The acquisition and use of adverbs in French as L2

Georges Daniel Véronique 1,* 1 UMR Parole et Langage, CNRS & Aix-Marseille Université; georges.veronique@univ-amu.fr * Correspondence: georges.veronique@univ-amu.fr. Abstract: The acquisition of adverbs in French as L2 (FSL) by learners with different L1s, at various levels of proficiency, has been extensively studied within various SLA projects (ESF, Interfra etc.). In the generativist tradition, it has been noted that “high category adverbs” such as those related to Mood-evaluative, Mood-epistemic and Tense are present early in FSL data. In the functionalist school, the early presence of time adverbs and scope particles in learner varieties has been analysed in relation with the informational and syntactic structures observed in these varieties and in terms of their contribution to temporal reference and other semantic domains in discourse. On the basis of oral data obtained from guided and unguided learners of FSL from different L1 backgrounds (Spanish, Moroccan Arabic, Swedish and Japanese), the paper discusses the acquisition of adverbs in FSL in a functionalist perspective. The development of this class of words is analysed through the positions they hold in learner utterances. The paper confirms that temporal adverbs of position, scope particles and connectives precede other time adverbs. Degree quantifiers are also present in the early developmental stages of FSL. L1 transfer modifies the order of acquisition of adverbs in FSL. Notwithstanding shared commonalities, the identification of a unique development sequence for all FSL learners is a challenging task.


Introduction
The purpose of this paper is to analyse some aspects of the acquisition of adverbs in French as a second language (FSL) and to describe the emergence of a class of adverbs in learner varieties of French. The paper surveys the acquisition of time and degree adverbs, which are among the first lexical and grammatical items to be observed in FSL together with connectives, and analyses the first tokens of scope particles such as aussi (also) and of -ment adverbs such as seulement (only) in learner varieties. The use of negation and of negative adverbs will not be examined in this paper but see Giuliano and Véronique (2005). The data for this paper were obtained from the Interfra (Bartning 2009) and ESF projects (Perdue 1993) on FSL and from transcribed oral data (narratives and conversation) from Japanese learners of FSL (Kerrou 2019, Klingler 1999, Renoud 2016. The acquisition of adverbs in the development of learner varieties in FSL is described in a functionalist framework (Klein 1986).
The focus of the paper is on commonalities in the acquisition and use of adverbs across FSL learners. However, variation due to L1 background or to the learning of French in naturalistic or guided settings are signalled in due course. Section 2 identifies some of the features of adverbs in French and the problems that arise in the course of the acquisition of this class of words in FSL. Section 3 outlines a functionalist perspective on development in SLA. Section 4 provides a short memorization. Besides, despite the specific -ment adverbial marker, adverbs are heterogenous in meaning and form. As pointed out by Bonami et al. (2004), placement and scope of adverbs vary according to whether they are incidental or integrated to the utterance. Alignment of adverbs is constrained; for instance, "manner and degree adverbs must follow all other adverbs except time adverbs" (Bonami et al. 2004: 158). In French, the pre-verbal and post-verbal positions (between the auxiliary and the verb and after the main finite verb) are the positions where adverbs are placed most frequently (Guimier 1993: 36). Most types of integrated adverbs are inserted in post-verbal position.
Thus, the acquisition of adverbs in French implies the identification and development of key positions in utterances, and of the semantic and syntactic intricacies of these lexical and grammatical items. However, because adverbs are more salient than inflection on the verb for instance, they are acquired before inflection on the verb (Ellis 2008).

Stages in the acquisition of French as a second language: a functionalist perspective
A functionalist account of SLA posits that the development of proficiency in L2 is a gradual process shared by all the learners of a specific target language (TL), bar some measure of individual variation.
The acquisition of grammar in TL is shaped by semantic and pragmatic constraints of transparency, salience, diagrammatic iconicity and isomorphism (Klein 1986). According to Andersen (1984), during early SLA, learners seek a one -to-one relation between TL forms and semantic and pragmatic functions. The learners evolve from some form of iconic syntax to more arbitrary constructions.

Towards the basic variety
According to Perdue (1992, 1997), three major stages may be identified in the development of L2 learner varieties. At stage 1, the Nominal Utterance Organization (NUO), learner utterances are mainly lexically based and ordered following pragmatic (focus last) and semantic (Controller first) constraints. Interlanguages exhibiting such properties are known as pre-basic learner varieties (Perdue 1996 In the first stages of L2 acquisition, i) discourse types are produced following a prototypical course of action; ii) utterances are organized by a small number of constructions; iii) the order of the constituents is determined by semantic and pragmatic constraints. Perdue (1996: 143) shows that at NUO, pre-basic learner utterances are highly structured and that "this structuring is the very basis for further development". In pre-basic variety (pre-BV) utterances, alignment is determined by a Topic-Focus order. Based on the same semantic and pragmatic principles as pre-BV, the basic variety (BV) is largely independent of the target language or the first language, reflecting the "construction faculty" more than […] (the) "copying faculty" (Klein 2009: 335) of the language acquirers (Dimroth 2018: 3). An adaptation to the properties of the target language, as well as more substantial L1 influence, are assumed to happen only when this simple and self-contained system evolves because of communicative shortcomings and starts to develop finiteness (Klein and Perdue 1997). Thus, according to Klein and Perdue (1997), L1 transfer is rare in the very early stages of L2 acquisition. It may be observed as from the IUO stage and varies according to the linguistic domains concerned (NP, VP, PrepP, etc.) (see Ellis 2008).

Beyond the basic variety in FSL
Most of the naturalistic FSL learners studied in the course of the ESF project developed no further than the post-basic variety (IUO). In the wake of Klein and Perdue (1997), Bartning and Schlyter (2004) defined a more complete scale of developmental stages in FSL for Swedish L1 learners based on five major set of features: the degree of inflection of VP, the expression of Temporality (Time, Modality and Aspect), the placement of negation and the development of NP and of subordination.
In Bartning and Schlyter's scale, the "initial stage" (stage 1) is similar to the pre-basic variety and the "post-initial stage" (stage 2) to the basic variety of Klein and Perdue's (1997) model. The proficiency in FSL are identified as "advanced low" (stage 4), "advanced medium" (stage 5) and "advanced high" (stage 6). These latter stages describe L2 development beyond the three stages identified by Klein and Perdue (1997).

Previous research work on the acquisition of adverbs in FSL
Previous work on the acquisition of adverbs in FSL has been primarily devoted to time-adverbs and scope particles, mainly additive and restrictive particles. Besides the items listed in table 1, alors (then), après (after), maintenant (now), peut-être (perhaps) and vraiment (really, truly) have also been frequently analysed.
Within the generativist framework developed by Cinque (1999), Schlyter (2005) notes that "high category adverbs" such as those related to Mood-evaluative, Mood-epistemic and Tense are present early in Swedish L1 FSL data produced by low level proficiency learners of French. Schlyter (2005) points out that these adverbs, potential functional heads for adverbial phrases, are present in early FSL grammar, although syntactic finiteness is not yet marked on the verb. Thus, they provide an early means of marking temporality in learner varieties of FSL (see Prévost & White 2000).
In a functionalist perspective, adverbs have been analysed in terms of scope, and in terms of their contribution to informational structure, to the semantics of learner utterances and to learners' L2 discourse. Dimroth and Watorek (2000) analysed the use of additive particle aussi in the data produced by three Spanish L1 and three Moroccan Arabic (MA) learners of FSL. They showed that in the early acquisition of FSL, scope particles such as aussi are adjacent to the item in their Domain of application (DA), and that the direction of scope is variable, with a dominant left orientation. The placement of scope particles is partly determined by the existence of contrastive topics within their DA. Starren and Van Hout (1996) provided a first analysis of the relation between the placement of temporal adverb of duration (TAD) toujours and aspectual meaning in learner varieties. Benazzo and Giuliano (1998) and Benazzo (2005) analysing FSL data collected from Spanish L1 participants of the ESF project showed that in the course of acquisition, the scope of the particles narrows down from scope over the whole utterance to a narrow scope where the adverb is adjacent to the item within its DA. Later in the course of acquisition, after integration within the finite utterance, wide scope develops again. Sanell (2007: 121) partly confirmed Benazzo's results for aussi in pre-basic and basic varieties (stade initial and post-initial) for Swedish learners of French L2, especially in terms of adjacency. Benazzo (2005) and Sanell (2007: 128-129) have shown that additive particles are acquired before iterative particles. Benazzo (2005) analyzed the acquisition of aussi, encore, déjà, toujours and seulement by Spanish L1 ESF learners. She observed a correlation between the acquisitional sequence of scope particles and the developmental stages of learner varieties in FSL. At the pre-basic stage, scope particles with additive and restrictive meanings bear narrow scope on NP and PP. At the basic level, scope particles carry temporal meanings such as iteration and have scope on the non-inflected verbs and its complements.
At the post-basic stage, the same particles carry resultative and continuative meaning with the inflected verb within their scope. These results were confirmed by Starren (2001)  This developmental sequence will be further discussed in section 7.
Although the placement of scope particles is sensitive to the development of finiteness on the verb, finiteness on VP does not play the same role in the MA L1 learners' varieties of FSL as in the Spanish data studied by Benazzo (2005: 87-88). Scope particles are found both in the thetic and categorical constructions in MA FSL learner varieties. They occur with contrastive topics and contrastive focus.
The development of scope particles in MA L1 learners of FSL seems to be determined by cognitive saliency and semantic complexity rather than by the information structure of the constructions where they occur or by the development of finiteness on VP (Véronique 2014).
Hancock and Sanell (2009) studied four French adverbs vraiment (really), peut-être (perhaps), aussi (also) and seulement (only) in the Interfra database. These adverbs were analysed in terms of the syntactic position they occupy and in terms of the semantic and pragmatic functions they fulfil within discourse. Hancock and Sanell (2009) noted that as from the 'advanced medium stage' (Bartning & Schlyter 2004), peut-etre, vraiment and seulement are used in a native-like manner by the learners whereas aussi exhibits a non-native-like frequency. Thörle (2020) in a study of the use of aussi by German L1 advanced learners of FSL noted one non-native-like peculiarity in her data, recourse to a full pronoun bound to the NP in the scope of aussi in topic position (e.g le père lui aussi jette le repas 'the father throws away the meal too'). In a nutshell, previous studies on the acquisition of adverbs in FSL have investigated mainly the use and acquisition of time adverbs, scope particles and some -ment adverbs bearing restrictive meaning or classified as evaluatives and modals. These studies have established the interplay between the development of adverbs, the organisation of information structure and syntactic and textual development in FSL. Both Benazzo (2005) and Hancock and Sanell (2018) have proposed acquisition orders for the following adverbs aussi, encore, déjà, toujours, peut-être, seulement, and vraiment in the data they analyzed. Although development orders have been identified for specific L1 groups of learners, no common sequence has been found yet. At least one focus particle, aussi, seems to be used in a non-native like way in learner varieties, even in near-native varieties. Relevant examples from the studies discussed above will be taken up in section 6 of this paper.

Research question and data base
Adverbs are present in learner varieties from their inception. They fulfil different functions at different stages of the development of learner varieties in FSL. Drawing on previous work and exploring new data, the paper analyses the development of the word class of adverbs in FSL learner varieties by describing the role of these lexical items at three key positions in the utterance (Bernini 2014): i) between the topic component and the rest of the utterance; ii) the utterance initial position and iii) the post-verbal position. As mentioned above (section 2), these are the positions where a wide range of adverbs is found in French. The presence of adverbs in these positions in learner varieties contributes to the definition of a word class of adverbs, distinct from nouns and verbs (Bernini 2014). The paper also wishes to test the claims of Benazzo (2005), Versteegh (2018) and Hancock and Sanell 2018 on the course of development of adverbs in SLA.

Data base
The survey of the development of adverbs in FSL is based on the following sources. Data produced by unguided FSL learners of the ESF project (Arabic L1 and Spanish L1 unguided learners of French) (see table 1), and by guided Japanese L1, English L2 learners (Kerrou 2019, Klingler 1999, Renoud 2016) and from the Interfra project (Swedish L1 guided learners of French) (see table 2) will be discussed.

Adverbs in FSL: results of data analysis
In the first place, the role of degree, time, additive and restrictive adverbs at various stages of development will be surveyed. Then, some of the time adverbs and scalar adverbs will also be discussed under their guise of scope particles (König 1991), together with seulement (always). Finally, most frequent -ment adverbs will be analysed. Only significant examples shall be quoted; unless information is provided to the contrary; the uses of adverbs described below are shared by all the learners at the same stage of acquisition. Examples from guided and unguided learners of French will be provided (see Tables 2 and 3). The orthographies used for coding oral data in the various data bases explored have been maintained in the examples quoted below.

Time Adverbs and the expression of temporality
In this sub-section, time, frequency and duration adverbs (see 2 above) will be grouped under the In examples 1 to 5, TAPs provide temporal reference for the events depicted and act as inter-clausal connectives.
Temporal adverbs of duration (TAD) such as toujours (always), temporal adverbs of frequency (TAQ) such as souvent (often) and temporal adverbs of contrast (TAC) such as déjà (already) are also used early in FSL acquisition.
In early learner varieties, toujours (always) as a TAD specifies various aspectual values. In 6, toujours in initial position extends its scope over l'ardoise le premier which is within its domain of application (DA), to express habituality, 1 In the course of the ESF project, data were collected from each informant on a monthly basis. 3 cycles of 9 encounters were organised. The data collecting period lasted 3 years in toto. All the tasks were repeated 3 times at ca. one year interval. Learner utterances analysed in the paper indicate the cycle and encounter where they occurred (e.g. Cycle 1, encounter 1); see Giacomi, Stoffel, Véronique eds. (2000). In 7 and 8, toujours is in key position 1 between the topic moi and new information /telefone/ papa. The aspectual meaning marked by toujours is iteration because of the lexical meaning of /telefone/ which depicts an activity, 7. A. deux mois moi toujours /telefone/ papa (two months, me always phone dad) (2.4).

Z. toujours toujours
With the same type of placement, in 8 toujours expresses continuity because of the non-telic value of mal de tête: 8. Z. un mois moi toujours mal la tête (one month me always head ache) (1.8).
As a TAQ, encore placed in utterance initial position marks continuity of TT in early FSL grammar as in 9: 9. Z. encore Rachid mal le ventre (always Rachid belly ache) (2: 4) In utterance final position associated to the focus with scope to the left, TAQ encore (still, again) relates the TSit of the event to the preceding TSit as in 10, 10. Z. /i parti/ à l'école encore (he goes to school again) (3: 2).
(yes ++ when walk // when always sleep like this + it's OK in the morning it's OK and this afternoon again when the wind is warm + again (s)he is ill).
As illustrated in 11, various scalar time adverbs are mustered in Zahra's basic learner variety to convey the iterative relation between the prevailing heat, sleep and her state of health. Toujours and encore refer to the time sequence of events /dor/ (sleep) and /fi/ chaud (it's hot) while aussi whose scope bears on /li/ malade hinges this clause to the other utterances.
In the Swedish Interfra data (Sanell 2007: 155), encore is present as from the intermediate stage as in 12, 12. I. oui oui tu suis la télévision française? (yes yes do you follow on the French TV), E(melie). Oui parfois bien sûr mais c'est encore un peu plus difficile (yes sometimes obviously but it is still difficult) (Sanell 2007: 155).
Encore conveys first a continuative value and at more advanced stages, encore conveys iterative meaning. In 12, encore regularly associated to the degree adverb peu (little) is integrated in post-verbal position in a thetic construction.
13. M(ona). Ils ont perdu ma valise encore une fois et il a fallu attendre trois jours encore (they have lost my suitcase once again and I had to wait for three days further) (Sanell 2007: 162).
In the Interfra data, TAC déjà is present as from the intermediate stage, 14. O(skar) oui parce que vous / nous avons déjà regardé un peu (…) film (yes because we have already viewed a little (of) the film) (Sanell 2007: 155).
At a more advanced stage, déjà marks ongoing events in the past,

Ch(ristina). C'est avec Depardieu mais je l'ai déjà vu une fois (It is with Depardieu that I have watched him once)
Benazzo examined closely the development of déjà in Spanish L1 FSL (see Table 4

Degree adverbs
Amidst the three classes of degree adverbs identified by Abeillé, Doetjes, Molendijk and De Swart (2004: 193), FSL learners acquire degree quantifiers such as beaucoup at a very early stage, other degree adverbs such as intensity (légèrement (lightly)) or completion adverbs (partiellement (partly)) are acquired late, if at all, possibly because they are -ment adverbs. Très (very) and beaucoup (much) are confused by Japanese L1 FSL learners as in 20 and 21.
20. Ai. Aujourd'hui j'étudi français. J'étudiait français très. Je veni au France (To-day, I learn French. I learnt French much. I came to France) 21. Ai. Et je n'ai pas de memorise parce que j'étudiais beaucoup (I have no memory because I learn a lot)

Scope particles
Various types of adverbs exert scope as from the first stages of acquisition in learner varieties. As may be observed in Tables 4, 5 Table 6 provides information on the use of scope particles aussi, même (even) and encore in MA FSL learners. Aussi is used by all the MA L1 informants at various stages of the development of their learner varieties, from pre-basic to post-basic stages. It is used with both contrastive focus and contrastive topics in its DA. The particle exerts both narrow and wide scope although adjacency to the element within its DA is the most frequent placement. In MA L1 FSL aussi is not used first at the periphery of the clause before working its integration in sentence internal position. Moroccan learners differ on that count from Spanish learners (see Benazzo 2005). The scope of aussi goes mainly to the left. Example 22 illustrates the use of aussi with a contrastive topic in its DA, 22. I. est-ce que tu penses que il était un peu raciste avec toi ('do you believe he was a little racist with you') A. cent pour cent la police aussi /e/ raciste ('hundred per cent (sure) the police also is racist') (2.7) In the Interfra data, aussi has been analysed by Sanell (2007) and Hancock and Sanell (2009). In Swedish L1 FSL, aussi is used as from the initial stage of acquisition with an additive meaning as observed also in data from the ESF project (Benazzo, 2000(Benazzo, , 2005Véronique 2012).  In MA L1 FSL, encore may be used with NPs, presentationals and clauses containing full finite verbs.
It conveys additive meaning (more) and it marks also iteration and continuity.
Besides, there are also some tokens of encore as a textual marker in the Interfra data, at more advanced stages of FSL acquisition (Hancock 2014).
In a nutshell, scope particles such as aussi, encore, déjà, même etc. are found among all the learner varieties examined. These particles convey nearly similar temporal and textual values in the different learner varieties. It should be noted that aussi is more frequently used in learner varieties than in native speaker data (Hancock & Sanell 2009 -ment adverbs are acquired late as from the advanced low stage of FSL acquisition. The only exception is the use of seulement in Spanish L1 learners which offers a clear case of interference from L1.

Summary
Learners with different L1 backgrounds exhibit some degree of variation in the use of the adverbs investigated. Sanell (2007: 154), for instance, shows that the first tokens of encore in learner varieties (L1 Swedish) at intermediate stage bear aspectual "continuative" meaning whereas the first tokens of encore in L1 Spanish learners of the ESF project bear "iterative meaning". Sun (2012: 123) emphasizes the fact that her Chinese L1 learners made little use of the diversity of time adverbs available in French, using aussi extensively. French learners with a Romance language background use items like Fr. seulement extensively.

Discussion
Given the heterogeneous lexical nature of this class of words, various types of adverbs are present as from the early stages of L2 acquisition. Early acquisition of adverbs is probably due to the salience of this class of items (Ellis 2008). Given that adverbs pertain to various semantic classes, they are more liable to L1 or L2 transfer than strict grammatical functors; the use of Fr. seulement provides a good case in point. In the course of the acquisition of FSL, adverbs are placed in various positions in finite or non-finite utterances, exerting wide or narrow scope in their domain of application. The type of scope exerted and the placement of adverbs vis à vis the constituents in their DA as illustrated for instance by aussi, exhibit some variation according to the groups of learners involved. This may be attributed to L1 transfer. It must be added that various groups of FSL learners tend also to overuse aussi as compared to native speakers (Sun 2014, Thörle 2020. As from the first stages of acquisition, adverbs are placed in the three key positions identified by Bernini (2014) in early learner varieties. In French L1, these positions are frequently occupied by adverbs as adjuncts.
As a syntactic word class, adverbs in FSL do not exhibit the same properties as in French L1. The constraints on the alignment of integrated adverbs described by Bonami et al. (2004: 158) rarely apply in FSL except for encore un peu (a little again) for instance, in the Interfra data because the alignment of adverbs is a rare phenomenon in FSL.
The attempts of Benazzo (2004) and Hancock and Sannell (2009) to sketch an order of acquisition of a set of adverbs in their data bases according to the stage of proficiency reached are challenging. From the data investigated in section 7, some regularities are observed: i) temporal adverbs of position, connectives and degree quantifiers are the first type of adverbs to be used together with additive aussi; ii) temporal adverbs of frequency, of contrast and of duration are used later with different semantic values according to the L1 and L2 of the learners ; -ment adverbs tend to be acquired late except for specific cases of L1 transfer as for seulement in the Spanish L1 FSL learner varieties. Certain semantic classes of adverbs, mainly marked by -ment adverbs, such as speech act adverbs are rarely found in FSL learner varieties. The acquisition of adverbs in FSL is gradual and orderly but transfer from L1 interferes with a common acquisitional sequence for all groups of learners.