Understanding Second Language Acquisition (2nd ed)

by Rod Ellis

  Irregular morphology precedes regular morphology, as illustrated by the prior acquisition of irregular past tense forms in English. This reflects the likelihood that irregular forms are acquired as distinct lexical items (i.e. through item-based learning).

  When learners are acquiring compound verb tenses, such as the present/past progressive and the present/past perfect, they begin by using a verb with verbal suffix (for example, ‘eating’) and only subsequently produce verbs with auxiliaries (for example, ‘is eating’).

  Bardovi-Harlig also noted that many of the learners she investigated had not acquired the full L2 tense-aspect system.

  Sequence of acquisition

  Specific grammatical forms manifest their own sequence of development as learners gain gradual control over them and slowly fit them into their existing interlanguage system. Arguably, then, what is needed is an approach that acknowledges the fundamental organic nature of L2 development. In this section, we will explore research that has examined L2 negation—probably the most researched grammatical property. In so doing, I will illustrate the two different approaches I introduced in the opening section of the chapter—frequency analysis and usage-based analysis.

  Stages in the acquisition of L2 negation

  The acquisition of L2 negation has attracted the attention of SLA researchers from early days. Studies by Ravem (1968), Cancino et al. (1978), and Wode (1976) examined how negation develops in Norwegian, Spanish, and German learners of English. Subsequently, researchers investigated other languages. De Swart (2006), for example, provided a composite account of how negation developed in all the languages investigated in the ESF project. These studies all involved naturalistic learners, but there were also longitudinal studies of tutored learners (for example, my own case study of three classroom learners). The research points to two general conclusions: there is surprising uniformity in how learners with different L1s acquire negation and this uniformity can be described in terms of a sequence of stages that learners pass through en route to the target structure.

  In the following account of the different stages in the L2 acquisition of negation, I offer a composite picture derived from a number of studies. I will provide examples of typical learner utterances using both my own data and data from other sources. In these examples ‘L’ stands for ‘learner’ and ‘S’ for some other speaker, usually a native speaker.

  The earliest negative utterances typically consist of just a lexical particle (for example, ‘no’) and express denial, refusal, rejection, or correction:

  S Mariana is Spanish.

  L No, Portuguese.

  Early on, ‘no’ is combined with other non-verb elements (i.e. a noun or an adjective) with the negating particle appearing in either utterance-initial or final position:

  L Nee hier haus. (= ‘No here house.’)

  S Il y a des taxis. (= ‘There are some taxis.’).

  L Non, taxi non. (= ‘No, taxi no.’).

  These utterances correspond to the pre-basic variety. Verb negation does not appear until the basic variety. Here, the negative particle typically precedes the verb:

  L Mariana no coming. (= ‘Mariana is not coming.’)

  L Me no drawing in here. (= ‘I won’t draw here.’)

  L Ils non comprehend. (= ‘They don’t understand.’)

  As learners enter the negative + verb stage, learners become responsive to the negative particles that occur in different sentence types (i.e. English ‘no’, ‘not’, and ‘don’t’) and thus reduce their reliance on a single negative particle.

  L Not finished (‘He has not finished his work.’)

  L Nein spielen Katze (= ‘No play cat.’)

  L Don’t look my card.

  The last utterance seems target-like but, at this stage, ‘don’t’ is morphologically unanalysed (i.e. it functions as a negative particle in its own right). The prevalence of negative + verb—irrespective of whether the source language is pre-verb-negating (like Spanish) or post-verb negating (like German)—suggests that it constitutes a basic option.

  Post-verbal negation in L2 English typically appears first when learners begin to use modal verbs:

  L I can’t play this one.

  L I won’t go.

  However, as with ‘don’t’, it is likely that these negative forms are unanalysed initially. That is, they serve as alternatives as learners expand the lexical range of their negating words. Nevertheless, they constitute a stepping stone to the final stage which occurs when learners enter the post-basic variety when they are finally able to mark verbs for tense and aspect. In English, ‘not’ is now positioned between the auxiliary and main verb. At this stage the learner’s negatives become more or less target-like:

  L He doesn’t know anything.

  L He didn’t said it.

  L She didn’t believe me.

  In L2 German, the negating particle is positioned after the verb in accordance with the target language:

  L Ich falle nicht runter. (= I fall not down)

  Learners’ development of negation, therefore, closely mirrors their progression through the learner varieties. As we noted earlier, however, this progression is not smooth. The stages—like the varieties—are not clearly defined, but overlap considerably. Development does not consist of sudden jumps, but of the gradual replacement of early patterns by later ones and not all learners reach the final stage. The learners’ first language also has an effect (see Chapter 6). Similar stages of development have been observed in other grammatical structures—for example, in English interrogatives (Pienemann et al. 1988) and relative clauses (Schumann 1980; Mellow 2006). The acquisition of syntactical features—and also many morphological features—entails a stage-like progression towards the target structure.

  Formulaic sequences

  A characteristic of learners’ negative utterances, especially in the initial stages, is that they are often formulaic. However, in the studies of L2 negation we have just considered, formulaic chunks were discounted as the aim was to identify stages of acquisition in terms of learners’ creative utterances. While researchers recognized that negative particles such as ‘don’t’ were probably und units, they did not examine the possibility that such units were extracted from larger wholes such as ‘I don’t know’. We will now consider research that has looked at formulaic sequences in L2 acquisition and then consider how such sequences figure in the acquisition of negation.

  Wray (2000) defines a formulaic sequence as:

  ‘a sequence, continuous or discontinuous, of words or other meaning elements, which is, or appears to be, prefabricated; that is stored and retrieved whole from memory at the time of use, rather than being subject to generation or analysis by the language grammar’

  (Wray 2000: 465).

  Formulaic sequences are of two kinds—routines which are totally unanalysed units and learned as wholes (for example, ‘I don’t know’), and patterns, which consist of a chunk with one or more open slots (for example, ‘Can I have a …?’).

  Myles, Hooper, and Mitchell (1998) proposed a number of criteria for determining whether an utterance is formulaic—it is:

  at least two morphemes in length

  phonologically coherent (i.e. fluently articulated, non-hesitant)

  unrelated to productive patterns in the learner’s speech

  more complex in comparison with the learner’s other output

  used repeatedly in the same form

  possibly inappropriate (syntactically, semantically, or pragmatically) or otherwise idiosyncratic

  situationally dependent

  community-wide in use.

  Lin (2012) emphasized the second of these criteria. She proposed that formulaic expressions have distinctive prosodic qualities (for example, a fast speech rhythm). However, it is not easy to decide whether a given utterance is a fixed routine, a pattern, or a creative construction. Indeed, the usage-based account of negation we consider in the next section assumes that many learner utterances are indete
rminate in nature.

  Formulaic sequences are common in both adult and child L2 learners, especially in the early stages. Specific sequences emerge tied to the performance of specific language functions. For example, ‘Can I have …?’ is used to perform a request, ‘Why don’t we …?’ to perform a suggestion, and ‘I’m sorry’ an apology. Both naturalistic and classroom learners rapidly acquire a repertoire of formulas which they use to perform the communicative functions important for them.

  While there is wide acceptance that formulaic sequences play an important role in L2 use, there is less agreement about the role they play in L2 acquisition. Hakuta (1976) questioned whether routines and patterns facilitate or hinder the acquisition of grammar. Krashen and Scarcella (1978) argued that formulaic speech and rule-created speech are unrelated. The alternative position is that formulaic sequences constitute the building blocks for the construction of an L2 grammar. In a seminal article, Pawley and Syder (1983) argued that achieving native-like control of a language involves not only learning a rule system that will generate an infinite number of sentences, but also ‘memorized sequences’ and ‘lexicalized sentence stems’. They saw the language learner’s task as that of discovering precisely what permutations of a sentence stem are possible. It is this view of L2 learning that underlies the emergentist, usage-based theories of the kind promulgated by N. Ellis (2002) (see Chapter 1).

  There is now ample evidence that learners are able to unpack the parts that comprise a sequence and, in this way, bootstrap their way to more creative constructions. Wong-Fillmore’s (1976) longitudinal case study illustrated how this was done. For example, Nora, the fastest of the five child learners she investigated, used two formulas:

  I wanna play wi’ dese.

  I don’t wanna do dese.

  She then discovered that the constituents following ‘wanna’ were interchangeable:

  I don’t wanna play dese.

  I wanna do dese.

  Wong-Fillmore commented that this ‘formula-based analytical process … was repeated in case after case’ (p. 645). In my case study, I showed that the classroom learners I studied were able to extract the constituents from initial routines. For example, by analyzing the routines:

  I don’t know

  I don’t understand

  I don’t like

  the learners arrived at the construction ‘don’t’ + verb. Myles (2004) also argued that formulas constitute an important starting point. She found that those classroom learners who failed to acquire a set of formulas at the start showed very little subsequent development.

  Individual learners vary in the extent to which they deconstruct formulaic sequences. Schmidt’s (1983) case study of Wes showed that although Wes made advances in fluency by acquiring and using formulaic sequences, he made very little progress grammatically. Functionally-oriented learners like Wes may see no need to decompose chunks if their communicative needs are satisfied by their repertoire of formulaic sequences.

  A usage-based account of L2 negation

  Eskildsen (2012) investigated the acquisition of L2 negation in a longitudinal study of two adult classroom learners (Carlos and Valerio). He had two aims. One was to explore what a usage-based account of the learners’ utterances showed about their development of negation. The other was to investigate ‘the situated nature’ of these learners’ ‘emergent linguistic inventory’ by examining the specific classroom interactions in which their negative utterances occurred.

  Eskildsen showed how the learners’ development started with recurring expressions and then evolved toward ‘an increasingly, schematic, dynamic inventory of linguistic resources’ (p. 363). Carlos, for example, demonstrated early use of the formulaic sequence ‘I don’t know’ and then expanded this formula by employing an array of personal pronouns in the subject position. In so doing, he was able to overcome the entrenched ‘no’ + verb construction that also figured in his early utterances. Ultimately, he was also able to mark the auxiliary ‘do’ for tense. However, Eskildsen noted that Carlos’s development was not linear and did not manifest the relatively discrete stages identified in the earlier studies of negation discussed above. He concluded ‘the data … showed no acquisitional stage-defining dominance’ (p. 364). He also found differences in how Carlos and the other learner, Valerio, developed L2 negation.

  Eskildsen illustrated how the interactions the learners participated in contributed to the development of L2 negation. For example, the sudden increase in Valerio’s use of the subject + ‘no’ + verb pattern was traceable to a particular lesson where an exemplar of this pattern (‘you no write’) arose out of an interaction with another student and was used with the specific purpose of instructing the student about the task they had been given. Valerio went on to use this pattern eight more times on the same day.

  Eskildsen’s study is important for a number of reasons. First, it demonstrates the need to consider formulaic sequences when investigating developmental sequences. The ‘I don’t know’ sequence served as an exemplar that sparked the learning of a pattern. Second, it suggests that there is variation in the learning trajectories of different negative patterns (for example, main verb negation as opposed to copula negation) and also differences in learners. Third, the study illustrates that ‘interaction and learning … cannot be kept apart’ (p. 366).

  An emergentist view of L2 development casts doubt on the existence of distinct developmental sequences, such as those reported for negation. It sees such sequences as idealized abstractions derived from the researchers’ interpretations of learner language and their commitment to a view of interlanguage as a series of interlocking learner grammars. Instead, in an emergentist account development is seen as organic, dynamic, and variable with no clearly delineated stages. Development consists of a ‘continuum of schematicity’ ranging from concrete exemplars to ‘generalizable schematic constructions’. (Eskildsen 2012: 33).

  Development of other linguistics systems

  Much of the early work in SLA focused on grammatical development and, as we have seen, interest in this aspect has continued up to today. However, SLA researchers have also investigated how the learner’s phonological and lexical systems evolve. We will briefly examine some of the key findings.

  Acquisition of L2 phonology

  As with grammar, the acquisition of L2 phonology is a gradual, dynamic process but also displays some general tendencies:

  Learners draw on the phonological features of their first language when speaking in the L2.

  However, not all L1-target language phonological differences cause equal difficulty to learners. Some L2 features are ‘marked’ relative to the learner’s first language and so will be difficult to acquire. For example, English learners of German have no problem devoicing ‘d’ in word final position (i.e. they pronounce ‘und’ correctly as ‘unt’) whereas German learners of English frequently substitute /t/ for /d/ in this position (i.e. they say ‘bet’ instead of ‘bed’). The explanation according to Eckman (1977) is that the voiceless/voiced distinction in word final position in English is highly marked (i.e. it is only found in a few languages such as English). Thus, English learners of German have no problem with the unmarked devoicing in word final position whereas German learners of English find difficulty with the marked usage. See the discussion of markedness in Chapter 6.

  Similarities between the first language and the target language do not always benefit the learner (Flege 1987). Beginner learners may incorporate an L1 equivalent feature into their interlanguage but this may prevent them from establishing a completely new phonetic category for the target language and so slow down acquisition. In contrast, learners will need to set up new phonetic categories for those features they perceive as different from their first language and consequently progress more rapidly in acquiring them. In other words, dissimilar sounds can be easier to acquire than similar sounds.

  Learners also manifest unique phonological forms in their production. That is, L2 phonological
patterns that are independent of both the first language and the target language have been observed. In some cases, these patterns are the same as those found in L1 acquisition. Wode (1977), for instance, found that the German children he studied followed the same developmental sequence for /r/ as native English-speaking children.

  In general, learners’ ability to perceive sound contrasts that do not exist in their first language precedes their ability to produce the same contrastsNOTE 5. Over time, as a result of exposure to the target language, learners acquire productive ability although their production of the sound contrasts may not be totally native-like (Hayes-Harb and Matsuda 2008).

  To illustrate the dynamic and systematic nature of L2 phonological development we will consider Abrahamsson’s (2003) study of three Chinese learners’ acquisition of L2 Swedish syllable structure. This study, unlike many studies of L2 phonology, was longitudinal, covering a period of nearly two years. Syllables can be open or closed. An open syllable consists of an onset (one or more consonants) and a rhyme (a vowel) (for example, ‘slow’ in English). Closed syllables contain a coda (i.e. one or more final consonants) (for example, ‘quick’). Swedish is predominantly a closed-syllable language whereas Chinese typically has open syllables. L2 learners of languages like Swedish frequently resort to simplification strategies when attempting to produce closed syllables. They may delete the final syllable, add a vowel (a process known as epenthesis) or devoice the final consonant. These processes can occur in all learners irrespective of the syllable structure of their first language, but they are especially prevalent in learners with an open-syllable first language.