From The Seventh Lacus Forum,1980, edited by James Copeland and Philip Davis, Columbia, Sth Carolina: Hornbeam Press

0. Introduction
   In this paper we report on a cross-sectional study focusing on the acquisition of the past participles of French verbs by English-speaking students enrolled in a French immersion program.¹ The paper shows that from a formal linguistic point of view, the morphology of the past participles of French verbs is rather complex and that this complexity results in errors. The study examines the underlying sources of these errors and whether the contribution of each error source changes in a predictable manner over time. It also analyzes the errors with a view to testing the psychological reality of a particular morphological analysis which has been proposed for past participle formation in French. Finally, it illustrates the fruitfulness of the application of a variationist approach to the study of the acquisition of verbal morphology.

1. Method
   Our subject sample consists of grades 2 to 6 English-speaking Canadian students who at the time of data collection (Spring 1977) were enrolled in a French immersion program in Montreal. There were 15 students per grade level. Our subject sample also includes a control group of grades 2 to 6 French-speaking students who were attending a unilingual French language school also in Montreal. There were three such students per grade level. All the students ranged in age from 6 to 12 and were approximately equally divided into males and females. The comparison group was included since we were concerned with establishing a realistic gauge for the immersion students' acquisition of the past participles. It was not known whether the native speakers would also err in their use of the past participles as the available literature did not provide information on the age at which native speakers of French acquire the past participle forms.

   Each student in both groups was interviewed for approximately 20 minutes in French. The interview questions were designed to elicit a spontaneous speech style. The interviews were recorded and then transcribed. The transcripts were checked for all contexts requiring the use of the compound past (either the passé composé or the plus-que-parfait).² Correct and incorrect past participles were tallied for each verb.

2. Past participles in French
   In this section we present some necessary facts concerning past participle formation in French.

   Although we do so from a formal linguistic point of view, this should not be taken to mean that learners of French operate in the same fashion as the linguist. It is a matter of empirical investigation whether they do or not.

   From the point of view of morphology, past participles in French are of the form stem + inflection. In this way does Martinet (1969) analyze all past participles in French. Like Martinet, we shall distinguish three reflexes of the past participle inflection. These are illustrated below.

   é [e]	Ø	u [y]
   mang – é [mãže] trouv – é [truve] all – é [ale] ét – é [ete]	dit –Ø [di] fait –Ø [fe] pris –Ø [pri] eu –Ø [y]	ven – u [vœny] cour – u [kury] v – u [vy] 1 – u [ly]

   It should be noted that the past participle of some of the Ø inflection verbs is subject to gender agreement (e.g., pris [pri] ~ prise [priz]; dit [di] ~ dite [dit]; etc.). Thus Martinet's three-way classification actually represents a simplification since it does not include the inflections [z] and [t]. Nevertheless we adopted his trichotomy as we disregarded gender agreement errors in our investigation of past participle formation.

   As is well known, most (90% according to Guiraud, 1973: 107) French verbs belong to the so-called -er group, that is, their infinitival and past participle forms are homophonous and end in the inflection [e]. This makes [e] by far the most generalized reflex of the past participle inflection. The remaining 10% of French verbs form their past participle with [y] or Ø. Another distinguishing feature of the verbs belonging to the -er group is that the great majority of them have only one stem, which is homophonous with the present 1^st, 2^nd, 3^rd sg. and 3^rd pl. forms (e.g., je mange [mãž] ~ j'ai mange [mãž-e]). In contrast, most of the other verbs have multiple stems (e.g., il meurt [mœr]; nous mourons [mur-õ]; mourir [muri-r]; elle est morte [mor-t] ).

   In summary, from a formal linguistic point of view, past participle acquisition in French involves learning: (a) the reflexes of the past participle inflection, (b) the stem(s) of the verb and (c) suffixation of the correct reflex to the proper verb stem. As such, accurate past participle formation is intricate and on this basis one can expect learner errors.

3. Previous research
   In his longitudinal study of the acquisition of French by two native speakers during the years 0 to 4, Grégoire (1968: 130, 356) found a tendency on their part to assimilate the morphology of non -er verbs to that of the -er verbs. As concerns the past participle, Grégoire reported his subjects using forms such as metté and répondé instead of the appropriate forms mis and répondu. He attributed these erroneous past participles to the powerful analogical effect of the morphology of the previously acquired past participles of the dominant -er verbs. In their analysis of the verb system of Toronto French immersion students, Harley and Swain (1978; 56, 57) reported similar analogical past participles (e.g., vené instead of venu). However, they also reported past participle errors not attested by Grégoire in the speech of his two subjects (e.g., [prã] instead of pris and courir/mettre instead of couru/mis). That such errors were not attested by Grégoire does not necessarily mean that they are particular to the French immersion students. This is due to the very limited size of Grégoire's sample. A larger sample of speakers might have revealed the existence of these new errors given that learners have been found to resort to different personal strategies in their acquisition of a given target language item (Hatch, 1974). On the other hand, the immersion students' acquisition of French at a later age than native speakers may result in unique errors due to heightened hypothesis testing ability (Delay and Burt, 1974; Taylor, 1975). A detailed study of the acquisition of past participles by a larger sample of very young native speakers of French is needed to resolve this issue.

4. Results
   We will present the most general findings first. The error analysis revealed no past participle errors in the speech of the native speakers. We conclude from this finding that the native speakers had already completed their acquisition of the past participles of French verbs by the second grade. The immersion students performed to some extent like the native speakers since as early as grade 2 we found only an insignificant proportion of errors in their formation of the past participle of the -er verbs. The few errors that we did observe all seemed to be cases of substitution of a form which can be interpreted either as a present tense 1^st, 2^nd, 3^rd sg. or 3^rd pl. or as an imperative 2^nd sg. (e.g.,* j'ai trouve,* j'ai va, etc.). Both interpretations are plausible since there is evidence from our corpus as well as Grégoire (1968) that both of these forms are acquired before the past participle forms. Thus, like Grégoire's young native speakers, the immersion students proceeded to acquire first the past participles of the dominant verb group. They were also undoubtedly aided in this by the felicitous homophony of the past participle and infinitive of -er verbs, the latter form being acquired, again according to Grégoire and our own corpus, at an early stage.

   In contrast to the comparison students however, the immersion students were found to commit, as late as grade 6, a significant proportion of errors affecting the past participle of non -er verbs. This finding is consonant, on the one hand, with the fact that these verbs are in the minority and on the whole have past participles which are morphologically less predictable than those of -er verbs and, on the other hand, with the absence of homophony between their past participial and infinitival forms. That the immersion students are still making past participle errors in grade 6 is of interest since it means that after seven years of schooling in French they have not yet reached a level of grammatical proficiency comparable to that of grade 2 native learners of French. Other researchers (Beniak, Mougeon & Côté 1980; Canale, Mougeon & Beniak 1978; Harley and Swain 1978; Spilka 1976; etc.) arrived at similar results as regards other aspects of the grammatical proficiency of early French immersion students. One explanation for this is that the immersion schools provide lower levels of exposure to French and opportunities for use of French than the natural acquisition setting in which the native speakers learn their language from infancy. We would like to point out, however, that while such findings may be taken as an indication that immersion programs fail to foster native-like grammatical proficiency, a recently completed study (Lepicq 1980) reports that many of the errors committed by immersion students at the end of elementary school (especially morphological errors) are rated relatively high by native speakers on a scale of acceptability and intelligibility. Lepicq's findings therefore suggest that the immersion students' interlanguage, albeit deviant with respect to the target language, is nonetheless suitable for the purpose of communicating with native speakers of French.

   In the remainder of this paper we wish to examine the results concerning the past participles of non -er verbs from each of the following language learning perspectives: (a) linguistic variability, (b) underlying sources of error, (c) changes in error source over time and (d) the psychological reality of linguistic descriptions.

   4.1 Linguistic variability
   We investigated the acquisition of each reflex of the past participle inflection and we did so as a function of the verb. We followed this verb by verb approach given that it had proved fruitful in two previous studies which we devoted to the acquisition of elements of French verbal morphology (Beniak, Mougeon and Côté, 1980; Canale, Mougeon and Bélanger, 1978). Through this approach we were able to show that the rate of acquisition of two morphemes (i.e., reflexive pronoun, auxiliary) varies considerably from one verb to another and that it is possible to account for this linguistic variability. The results displayed in Table 1 below reveal the same kind of variability, i.e., that there is internal variation in the rate of acquisition of each reflex of the past participle inflection. For example, within the category of past participles ending in [y], one finds lire which is the object of high proportions of past participle errors and voir which is the object of very few or no past participle errors. Another interesting finding of our analysis is the fact that we have identified several non -er verbs (faire, dire, voir and avoir) whose past participles are acquired (≤ 10% of errors being our acquisition threshold) as early as grade 2, just like the past participles of the -er verbs as reported earlier. These four verbs stand in sharp contrast to verbs like lire or mettre whose past participles are the object of higher than 10% of errors at all of the grade levels.

   The above findings imply that it would have been invalid to establish an order of acquisition of the different reflexes of the past participle inflection on the basis of general percentages of error. Further in this connection, looking back on the morpheme order acquisition studies, the above findings also suggest that these studies' reliance on general percentages of error calculated for each one of the several morphemes which were examined may have hidden much in the way of interesting linguistic variability and therefore possibly led to morpheme rank orderings which were the artifact of these studies' methodology. Those familiar with sociolinguistics will have noticed the similarity between the methodology we have adopted and are advocating and the variationist approach to language use, which is slowly gaining wider application in the field of language learning research.³

   We will not try in this study to account in a detailed fashion for the observed variation in the rates of acquisition of the past participles displayed in Table 1. Let us simply mention that it would appear that verb frequency in the students' linguistic input and production and whether the past participle is homophonous or not with a previously learned verb form are factors which have a bearing on rate of acquisition. For example, the fact that dire and faire are verbs whose past participles are homophonous with the present tense 1^st, 2^nd and 3^rd sg. forms and are high frequency verbs may explain why the past participles dit and fait were the object of very few or no errors at each grade level. However, past participles such as vu and eu, which were also the object of very few or no errors in spite of the fact that they lack previously learned homophonous counterparts and are probably less frequent than fait and dit, are counter-cases which suggest that it is necessary to look for other explanatory factors.

   Table 1: Percentages of error per past participle and grade level

   Past participle	In­flection	Gr 2	Gr 3	Gr 4	Gr 5	Gr 6
   dit fait parti pris fini mis dormi eu	–Ø –Ø –Ø –Ø –Ø –Ø –Ø –Ø	(12)a 0% (25) 4% (0) – (4) 50% (6) 17% (5) 40% (12) 75% (10) 0%	(16) 0% (34) 6% (1) – b (15) 27% (1) – (17) 24% (6) 50% (10) 0%	(20) 0% (45) 2% (4) 0% (20) 5% (3) – (13) 15% (3) – (18) 6%	(14) 0% (46) 0% (7) 0% (22) 0% (5) 20% (9) 11% (2) – (8) 0%	(8) 0% (55) 2% (11) 9% (9) 0% (5) 0% (4) 25% (2) – (3) –
   vu venu revenu bu lu	–u –u –u –u –u	(6) 0% (8) 37% (5) 60% (0) – (4) 75%	(29) 3% (7) 14% (9) 11% (0) – (1) –	(37) 5% (12) 25% (14) 50% (0) – (1) –	(17) 0% (12) 8% (12) 17% (0) – (5) 40%	(6) 0% (15) 7% (31) 0% (4) 0% (7) 43%
   été	–é	(1) –	(0) –	(3) –	(1) –	(4) 0%

   1. The numbers in parentheses indicate total number (N) of correct and incorrect past participles.
   2. We calculated error percentages only when N was = 4.

   4.2 Underlying sources of error
   Taylor (1975) has argued that language learning errors can be attributed to the reliance by learners on previously acquired forms to compensate for a lack of knowledge of the correct forms. Thus we expected some proportion of the past participle errors committed by the immersion students to reflect this strategy. Specifically, we expected substitutions of present tense 1^st, 2^nd, 3^rd sg. and 3^rd pl., imperative 2^nd sg. and infinitival forms for the past participle since, as stated earlier, these forms are acquired before the compound past forms by native as well as L2 learners. Our expectations were borne out. Out of the total number of errors (92) committed by the immersion students in grades 2 to 6 with respect to the formation of the past participle of non -er verbs, 51% reflected the above-mentioned strategy. In fact, reliance on the infinitive (e.g., *j'ai prendre, *il a avoir, etc.) was the single most productive strategy, accounting alone for 36% of the past participle errors. Reliance on the abovementioned present tense or imperative forms (e.g., *tu as [me]; *tu es [vjeε̃], etc.) accounted for 15% of the past participle errors and was the least productive strategy. A possible reason for the immersion students heavier reliance on the infinitive than on the present or imperative may be the homophony which characterizes the infinitive and past participle of the dominant -er verbs. Such homophony may have been noticed by the immersion students, used as a strategy for the formation of the past participle of -er verbs and applied to the formation of the past participle of non -er verbs.⁴

   Given the great predominance of -er verbs in French, we also expected that our data would evidence attempts on the part of the immersion students to form the past participle of non -er verbs as they would that of -er verbs (i.e., use an [e] ending). Errors of the type *il a_[vœne], *j'ai [mεte], etc., which represented 28% of the total number of past participle errors, are a confirmation of this hypothesis. Although these errors are analogical in nature, there are several ways in which the process of analogy may be operating as we shall argue in the last section of this paper. In any case, it should be pointed out that, unlike the first type of error examined above, such analogical errors constitute evidence that the learners are coming to grips with the morphology of the past participle.

   There is one other type of past participle error which the students committed. These were errors like *il a ouvri/reveni, *j'ai prenu/moru, etc. They accounted for the remaining 21% of the past participle errors of non -er verbs. This type of error was not attested by Harley and Swain (1978) in the French of Toronto immersion students nor was it attested by Grégoire in the speech of his two young native speakers. We find this error-type interesting because it evidences the fact that the past participles of non -er verbs can also serve as the basis for the creation of analogical past participles within that class of verbs. It is also noteworthy that the inflections of the past participles of non -er verbs were not used to form the past participle of -er verbs, i.e., we did not observe cases such as *j'ai chantu or *il a trouvi, etc. This is probably due to the regularity of the morphology of the -er verb past participles and to the fact that the non -er verbs, being in the minority, cannot serve as the basis for the creation of analogical past participles outside their own class.

   In our discussion of the underlying sources of past participle errors we have so far not considered the possible role of mother tongue interference. Dumas, Selinker and Swain (1973) have made the general observation that interference is one of the major sources of error in the speech of French immersion students. We would like to see whether this observation applies in the case of past participle formation.

   It has been reported that inflectional morphology doesn't lend itself to borrowing (Weinreich 1968). Consonant with this finding, there were no inflectional borrowings in the speech of the immersion students, that is, we found no examples where they had coupled one of the English allomorphs of the past participle inflection to a French verb stem (e.g., *sauted ~ sauté, *vend ~ venu, *frappe ~ frappé, etc.). One possible explanation for this lack of transferability is the relationship of dependence which inflections hold vis-à-vis stems. In other words, inflections are not perceived by learners or speakers (contact situation) as items which are free and therefore transferable. Cases of language switch such as j'ai 'opened' la porte for j'ai ouvert la porte, found occasionally in our corpus and in that of other researchers, are not counter-examples but rather constitute further supporting evidence for our claim that inflections are dependent on stems and for the observation that the inflections of a given language are rarely, if ever, integrated into the morphology of another language.⁵

   Another potential case of interference might follow from the fact that the past participle of a good number of English verbs is homophonous with the simple past from (e.g., he stopped ~ he has stopped, I stood ~ I have stood, etc.). If native speakers of English are able to perceive this generalized homophony, perhaps they can transfer this correspondence to the learning of French past participles. Such transfer would be negative as it would yield errors like *il a [finisε] , *j'ai [avε], etc. The spoken French of the immersion students contained no such errors and therefore we conclude that such negative transfer did not take place.

   We would like to take up briefly now the question of possible positive transfer from English. Like French, English has past participles which can be broken down morphologically into a stem plus a suffix, i.e., one of the allomorphs of the past participle inflection. This cross-linguistic similarity in the morphology of past participles can only be of help to the immersion students. Because of this similarity, the immersion students do not have to discover the morphology (i.e., stem + suffix) of past participles in French. In addition, neither do they have to discover the syntax (i.e., auxiliary + past participle) of the compound past. In this sense the immersion students enjoy a head-start on the native learner of French and might also enjoy a head-start on L2 learners whose mother tongue lacks past participles, an hypothesis which could be empirically tested.

   4.3 Changes in error source over time
   In two studies of the acquisition of English as a second or third language, Taylor (1975) and Lococo (1976) showed that the proportion of learner errors attributable to interference and overgeneralization varied over time in a non random fashion. More specifically, the proportion of intersystemic errors decreased and that of intrasystemic errors increased as acquisition progressed from the early to the intermediate stages. Given the cross-sectional nature of our study, we thought it would be of interest to investigate whether the two intralingual sources of past participle errors that we have identified could be predicted to change in importance over time and, if so, whether there was indeed a change.

   We have indicated in the previous section that overgeneralization of certain forms of the infinitive or present/imperative is indicative of ignorance of the morphology of the past participle. As such, it is reasonable to assume that this strategy should be more characteristic of the interlanguage of the younger students than of that of the older ones. Alternatively, the analogical errors evidencing the students' coming to grips with the morphology of the past participle should be expected to be more characteristic of the approximative system of the older students than of that of the younger ones.

   A look at Table 2 indicates that these expectations are largely borne out. We can see that at the Grade 2 and 3 levels the percentage of errors due to overgeneralization of the infinitive or present/ imperative accounts for a greater proportion of errors than do the other two sources of error combined. From grade 4 on however, errors showing that the students are coping with the morphology of the past participles are proportionately more frequent than those resulting from overgeneralization of non-participial verb forms.

   Table 2: Change in the contribution of error sources by grade levels.

   Grade	Analogy with PP of -er verbs		Overgeneralization of Infinitive or Present/Imperative		Analogy with PP of non -er verbs
   	N of incorrect PP	%	N of incorrect PP	%	N of incorrect PP	%
   2	6	21	15	52	8	27
   3	3	14	13	62	5	24
   4	9	41	10	45	3	14
   5	4	50	3	38	1	12
   6	4	33	5	42	3	25

   The findings of the present study and those of Taylor and Lococo suggest that in instances where there are several competing sources of L2 errors, the search for changes in the contribution of each error source over time is revealing, whether it is done for one class of items (as in the present study) or for the whole target language (as in the case of Taylor and Lococo's studies). However we feel that from a methodological standpoint, such a search is best undertaken separately for each class of target language items (e.g., prepositions, auxiliaries, etc.) under investigation. On the one hand, this discrete rather than general approach renders possible the identification and exclusion of those classes of target language items which give rise to errors which are attributable to only one underlying cause. On the other hand, only the discrete approach can reveal whether the contribution of a given error source necessarily shows the same evolution over time across different classes of target language items, an assumption which seems to be implicit in the general approach.

   4.4 The psychological reality of linguistic descriptions
   One of the aims of psycholinguistics has been the experimental verification of some of the abstract constructs posited by linguists (e.g., transformational rules, word derivation processes, phonemes, etc.) in their analyses of various languages.

   In the particular area of morphology, Derwing and Baker (1977) designed an experiment to test which one, if any, of the several alternative analyses which have been proposed for plural formation in English, had in their words "an empirically defensible claim to psychological reality". Similarly, in this section we would like to attempt to test the "psychological reality" of Martinet's morphological analysis of past participle formation in French. Our testing ground will be the past participle errors committed by the immersion students.

   The errors which involve use of a non-participial form are excluded here since they do not enlighten us on the issue at hand. As concerns the analogical past participles ending in [e], let us examine how the process of analogy which underlies them might be working. On the one hand, one may suppose that the students have isolated, on the basis of the past participles of -er verbs (which they master very early), the invariant ending [e] , that they have also isolated some stems of non -er verbs on the basis of certain familiar tenses (e.g., nous ven-ons, vous ven-ez, je ven-ais) and that they are affixing the ending [e] to such stems (e.g., *il a ven- [e] ) by analogy with the past participles of the majority -er verbs. On the other hand, the analogical past participles ending in [e] may also be interpreted as substitutions of previously learned simple verb forms ending in [e] (e.g., 2^nd pl. present or imperative). This type of analogy would seem to be based on rhyme, i.e., the form ven- [e] (2^nd pl. present or imperative) rhymes with the past participle of any -er verb and is therefore a prime candidate for use as a past participle of venir in the absence of knowledge of venu. Note that although this rhyming analogy operates at a word level, it involves the perception of a particular word ending: [e].

   Turning to the analogical past participles not ending in [e] , it is obvious that they cannot be the result of rhyming analogy as forms like *reveni, *prenu, etc. do not have any homophonous counterparts in French. We are left therefore with an explanation resting on analogy operating at a morphemic level, i.e., involving isolation of an ending and a stem and affixing of one to the other. It is clear that in the case of analogical past participles such as *prenu, [y] is the ending which has been isolated and affixed to the stem pren- . Matters are less clear, however, in the case of analogical past participles ending in [i]. Two formation strategies might be invoked, one which involves isolation of the stem reveni- and affixation of a Ø ending to it, the other which involves isolation of the stem reven- and affixation of an [i] ending. We believe that the second strategy is more plausible since as Slobin (1977) has observed, learners tend to avoid Ø morphemes. Finally, one could also argue that the analogical past participles ending in [i] are the result of a heuristic of the type 'Drop the -r of the infinitive'. This heuristic could result from an analogy of the form revenir: *reveni on the basis of partir: parti, finir: fini, sortir: sorti, etc.⁶

   In conclusion, the discussion above has shown that learners of French evidence the capacity to form analogical past participles by combining stems and endings, by relying on rhyming analogy or by resorting to a heuristic of some kind. Further, it has shown that when a morphological strategy-is being followed, the endings selected by the learners need not correspond to those posited by Martinet (1969). Whether learners and speakers, for that matter, follow morphological rules when they produce correct past participle forms is another question. We leave it to future empirical research.

Notes

¹ In such programs Anglophone students are schooled entirely or partially in French. Immersion programs were set up to bring about higher levels of French language proficiency than those attained by students enrolled in the traditional core French programs.

² Although in such contexts the students primarily used compound past forms, they also used simple verb forms. These simple verb forms resembled the present, the infinitive or the imparfait and probably reflect the immersion students' unfamiliarity with the compound past. Interference from the rules of tense usage in English may be at the root of the use of the imparfait in this way (e.g., he said good-bye ? *il disait au revoir). These errors were excluded from the present study since they do not bear on the question of past participle formation.

³ Sociolinguistic theory is not only equipped to tackle the problem of linguistic variability but that of inter-individual variability as well (via the use of implicational scales for instance) Two recent studies in the literature on second language learning (Hyltensham 1977; Andersen 1978) go a long way toward extending sociolinguistic theory to the solution of the latter problem. In the case of the present study, the data base was too sparse to permit any meaningful study of the acquisition the past participle inflection on an individual basis.

⁴ It is true that the past participle of some very frequent non -er verbs (i.e., faire, dire) is homophonous with some present or imperative forms. However, this homophony is characteristic of only a few verbs in French and therefore it may not be sufficient to induce learners to rely as heavily on the present tense/imperative overgeneralization as on the overgeneralization of the infinitive in forming the past participles of non -er verbs.

⁵ Verb stems, however, can be borrowed as evidenced in our corpus as well as in that of Harley-and Swain (e.g., j'ai baké un gâteau). Such forms show that in contrast to inflections (bound morphemes), verb stems are free morphemes and therefore transferable.

⁶ It might also be the case that these erroneous past participles are substitutions of a form of the imparfait (i.e., 1^st, 2^nd, 3^rd sg. or 3^rd pl.) which certain speakers pronounce with the ending [e] rather than the prescribed [e].

References

Andersen, R. 1978. An Implicational Model for Second Language Research. Language Learning, Vol 28, No 2.

Beniak, E., Mougeon, R. & Côté, N. 1980. Acquisition of French pronominal verbs by groups of young monolingual and bilingual Canadian students. In The sixth Lacus forum. Columbia, Sth. Carolina: Hornbeam Press.

Canale, M., Mougeon, R. and Bélanger, M. 1978. Analogical leveling of the auxiliary être in Ontarian French. In Contemporary studies in Romance linguistics. (Ed. M. Suñer). Washington, D.C.: Georgetown University Press.

Canale, M., Mougeon, R. Beniak, E. 1978. Acquisition of some grammatical elements in English and French by monolingual and bilingual Canadian students. Canadian Modern Language Review, Vol. 34, No 3.

Derwing, B. and Baker, W. 1977. The psychological basis for morphological rules. In Language learning and thought. (Ed. J. Macnamara). New York: Academic Press.

Dulay, H. and Burt, M. 1974. Errors and strategies in child second language acquisition. Tesol Quaterly, Vol 8, No 2.

Dumas, G., Selinker, L. and Swain, M. 1974. L'apprentissage du français langue seconde en classe d'immersion dans un milieu torontois. Working Papers on Bilingualism, No 1.

Grégoire, A. 1968. L'apprentissage du langage II. 2^nd ed. Paris: Les Belles Lettres.

Guiraud, P. 1973. Le français populaire. Paris: Presses Universitaires de France.

Harley, B. and Swain, M. 1978. An analysis of verb form and function in the speech of French immersion pupils. Working Papers on Bilingualism, No 14.

Hatch, E. 1974. Second language learning—universals? Working Papers on Bilingualism, No 3.

Hyltensham, K. 1977. Implicational Patterns in Inter-language Syntax Variation. Language Learning, Vol 27, No 2.

Lepicq, D. 1980. Aspects théoriques et empiriques de l'acceptabilité linguistique: le cas du français des élèves des classes d'immersion au Canada. Unpublished Ph.D. thesis. University of Toronto.

Lococo, V. 1976. A cross-sectional study on L3 acquisition. Working Papers on Bilingualism, No 9.

Martinet, A. 1969. Le français sans fard. Paris: Presses Universitaires de France.

Slobin, D. 1977. Language change in childhood and in history. In Language learning and thought. (Ed. J. Macnamara.) New York: Academic Press.

Spilka, I. 1976. Assessment of second language performance in immersion programs. Canadian Modern Language Review, Vol 32, No 5.

Taylor, B. 1975. The use of overgeneralization and transfer learning strategies by elementary and intermediate students of E.S.L. Language Learning, Vol 25, No 1.

Weinreich, U. 1968. Languages in contact. The Hague: Mouton.