The production of lexical categories (VP) and functional categories (copula) at the initial stage of child L2 acquisition (Research Paper)

The production of lexical categories (VP) and functional categories (copula) at the initial stage of child L2 acquisition

Assistant Professor of Applied Linguistics

Abstract

This is a longitudinal case study of two Farsi-speaking children learning English: ‘Bernard’ and ‘Melissa’, who were 7;4 and 8;4 at the start of data collection. The research deals with the initial state and further development in the child second language (L2) acquisition of syntax regarding the presence or absence of copula as a functional category, as well as the role and degree of L1 influence in lexical and functional categories. Some studies in the field of child first language (L1) acquisition are discussed to determine similarities or differences between child L1 and child L2 acquisition. Examining data collected from the children’s spontaneous speech, the researcher’s diaries and translation, and other tasks over a period of 20 months. The competing claims of the two most prominent hypotheses about early L2 grammars are tested: Vainikka & Young-Scholten’s (1996) Minimal Trees/Structure Building hypothesis and Schwartz & Sprouse’s (1996) Full Transfer/Full Access hypothesis. Word order, use of rote-learned formulae, and suppliance of verbs (lexical category) and copula (functional category) are investigated, and the conclusion is reached that lexical categories are influenced by L1 whereas functional categories are absent at the initial state and that they emerge without the learners’ reliance on their L1, consistent with Minimal Trees/Structure Building.

Keywords: Lexical categories, Functional categories, Child L1 and L2 acquisition, Copula, Full transfer/full access, Minimal trees/structure building

1. Introduction

The first studies on child language acquisition began to appear over one hundred years ago. These were part of a general interest in child development that occurred at that time led in many respects by the work of G. Stanley Hall in North America and William Preyer in Europe. For the first time in the history children became the focus of study to determine the way in which they develop in general. The method selected was parental diary. The linguist or psychologist parent would keep a diary of his/her child’s learning over some period of time (see L1 studies discussed in Ingram, (1989) as well as Wode’s (1978) study of his four L1 German, L2 English children). Collecting data from children is a challenging and demanding activity which requires patience and accuracy. The investigator should make the data collection a pleasant task for the children to feel comfortable while being studied. The questions should be related to their interests and free of repetitions. If the children are given lots of input regarding a specific structure through repetition, their production will likely be unnatural and based on memorization. There should be, on the other hand, enough production by the learners of a construction under study since a small number of productions cannot be a good indication of the subjects’ underlying grammars related to that structure (see Cox, 2005). This contradiction makes data collection a difficult task.

One of the first approaches to L1 influence in the domain of second language acquisition based on structural linguistics and behaviourist psychology was the Contrastive Analysis Hypothesis (CAH) of Lado (1957). It claimed that individuals tend to transfer the forms and meanings, and the distribution of forms and meanings of their native or first language and culture to the foreign language (FL) culture. Those properties of the L2 which are similar to the L1 were held to be easily learned, whereas the different features were hypothesized to be difficult to learn. The plausibility of CAH was rejected by many researchers (Lococo 1975; among others). Inability of CAH to predict some phenomena in L2 acquisition led researchers in late 1960s and early 1970s to change their focus from L1 transfer and devote most of their attention to stage-like development and cross-learner systematicity.

The term ‘initial state’ which was largely neglected from the mid-1970s to the early 1990s refers to the unconscious linguistic knowledge L2 learners start with. Differences between child L1 and child L2 acquisition, the issue of L2 initial state and the extent of L1 influence have been addressed in a number of studies since then (Eubank 1996, Hawkins 2001, Schwartz & Sprouse 1996, Vainikka & Young- Scholten 1994, 1996a, 1996b; 2005,). These studies are mainly concerned with L1 influence in adult L2 acquisition but there are also studies dealing with L2 children in this regard (Haznedar 1997, Lakshmanan 1993/1994, Lakshmanan 1994, Lakshmanan & Selinker 1994, Unsworth 2005). There is no agreement among L2 researchers regarding the characteristics of the initial state in terms of the role of the L1. According to the Full Transfer/Full Access hypothesis of Schwartz & Sprouse (1996) which is based on the ‘strong continuity hypothesis’ in L1 acquisition, the entire L1 grammar (in the sense of all abstract properties but excluding specific lexical items) constitutes the initial state. This means that all the principles and parameter values as instantiated in the L1 grammar immediately carry over as the initial state of a new grammatical system on first exposure to input from the target language. The initial state of the L2 system will change in light of Target Language (TL) input that cannot be generated by this L1 grammar. As restructuring continues, each intermediate system is a distinct interlanguage (grammar). The idea under FT/FA is that the course that L2 development takes is determined in part by the initial state, in part by input, in part by the apparatus of Universal Grammar (UG) and in part by learnability considerations. Under the Valueless Features hypothesis of Eubank (1993/1994, 1994, 1996) the initial state is a grammar where L1 lexical and functional categories are present in the earliest interlanguage grammar, but their feature values are claimed not to be. Features are not thought of as being either strong or weak, but are valueless or inert in the initial state, i. e., feature strength does not transfer. The Minimal Trees/Structure Building hypothesis is based on the ‘weak continuity hypothesis’ for L1 acquisition (Clahsen, Eisenbeiss & Penke 1996; Clahsen, Eisenbeiss & Vainikka 1994; Pinker 1984; Vainikka 1993/1994). Vainikka & Young-Scholten (1994; 1996a, b) claim that the initial state is a grammar with early representations based on the L1, but the L1 grammar is only partially involved in constituting the initial state. The initial grammars lack the full complement of functional categories, whereas these categories exist in the UG inventory. This means that the initial grammar includes only lexical categories and lacks functional categories of both the L1 and any other source. Vainikka & Young-Scholten also claim that grammars in the earliest stage of development are different from later grammars, lacking certain subsequent properties. Under this approach functional categories emerge gradually. According to Vainikka & Young-Scholten’s proposal, the initial state in L2 acquisition consists of a grammar partly based on the L1 where lexical categories, together with associated properties, particularly headedness, are found in the initial interlanguage grammar. Organic Grammar is Vainikka & Young- Scholten’s more recent hypothesis. They introduce it in a 2005 paper criticizing the Basic Variety (Klein & Perdue, 1992, 1997) which claims that in the earliest interlanguage, the sentences are basically in SVO word order. They also criticize Processability theory (PT) (Pienemann, 1998, 2003) which claims that Processing is the same for both L1 and L2, but adult L2 learners access UG via their L1. PT also claims that at the earliest stages the lexical items are syntax-independent. Vainikka and Young-Scholten point out that language learners whether L1 or L2, build up phrase structure in a similar way starting with lexical projection but L2 learners’ initial state includes the L1 lexical projection; or, is based on L1-based minimal trees as well. In another article, Vainikka & Young-Scholten (2006) argue against Prévost & White (2000a, b, c) who claim that their child L2 data provide evidence for the Truncation Hypothesis, while the adult L2 data support the Missing Surface Inflection Hypothesis. Vainikka & Young-Scholten reanalyze the data and observe that Prévost & White’s data support Organic Grammar. The last hypothesis to discuss is Modulated Structure Building (Hawkins, 2001) where L2 learners start building syntactic representations for clauses with projections of thematic verbs but without an Inflections Projection (IP). He argues that learners start with minimal trees (as described above) which is the structure building part of the theory, but first language functional features transfer to the second language (the modulated part). He claims that parameters involving functional features cannot be reset in the second language and learners re-analyse the input on the basis of first language settings.  The studies mentioned above in the domain of the initial state of L2 acquisition have generally focused on the role of L1 influence in adult L2 acquisition and little work has been done on child L2 acquisition in this regard. The present study looks at the acquisition of L2 English morphosyntax by two L1 Farsi children to find out how UG, the L1 and the input interact when age is not a factor. There are few if any longitudinal studies of the acquisition of English morph-syntax by more than one child at nearly the same age, but different sex and in the same environment. The spontaneous data (S) in this study is based on oral production gathered longitudinally from two Farsi-speaking children, Melissa 7;4 and Bernard 8;420 (sister and brother; from now on in this study M & B) who at the start of data collection had not been exposed to English upon their arrival in the UK. They lived in university student family accommodation where there were many native and nonnative children available to talk to. They started going to an English school for six hours a day immediately after their arrival. There were no other Farsi speaking children in that school and they had to interact in English during six hours of school each day. Their father (the present researcher) was a PhD student in the UK at the time of the study and they expected to stay in the UK at least for three years. Their physical and mental development was age-consistent and they did not suffer from any speech or language impairments. At home, they watched British television and became interested in reading different kinds of books in English. Data collection started on 20 April 2003 which is about 50 days after the learners’ arrival and the learners can be considered as being in their initial states of L2 acquisition. This study is different from some child L2 studies (e. g. Grondin & White, Lakshmanan & Selinker) based on data being collected relatively long after initial exposure. Since the researcher was not a native speaker of English, two native speakers who were teachers as well as linguistics students at the time helped him in the data collection process. The data were collected for 20 months. Audio-recordings were made roughly once a week, but sometimes every other week or even once a month when the assistants were away. Recording would start after five or ten minutes of greetings and warm-up. Each recording varied in length from 90 to 120 minutes. 41 samples were audio-recorded, transcribed and analyzed.

Research questions

The present study aims at finding answers to the following questions:

1. Is there L1 influence in the early lexical and functional categories?
2. What is the status of copula as a functional category in the initial grammars of the two Farsi-speaking children in this study and when do they emerge?

To suggest answers for the two above-mentioned questions, Farsi and English are compared regarding the headedness of VP (lexical category) and copula (functional category) to find out the presence or absence of L1 influence.

The VP in Farsi

Farsi is an Indo-European language. The standard analyses of Farsi show that VP is always head final both in main clauses (1) and embedded clauses (2) and it has a SOV word order (Mahootian, 1997). When a prepositional phrase is present it typically occurs between the subject and direct object, therefore, a more complete description of constituent order is S PP O V. Verbs are marked for tense and aspect and agree with the subject in person and number and the subject is derivable from both agreement marking on the verb and from pragmatic clues in the discourse and can be empty. Although Persian is verb-final at the sentential level, it behaves like head initial languages in noun phrases and prepositional phrases. The head noun in an NP is often followed by the modifiers and possessors and the preposition precedes the complement NP. Certain prepositional phrases such as locative and directional PPs can follow the verb.

(1) Ali ketab mikhanæd.

Ali book pres-read-3sg

‘Ali reads book.’

(2) Ma midanim ke Ali ketab mikhanæd.

We pres-know-1pl that Ali book pres-read-3sg

‘We know that Ali reads book.’

Whereas in a simple transitive clause the default or neutral word order is SOV, other orders are possible as a result of scrambling. This makes Farsi have relatively free word order. All scrambled results are grammatical and are semantically equal, but movement of elements is not without pragmatic consequences or implications.

(3) Ali medad ra bærdašt (SOV)

     Ali pen          OM^[2]  took

‘Ali took the pen.’

(4) medad ra Ali bærdašt (OSV)

      Pen        OM Ali took

    ‘Ali took the pen.’

(5) medad ra bærdašt Ali (OVS)

     Pen       OM took      Ali

  ‘Ali took the pen.’

(6) Ali bærdašt medad ra (SVO)

    Ali      took      pen OM

   ‘Ali took the pen.’

Although there are modals and auxiliaries in Farsi, they do not work in the same way as in English, for all verbs in Farsi are inflected. Farsi does not have the range of meanings expressed through modals in English:

(7) Man mi-tævan-æm be-nevis-æm

     I pre-can-1sg subjunctive-write-1sg

   ‘I can write.’

Early verb phrases

This study tries to make as strict as possible the criteria to decide whether the verb in the initial state is in the VP or it has been raised to a functional projection. In Vainikka & Young-Scholten’s (1994) study of (adult) L2 acquisition of German by 107 speakers of different L1s, they concluded that the verb is considered to be in the VP if it is preceded by an object, an adverb, or other PP arguments or adjuncts. If the verb is followed only by one of the following elements, it is considered as being raised from the VP. As was already mentioned, the canonical word order in Farsi is SOV. Although there were not many (see below) sentences containing thematic verbs in the earliest samples, a high percentage of the early verbs followed their complements in these sentences. To calculate the extent of XV vs. VX utterances produced in the early productions of the two learners, those utterances which contain a thematic verb and at least another VP-internal constituent were counted. Early thematic verbs were produced while asking the subjects to translate some sentences from Farsi into English. A high proportion of the thematic verbs produced up to Sample 7, 21 out of 23 (91.30%), are in an SOV word order. In (8d) and (8e), the verb have which is also placed in final position in the VP, semantically means to be (3sg) and has been produced by both subjects most probably due to similar phonological appearance of have with the verb ‘hast’ meaning ‘to be’ in Farsi. Some of the early utterances are given in (8):

(8)  a. B: My ice cream like. (T 4)

        ‘I like ice cream’.

      b M: My can football play. (T 4)

        ‘I can play football.’

      c. B: Mum salad food # fooding. (T 4)

         ‘Mum is eating salad.’

      d. B: Shell water has. (S 7)

        ‘The shell is in the water.’

      e. M: Spot cupboard have. (S 7)

        ‘Spot is in the cupboard.’

      f. M: The chicken on the tractor sitting. (S 8)

        ‘The chicken is sitting on the tractor.’

The reason behind the strategy of giving the subjects the translation task (T) is that restructuring towards the L2 may be so rapid that makes the detecting of the initial transfer next to impossible, especially when the subjects do not have enough thematic verbs in their vocabulary to use. Moreover, the researcher wanted to explore the idea that word order is implicitly acquired by the learners by giving the learners two possible forms in the L1 to find the probable different production in their L2. For example, when the learners were given a null subject sentence, since Farsi is a prodrop language, the English equivalent would be without a subject as in (9). When they were given a Farsi sentence with a subject, the English equivalent was produced with the subject although these subjects do not often assign case as in (10a). Although this may cause to plugging English words into Farsi syntax, it shows, at least, that the learners have no idea about the English syntactic structure.

(9) a. M: Can football play. (T 4)

     b. B: Tennis play. (T 4)

(10) a. M: My can football play. (T 4)

       b. B: We tennis play. (T 4)

Figure 1: Percentage of VX vs. XV utterances

Up until Sample 8, B. produced 17 out of 35 (48.5%) XV utterances and M produced 18 out of 34 (52.9%). The first VX utterances are produced in Sample 6. The XV production rate gradually decreases from Sample 9 to 11 (24% and 27.9% for B and M respectively). This is very similar to Haznedar’s study as Erdem uses VX order consistently from Sample 9 on. The headedness of the VP in the present study completely changes to head initial as in the English by Sample 14. There are no XV utterances produced in the whole study after Sample 14 except for the example given in (11).

(11) M: I likes sleepy story read

This is again similar to Haznedar as Erdem stops using XV utterances in Sample 12. From this sample on, Erdem uses XV order once in each sample from Sample 14 to 17 out of 244 utterances.  The similarity of these two studies with regard to shift of VP headedness could count as a robust finding that child L2 learners switch headedness around the same time. It should be mentioned that functional categories (IP in our case) are related to syntactic properties. Lack of copulas (except those produced as a chunk or as a result of repetition), and non-existence of T(ense) and Agr(eement) during these 14 samples argue for the absence of functional categories in children’s early grammars. In the next part the production of copula as inflectional categories will be discussed

Copula be production

Although some copulas emerge even before Sample 14 they are not used consistently and some of them are not target-like. SOV order is not observed in copular constructions. This may have two reasons. The first reason is that the majority of obligatory contexts for the copula lack the copula. The second reason may be that learners in the initial stages do not use non-finite copula forms. Finally early copula constructions may be readily available in the chunks memorized by the learners where the learners do not recognize these forms appearing in chunks as finite verbs. The following examples show how the learners copy the investigator by using the same word order used in questions for answering her:

(12) a. R. Is this a notebook?

           B: No, is this a book. (S 10)

       b. R: Whose bicycle is this?

          M: My bicycle is this. (S 10)

This is in line with what Myles (2004) claims regarding the predominant use of nonfinite forms in main clauses in the earliest stages of productive use. It is evident in the present study that B and M initially used pronominal subjects plus copula as chunks and if the subject, for example, is lexical rather than pronominal, the copula is missing as well. The few (two) VX utterances produced in Sample 4 are with auxiliary ‘is’ and can be analyzed as rote learning since the learners produced them right away after being given lots of sentences with –ing. The high percentage of XV utterances before and after these utterances further indicates their rote nature:

(13) a. B: She is open the door. (S 4)

       b. M: She is brushing her teeth. (S 4)

Under structure building, there is a direct relationship between non-raised verbs and lack of subjects in the utterances. Although there were few null subjects with thematic verbs in this study (only 41), 19 out of 28 (67.85%) bare VP produced up to Sample 14 include null subjects, as shown in examples in (14). This also emphasizes the correlation between the existence of non-finite verbs and utterances with null subjects:

(14)  a. M: Can football play. (S 4)

        b. B: Monday apple eat. (S 9)

        c. B: One sister have. (S 9)

As has been found by many researchers (V & Y-S, 1994, 1996a, b, Haznedar, 1997, among others) the fact that an L1 (Farsi in our case) has a subject-verb agreement paradigm does not give the learners the advantage in putting agreement markers at the end of the verbs at this stage; Verbs are not inflected with agreement markers. Under structure building/organic grammar, if copulas, auxiliaries and modals are base-generated in INFL, their production is unexpected at this stage.

Early production of copulas also shows the nature of early stages of L2 acquisition. Copula be is among the first verbs appearing in the earliest production of both subjects mostly in the form of It’s a…., It is a…… as in (15).

(15)  a. B: It is a table. (S 3)

        b. B: It’s a bed. (S 4)

        c. M: It is a butter. (S 4)

        d. B: It’s a flower. (S 5)

        e. M: It is a octopus. (S 5)

        f. M: It is a duck. (S 6)

Despite the high frequency of these two forms from Sample 3 on, a high percentage of copulas were non-target-like (inappropriate use, lacking consistent agreement with the subject), which may indicate the unanalyzed nature of early copulas. It seems that they have got some sort of communicative strategy through using it is (a) as they were more or less forced to produce English before they knew much:

(16) a. R: What is the doll doing?

           B: It is a cry. (S 3)

       b. R: What are these?

           B: It is a cars. (S 3)

       c. R: What is the lion doing?

          B: It is a food. (S 3)

      d. R: What time is it?

          B: It is a # clock. (S 4)

Figure 2: Percentage of copula

As mentioned before, the learners’ copulas included rote learned forms which did not show any IP projection. From Sample 10 to 14, the emergence of copulas with lexical subjects for B and M reaches 50% and 65% respectively but despite this high percentage, either the position of the copula is not target-like or there is lack of agreement. Following Vainikka & Young-Scholten, it is concluded that following the VP- stage, between Samples 10 and 14, learners have projected an underspecified IP-level functional projection as learners show evidence of properties representing IP emerging, i. e. copulas modals, auxiliaries and verb raising, but there is lack of overt agreement which suggests that this projection is different from IP.

(17)  a. B: Your trousers is green. (S 10)

        b. B: She book are they. (S 10)

        c. M: Mummy is shirt a yellow and green. (S 10)

        d. M: Daddy is trousers is black. (S 10)

        e. M: Mum hat is this. (S 12)

        f. B: Mum book is this. (S 12)

        g. M: No, not Ali and Abolfazl it’s the # short. (S 12)

        h. B: They apples is on the tree. (S 13)

        i. M: Monkeys is not a beautiful. (S 13)

        j. B: I is your son. (S 14)

What is more important regarding missing copulas with lexical subjects is that it casts doubt on the idea that any nominative subjects that are produced in the early stages are really positioned in the specifier of IP since if it were the case, there should have been a stable IP node to put the copula in even in the absence of nominative subjects. This may indicate that he is, they are and you are produced early on are all chunks rote-learned as a result of input overloaded to the learners. During Samples 15-18, the production of copulas with lexical subjects gradually increases and reaches 100% in Sample 18 for both learners. Looking at Figure 2, it can be noted that it is from Sample 18 on that the production of copula is almost always target-like and there is not a high fluctuation in the learners’ productions. This seems to show a high correlation between producing copulas with lexical subjects and producing copulas as a whole. The question here is why there is still a little fluctuation in learners’ production regarding this morpheme in some files of this study. Figure 2 shows this fluctuation and some of the non-target-like copulas are presented in (18):

(18) a. M: What colour his hat? (S 19)

       b. B: What they mum name? (S 19)

       c. B: What in your bag? (S 19)

       d. M: What in your bag? (S 19)

       e. B: He said, ‘You friend with me?’ (S 19)

       f. M: What time for cartoon? (T 19)

       g. B: Because the bee hive just for the bee. (S 22)

All the above sentences show that while copula is already productive in the grammar of both learners, they have problem to raise this to a higher projection to make questions or embedded clauses. Moreover, they have also problem with tense and it is not yet operative in their grammars either. The researcher noticed that copula is is used as a master key and is produced in many contexts mostly in non-target-like forms in the initial stages of L2 acquisition. Even when the learners have a good command of copula forms, there are still some oversuppliances of copula is. From Sample 12 up to Sample 17 where the children acquire the morphemes for the different pronominal subjects, the percentage of non-agreement increases and is stands for other copula forms as well. The rate of copula production in this study depends mostly on two factors. First, data collection method had an impact on the rate of copula production. The rate of copula production is higher in spontaneous speech production than in an elicited production task such as translation in this study. This shows that input received during the data collection method in spontaneous speech leads the learners to produce some cliché forms which will not exist when they produce the same structures by themselves. This is in line with Cox (2005), who claims different methods of data collection (elicited production vs. spontaneous and cross-sectional vs. longitudinal) is the source of discrepancy noticed in different studies. Second, the type of subject (pronominal versus non-pronominal) influences the rate of copula production. Copulas often followed pronominal subjects whereas non-pronominal subjects at the initial stages mostly lack copulas which again indicate that rote learning is involved. This provides strong counter-evidence for Haznedar (1997, 2001, 2003) and L & S (1994) who take only the presence of copula as evidence supporting the presence of IP (functional projections) in the initial state and strongly argues for V & Y-S (1994; 1996a, b) who claim that initial grammars lack the full complement of functional categories. Following V & Y-S the mere presence of copula indicates that learners project an underspecified IP-level functional projection since despite the presence of copula there is a lack of an agreement paradigm for that.

Similar to the present study, Ionin & Wexler (2002) also observed oversuppliance of is in their study of 20 L1 Russian children ranging in age from 3;9 to 13;10 acquiring L2 English. Some of the learners used forms of be in utterances containing an uninflected thematic verb. Above all, they concluded that the majority of utterances were not intended as progressive (19a) and include generic (b), stative (c), past (d) and future (e) meanings as well (see example 17). Wagner-Gough (1978), discusses data from a Farsi-speaking child (Homer) acquiring L2 English and observes that progressive -ing indicates 4 different time periods (see below in this section).

(19) a. The cats are pull mouse’s tail. (AN, 10; 1)

       b. They are help people when people in trouble. (DA, 1, 9; 7)

       c. He is want go up then. (GU, 3; 9)

      d. He is run away, I stayed there. (GU, 3; 9)

      e. I’m buy for my mother something. (AY, sample 2, 10; 4)

In line with the Minimal Trees (Vainikka & Young-Scholten, 1994, 1996a,b), Modulated Structure Building (Hawkins, 2001), Valueless Features hypotheses (Eubank, 1993/1994) and Myles (2004, 2005), arguing that syntactic features are not established until speakers show productive use of the related morphology in their utterances, where tense and agreement morphology appearing initially on verbs is just noise, the researcher noticed that both subjects produced lots of non-target-like forms and noise:

(20) a. R: Is this grass?

           B: No, is not a grass. (S 6)

       b. R: Where are the teddy bears?

          M: whe are two teddy bears. (S 6)

         ‘There are two teddy bears.’

       c. B: These a green (S 6)

         ‘These are green.’

       d. M: They I two daisy. (S 7)

          ‘They are two daisies.’

       e. R: Does she have a bag? (S 8)

           M: She is handbag. (have a bag = handbag)

Schwartz & Sprouse (1996) and Epstein et al. (1998) assume that any production of functional elements indicates that learners have full competence with respect to these functional categories. Grondin & White 1996; White 1996; Lardiere (1998a) also argue that if a morpheme is productively used in performance even in a small number of instances, it indicates that the underlying syntactic structure of the morpheme has emerged. Haznedar (1997, 2001, 2003) (see 2.8.3.1) takes suppliance rather than the target-like use of copula as the only reasoning for existing IP. She states that the majority of the copulas produced were is although she observed some am and are as well in her data. Hawkins (2001), on the other hand, states that the suppliance of morphemes on obligatory occasions does not necessarily mean that the learner has assigned the same interpretation to the morphemes as native speakers of the target language since the morpheme may also be used in a context where it should have not been used as the data in the present study show. Using different methods of data collection allows us to more confidently claim that early is does not represent that the functional categories have been projected in the learners’ grammars. Looking at the examples in (17), if the mere suppliance of the copula indicates the emergence of the underlying syntactic structure, the rate of production in Sample 3 for both subjects is 100%, but if the correct suppliance is the criterion, the production rate is 0%. To address this issue and to find the reason behind fluctuation in producing copula especially during the early samples, the researcher adopted two criteria. As mentioned in 3.3.1, there were also some data gathered through diary collection (D) and translation (interpretation) tasks in this study. Up to Sample 6, there were eight sentences collected in these two ways where there is an obligatory context for copula production. This is the first criterion to see if there is any difference in copula production as a result of different data collection methods.

Context: B. and M. were disputing and the R. was overhearing:

(21) a. B: Not funny. (D 4)

          ‘It is not funny.’

       b. B: M., now you bad. (D 4)

          ‘Now you are bad.’

        c. B: You bad don’t. (D 4)

           ‘You are not bad.’

        d. B: M. girl good. (D 4)

           ‘M. is a good girl.’

        e. M: B. boy good (D 4)

           ‘B. is a good boy.’

        f. B: Don’t funny. (D 4)

           ‘It’s not funny.

Context: Showing the learners some pictures and asking them to say something by giving them Farsi sentences:

     g. R.: Say this woman is not happy:

        M: Happy not. (T 5)

       ‘She is not happy.’

     h. R.: Say no, it is not a pen:

        M: No, pen not. (T 5)

       ‘No, it is not a pen.’

In all the above sentences in (21) copula is missing, whereas in (16-20) it has been produced, but in an inappropriate way. The researcher then compared copulas produced where the subject was a pronominal one versus those with lexical subjects to see whether this plays any role in subjects’ copula production. During Samples 6 to 9, the production of copula with pronominal subjects for B and M is 45.21% and 38.29% respectively whereas the rate of copula produced with lexical subjects during the same period for the subjects is 3.07% and 17.02%. Some examples are given in (22).

 (22) a. B: Book on the table. (S 6)

        b. M: Boat small. (S 6)

        c. B: They are four cats. (S 6)

        d. M: You are not girl. (S 6)

       e. B: Clown on the box. (S 7)

       f. M: Pen on the radio. (S 7)

       g. B: They are two dogs. (S 7)

       h. M: It is a blue. (S 7)

       i. B: Her hair yellow. (S 8)

       j. B: They are three bread. (S 8)

       k. M: Her skirt black. (S 8)

       l. M: They are pictures. (S 8)

       m. B: His face a red. (S 9)

       n. M: Honey on the table. (S 9)

       o. B: It is a blue. (S 9)

       p. M: No, it is a moon. (S 9)

Conclusion

Although the two subjects produce sentences with copula from the very early stages, the nature of these early emergences shows that they are not base-generated in INFL. Based on what mentioned above, the present study comes to the conclusion that lexical categories are influenced by L1 whereas functional categories are absent at the initial state and that they emerge without the learners’ reliance on their L1, consistent with Minimal Trees/Structure Building.