The Advanced L2 Writer of French: a Study of Number Agreement in Swedish Learners

The Advanced L2 Writer of French

MALIN ÅGREN

The Advanced L2 Writer of French: A Study of Number Agreement in Swedish Learners

The literature in the field of L2[1] acquisition has so far mainly described the advanced L2 speaker (cf. Towell et al, 1996; Bartning, 1997; Howard, 2002; Labeau, 2005, etc). However, it is insufficient to analyze the instructed L2 learner of French exclusively in terms of oral production. In the instructional setting, written input and output play an important role very early in the acquisition process (Harklau, 2002). This paper proposes a widening of the notion of advanced learner to the domain of the written interlanguage and emphasizes the advanced L2 writer of French compared to the native writer.

In the context of this study, a group of young Swedish L2 learners of French at an advanced level is compared to a group of native writers of approximately the same age (16-19 years). We present a detailed analysis of the morphological number marking and agreement in nominal and verbal phrases (NP and VP). The phenomenon of number agreement is particularly interesting in French where number morphology often is silent in speech, and therefore potentially difficult to carry out in writing.

In a wider perspective, this article addresses the question of the different relations of written/spoken French in the L1 and L2 settings and their possible influence on the production of written morphology. In L1, the written system is learnt when the oral language is well established, while oral and written language are learnt in parallelin the instructed L2 setting. In addition, the tutored L2 learners receive a massive written input from the beginning of acquisition. We will show that there are differences in how number morphology is produced in advanced L2 and in the age matched L1 writers. However, we will argue that the underlying L2 grammatical system is not part of the explanation of these differences. Rather, we believe that a combination of external factors, like audible cues, semantic and syntactic complexity as well as type/token frequencies can account for the findings.

General background

Who is an advanced L2 learner of French?

According to Kihlstedt (1998), the study of the advanced L2 learner suffers from a confusion of terms, e.g. advanced learner, qualified learner, competent learner or instructed learner. In her summaryof the literature of the field, Bartning (1997) also raises the same problem of conflicting terms. She suggests the concept of “apprenant à un niveau d’instruction élevé” (‘learner on a high instructional level’) to address a person studying an L2 in an instructional setting that, among other things, emphasizes metalinguistic knowledge. For this group of learners, education at university level is often combined with a stay in the country where the language is spoken.

However, all instructed learners do not reach the advanced level (Kihlstedt, 1998: 19). In fact, there is an important distinction to be made between context of acquisition, on the one hand, and level of acquisition, on the other hand. First, the instructed (tutored) learner can be contrasted to the natural (untutored) learner and in this connection the number of years at university-level is often taken as the main indicator of level of acquisition. An example is Labeau’s study (2005) that compares English L2 learners of French in the first, second and fourth year at the university (see Howard, 2006, for a discussion). Second, when focusing mainly on the stage of development, i.e. initial, intermediate, advanced, etc., it is often desirable to specify more precise linguistic criteria for this categorization. The main idea is that, in many instructional settings,length of study is an uninformative variable (Bardovi-Harlig, 2000: 433). Thus, the present study of advanced Swedish learners of French is based on a more fine-grained notion of level of acquisition, which has little to do with the number of years of instruction. The grouping of learners in different stages of development is based on independent linguistic criteria, such as sentence structure, finiteness, tense,mode and aspect (see Method). These measures are defined in Bartning and Schlyter (2004), see also Klein and Perdue (1997), and we use them to be able to investigate the development of a new variable, in this case number agreement in NP and VP[2].

When presenting the idea of a continuum of acquisition, Bartning (1997) puts forward several specific advanced stages located above the Basic Variety (Klein and Perdue, 1997) and characterized by a “grammaticalized” language, although sometimes weak in morphology. However, the language of the advanced learner is still developing and can, at the upper end, be distinguished from the native-likelanguage by the persistence of non-native patterns (over and under use of certain structures) and a non-native organisation of discourse. Thus, Bartning and Schlyter (2004) suggest three stages of advanced L2 acquisition of French by Swedish learners: le stade avancé-bas (low-advanced), le stade avancé-moyen (intermediate-advanced) and le stade avancé supérieur (native-like). Following this categorization, the learners of the present study belong to the low-advanced stage of L2French.

Written French and its L1 and L2 acquisition

According to Jaffré (1992),the French language counts approximately 30 phonemes that are matched with more than 130 graphemes. This is far from being an economical system. French, like English, is known as a deep orthography, where the “polyvalence” between oral and written language is important and difficult to acquire, especially if you proceed from speech to writing (as in L1).

Written French is known for its specific morphological distinctions in for example number and gender agreement or verbal morphology, which are not distinguishable in the spoken languge. These constitute “fragile orthographic zones” that are difficult to master for L1 learners of French and even for more experienced writers (Brissaud and Jaffré, 2003; Fayol and Largy, 1992; Jaffré and David, 1999). The problem in this domain is that the written language cannot take advantage of its oral counterpart. The difficulty is that the so calledsilent morphology must be derived from the context and not from phonological cues. It has been shown that the mastery of grammatical morphology is crucial for the acquisition of written French. Moreover, the learning and automatization process of the written sub-system is often long and laborious. Fayol and Largy (1992) point out that the real time processing of this written morphology is very costly in cognitive terms and that the system is vulnerable even for expert writers.

For the purpose of the present study, it is important to stress the fact that the context of acquisition of written French is fundamentally different in L1 and L2. We believe that a complete discussion of the results from a comparative study on L1 and L2 written production must include a discussion of the relation between oral and written systems in the two settings respectively. Granfeldt (2006, manuscript) emphasizes the differences in sequencing of the four language activities and in differentiation of input and output. In the L1 setting, listening and speaking clearly precede reading and writing and, in addition, input precedes output in time.In the L2 classroom, the situation is essentially the opposite. The L2 learner is simultaneously confronted with the oral and the written language. Output is often just as valuable as inputand listening and speaking activities arecombined with reading and writingalready at the beginning of acquisition. Other characteristics of the traditional L2 setting are a large amount of written input and a focus on form perspective. Thus, there is reason to believe that the L2 system in general could be influenced by the written language and that L2 learners of written French are used to paying attention to formal aspects of the language. Finally, the concurrent learning processes of speaking and writing could indeed increase the cognitive effort of L2 learning[3].

Number in written French

In the case of written French, number marking and agreement[4]are facilitated by their semantic transparency and the great regularity of the plural morphemes: -s in NP and –nt in VP. In the majority of cases, plural is expressed with the addition of –s to the noun and to the elements in NP that agree with it. Furthermore, according to Riegel et al. (1994: 263), about 90 % of the French verbs agree in number (3rd person plural) with an –nt morpheme added to the verb stem[5]. The regular plural agreement in NP and VP is illustrated in (Ex-1).

(Ex-1)Le|s garçon|s italien|s adore|nt la bière.

‘The Italian boys love beer’

Considering the interrelation of these two plural morphemes, Fayol (2003) notes that –s in NP is much more frequent than –nt in VP. This claim is valid for both type frequency: -s occurs on several constituents while –nt only occurs on verbs, and token frequency: nouns in the plural are more frequent than verbs in the 3rd person plural.

In the discussion of number agreement, we sometimes refer to the learners’ L1 (Swedish) as compared to the L2 of the present study (French). However, it should be mentioned that English is the first L2 learned in Swedish schools from the age of eight/nine years. This is a language that is well mastered by most Swedish adolescents. We present Table 1 in order to illustrate similarities and differences between the three relevant languages for the learners in this study[6].

As illustrated in Table 1, in a small group of French nouns and adjectives (2 %), plural is marked with an irregular –x (New et al., 2004). Due to limited space, these exceptions will not be discussed in the present paper (see Ågren, 2005, for a detailed analysis of their use in written L2 French).

In VP, the analysis is based on a division of the verbs into three groups, according to their written agreement pattern: 1) regular verbs (stem+-nt: ils parlent), 2) four very frequent irregular verbs that agree in 3rd person plural with a suppletive form: il est / ils sont (être, ’be’), il a / ils ont (avoir, ’have’), il fait / ils font (faire, ’do’) and il va / ils vont (aller, ’go’), and 3) a mixed group of irregular verbs (e.g. il prend / ils prennent,‘take’, il veut / ils veulent,‘want’,etc.) that mark plural with an audible stem alternation and the addition of the morpheme –nt.Earlier research has shown that L1 writers are sensitive to oral cues in written plural agreement (Largy and Fayol, 2001). Hence, they wouldproduce the number agreement inils finissentmore easily than in ils parlent, since the singular-plural alternation is phonologically distinguishable in the former but not in the latter. In other words, since L1 learners have gone from speech to writing, their “fragile zone” seems to be constituents where there is no audible difference between singular and plural.

Table 1: Number marking on nouns and pronouns, and number agreement on verbs and adjectives in French, Swedish and English

Caption: grey fieldsindicate domains where the plural form in written French is not phonologically distinguishable from the singular form.

Note that the L1 of the learners, Swedish, is characterized by a lack of verb agreement but, on the other hand, by the presence of an audible number marking within NP; on nouns, pronouns and adjectives. This fact could have a possible influence on the L2 production in French.

Rationale and hypotheses

The rationale for this study is above all empirical. We would like to discover and describe the similarities and differences between the written production of L1 and advanced L2 writers. This work is part of an ongoing project of establishing specific levels of correct plural marking and agreement in written production of L2 French[7].

Before formulating the hypotheses for this study, we conclude thatthere is reason to believe that the number feature as such should not cause great problems to the L2 learners of our study. They belong to the advanced level of acquisition and we know that the number distinction is present in their L1. Furthermore, several studies confirm that number is an early feature in both L1 and L2 acquisition (Fayol, 2003; Hedbor and Ågren, 2006; etc.). However, when it comes to the acquisition setting of writing, the circumstances are, as mentioned above, very different in L1 and L2. The L2 learners receive a great amount of written input from early on and are used to focusing on form and correctness.

In the L1 setting, it has been observed that learners build on the oral language system when learning to write. Written production can be influenced by the spoken language, even at advanced levels. Concerning number agreement, Largy and Fayol (2001) have shown that French L1 writers are sensible to oral cues in written plural formation.

In sum, previous sections give rise to the following hypotheses:

Hypothesis A: Since L1 learners have gone from speech to writing, their “fragile zone” seems to be constituents where there is no audible difference between singular and plural.Therefore, we expect certain omissions of the plural marking and agreement in L1 French in domains that express homonymy between singular and plural.

Hypothesis B: The L2 learners of this study express number in their L1 and previous studies have shown a rather good and early mastery of written morphology intutored L2 French. These two factspredict a high level of correctness in plural marking and agreement at the advanced L2 level. In addition, we postulate that the linguistic system in tutored L2 French to a larger extent than in L1 is influenced by the written language. Thuswe test the hypothesis that the degree of sensitivity to oral cues in plural marking and agreement is weaker in L2 than in L1 writers.An interesting effect of this hypothesis could be that L2 writers at advanced levels have less difficulty in domains that express homonymy between singular and plural.

Method

The present analysis is based on a comparison of 60 writers from the CEFLE corpus[8]: 30 texts from the most advanced L2 learners in the corpus and 30 texts written by the French control group. The L2 learners are Swedish students at high school level, in their 5th or 6th year of French (age 17-19 years). The French L1 writers are of approximately the same age and come from two different high schools in France. The grouping of the L2 learners at an advanced level is based on independent linguistic criteria in several key domains for Swedish L2 learners of French, such as sentence structure, finiteness, mode and expression of time and aspect (see Bartning and Schlyter, 2004). The L2 texts are rich in various types of subordinate clauses, non-finite forms in finite contexts are virtually absent and person agreement is well mastered. Furthermore, these learners express tense and aspect on the verb. Thus, the specific characteristics indicated above distinguish the advanced learners from those at an intermediate or low level of L2 French.

The texts are narratives based on a picture story called Le voyage en Italie (‘The trip to Italy’) where the learners’ attention is focused on content and transmission of the message rather than on form. The pictures tell a (rather classical) story of two Swedish girls on holiday in Italy. They meet two nice Italian boys in a bar and spend a lovely time visiting the country with their new boyfriends. This story was created to elicit gender and number in NP and VP. Differences in nationalities, sizes and colours were used to stimulate the use of adjectives that are often very rare in spontaneous language production. The learners were instructed to tell the story in as much detail as possible to someone who could not see the pictures. They had 50 minutes at their disposal to accomplish the task, which was computerized (cf. Bardovi-Harlig and Bofman, 1989).

The plural contexts have been studied in detail with focus on nouns, pronouns, adjectives and verbs. As already mentioned, the determiners and the irregular nouns and adjectives are not discussed in this paper. Note that our focus is on grammatical morphology and therefore spelling problems are not accounted for. Furthermore, all Swedish or English words have been excluded from the analysis.

The picture story is very important in our analysis of number agreement, since it gives us the context that the writers refer to in their texts. Omissions of plural forms (*il *parle avec les filles, ‘theySG talkSG to the girls’) are easily detected since the specific context is given in the pictures (cf. two boys are talking to the girls). Each writer was analyzed individually before the results of the two groups were gathered. We count the percentage of correct marking or agreement (based on tokens) for the constituents indicated above in order to test our hypotheses. In addition, a more detailed analysis of the omissions is presented.

The linguistic analysis of the material has partly been carried out by the CLAN tools in the CHILDES system (MacWhinney, 2000) and completed by manual analysis. We also used Direkt Profil, an automatic tool for the analysis of written L2 French (Granfeldt et al., 2006) to extract the elements that agree (or not) in number. The statistical analysis where we compare percentages of correct agreement is based on the non parametric Mann-Whitney U-test.