Language Evolution: The Windows Approach addresses the question: "How can we unravel the evolution of language, given that there is no direct evidence about it?"
The LINGUIST List is dedicated to providing information on language and language analysis, and to providing the discipline of linguistics with the infrastructure necessary to function in the digital world. LINGUIST is a free resource, run by linguistics students and faculty, and supported primarily by your donations. Please support LINGUIST List during the 2016 Fund Drive.
SUMMARY “Irregularity in Morphology (and beyond)” is intended for readers interested in morphology, (ir)regularity in morphology, and for linguists generally. It offers detailed analyses of various morphological phenomena from European languages including French, Spanish, Italian, Catalan, and Latin as well as Greek, German, Turkish, English, Russian, but also Native American languages, including Thompson Salish and Iroquoian languages. The volume follows the 2009 international conference on “Irregularity in Morphology (and Beyond)” held in Bremen, Germany.
The book includes descriptive, empirical and theoretical reviews of irregularity in morphology. Empirical data are presented from children and adults. Neurological disorders are also discussed, covering Broca’s aphasia, Wernicke’s aphasia, and Parkinson’s disease. Concepts such as markedness, redundancy, suppletion, and recursiveness are investigated throughout the volume. A short description of each paper follows.
Thomas Stolz, Hitomi Otsuka, Aina Urdze, and Johan van der Auwera authored “Introduction: Irregularity -- glimpses of a ubiquitous phenomenon” (pp.7-37). Irregularity is attested in a variety of languages and language families, but receives little attention in handbooks. A summary of Kiefer’s (2000) major statements (given on p.12, e.g., “irregularities are violations of rules of grammar» or «irregularity is supported by high frequency”) provides a good opening on what is irregularity and some of the properties discussed in the book. Proofs of the universality quality of irregularity are given with evidence from Indo-European (e.g., Armenian and French) and unrelated languages (e.g., Dargwa and Basque) grammars. Importance is placed on irregularity beyond morphology. Then, an extensive and careful review of literature on suppletion is provided, comparing 100 Indo-European and non-Indo-European languages leading to the conclusion that suppletion is particularly associated with Indo-European. The authors conclude by briefly introducing the volume’s organization.
Marianne Mithun examines Mohawk, Tuscora and Cherokee (all Iroquoian) in “The deeper regularities behind irregularities” (pp.39-59). Iroquoian languages are polysynthetic languages, with complex derivational and inflectional morphology concentrated in the verb. The first example from Mohawk (p.40), Tha’-t-onta-ho-ate-nenneri’t-ate-’sere-ht-at-kehront-ako-hatie’ (‘He’s just on his way back from buying himself another dang car’), is composed of 13 different parts or morphemes (indicated by the hyphens). Another instance of this complexity is that a term like ‘younger sibling’ can take 35 different forms. Mithun covers irregularity in pronominal inflection, tense inflection, aspect inflection, and derivation in these morphologically rich languages. She first addresses Mohawk instrumental applicatives -hst and -hkw and concludes that their irregular distribution is not phonological but rather a matter of lexical choice. The pronominal prefixes k- and wak- further strengthen her conclusion of semantic influence of lexical choice in Mohawk. Then, the imperative pronominal prefix alternation found in Mohawk s-/ts-,Tuscarora s-/tš, and Seneca s-/ts- would come from Proto-Northern-Iroquoian, while the tak-/-hsk- irregularity in Mohawk would come from use of the cislocative ta- rather than the pronominal prefix, possibly for politeness reasons. The irregular prefix for ‘they/me’ is composed of a singular prefix plus a distributive enclitic, actually a generic third person pronominal similar to English ‘one’. Then, semantic irregularities are presented for pronominal and tense inflection and Mithun closes with irregularity on root-aspect collocations that indicates that aspect marker is triggered by the root rather than by word function (p.58).
“Sturtevant’s paradox revisited”, by Paolo Ramat (pp.61-79), deals with possible diachronic sources of irregularity, by analyzing Sturtevant’s Paradox, which claims that sound change is regular but creates irregularity, while analogical change is irregular but creates regularity. Ramat covers the concepts of complexity, markedness, redundancy, analogy, and predictability. Examples are introduced from Latin to Italian -am/-as/-at to -o/-i/-a imperfect endings and German war/waren opposed to English was/were singular versus plural past forms of “to be”. Ramat then reminds us that Sturtevant’s formulation has two layers, one for phonology and one for morphology, with different chronological levels. Thus, a later rule can affect outputs of a previous rule, but not always, which gives rise to irregular patterns (e.g., was/were versus drove/drove). The author associates regularity with unmarkedness and irregularity with markedness, while making a distinction with complexity (for irregularity/markedness). Examples are provided from comparing English to Turkish and Hungarian, arguing that complexity and markedness values are language specific (i.e., constructions can be complex yet unmarked in a given language). Redundancy helps to detect linguistic transparency (e.g., double marking like German singular/plural Gast/Gäste). Further, redundancy-economy changes would explain why old and new forms coexist in paradigms. Analogical rules introduce regularized forms that go against paradigm regularity. An example of such rules is taken from children productions of bring/*brang by analogy to sing/sang as opposed to *bringed. The author concludes that while irregularity is not predictable, it is part of all languages.
“Paradigm gaps in Whole Word Morphology” by Luc Baronian and Elena Kulinich (pp.81-100) considers paradigm gaps of defective verbs in French, English, Spanish, and Russian based on Whole Word Morphology (WWM). Irregularity is approached in defective verbs, verbs missing at least one form in the paradigm (e.g., the French verb frire ‘to fry’ does not have plural forms; the verb beware in English has no past or present). WWM theory does not provide general default forms and is a word-, not primitive-based theory of morphology. Their proposal is that for Word-Formation Strategies (WFS, seven in total) speakers favor network representation and make phonemic subgeneralizations. As a word-based theory built on existing phonological forms (e.g., goose, tooth <--> geese, teeth), WFS will limit formations outside the model (e.g., proof/*preef). A survey with Russian speakers to induce defective and non-defective real, borrowed, low-frequency and nonce 1SG non-past verbs reveals a significant difference between low-frequency and defectives. As they note, the survey fails in ecological validity by its design, forcing a response when a ‘no form’ response would be expected for defectives. The authors conclude that WWM accounts for paradigm gaps of defective verbs where small classes (of verbs) are separated into phonological subgeneralizations.
“Irregularity in inflectional morphology -- where language deficits strike” by Martina Penke and Eva Wimmer (pp.101-125) treats irregularity in inflectional morphology from a neurological disorders perspective. Inflectional morphology is often affected in developmental or acquired language deficits. From a dualistic point of view (e.g., Pinker, 1999), regular forms (e.g., love-loved) would be computed by combining stems and affixes, while irregular forms (e.g., sing-sang) would be retrieved as is in the lexicon. This implies that selective deficits should be found between speakers (i.e., some would be impaired on regular forms with irregular forms spared and others would demonstrate the complete opposite pattern). Data for such deficits are presented from Ullman et al. (2005) where on the one hand a Broca’s aphasic English patient showed selective impairment with regular inflection and on the other hand a fluent anomic aphasic English patient showed irregular inflection deficits (and spared regular inflection). These two cases were also observed in distinct lesion sites, frontal in the first case and posterior in the later. These selective deficits are discussed in terms of the Procedural/Declarative Model (e.g., Ullman, 2001) where regular inflection is part of the procedural memory system located in left frontal brain areas (Broca’s aphasia) as opposed to irregular inflection that is part of the declarative memory system located in left temporo-parietal brain areas (i.e., Wernicke’s aphasia). Then, however, evidence from other languages (e.g., German, Italian) suggests that language typology plays a role in such a dualistic distinction, probably because in these languages, irregulars will also have inflectional endings, while in English, regularity is confounded with inflection. Empirical data are then presented where elicited productions of regular (e.g., lachen-gelacht ‘to laugh-laughed’; Blume-Blumen ‘flower-s’) and irregular (e.g., trinken-getrunken ‘to drink-drunk’; Muskel-Muskeln) verbal and nominal forms in German-speaking participants with Broca’s aphasia, Wernicke’s aphasia, Parkinson’s disease, and without deficits are compared. Results show that in German, irregularity correlates with more errors than regularity regardless of neurological syndromes. Participants with neurological syndromes demonstrate a selective deficit affecting irregular inflection (i.e., no dual patterns were found in German speakers between Broca’s and Wernicke’s aphasics unlike English speakers in Ullman et al., 2005). However, frequency effects are found in both aphasic groups. These results are thus incompatible with the Procedural/Declarative Model (e.g., Ullman, 2001) in which Broca’s and Wernicke’s aphasics should differ in regular/irregular inflectional morphology and confirm the importance of investigating typologically different languages.
The paper by Anna Anastassiadis-Symeonidis and Elvira Masoura, “Word ending-part and phonological memory: a theoretical approach” (pp.127-140), presents empirical evidence from two experiments in Greek addressing class marking (in word final position e.g., -ιά [-ia]; -άκι [aki]) or phonological similarity (e.g., αυτά-πουφτά [afta]-[poufta]; μήνας-μήδρο [minas]-[midro]). In Greek, class markings play synchronic (integration into a grammatical category) and diachronic (regulation of categories) roles. The authors adopt Corbin’s (1987) framework for case markers in which new words entering a language are created following rules and that when (sound) irregularities enter, it triggers the creation of regularity by means of case marking to respect to conformity to a category. This would be in line with semantic networks (Collins & Loftus, 1975) where concepts are part of a network. Also, phonological similarity (Baddeley, 1966) would impact participants’ recall capacities depending if the tasks imply immediate (disruption effects) or delayed (facilitation effects) recall. A series of experiments examined class marking roles. In the first set of experiments on case markings, participants had to create a new category, words belonging to a category or a new verb. For the second experiment, participants had to recall pairs of words that were phonologically similar initially or finally or phonologically dissimilar. Results show that participants tended to create words with an ending phonologically resembling the dominant ending of the category (e.g., ending in -ιά [-ia]). Also, participants remembered phonologically similar words more accurately than dissimilar ones (word position did not quite significantly differ). The authors conclude that results of their study are consistent with Corbin’s (1987) model.
Leah Bauke’s paper “(Ir)regularity in nominal root compounds” (pp.141-165) treats word formation in French, English, and German. She begins by noting that most irregularities are limited to morphology in the lexicon, while syntax is regular. She cites Chomsky’s (1970) claim that irregularities arise particularly in derivational morphology, while inflectional morphology is (like syntax) regular. Part two discusses nominal root compounding. In Germanic languages like German and English, this process is believed to be productive and recursive where new compounds can be formed. In contrast, Romance languages like French, for example, compounding lacks productivity and recursivity. The difference could reflect an abstract clitic position allowing productive compounding that exists in Germanic but is missing in Romance. Moreover, it is argued that German has two types of nominal root compounds, with or without inflectional elements i.e., recursive and non-recursive. Inflectional recursive compounds resemble English compounds and allow compositional interpretation, while non-inflectional non-recursive compounds resemble French compounds, where meanings are ‘drifted’ or non-compositional. Moreover, recursive compounds would be built in the syntax, while non-recursive compounds would be built in the lexicon (cf. Roeper, Snyder, & Hiramatsu, 2002). Also, compositional compounds are specified for lexical category (i.e., are not roots), while non-compositional compounds are (bare) roots. In conclusion, phase instantiations are created with compositional compounds that give rise to abstract clitic positions that explain the recursive nature of such compounds.
Volha Kharytonava’s “Taming affixes in Turkish: with or without residue?” (pp.167-185) deals with the Suspended Affixation (SA) phenomenon which relates to a situation where in a coordinate construction, an affix is omitted from one of the coordinands that another coordinand has, so that the affix has scope over both coordinands. The phenomenon is examined with noun compound data from Turkish (where, e.g., the first two words lack subject agreement marking that the third possesses: gid-er-, gör-ür-, al-ιr-Ø-ιz). The phenomenon consists of two distinct processes (coordination and affixation). Two existing approaches differ in their ordering of these processes: Approach A in which coordination precedes affixation versus Approach B in which affixation precedes coordination (this process also implies deletion of preexisting affixes). In Turkish, three possible outcomes exist for noun compound coordination that share interpretation: no SA, total SA or partial SA (non-final conjuncts have -(s)I morpheme, note that -(s)I morpheme is in complementary distribution with the possessive morpheme). Kharytonava assumes Distributed Morphology and Feature Geometry accounts and proposes that SA is a deletion process from the possessive suffix. By an Impoverishment process of features in a terminal morpheme, both partial and total SA are made possible for non-final conjuncts. Impoverishment would be a property of noun compounds only and would not involve suffixes other than possessives in Turkish.
Francesco Rovai’s “On some Latin morphological (ir)regularities” (pp.187-211) discusses the unproductive irregular 2nd-declension nominative plural in Latin -eis (e.g., thurari-eis ‘incense-sellers’) and 3rd-declension genitive singular in -us (e.g., patr-us ‘father’) inflections. They contrast with their productive counterparts -i (2nd-declension regular nominative plural) and -is (3rd-declension regular genitive singular). The paper tests whether regular constraints exist for such irregulars using a usage-based framework (Hopper, 1987) in which linguistic structures (like categorization) are part of a general cognitive capacity. Morphological (ir)regularity is said to be best explained as a continuum from productive to unpredictable forms (p.187 citing Ramat, 1985). Bybee & Slobin’s work (1982) is taken as evidence for such a view because speakers can generalize about irregular forms or schemas Morphological classes are gradient in productivity and applications to novel items are related to type frequency and similarity between items. The -eis ending (sometimes spelled -is or -es), an innovation that could have been triggered by contact between Latin and Oscan, is discussed based on 47 forms (26 family names and 21 common nouns). The 2nd-declension nominative plural -eis appears to share semantic (it seems to be limited to [+/-definite]/[+human] nouns) and phonetic (it occurs in words ending in [alveolar sonorant]+[j], e.g., -[lj]-, -[rj]-, -[nj]-) attributes. Rovai says that -eis can thus be viewed as a prototypically structured morphological class where marginal members may be associated and items sharing none of the features with the prototype are not included (also, semantic features seem more relevant than phonetic features). The 3rd-declension genitive singular in -us ending, that would be a relic from Indo-European, is discussed with 26 forms. It too could be viewed as a prototypically structured morphological class that shares semantic (also limited to [+/-definite]/[+human] nouns) and phonetic (alveolar sonorant -[n]- or -[r]- to the exception of 3 forms) attributes among forms it applies to. Both irregularities suggest a scalar status for irregular morphology since they are neither completely unpredictable nor unproductive.
Lucia Aliffi’s “Irregularity in Latin: gender and inflexional class” (pp.213-225) provides a broader picture of gender and inflexional class irregularities in Latin, namely, feminine nouns in -us and masculine nouns in -a. Feminine nouns in -us are unproductive (58 nouns, of which 37 refer to plants, 31 are only feminine, while the remaining can also be masculine and/or neuter) belong mostly to 2nd-declension (e.g., hum-us ‘earth/soil’) and 4th-declension (e.g., dom-us ‘house’) classes. Masculine nouns in -a of the 1st-declension class are more common (101 are only masculine e.g., gumi-a ‘glutton’ and 15 are both masculine and feminine, e.g., hybrid-a ‘hybrid/half-breed’). The majority of masculine nouns in -a are Graecisms (55 total, e.g., achet-a from ἀχέτης ‘male singing cicada’) or Latin (a total of 42 masculine gender only, e.g., andabat-a ‘blind-folded gladiator’) or compounds (a total of 47 of Latin origin masculine gender only or common masculine/feminine gender, e.g., agricol-a ‘farmer’). According to the author, masculine nouns in -a are not irregular since they are productive and have ancient or Greek origins. Aliffi’s conclusion is that grammatical gender is arbitrary.
“Regularity, sub-regularity and irregularity in French acquisition” by Phaedra Royle, Gustavo Beritognolo and Eve Bergeron (pp.227-250) presents data from French children on morphological (ir)regularity. The authors compare French, Spanish, and Italian for sub-regularity. Single- and dual-route models of acquisition are discussed where type frequency is taken as evidence for the former model and overregularizations (e.g., *foots for ‘feet’) as evidence for the latter. Spontaneous verb production in French is presented as evidence of the single-route model because bilingual children produced regular (type frequency effect) verbs better than irregular (token frequency effect) ones (i.e., Nicoladis, Palmer, & Marentete, 2007). Sub-regularity is then discussed in Romance (Spanish, Italian, and French). An elicited verb production study shows that French children are better at producing regular verbs than irregular (Royle, 2007; Kresh, 2008). Both studies also indicate influences of verb frequency, where infrequent verbs are more problematic than frequent ones and regular verbs better produced than sub-regular ones (for 2-4 year olds for Royle, 2007; kindergarten and second grade children for Kresh, 2008). These data do not support the single-route model. A corpus analysis of children’s (3-4 year olds) verb usage is then presented to give further information on type-token frequencies of French verbs. It reveals that French children spontaneously produce more irregular tokens than regular and sub-regular ones combined, while the regular type is more present overall. Then follows a discussion about size and color variable (e.g., vert-verte ‘green’) and invariable (e.g., rouge-rouge ‘red’) French adjectives where invariable adjectives are better mastered than variable ones. The take-home message is that French children are sensitive to productive morphological rules. The presence of specific patterns in the corpus does not automatically trigger rule-like productions in children (i.e., French variable adjectives and sub-regular -u verbs).
“Overabundance in Italian verb morphology and its interactions with other non-canonical phenomena” by Anna Thornton (pp.251-269) deals with overabundance in Italian. In a canonical morphological paradigm, all cells are filled, while defective paradigms are non-canonical. Stem allomorphy is a type of canonical deviation. Thornton addresses overabundance, another type of canonical deviation. She defines overabundance as follows, when a lexeme paradigm cell is filled by two synonymous forms realizing the same morphosyntactic properties that can be used interchangeably (e.g., burnt-burned in English). Forms that share a cell in a lexeme/paradigm are called cell-mates (they are traditionally labeled doublets, but Thornton points out that more than two forms are possible e.g., Italian apparve-apparse-apparì ‘appear.PRF.IND.3SG’). Italian is a particularly rich language in morphological overabundance, as is demonstrated throughout the article. Overabundance interacts with other canonical deviations (e.g., allomorphy, heteroclisis). In particular, the phenomenon suggests, at least in Italian, a strict interaction between overabundance and heteroclisis since two cell-mates can belong to different inflectional classes.
Claudi Balaguer studies two aspects of verb morphology in Northern Catalan in “Fighting irregularity: the reconstruction of verb morphology in Northern Catalan” (pp.271-284), namely the degrees of reconstruction and regularization of the indicative present 1SG and 2SG person markers and the -i ending generalization to the imperfect and conditional tenses. Balaguer focuses on the Roussillon-Conflent-Vellespir areas to analyze verb morphological changes and evolution in this set of Catalan varieties. Catalan has lost all Latin final vowels (except -a), giving rise to suffix-less forms. For the indicative present 1SG, -Ø ending from Old Catalan evolved in a first phase to -e (in Valencian), -i (used in northeastern Catalan areas e.g., canti ‘I sing’), -o (realized as [u] in the oriental area or [o] in the Occidental dialect), and -c (e.g., bec ‘I drink’ from Latin BIBŌ) endings depending of the area. In a second phase, northern and central dialects displayed -o (central) and -i (northern) endings. Another phase of regularization took place, one of an inchoative transformation that made forms more distinguishable from one another (e.g., moreixo instead of moro ‘I die’). Changes also occurred with the 2SG indicative present from Latin to Old Catalan to Standard Catalan (e.g., CURRIS --> cors --> corres ‘you run’). Changes from Latin also gave rise to ambiguity between 1SG and 3SG in the imperfect and conditional tenses (e.g., cantava ‘I/he sang) that underwent an -i change in 1SG (e.g., cantavi ‘I sang). These processes reveal attempts at regularization for pronunciation and ambiguity reasons that occurred in the endangered Northern Catalan dialect but that are slowly disappearing under pressure from central Catalan.
Karsten Koch examines agreement marking irregularity in Thompson Salish clefts by appealing to semantic interpretation “Two levels of semantics and irregular morphology in Thompson Salish clefts” (pp.285-303). In Thompson Salish, person agreement morphology can optionally match clefted 1st or 2nd person focus that may show two types of irregularity, optionality and multiple exponence of agreement (the latter may be manifested by a -ne suffix but are expected to use a subject gap marker or default morphology). Semantic interpretations posit two levels of interpretation, an ordinary meaning and a focus one. Linking the two levels of interpretation leads to a discourse updating process that may result in morphological irregularity because of competing semantic representations. Koch uses the Structured Meaning approach for focus semantic meaning that allows three focus semantic objects, namely Focus, Alternatives, and Background. His two observations are that the irregularity is optional and is restricted to agreement features.
EVALUATION These papers fit together, revolving around the topic of irregularity in morphology. This subject is vast in itself, and an appropriate examination must cover a wide variety of languages and linguistic phenomena. Theoretical approaches vary greatly from one paper to another, which some readers may find difficult to follow. Regardless, papers are well-written in general and extensive enough for the reader to grasp the authors’ full intent. However, readers eager to find recent research data or child acquisition data may not be satisfied; such research is not well covered in the volume. Even though all papers treat irregularity in morphology, they can be better appreciated individually because each paper is self-contained and they do not refer to each other. This sparseness of specific matters could be problematic for an audience looking for a cohesive treatment of irregularity in morphology.
The book achieves the goal of enriching our understanding of irregularity in morphology. Its most interesting aspects are that together, the papers cover a large array of languages, from Native American to European languages. They also address a great number of linguistic concepts and singularities which contribute together or individually to shed light on the complex topic of irregularity in morphology.
Linguists, researchers and language specialists may find in this book answers to some of their questions concerning morphology and (ir)regularity. It is comparable with other linguistic volumes about morphology with the particularity of focusing on morphological regularity, sub-regularity and irregularity. For such a complex subject, the book coheres well, although there is no single direct line of argument beyond its main subject. In sum, irregularity in morphology leaves open a world of potential future research topics which may never be fully understood.
REFERENCES Baddeley, Allan. 1966. The influence of acoustic and semantic similarity on long-term memory for word sequences. Quarterly Journal of Experimental Psychology 18. 302-309.
Bybee, Joan, & Slobin, Dan. 1982. Rules and schemas in the development and use of the English past tense. Language 58. 265-289.
Chomsky, Noam. 1970. Remarks on nominalization. In Roderick Jacobs and Peter Rosenbaum (eds.), Readings in transformational grammar. Washington: Georgetown University Press. 184-221.
Collins, Allan, & Loftus, Elizabeth. 1975. A spreading activation theory of semantic processing. Psychological Review 82. 407-428.
Corbin, Danielle. 1987. Morphologie dérivationnelle et structuration du lexique, 2 vol. Tübingen: Max Niemeyer.
Hopper, Paul. 1987. Emergent grammar. Berkeley Linguistics Society 13. 139-157.
Kiefer, Ferenc. 2000. Regularity. In Geert Booij, Christian Lehmann, & Joachim Mogdan (eds.), Morphologie/Morphology. Ein internationales Handbuch zur Flexion und Wortbildung/An International Handbook on Inflection and Word-Formation. Berlin: Mouton de Gruyter. 296-302.
Kresh, Sarah. 2008. L’acquisition et le traitement de la morphologie du participe passé en français. M.A. Thesis. Université du Québec à Montréal.
Nicoladis, Elena, Palmer, Andrea, & Marentete, Paula. 2007. The role of type and token frequency in using past tense morphemes correctly. Developmental Science 10(2). 237-254.
Pinker, Steven. 1999. Words and rules. New York: Basic Books.
Roeper, Thomas, Snyder, William, & Hiramatsu, Kazuko. 2002. Learnability in a minimalist framework: root compounds, merger, and the syntax-morphology interface. In Ingeborg Lasser (ed.), The process of language acquisition. Frankfurt: Peter Lang. 25-37.
Royle, Phaedra. 2007. Variable effects of morphology and frequency on inflection patterns of French preschoolers. The Mental Lexicon Journal 2(1). 103-125.
Ullman, Michael T. 2001. A neurocognitive perspective on language: The declarative/procedural model. Nature reviews Neuroscience 2. 717-726.
Ullman, Michael T., Pancheva, Roumyana, Love, Tracy, Yee, Eiling, Swinney, David, & Hickok, Gregory. 2005. Neural correlates of lexicon and grammar. Brain and Language 93(2). 185-238.
ABOUT THE REVIEWER:
Alexandra Marquis holds a Ph.D. in Psychology and is currently a Postdoctoral Researcher at École d’orthophonie et d’audiologie (School of Speech Language Pathology and Audiology) at the Université de Montréal (Canada). Her research interests revolve around psycho- and neurolinguistics, language acquisition in infancy, child language development before schooling, and the comparison between first (L1) and second (L2) language processing in both children and adults with or without aphasia. Her present research project deals with the role of verb groups in language development for francophone L1 and sequential bilingual L2 children in Quebec.