participant from the L3_NMG group did not
respond to this question.
12 One participant from the L3_NMG group did not respond to this question.
58
Table 5. Frequency Distribution of Language Background
Language Frequency Percent (%)
Child's most often spoken language in the home
English 21 77.8
Other 6 22.2
Total 27 100
Child's first language he/she learned to speak at home
Arabic 1 3.7
English 9 33.3
Farsi 1 3.7
Filipino 1 3.7
Greek 1 3.7
Korean 1 3.7
Mandarin 2 7.4
Polish 2 7.4
Portuguese 1 3.7
Russian 1 3.7
Slovak 1 3.7
Spanish 2 7.4
Tamil 3 11.1
Tibetan 1 3.7
Total 27 100
Table 6. Frequency Distribution of Maternal Education
Level of Education Frequency Percent (%)
Some High School
1
3.7
Completed High School
1
3.7
Some College/University
2
7.4
Completed College/University
11
40.7
Professional Degree/Certificate
4
14.8
Post-Graduate Degree (Master's, Doctorate, Law, Medicine)
6
22.2
Missing
2
7.4
Total 27 100
Measures
The test battery was made up of standardized and experimental measures. In the present
study, children were assessed on the picture task and the determiner task described in Study 1. In
addition, parents completed the same demographic questionnaire as in the first study. Two
control measures (i.e., French receptive vocabulary and French word reading described in Study
59
1) and four demographic variables (i.e., maternal education, grade, gender, place of birth) were
used as matching variables.
Procedure
The methodology employed in the current study was the same as the one reported in
Study 1.
Results
For each child, proportions of correctly marked masculine and feminine pseudowords were
calculated per language by dividing the number of correctly marked pseudowords in each
category by the total number of noun endings in that category (i.e., masculine proportion = total
correct / 6, feminine proportion = total correct / 6). Means and standard deviations of ending-
consistent responses according to language, task (determiner or picture), and type of ending bias
(masculine or feminine) are given in Table 7.
There were no univariate outliers in the sample based on standard scores with a value
greater than 2.2 standard deviation units away from the mean (Hoaglin & Iglewicz, 1987). The
data was examined for normality by inspecting skewness and kurtosis values for each variable
following the guidelines proposed in Kirby et al., (2012) (i.e. statistic/SE < ± 3.09). The data fell
within the acceptable range for all experimental measures. A one-way multivariate analysis of
variance (MANOVA) examined whether age, French receptive vocabulary, and French word
reading differed across a categorical variable, language group (i.e., L3_MG, L3_NMG, and
L2_NMG). As indicated in Table 7, the L2_NMG group scored the highest in French receptive
vocabulary and French word reading when compared to the L3_MG and the L3_NMG groups.
However, the MANOVA revealed that there were no significant differences among the three
groups with respect to age F(2, 24) = .04, p = .965, French receptive vocabulary F(2, 24) = .20, p
60
= .818, and French word reading F(2, 24) = .04, p = .961. A one-way analysis of variance
(ANOVA) found no significant difference between the three groups (i.e., L3_MG, L3_NMG,
and L2_NMG) with respect to the distribution of maternal education, F(2, 22) = 3.22, p = .059.
61
Table 7. Descriptive Statistics of Measures
Variable L3_MG (n = 9) L3_NMG (n = 9) L2_NMG (n = 9) Combined (n = 27)
M SD M SD M SD M SD
Agea 121.56 17.26 119.56 17.80 121.11 15.13 120.74 16.14
Maternal Educationb 5.13 .99 3.63 1.60 4.33 .87 4.36 1.29
French Receptive Vocabularyc 76.78 46.23 80.89 24.67 87.56 34.63 81.74 35.13
French Receptive Vocabulary SS 83.63 24.72 80.78 20.14 85.11 22.71 83.15 21.66
French Word Readingd 101.89 40.68 98.33 39.17 102.56 18.62 100.93 33.04
French Word Reading SS 98.88 14.94 98.25 15.35 88.44 11.52 94.92 14.24
Picture Feminine Correcte .78* .08 .63* .11 .63 .25 .68 .17
Picture Masculine Correcte .76* .21 .65* .10 .61 .24 .67 .19
Determiner Feminine Correcte .80* .22 .69* .26 .50 .26 .66 .27
Determiner Masculine Correcte .70* .18 .54 .22 .57 .15 .60 .19
Note. SS = Standard Scores. aIn months. b Mother’s highest level of education based on a 6-point scale (some high school (1),
completed high school, some college/university, completed college/university, professional degree/certificate, and post-graduate
degree (6)). Maximum score = 32. cNumber of known words. Maximum score = 170. dSum of the number of correctly identified
sounds and letters, and number of words read correctly. Maximum score = 131. eProportion of correct trials (chance = .50).
*Performance significantly above chance.
62
Picture Task
Typically Masculine and Feminine Endings. An a priori contrast analysis, found a
significant difference in favour of the L3_MG group when compared with the L2_NMG group in
the proportions of correctly marked masculine and feminine pseudowords that were assigned to
female and male entities, p = .042, 95% CI [-.29, -.01]. However, no significant difference was
found between the L3_NMG group and the L3_MG group, p = .072, 95% CI [-.27, .01]. In
addition, t tests performed on these values revealed that performance was greater than chance for
the feminine pseudowords (p < .001 for L3_MG, p = .008 for L3_NMG, but not significant for
L2_NMG p = .154) and for masculine pseudowords (p = .005 for L3_MG, p = .002 for
L3_NMG, but not significant for L2_NMG group p = .195). As shown in Figure 9, the L3_MG
group made significantly more ending-consistent responses for both masculine and feminine
nouns when compared to the L2_NMG group.
Figure 9. The proportion of masculine and feminine responses (with standard error bars) in the
picture task for biased pseudowords as a function of language.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
L3_MG L3_NMG L2_NMG
Pro
port
ion
of
end
ing
-con
sist
ent
resp
on
ses
Language
Masculine
Feminine
**: p = .042**
**
63
Determiner Task
Typically Masculine and Feminine Endings. An a priori contrast analysis found a
significant difference in favour of the L3_MG group when compared with the L2_NMG group in
the proportions of correctly marked masculine and feminine pseudowords, p = .005, 95% CI [-
.36, -.07]. There was a trend toward a significant difference between the L3_NMG group and the
L3_MG group, p = .056, 95% CI [-.28, .004]. In addition, t tests performed on these values
revealed that performance was greater than chance for the feminine pseudowords (p = .003 for
L3_MG, p = .062 for L3_NMG, but not significant for L2_NMG p = 1.00) and for masculine
pseudowords (p = .010 for L3_MG, but not significant for L3_NMG p = .622, and L2_NMG
group p = .169). As shown in Figure 10, the L3_MG group made significantly more ending-
consistent responses for both masculine and feminine nouns when compared to the L2_NMG
group.
Figure 10. The proportion of masculine and feminine responses (with standard errors bars) in the
determiner task for biased pseudowords as a function of language.
64
Discussion
Due to the relative dearth of studies, the process by which French L3 children learn to
mark gender remains unclear. To fill the gap of previous research, the objective of Study 2 was
to examine gender-marking in French, as a function of home language status among three groups
of French immersion students: children whose home language marks gender (L3_MG), children
whose home language is a language other than English that does not mark gender (L3_NMG)
and English-first language children (L2_NMG). Building on the foundational research in cross-
language transfer, Lado’s Contrastive Analysis hypothesis (1957) stated that it is possible to
determine which elements of the target language will be difficult for learners from a given L1
background based on structural similarities and differences between the L1 and the additional
language. Based on this theory, positive transfer was expected for the L3_MG group learning
French as the presence of the gender system in both the home language and the language of
instruction was predicted to facilitate category acquisition. On the other hand, negative transfer
was anticipated for the L3_NMG and the L2_NMG groups, based on the absence of gender in
the L1.
To verify these predictions, a priori contrast analyses were conducted for both tasks in
which I found a significant difference in favour of the L3_MG group when compared to the
L2_NMG group in the proportion of correctly marked pseudowords; none of the remaining
comparisons were significant although there was a trend toward a significant difference between
the L3_NMG and the L3_MG groups. In other words, the results from the current study suggest
that there is some evidence that gender-marking ability acquired in the L1 (e.g., Spanish or
Russian) supports the acquisition of gender in a third language (e.g., French), as indicated by the
superior performance of the L3_MG group. Therefore, it appears that structural features that are
absent from the L1 may not be readily learned in a second language as performance for the
65
English-first language group was below chance in both tasks. Interestingly, in this study, the
L3_NMG group almost always assigned pseudowords at a level above chance. Thus, it seems
that gender can be acquired to some extent for French L3 children whose L1 does not mark
gender but, its development appears to be slower than that of French L3 learners whose L1 marks
gender. Ultimately, to the best of my knowledge, this study is the first to support cross-transfer of
gender-marking ability between the L1 and the L3, consistent with Lado’s Contrastive Analysis
Hypothesis (1957).
One possible explanation for the L3_MG group’s greater performance in the determiner
task lies in the “chunking” strategy, which is the product of exposure to the oral form of the word
and its determiner. According to Carroll (1989), native French-speaking children learn new
words as one unit (e.g., labicyclette, ‘the bicycle’), whereas, English L1 children learning French
process nouns and their modifiers as separate lexical items (e.g., laFEM bicyclette). According to
Carroll (1989), “endings do not exist.” (p. 563) However, the results in the current study suggest
that morphophonological cues are important. By analogy, the L3_MG group may have used the
“chunking strategy” in the determiner task by associating the correct article after having
processed the pseudoword’s morphophonological ending from right to left (i.e., relied on the
“backward processing strategy” Tucker et al., 1977) (e.g., unesinourette*, ‘a sinourette*’).
Furthermore, the L3_MG group may have relied more heavily on morphophonological cues
(given the presence of gender marking in their L1) when compared to the L3_NMG and the
L2_MG groups.
Applying Kuo and Anderson’s structural sensitivity theory (2010) to language
acquisition, it appears that the L3_MG group’s heightened awareness to language across diverse
contexts made them more sensitive to gender marking, and they were able to learn more quickly
66
which endings were more likely to be masculine or feminine, facilitated by the presence of
gender in the L1. More recently, Kahn-Horwitz, Kuash, Ibrahim and Schwartz (2014), found that
compared to bilingual children (e.g., Hebrew L1, English L2 learners), “multilingual children can
draw on a larger and more varied linguistic and cognitive repertoire of resources” that exist in
their first or additional language when they acquire specific elements in a new language (Geva,
2014, p. 10). In other words, studies, including the current one, suggest that the L3_NMG group
may have had a slight advantage in gender-marking because they had access to two linguistic
systems (i.e., more language experience) that shared the same linguistic feature (i.e., gender-
marking) when acquiring a third language as opposed to their English monolingual counterparts
(Cenoz, 2003). Future research on L3 acquisition is necessary to support these results.
67
Chapter 7: Conclusion
Limitations and Future Research
When evaluating the results, a few limitations of the current study have to be taken into
account. First, longitudinal follow-up is necessary to examine the development of gender
performance and whether the gap between students whose first language marks gender and
students whose L1 does not mark gender changes over time. Second, our sample size (especially
in Study 2) was small which limits the generalizability of the results. Third, a monolingual
French comparison group would have provided a clearer picture of the nature of the relationship
between gender assignment in native speakers compared to French L2 learners. Finally, the
linguistic backgrounds of the participants in Study 2 were scattered among a variety of languages
(ns = 1-3 children per language), which prevented the researcher from examining the results in
terms of their typological profiles. Future research will follow up on this study by investigating
gender-marking abilities across more homogenous groups of children in French immersion:
Romance language speakers (e.g., Italian), students whose home language is a non-Romance
language but marks gender (e.g., Hebrew) and students whose home language does not mark
gender and is not English (e.g., Chinese). These studies may show that second language
acquisition of grammatical gender is affected more by the structural similarity of gender marking
between the first language and an additional language than by the mere presence of gender
features in the L1.
According to Lyster (2006), final phonemes are not the most reliable predictors of
phonological endings when compared to orthographic representations. For example, in French,
/ε/ is a feminine ending when spelled -aie but, a masculine ending when spelled -ais. Hence, a
potential avenue for future research includes investigating the relationship between
68
morphophonological cues and orthographic regularities in French using both spoken and written
measures of gender assignment. Furthermore, it would be interesting to explore gender-marking
acquisition across different types of French programs (e.g., core French vs. French immersion).
Likewise, it would be fascinating to examine gender development in multi-language (French-
Spanish-German-Mandarin) immersion programs, an innovative program that is under
negotiation in the city from which this thesis sample was drawn.
Educational and Theoretical Implications
Typically, French immersion students recognize that they can use many anglicisms and
commit grammatical errors in their speech and writing and still be understood by their peers
(Lyster, 1987 as cited in Warden, 1997). In fact, teachers may be unwilling to interrupt the flow
of the conversation in order to correct for gender mistakes, as error correction is not often
emphasized (Harley, 1993). Furthermore, although there is a large supply of French grammatical
textbooks, typically attention to gender is sparse. From a more practical point of view, teachers
can teach students effective strategies to solve the problem of gender assignment by making
them more aware of the orthographic patterns at the ends of nouns that serve as cues for
predicting grammatical gender. The mastery of grammatical gender can improve the fluency of
communication in everyday language use. Ultimately, analytic strategies such as drawing
children’s attention to word ending regularities may be necessary to complement the experiential
learning in French immersion programs (Harley, 1993).
From a more theoretical point of view, although the French gender attribution system is
opaque and arbitrary (Corbett, 1991), morphophonological rules based on a noun’s ending serve
as a useful strategy in noun subclass assignment (Lyster, 2006). According to the FFFH
(Hawkins & Chan, 1997) the children in the aforementioned study missed the sensitive learning
69
period for gender acquisition and were expected to face difficulties with gender attribution. That
being said, from a developmental perspective, the findings from Study 1 do not support the
assumptions of the FFFH. The results provided evidence that correct gender assignment
improved across grade. Thus, in accordance with the FTFA (e.g., Schwartz & Sprouse, 1994),
with sufficient input in the French language, the learning trajectory for gender marking is
comparable between L1 and L2 speakers in French. Furthermore, in Study 2, there was some
evidence that gender-marking ability acquired in the L1 supports the acquisition of gender in a
third language. Given their diversity of linguistic experience, multilingual children should not be
considered as a uniform group. For children whose L1 marks gender, French immersion teachers
may be able to draw on the similarities and differences in gender-marking patterns between the
L1 and French. Consequently, children whose first language does not mark gender may require
additional support. Taken together, explicit form-focused gender-marking instruction (based on
word ending regularities) is essential for attaining native-like mastery in gender assignment.
70
References
Allen, M. (2004). Reading achievement of students in French immersion programs. Educational
Quarterly Review, 9(4), 25-30.
Andersen, R. (1984). What’s gender good for, anyway? In R. Andersen (ed.), Second languages:
A cross-linguistic perspective, pp. 77–99. Rowley, MA: Newbury House.
Andriamamonjy, P. (2000). Le rôle du genre grammatical au cours de la reconnaissance de
noms. L’Année Psychologique, 100(3), 419–442.
Au‐Yeung, K., Hipfner‐Boucher, K., Chen, X., Pasquarella, A., D'Angelo, N., & Hélène Deacon,
S. (2015). Development of English and French Language and Literacy Skills in EL1 and
EL French Immersion Students in the Early Grades. Reading Research Quarterly, 50(2),
233-254.
Bartning, I. (2000). Gender agreement in L2 French: Pre‐advanced vs advanced learners. Studia
linguistica, 54(2), 225-237.
Bérubé, D., & Marinova-Todd, S. H. (2012). The development of language and reading skills in
the second and third languages of multilingual children in French
Immersion. International Journal of Multilingualism, 9(3), 272-293.
Boloh, Y., & Ibernon, L. (2010). Gender attribution and gender agreement in 4- to 10-year-old
French children. Cognitive Development, 25(1), 1–25.
Boloh, Y., & Ibernon, L. (2013). Natural gender, phonological cues and the default grammatical
gender in French children. First Language, 33(5), 449–468.
Bruhn de Garavito, J.,&White, L. (2002). The second language acquisition of Spanish DPs: The
71
status of grammatical features. In A. T. Pérez-Leroux & J. M. Liceras (eds.), The
acquisition of Spanish morphosyntax: The L1/L2 connection, pp. 153–178. Dordrecht:
Kluwer.
Canadian Parents for French. (2011). Retrieved from
http://cpf.ca/en/files/Prov-Enrol-Stats-2010-20113.pdf.
Carroll, S. (1989). Second-language acquisition and the computational paradigm*. Language
Learning, 39 (4), 535–94.
Cenoz, J. (2003). The additive effect of bilingualism on third language acquisition: A review.
The International Journal of Bilingualism, 7, 71-87.
Clark, E. V. (1985). The acquisition of Romance, with special reference to French. In D. Slobin
(Ed.) The crosslinguistic study of language acquisition: Hillsdale, NJ: Erlbaum.
Comrie, B. (1999). Grammatical gender systems: A linguist’s assessment. Journal of
Psycholinguistic Research, 28(5), 457–466.
Corbett, G. G. (1991). Gender. Cambridge textbooks in linguistics. Cambridge: Cambridge.
Corbett, G. G. (2006). Gender, Grammatical. The Encyclopedia of Language & Linguistics.
2nd Edition. Oxford: Elsevier, 749-756.
Cummins, J. (1981). The role of primary language development in promoting educational
success for language minority students. In California State Department of Education
(Ed.), Schooling and language minority students: A theoretical framework (pp. 3-49). Los
Angeles, CA: National Dissemination and Assessment Center.
Cuza, A., & Pérez-Tattam, R. (2014). Grammatical gender selection and phrasal word order in
child heritage Spanish: A feature re-assembly approach. Bilingualism: Language and
Cognition, 1–19.
Desrochers, A., & Paivio, A. (1990). Le phonème initial des noms inanimés et son effet sur
72
l'identification du genre grammatical. Canadian Journal of Psychology/Revue canadienne
de psychologie, 44(1), 44.
Dewaele, J. M., & Véronique, D. (2001). Gender assignment and gender agreement in advanced
French interlanguage: A cross-sectional study. Bilingualism: Language and
Cognition, 4(03), 275-297.
Dunn, L. M., Thériault-Whalen, C. M., & Dunn, L. M. (1993). Échelle de vocabulaire
en images Peabody. Toronto: Psycan.
Fleischman, H. L., Hopstock, P. J., Stephenson, T. G., Pendzick, M. L., & Sapru, S.
(2003). Descriptive study of services to LEP students and LEP students with
disabilities (Vol. 4). US Department of Education, Office of English Language
Acquisition, Language Enhancement, and Academic Achievement for Limited English
Proficient Students.
Franceschina, F. (2005). Fossilised second language grammars: the acquisition of grammatical
gender. Amsterdam: John Benjamins.
Geva, E. (2014). The Cross-Language Transfer Journey – A Guide to the Perplexed. Introduction
to the special issue: Crosslinguistic Transfer in Reading in Multilingual Contexts –
Recent Research Trends, Written Language & Literacy 17 (1), 1–15.
Gollan, T. H., & Frost, R. (2001). Two routes to grammatical gender: Evidence from
Hebrew. Journal of psycholinguistic research, 30(6), 627-651.
Grevisse, M. (1980). Le bon usage. (11 e édit). Bruxelles: Duculot.
Grüter, T., Lew-Williams, C., & Fernald, A. (2012). Grammatical gender in L2: A production or
a real-time processing problem? Second Language Research, 28(2), 191–215.
Guillelmon, D., & Grosjean, F. (2001). The gender marking effect in spoken word recognition:
The case of bilinguals. Memory & cognition, (29) (3), 503-511.
73
Harley, B. (1993). Instructional Strategies and SLA in Early French Immersion. Studies in
Second Language Acquisition, 15(02), 245–259.
Harris, J. (1991). The exponence of gender in Spanish. Linguistic Inquiry, 22, 27–62.
Hawkins, R., & Chan, C. (1997). The partial availability of Universal Grammar in second
language acquisition: The 'failed functional features hypothesis.' Second Language
Research, 13(3), 187–226.
Hawkins, R., & Franceschina, F. (2004). Explaining the acquisition and non-acquisition of
determiner-noun gender concord in French and Spanish. Language Acquisition and
Language Disorders, 32, 175-206.
Hoaglin, D., & Iglewicz, B. (1987). Fine-Tuning Some Resistant Rules for Outlier Labeling,
Journal of the American Statistical Association, 82, 1147-1149.
Hockett, C. F. (1958). A course in modern linguistics. Language Learning, 8(3‐4), 73-75.
Hohlfeld, A. (2006). Accessing grammatical gender in German: The impact of gender-marking
regularities. Applied Psycholinguistics, 27(2), 127-142.
Holmes, V. M., & Dejean de la Bâtie, B. (1999). Assignment of grammatical gender by native
speakers and foreign learners of French. Applied Psycholinguistics, 20(4), 479–506.
Holmes, V. M., & Segui, J. (2004). Sublexical and lexical influences on gender assignment in
French. Journal of Psycholinguistic Research, 33(6), 425–457.
l’Huillier, M. (1999). Advanced French Grammar. Cambridge University Press.
Hulk, A., &Tellier, C. (1999). Conflictual agreement inRomancenominals. In J.M.Authier, B.
Bullock, & L. Reed (Eds.), Formal perspectives on Romance linguistics (pp. 179–195).
Amsterdam: Benjamins.
Jedwab, J. (2003). It pays to be bilingual in Canada: Though not everywhere. Retrieved from
74
Office of the Commissioner of Official Languages website: http://www.ocol-
clo.gc.ca/sites/default/files/Pays_to_be_Bilingual.pdf
Jescheniak, J. D., & Levelt, W. J. (1994). Word frequency effects in speech production: Retrieval
of syntactic information and of phonological form. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 20(4), 824.
Kahn-Horwitz, J., Kuash, S., Ibrahim, R., & Schwartz, M. (2014). How do previously acquired
languages affect acquisition of English as a foreign language: The case of
Circassian. Written Language & Literacy, 17(1), 40-61.
Karmiloff-Smith, A. (1979). A functional approach to child language: A study of determiners
and reference. Cambridge, UK: Cambridge University Press.
Kirby, J. R., Deacon, S. H., Bowers, P. N., Izenberg, L., Wade-Woolley, L., & Parrila, R. (2012).
Children’s morphological awareness and reading ability. Reading and Writing, 25(2),
389-410.
Kuo, L. J., & Anderson, R. C. (2010). Beyond cross-language transfer: Reconceptualizing the
impact of early bilingualism on phonological awareness. Scientific Studies of
Reading, 14(4), 365-385.
Laaha, S., & Gillis, S. (Eds.). (2007). Typological perspectives on the acquisition of noun and
verb morphology. Universiteit Antwerpen.
Lado, R. (1957). Linguistics Across Cultures. Applied Linguistics for Language Teachers. Ann
Arbor: University of Michigan Press.
Levelt, W. J., Roelofs, A., & Meyer, A. S. (1999). A theory of lexical access in speech
production. Behavioral and brain sciences, 22(1), 1-38.
Larrañaga, P., & Guijarro-Fuentes, P. (2013). The acquisition of copula verbs in Basque by
bilinguals. International Journal of Bilingualism, 17(3), 284-308.
75
Liceras, J. (2010). Second language acquisition and syntactic theory in the 21st century. Annual
Review of Applied Linguistics, 30, 248–269.
Lyster, R., (1987). Speaking immerion. The Canadian Modern Language Review, 43, (4), 701-
717.
Lyster, R. (2006). Predictability in French gender attribution: A corpus analysis. Journal of
French Language Studies, 16(1), 69–92.
Lyster, R. (2008). Learning French as a second language through immersion. In D. Ayoun (Ed.),
Studies in French applied linguistics (pp. 3-36). Amsterdam/Philadelphia: John
Benjamins.
Mastropavlou, M., & Tsimpli, I. M. (2011). The role of suffixes in grammatical gender
assignment in Modern Greek: A psycholinguistic study. Journal of Greek
Linguistics, 11(1), 27-55.
Meisel, J. (1983). Transfer as a second language strategy. Language and Communication,3, 11–
46.
Meisel, J. (2001). The simultaneous acquisition of two first languages: Early differentiation and
subsequent development of grammars. In J. Cenoz & F. Genesee (Eds.), Trends in
bilingual acquisition (pp.11 – 42). Amsterdam: John Benjamins.
Melançon, A., & Shi, R. (2015). Representations of abstract grammatical feature agreement in
young children. Journal of child language, 1-15.
Mitchell, R., Myles, F., & Marsden, E. (2013). Second language learning theories. Routledge.
Montrul, S., & Potowski, K. (2007). Command of gender agreement in school-age Spanish
English Bilingual Children. International Journal of Bilingualism, 11(3), 301–328.
New B., Pallier C., Ferrand L., Matos R. (2001) Une base de données lexicales du français
76
contemporain sur internet: LEXIQUE, L'Année Psychologique, 101, 447-462.
http://www.lexique.org
Nichols, J. (1992). Linguistic diversity in space and time. Chicago: University of Chicago Press.
Nicoladis, E. (2006). Cross-linguistic transfer in adjective–noun strings by preschool bilingual
children. Bilingualism: Language and Cognition, 9(1), 15-32.
Odlin, T. (2003) Cross-linguistic influence. In C. Doughty & M. Long (Eds.), Handbook of
Second Language Acquisition (pp. 436-486). Oxford: Blackwell.
Ontario Ministry of Education. (2008). Supporting English language learners: A practical guide
for Ontario educators, Grades 1 to 8. Toronto, ON: Queen’s Printer for Ontario.
Presson, N., MacWhinney, B., & Tokowicz, N. (2014). Learning grammatical gender : The use
of rules by novice learners. Applied Psycholinguistics, 35(4), 709–737.
Price, G. (2008). A comprehensive French grammar. Malden MA: Blackwell.
Quemada, B. (1971). An Inverse Dictionary of All Nouns Listed in the Petit Larousse
[microfilm]. Rochester, NY: Eastman Kodak.
Rebuffot, J. (1993). Le point sur l’immersion au Canada. Montréal: Éditions CEC.
Riente, L. (2003). Ladies first: The pivotal role of gender in the Italian nominal inflection
system. McGill Working Papers in Linguistics, 17(2), 1–53.
Roca, I. M. (1989). The organisation of grammatical gender. Transactions of the Philological
Society, 87, 1–32.
Roelofs, A. (1992). A spreading-activation theory of lemma retrieval in speaking. Cognition, 42,
(1), 107-142.
Sabourin, L., Stowe, L. A., & de Haan, G. J. (2006). Transfer effects in learning a second
language grammatical gender system. Second Language Research, 22(1), 1-29.
Schriefers, H., & Jescheniak, J. D. (1999). Representation and processing of grammatical gender
77
in language production: A review. Journal of psycholinguistic research, 28(6), 575-600.
Schwartz, M., Minkov, M., Dieser, E., Protassova, E., Moin, V., & Polinsky, M. (2014).
Acquisition of Russian gender agreement by monolingual and bilingual children.
International Journal of Bilingualism, 1–27.
Schwartz, B. D., & Sprouse, R. (1994). Word order and nominative case in nonnative language
acquisition: a longitudinal study of (L1 Turkish) German interlanguage. Language
acquisition studies in generative grammar, 31(4), 71-89.
Séguin, H. (1969). Les marques du genre dans le lexique français écrit contemporain:
Compilation des cas et essai de classement. (Unpublished doctoral disseratation).
University de Montréal. Montréal.
Seigneuric, A., Zagar, D., Meunier, F., & Spinelli, E. (2007). The relation between language and
cognition in 3- to 9-year-olds: The acquisition of grammatical gender in French. Journal
of Experimental Child Psychology, 96(3), 229–246.
Selinker, L. (1972). Interlanguage. IRAL-International Review of Applied Linguistics in
Language Teaching, 10(1-4), 209-232.
Sera, M. D., Elieff, C., Forbes, J., Burch, M. C., Rodríguez, W., & Dubois, D. P. (2002). When
language affects cognition and when it does not: an analysis of grammatical gender and
classification. Journal of Experimental Psychology: General, 131(3), 377–397.
Serratrice, L. (2013). Acquisition of features in the nominal domain in bilingual acquisition.
International Journal of Bilingualism, 17(5), 657–664.
Sinay, E. (2010). Programs of choice in the TDSB: Characteristics of students in French
immersion, alternative schools, and other specialized schools and programs. Retrieved
from Toronto District School Board website:
http://www.tdsb.on.ca/Portals/0/community/community%20advisory%20committees/fsla
78
c/ support%20staff/programsofchoicestudentcharacteristics.pdf
Slevc, L. R., & Miyake, A. (2006). Individual Differences in Second- Language Proficiency
Does Musical Ability Matters?. Psychological Science, 17(8), 675–681.
Spalek, K., Franck, J., Schriefers, H., & Frauenfelder, U. H. (2008). Phonological regularities
and grammatical gender retrieval in spoken word recognition and word production.
Journal of Psycholinguistic Research, 37, 419–442.
Surridge, M. E. (1993). Gender assignment in French: The hierarchy of rules and the chronology
of acquisition. International Review of Applied Linguistics, 31, 77–95.
Surridge, M. & Lesssard, G. (2008). Le Genre en Français: Quasi-Regularites de L'Attribution et
Fonction Grammaticale. Langues et Linguistique, 32, 181-218.
Swain, M., & Lapkin, S. (2005). The evolving sociopolitical context of immersion education in
Canada: some implications for program development. International Journal of Applied
Linguistics, 15(2), 169-186.
Taft, M., & Meunier, F. (1998). Lexical representation of gender: A quasiregular
domain. Journal of Psycholinguistic Research, 27(1), 23-45.
Thorndike, E. L. (1923). The Influence of First-Year Latin Upon Ability to Read English. School
& Society, 17, 165-168.
Thornton, R. (1996). Elicited production. Methods for Assessing Children’s Syntax. 77–102.
Tsimpli, I. M., & Hulk, A. (2013). Grammatical gender and the notion of default: Insights from
language acquisition. Lingua, 137, 128–144.
Tucker, G., Lambert, W., & Rigault, A. (1977). The French speaker’s skill with grammatical
gender: An example of rule-governed behavior (Vol. 8). Mouton De Gruyter.
Tucker, G. R., (1968). French Speakers’ Skill with Grammatical Gender: An 'Example of Rule
Governed Behavior. (Unpublished doctoral disseratation). McGill University. Montreal.
79
Tucker, G. R., Lambert, W. E., Rigault, A., & Segalowitz, N. (1968). A psychological
investigation of French speakers' skill with grammatical gender. Journal of verbal
learning and verbal behavior, 7(2), 312-316.
Turnbull, M., Lapkin, S., Hart, D., & Swain, M. (1998). Time on task and immersion graduates’
French proficiency. In S. Lapkin (ed.), French second language education in Canada:
Empirical studies (pp. 31-55). Toronto: University of Toronto Press.
Unsworth, S. (2008). Age and input in the acquisition of grammatical gender in Dutch
ProQuest. Second Language Research, 24(3), 365–395.
Warden, M. (1997). The Effect of Form-focused Instruction on Control Over Grammatical
Gender by French Immersion Students in Grade 11. (Unpublished doctoral
disseratation). Ontario Institute for Studies in Education of the University of Toronto.
Toronto.
Webster’s Ninth New Collegiate Dictionary. (1991). Springfield, MA: Merriam-Webster.
Wechsler, D. (2005). Wechsler individual achievement test (2nd ed). San Antonio, TX:
Psychological Corporation.
Wesche, M. B. (2002). Early French immersion: How has the original Canadian model stood the
test of time? In P. Burmeister, T. Piske & A. Rohde (Eds.), An integrated view of
language development: Papers in honor of Henning Wode (pp.357-379). Trier, Germany:
Wissenschaftlicher Verlag Trier.
White, L., Valenzuela, E., Kozlowska-Macgregor, M., & Leung, Y. K. I. (2004). Gender and
number agreement in nonnative Spanish. Applied Psycholinguistics, 25(1), 105-133.
80
Appendix A
List of pseudowords13
Pseudowords with neutral
endings
Pseudowords with masculine
endings
Pseudowords with feminine
endings
fergadèle⁺ colcheau fralocture
suticoque⁺ moritrier troprine
flaumide⁺ lopisme blaquotte
chonlige* savouguin* possidelle*
bormispule⁺ bilacron cralibaine
jurmone⁺ sarnour sinourette*
bicatique* lucadour nasprelle
vécostiode⁺ quastrin cabagine
prucergone⁺ pliquachon* ploumiette
noplide⁺ vadanguier* rounoture
stongode⁺ naurisme jelaine
mauléroque⁺ bloukeau* braivotte
spadique*
derivige
cergatule⁺
cuipèle⁺
13 * Pseudowords that were used in Seigneuric et al. (2007). ⁺ Pseudowords that were used in Holmes and Dejean
de la Bâtie (1999).
81
Appendix B
The Determiner Task
Materials : none
Scoring : Circle 1 if the child correctly identifies the pseudoword marked for gender (based on
its ending). Circle 0 if the child incorrectly identifies the pseudoword marked for gender (based
on its ending). The total score is the total number of correct test items. There is no right or wrong
answer for nouns that are not marked for gender. All of the responses for unmarked pseudowords
will be counted to calculate the percentages of the feminine and the masculine.
Please note: un and une must be emphasized by the experimenter when addressing the
children. Do NOT provide any feedback to the child on any of the items (i.e., practice and test
items). Circle ONE answer only.
PRACTICE ITEMS: (please circle the determiner provided by the child)
1. J’ai inventé des nouveaux mots. Voici chulote. Peux-tu répéter le mot s’il te plaît? Tu
préfèrerais dire un chulote ou une chulote? I have invented new words. This is chulote. Can you
please repeat the word? Do you prefer to say a (masculine) chulote or a (feminine) chulote?.
Child : une un chulote
Correct Answer : feminine masculine N/A
2. J’ai inventé des nouveaux mots. Voici jattise. Peux-tu répéter le mot s’il te plaît? Tu
préfèrerais dire une jattise ou un jattise? I have invented new words. This is jattise. Can you
please repeat the word? Do you prefer to say a (masculine) jattise or a (feminine) jattise?.
Child : une un jattise
Correct Answer : 1 0 Une jattise
3. J’ai inventé des nouveaux mots. Voici dostral. Peux-tu répéter le mot s’il te plaît? Tu
préfèrerais dire un dostral ou une dostral? I have invented new words. This is dostral. Can you
please repeat the word? Do you prefer to say a (masculine) dostral or a (feminine) dostral?.
Child : une un dostral
Correct Answer : 0 1 Un dostral
Est-ce que tu as des questions? Do you have any questions? You may repeat the words if the
child does not hear them. Repeat the procedure for TEST items 1-20.
TEST ITEMS : (If the child says they do not know the answer, please ask them to choose
one anyways)
1. Voici fralocture. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire une fralocture ou un
fralocture?
82
Child : une un fralocuture
Correct Answer : 1 0 Une fralocture
2. Voici colcheau. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire un colcheau ou une
colcheau?
Child : une un colcheau
Correct Answer : 0 1 Un colcheau
3. Voici troprine. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire une troprine ou un
troprine?
Child : une un troprine
Correct Answer : 1 0 Une troprine
4. Voici blaquotte. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire un blaquotte ou une
blaquotte?
Child : une un blaquotte
Correct Answer : 1 0 Une blaquotte
5. Voici fergadèle. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire une fergadèle ou un
fergadèle?
Child : une un fergadèle
Correct Answer : feminine masculine N/A
6. Voici moritrier. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire un moritrier ou une
moritrier?
Child : une un moritrier
Correct Answer : 0 1 Un moritrier
7. Voici suticoque. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire une suticoque ou un
suticoque?
Child : une un suticoque
Correct Answer : feminine masculine N/A
8. Voici flaumide. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire un flaumide ou une
flaumide?
Child : une un flaumide
Correct Answer : feminine masculine N/A
83
9. Voici lopisme. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire une lopisme ou un
lopisme?
Child : une un lopisme
Correct Answer : 0 1 Un lopisme
10. Voici chonlige. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire un chonlige ou une
chonlige?
Child : une un chonlige
Correct Answer : feminine masculine N/A
11. Voici bormispule. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire une bormispule ou
un bormispule?
Child : une un bormispule
Correct Answer : feminine masculine N/A
12. Voici possidelle. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire un possidelle ou une
possidelle?
Child : une un possidelle
Correct Answer : 1 0 Une possidelle
13. Voici cralibaine. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire une cralibaine ou un
cralibaine?
Child : une un cralibaine
Correct Answer : 1 0 Une cralibaine
14. Voici jurmone. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire un jurmone ou une
jurmone?
Child : une un jurmone
Correct Answer : feminine masculine N/A
15. Voici savouguin. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire une savouguin ou un
savouguin?
Child : une un savouguin
Correct Answer : 0 1 Un savouguin
84
16. Voici sinourette. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire un sinourette ou une
sinourette?
Child : une un sinourette
Correct Answer : 1 0 Une sinourette
17. Voici bilacron. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire une bilacron ou un
bilacron?
Child : une un bilacron
Correct Answer : 0 1 Un bilacron
18. Voici sarnour. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire un sarnour ou une
sarnour?
Child : une un sarnour
Correct Answer : 0 1 Un sarnour
19. Voici bicatique. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire une bicatique ou un
bicatique?
Child : une un bicatique
Correct Answer : feminine masculine N/A
20. Voici vécostiode. Peux-tu répéter le mot s’il te plaît? Tu préfèrerais dire un vécostiode ou
une vécostiode?
Child : une un vécostiode
Correct Answer : feminine masculine N/A
SCORING :
Total Feminine
Correct (/6)
*Question (1,
3, 4, 12, 13, 16)
Total
Masculine
Correct (/ 6)
*Question (2, 6,
9, 15, 17, 18)
Total Correct
(Feminine
+Masculine)
(/12)
Non- Marked
Nouns (/ 8)
*Question (5, 7,
8, 10, 11, 14, 19,
20)
Ratio (# of
masculine /8, # of
feminine /8)
Masculine
Feminine
85
Appendix C
The Picture Task
Materials : none
Scoring : Circle 1 if the child correctly identifies the pseudoword marked for gender (based on
its ending) with the gender of the stimulus. Circle 0 if the child incorrectly identifies the
pseudoword marked for gender (based on its ending) with the gender of the stimulus. The total
score is the total number of correct test items. There is no right or wrong answer for nouns that
are not marked for gender. All of the responses for unmarked pseudowords will be counted to
calculate the percentages of the feminine and the masculine.
Do NOT provide any feedback to the child on any of the items (i.e., practice and test items).
Circle ONE answer only.
PRACTICE ITEMS: (please circle the response provided by the child)
1. Regarde ces deux images. C’est un couple de personnes imaginaires. Montre-moi le garçon et
la fille. Écoute le mot «flaison ». C’est un mot que j’ai inventé. Montre-moi qui pourrait
s’appeler « flaison »? (Look at these pictures. This is a couple of imaginary beings (Martian-like
persons). Show me the girl and show me the boy. Listen to the invented word « flaison ». Who
would you prefer to call « flaison » between these two imaginary persons? »
Child : Male Martian Female Martian
Correct Answer : 0 1
2. Regarde ces deux images. C’est un couple de personnes imaginaires. Montre-moi le garçon et
la fille. Écoute le mot «laffrège ». C’est un mot que j’ai inventé. Montre-moi qui pourrait
s’appeler « laffrège »? (Look at these pictures. This is a couple of imaginary beings (Martian-like
persons). Show me the girl and show me the boy. Listen to the invented word « laffrège». Who
would you prefer to call « laffrège » between these two imaginary persons? »
Child : Male Martian Female Martian
Correct Answer : 1 0
3. Regarde ces deux images. C’est un couple de personnes imaginaires. Montre-moi le garçon et
la fille. Écoute le mot «tarcule ». C’est un mot que j’ai inventé. Montre-moi qui pourrait
s’appeler « tarcule »? (Look at these pictures. This is a couple of imaginary beings (Martian-like
persons). Show me the girl and show me the boy. Listen to the invented word « tarcule». Who
would you prefer to call « tarcule » between these two imaginary persons? »
Child : Male Martian Female Martian
Correct Answer : N/A N/A
Est-ce que tu as des questions? Do you have any questions? You may repeat the words if the
child does not hear them. Repeat the procedure for TEST items 1-20.
86
TEST ITEMS : (If the child says they do not know the answer, please ask them to choose
one anyways)
1. Écoute le mot «lucadour ». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« lucadour »?
Child : Male Martian Female Martian
Correct Answer : 1 0
2. Écoute le mot «nasprelle». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« nasprelle »?
Child : Male Martian Female Martian
Correct Answer : 0 1
3. Écoute le mot «cabagine». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« cabagine»?
Child : Male Martian Female Martian
Correct Answer : 0 1
4. Écoute le mot «quastrin ». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« quastrin »?
Child : Male Martian Female Martian
Correct Answer : 1 0
5. Écoute le mot «prucergone ». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« prucergone »?
Child : Male Martian Female Martian
Correct Answer : N/A N/A
6. Écoute le mot «pliquachon ». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« pliquachon »?
Child : Male Martian Female Martian
Correct Answer : 1 0
7. Écoute le mot «vadanguier ». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« vadanguier »?
Child : Male Martian Female Martian
87
Correct Answer : 1 0
8. Écoute le mot «ploumiette ». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« ploumiette »?
Child : Male Martian Female Martian
Correct Answer : 0 1
9. Écoute le mot «noplide ». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« noplide »?
Child : Male Martian Female Martian
Correct Answer : N/A N/A
10. Écoute le mot «rounoture ». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« rounoture »?
Child : Male Martian Female Martian
Correct Answer : 0 1
11. Écoute le mot «stongode ». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« stongode »?
Child : Male Martian Female Martian
Correct Answer : N/ A N/ A
12. Écoute le mot «jelaine ». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« jelaine »?
Child : Male Martian Female Martian
Correct Answer : 0 1
13. Écoute le mot « mauléroque ». C’est un mot que j’ai inventé. Montre-moi qui pourrait
s’appeler « mauléroque »?
Child : Male Martian Female Martian
Correct Answer : N/A N/ A
14. Écoute le mot « spadique ». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« spadique »?
Child : Male Martian Female Martian
Correct Answer : N/A N/A
88
15. Écoute le mot «braivotte ». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« braivotte »?
Child : Male Martian Female Martian
Correct Answer : 0 1
16. Écoute le mot « derivige ». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« derivige »?
Child : Male Martian Female Martian
Correct Answer : N/A N/A
17. Écoute le mot « naurisme ». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« naurisme »?
Child : Male Martian Female Martian
Correct Answer : 1 0
18. Écoute le mot « bloukeau ». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« bloukeau »?
Child : Male Martian Female Martian
Correct Answer : 1 0
19. Écoute le mot « cergatule ». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« cergatule »?
Child : Male Martian Female Martian
Correct Answer : N/A N/A
20. Écoute le mot « cuipèle ». C’est un mot que j’ai inventé. Montre-moi qui pourrait s’appeler
« cuipèle »?
Child : Male Martian Female Martian
Correct Answer : N/A N/A
89
SCORING :
Total Feminine
Correct (/6)
*Question (2,
3, 8, 10, 12, 15)
Total
Masculine
Correct (/ 6)
*Question (1, 4,
6, 7, 17, 18)
Total Correct
(Feminine
+Masculine)
(/12)
Non- Marked
Nouns (/ 8)
*Question (5, 9,
11, 13, 14, 16, 19,
20)
Ratio (# of
masculine /8, # of
feminine /8)
Masculine
Feminine
90
Appendix D
Martians used in the Picture Task
91
Appendix E
Bias Value Calculations Obtained from the Lexique Database
Ending
Number of
Entries
Number of Masculine
Entries
Number of Feminine
Entries
Proportion of Masculine
Entries
Proportion of Feminine
Entries
ine 508 38 470 .07 .93
ette 430 18 412 .04 .96
otte 59 6 53 .10 .90
elle 173 11 162 .06 .94
ure 353 40 313 .11 .89
aine 90 12 78 .13 .87
on 1039 745 294 .72 .28
eau 202 196 6 .97 .03
ier 545 543 2 .99 .01
in 505 494 11 .98 .02
isme 485 485 0 1 0
our 54 50 4 .93 .07
ode 24 12 12 .50 .50
oque 33 16 17 .48 .52
ique 193 65 128 .34 .66
èle 15 6 9 .40 .60
ule 132 51 81 .39 .61
ige 20 12 8 .60 .40
one 79 52 27 .66 .34
ide 67 46 21 .69 .31