Understanding Second Language Acquisition (2nd ed)

by Rod Ellis

  A usage-based account of L2 development seeks to plot how learners gradually and dynamically move from chunks to constructions, but without claiming any fundamental distinction between the two. Thus, it avoids the idealized picture of L2 development inherent in the idea that a grammatical structure—such as negatives—involves a series of relatively distinct stages of development. It acknowledges the variability and non-linearity inherent in L2 development by plotting the messy ‘trajectory of learning’ that learners follow as they gradually unpack chunks and then repackage the parts into new chunks and constructions. It is compatible with Complexity Theory, which was introduced in Chapter 1 and is considered further in Chapter 8.

  Case studies of L2 learners

  The case studies I will now consider all involved the analysis of oral data collected from learners over a lengthy period of time (more than a year). For each case study, I will provide background information about the learner(s) and the methods of data collection, summarize the main findings, and point out the key issues the studies raise.

  Schmidt’s (1983) study of Wes

  Wes was a 33-year-old Japanese learner of L2 English who left school at the age of fifteen and thus had had very little experience of formal instruction. He was a successful artist. He divided his time between living in Hawaii and Japan, spending increasing amounts of time in the former. He mixed predominantly with English speakers in Hawaii and thus experienced very little social distance from native speakers of English. Data were collected over a three year period mainly by means of one-hour tape-recorded monologues where Wes commented on his business, his daily activities, and his visits back to Japan. Schmidt also made recordings of informal conversations between Wes and native speakers.

  The focus of Schmidt’s study was the extent to which Wes’s acculturation to American society could explain his development of communicative competence. Communicative competence was considered in terms of four components: (1) linguistic competence (i.e. the ability to use grammatical structures with target-like accuracy); (2) sociolinguistic competence (i.e. the ability to use language in socially appropriate ways); (3) discourse competence (i.e. the ability to participate in coherent and cohesive conversations); and (4) strategic competence (i.e. the ability to deal with communication breakdown) as in Canale (1983).

  Schmidt’s main finding was that development of these abilities proceeded separately. Wes’s linguistic competence remained quite limited. His pronunciation was good (especially his intonation), but his grammar hardly developed over the three-year period. Of the nine grammatical morphemes Schmidt investigated, only three reached the 90 per cent criterion level of accuracy. There was greater evidence of development in his sociolinguistic competence. For example, initially Wes’s directives relied extensively on formulaic expressions (for example, ‘Can I have a …?’), but by the end of the three-year period, gross errors in his use of directives had been eliminated and his English utterances were largely socially appropriate although sometimes idiosyncratic. The aspect that showed the greatest development was Wes’s discourse competence. Wes also manifested considerable strategic competence. For example, he was able to repair communication breakdowns by making effective use of communication strategies such as paraphrase (for example, his use of ‘money-girl’ for ‘prostitute’). However, he rarely bothered to repair his utterances when he received feedback and seemed to operate on the principle that it was the responsibility of native speakers to understand him rather than his responsibility to make himself understood. Overall, Wes proved to be an effective communicator but a poor learner in terms of linguistic development.

  Schmidt’s study was notable in two principal ways. First it showed the partial independence of grammatical competence from other aspects of communicative competence. Second, the lack of linguistic development could not be explained by Wes’s failure to acculturate as in fact he became socially very integrated when he lived in Hawaii. One possible explanation is that, as a functionally-oriented learner Wes paid little attention to the input he was exposed to and also developed little metalinguistic awareness of English grammar.

  R. Ellis’s (1984, 1992) study of two classroom learners

  The next case study is my own. I investigated classroom rather than naturalistic L2 learners. There were three learners, all children aged ten to 13 years, and all complete beginners at the start of the study. They were learning English in a language centre in an outer suburb of London. The ten-year-old was Portuguese while the other two (a brother and sister) came from Pakistan. I collected data over a two-year period by sitting in their English classes and noting down all the utterances they produced together with contextual information relating to the function and audience of their utterances. I focused on their communicative speech rather than the language they produced in formal practice activities.

  My initial purpose in conducting these case studies was to examine whether the pattern of development evident in these classroom learners was the same as or different from the pattern reported for naturalistic learners (for example, Cancino et al. 1978). In other words, I wanted to know whether the instructional setting influenced the way in which the children learned English.

  In the 1984 publication, I focused on the learners’ linguistic development. One of the main findings was that all three children made extensive use of formulaic sequences as a means of performing the communicative acts required of them in the classroom, where English served as the medium as well as the object of instruction. Over time, the learners were able to modify and extend these formulas. For example, for the ‘I don’t know’ formula, they substituted other verbs (for example, ‘I don’t understand’), changed the subject (for example, ‘You don’t know’) and added a constituent (for example, ‘I don’t know this one’). In other words, the learners were slowly unpacking the formulas, releasing their constituents for creative language use, as well as learning how to combine simple formulas into a more complex whole. Another feature of the three children’s language was ‘semantic simplification’ (i.e. they omitted constituents that perform semantic roles that would normally be encoded by a native speaker). For example, one of the learners produced the utterance ‘Sir, sir, pencil’ after the teacher had taken his pencil, meaning ‘You have taken my pencil’). Such utterances were readily understood because the missing constituents could easily be recovered with the help of contextual clues. Over time, such simplification diminished in the children’s speech.

  The study also investigated the sequence of acquisition for negatives and interrogatives. The developmental profiles for these structures of the three children were very similar to each other and showed a striking similarity to that reported for naturalistic learners. For example, they all began by producing verbless negatives such as ‘No pencil’ (i.e. ‘I don’t have a pencil’), before moving on to ‘no’ + verb negatives (for example, ‘No looking my card’) and then gradually introducing negatives with auxiliaries, first using ‘don’t (for example, ‘Don’t look my card’) and then a wider range of auxiliary forms. After two years, however, these learners could still not consistently produce target-like negatives, although the Portuguese boy was clearly more advanced than the two Pakistani learners.

  In the 1992 publication, I re-examined the same data to see what development took place in a pragmatic feature—which was requests. Over time, the range of request types expanded (for example, requests involving modal verbs such as ‘could’ began to appear) and requests that encoded the hearer’s perspective (for example, ‘Could you …’) as opposed to the speaker’s emerged. However, the learners’ requests continued to be of the direct kind, more complex types of requests did not occur at all, and the range of formal devices for encoding requests remained limited. There was also no evidence of the learners systematically modifying their choice of request strategy according to addressee (i.e. when they addressed the teacher or another student in the class).

  Three main conclusions emerged from my study. First, like Wes, t
hese learners relied initially on formulaic chunks to express their communicative needs and gradually learned how to manipulate the linguistic elements in these chunks to produce more varied, novel utterances. Second, the general pattern of development was very similar to that reported for naturalistic learners, suggesting that the classroom setting did not have a major effect on how these learners’ linguistic competence developed. Third, after two years, their development was still quite limited. They did not consistently produce target-like negatives and they still possessed only a limited range of requesting strategies. There are two mutually compatible explanations for these developmental limitations. The first is that the classroom setting did not afford the appropriate communicative conditions for acquisition. The other is that L2 development is inevitably a slow and gradual process and that full grammatical and pragmatic competence cannot be acquired even in a two-year period.

  Jia and Fuse’s (2007) study of Chinese ESL learners

  This was a five-year study that investigated the acquisition of a set of six English grammatical morphemes (regular and irregular past tense, third-person singular -s, verb + -ing, copula be, and auxiliary do). There were ten learners in this study—five girls and five boys who were aged between five and 16 years when they first arrived in the US. They all attended English-speaking schools where 70% of the students were native speakers of English. They all received focused ESL instruction for a period of time. Jia and Fuse divided the learners into two groups—six who were early arrivals and four who were late arrivals. The main aims were to investigate the acquisition trajectories of these features and whether there were any age-related differences.

  Data were collected from the learners by means of language tasks involving story telling prompted by pictures and interviews about the learners’ activities in their schools and at home and their language use in various situations. It took place regularly throughout the five years of the study. Obligatory occasions for the six morphemes were identified and the percentage correct calculated. A morpheme was considered to have been mastered if it achieved the 80% criterion level across three consecutive data collection sessions. Order of acquisition of the six morphemes was determined in two ways: by examining the number of learners who demonstrated mastery of each morpheme and by calculating the total percentage accuracy of each morpheme for all the learners.

  The main findings were as follows. The age of arrival of the learners had no effect on the order of acquisition. The same structures fell into the low-, medium-, and high-accuracy levels for both the early and late arrival groups. The two easiest structures were progressive -ing and auxiliary ‘do’ and the two most difficult were regular past tense and third-person -s. However, there were some age-related effects. For example, at the end of the study, the early arrivals demonstrated greater accuracy than the late arrivals on the two most difficult morphemes. The acquisition trajectories of the morphemes differed markedly. Progressive -ing and plural -s, for example, showed accelerated learning initially and then levelled off. In contrast, third-person -s was acquired slowly but steadily with no plateauing, while regular past tense showed no significant growth over time, but with fluctuations from one point in time to the next.

  Jia and Fuse discussed these results in terms of the factors that can account for the same order of acquisition manifested by the ten learners. They concluded that the most likely explanation lay in word frequency and salience; that is, learners learn those features that are more frequent and/or more salient earlier than those features that are less frequent and/or less salient. They also concluded that the results lent no support to the Critical Period Hypothesis as there was no evidence of any sharp discontinuity in the early and late arrivals’ acquisition of English (see Chapter 2). They explained the advantage noted for the early arrivals in terms of the richer learning environment they experienced.

  Some general observations

  These three case studies vary in the approach they adopted to investigating L2 development. Schmidt (1983) and Jia and Fuse (2007) used obligatory occasion analysis to determine the order of acquisition of grammatical morphemes. Ellis (1984, 1992) employed frequency analysis to identify the sequence of acquisition of grammatical structures. Ellis also investigated L2 pragmatic development and explored how his learners decomposed formulaic sequences.

  The studies suggest a number of generalizations about how an L2 develops over time:

  For some learners (for example, Wes), little grammatical development appears to take place. Such learners appear to be functionally-oriented and not motivated to acquire target-language norms.

  L2 development is uneven. For Wes, development was evident in the sociolinguistic and discourse aspects of the L2 but not in grammar. Jia and Fuse showed that different grammatical features followed different trajectories, some developing steadily over time and others accelerating rapidly to begin with and then plateauing.

  Grammatical development is not linear; there are notable fluctuations in the accuracy with which grammatical features are used from one time to another.

  Learners’ early attempts to use the L2 are characterized by structural and semantic simplification.

  The studies reported that the learners made extensive use of formulaic expressions to communicate and that these were prevalent in the early stages. My own studies documented how the learners gradually unpacked the components of the formulas they had learned, constructing patterns that were less fixed and could be deployed more creatively.

  Learners appear to acquire grammatical morphemes in a relatively fixed order irrespective of the age of the learners. Jia and Fuse suggested that morphemes that are salient and frequent are acquired earlier than those that are less salient and frequent.

  There is also evidence of sequences of acquisition; grammatical features—such as negatives—and pragmatic features—such as requests—are acquired gradually in observable stages. This was evident in both the studies that investigated naturalistic acquisition (Jia and Fuse 2007) and the study that investigated classroom learners (Ellis 1984).

  Considerable differences exist in the rate and success of acquisition by individual learners. Various factors can contribute to this—the learner’s first language, the learner’s age, and the richness of the learner’s learning environment.

  In the following sections, we will explore these generalizations in greater depth, beginning with research that has attempted to characterize the principles that underlie the development of learner varieties and then moving on to examine research that has focused on the acquisition of specific grammatical features.

  Learner varieties

  Dimroth (2012) defined a learner variety as ‘a coherent linguistic system produced by a language learner’ and emphasized that it is to be seen as a language variety in the same way as a dialect. In other words, a learner variety constitutes an interlanguage.

  The ‘Second Language Acquisition of Adult Immigrants’ project funded by the European Science Foundation (ESF) (Klein and Perdue 1997) involved a longitudinal study of the naturalistic acquisition of five European languages by 40 adult immigrants with different first languages, all of whom had only recently arrived in the target language country. Data consisted of free conversations and also oral production elicited by means of narrative and description tasks. The varieties were seen as manifesting particular form-function mappings and development as driven by the learners’ need to communicate more effectively by means of increasingly complex language.

  The initial variety is the pre-basic variety, which is characterized by nominal utterance organization (i.e. there were no verbs). At this stage, utterances are scaffolded (i.e. constructed over more than one turn) and context dependent. In time, this gives way to the basic variety. Utterances now include verbs, but these are non-finite (i.e. they are not inflected for tense or aspect). It constitutes a much more effective communicative tool than the pre-basic variety—in fact, it proved so effective for one third of the learners that they did not progress
to the post-basic variety when finite verbal utterance organization finally occurs. Table 4.1 provides a more detailed description of the linguistic features that characterize each varietyNOTE 3.

  Learner variety Linguistic features

  Pre-basic variety—nominal utterance organization Small vocabulary (around 50 words)

  Two types of utterance structure:

  1 NP + NP/adj/PP

  e.g. ‘girl hunger’; ‘Marie old’; ‘man in the street’.

  2 NP + affirmative/negative particle e.g. ‘car no’; ‘bicycle yes’.

  A few adverbs and participles.

  Basic variety—non-finite verb organization Extended lexical repertoire

  Three phrasal patterns:

  1 NP1 + V + (NP2) + (NP2)

  e.g. ‘he come’; ‘the man eat meal’; ‘the man give girl a present’.

  2 NP1 + copula + NP2/PP/adj e.g. ‘the man is doctor’; ‘he is in the house’; ‘he is tired’.

  3 V + NP2

  e.g. ‘finish book’.

  Verbs are not marked for tense or aspect.

  Grammatical categories such as ‘subject’ and ‘object’ do not exist.

  Rich repertoire of adverbs that can precede or follow all the patterns.

  Post-basic variety—finite verb organization Pronominal forms become productive.

  Finite verb forms appear to mark grammatical agreement, tense and aspect.

  Devices for encoding focus occur (e.g. ‘It was the movie John liked best.’).