by Rod Ellis
Abrahamson identified an overall developmental profile in the three learners. Error frequencies were characterized by four phases: (1) an initial phase of relatively high error rates followed by a rapid decrease in error frequency; (2) a linear increase in error frequency; (3) a stable plateau of relatively high error frequencies; and (4) a final decrease in error rates. Thus—as has been observed in the acquisition of some grammatical features such as English irregular past tense—the learners manifested a U-shaped pattern of development. To explain this, Abrahamsson (2003) suggested that the learners initially focused on accurate production but then, as they paid more attention to fluency, their errors increased before decreasing when they had achieved sufficient fluency to pay attention to accuracy once againNOTE 6. The nature of the learners’ errors also changed over time. During the first year, deletion gradually gave way to epenthesis as a simplification strategy. In the second year, epenthesis also reduced. Abrahamsson proposed the following developmental sequence:
deletion > epenthesis > feature change (for example, devoicing) > target form
The acquisition of L2 phonology, then, is in many respects not dissimilar to the acquisition of L2 grammar. It is dynamic and complex, reflecting the variable impact of different factors (i.e. the learner’s first language, the inherent difficulty of specific phonological forms, exposure to L2 input, and the linguistic and situational context). These factors affect performance at different times and in different ways with different learners. Within this complexity, however, as with grammar, it is possible to identify a general learning trajectory.
Development of the L2 lexicon
Vocabulary development involves learning associations between lexemes (i.e. the phonological and graphological forms of words) and lemmas (i.e. the semantic-syntactic meanings of words). But it involves much more besides. For a start, as we have already noted, L2 learners acquire formulaic chunks (i.e. ‘lexical phrases’) as well as discrete words. Words and phrases are not stored as discrete items but in structured networks of semantic relations (Meara 2009). Individual words associate with other words paradigmatically (for example, ‘war’ associates with ‘guns’) and syntagmatically (for example, ‘war’ collocates with ‘declare’ in the phrase ‘declare war’). Thus, to investigate how the L2 lexicon develops it is necessary to consider both the breadth (i.e. how many words the learner knows) and the depth (i.e. how words are interconnected) of the learner’s lexicon and how both develop in the direction of the target-language lexicon.
There is now a large literature on L2 vocabulary learning (see, in particular, Nation 2001; Singleton 1999; Meara 2009). Here, I can only point to a number of generalizations that the research supports.
In general, receptive knowledge of individual words precedes productive knowledge. Meara (2009) argued that a word becomes available for productive use only when the learner has established connections with other words in the mental lexicon. In other words, productive ability is linked to depth of word learning.
Development involves not just knowing the meaning of a word and its connections to other words, but of being able to access this knowledge rapidly for both reception and production. Thus, learning vocabulary is not an instantaneous process and change is constant (Meara 2009). Also, words can fall in and out of prominence over time.
Words can be learned incidentally through exposure or intentionally, for example by memorizing lists of words. For incidental learning to take place learners need to be able to infer the meaning of a new word from context. Nation (2001) pointed out that this becomes easier the more words the learner knows. Learners need to know 95% of the words in a written text to successfully guess the meanings of the other five per cent. In the case of learning from oral input, however, learners can make use of the situational context as well as the linguistic context. Multiple exposures to a new word in a variety of different contexts are needed for incidental learning to take place.
In general, lexical units (words or formulaic sequences) that occur frequently in the input will be acquired earlier than those that occur less frequently.
N. Ellis (1997) argued that lexical acquisition does not just involve the learning of individual lexemes, but can also take place by segmenting formulaic sequences. For example, by segmenting the formula ‘I don’t know’, the learner discovers that ‘I don’t’ and ‘know’ constitute separate lexical units which can then be combined with other words in the L2 lexicon. The process can also work the other way round. Learners can construct an utterance from their knowledge of individual words and then store the utterance as a chunk. In this respect, the processes involved in vocabulary and grammar acquisition cannot be easily distinguished.
Lexical development can also push grammatical development, providing further evidence of the interplay between the lexical and grammatical systems. As Bell (2009) put it, ‘grammatical complexity can be fast forwarded by lexical formulae’ (p. 126) as when the acquisition of the formula ‘it seems that …’ prompts the use of embedded clauses.
Learners draw on their first language in various ways. Initially, the link between an L2 lexical form and meaning may be indirect via the equivalent first language lexical form (L2 form → L1 form → meaning). Later a direct connection may be made (L2 → meaning) (see Jiang 2000). Learners also make use of cognates (i.e. words that are formally the same or similar in their first language and the target language) and sometimes establish false cognates.
It is clear that vocabulary acquisition is not a simple linear process of adding new words to an existing lexicon, but also of adding ‘depth’ to existing words and incorporating new words into a network of form—meaning connections that grow more complex over time. This process is intimately connected with the development of grammar.
In contrast to research on L2 grammar learning, there have been few longitudinal studies of vocabulary development in either naturalistic or instructed settings. Yoshida’s (1978) longitudinal study of a young Japanese child’s acquisition of L2 English reported that nouns were learned before verbs—a characteristic of the pre-basic variety. Wode et al.’s (1992) longitudinal study of four German children pointed to a number of differences between first language and second language vocabulary acquisition. For example, whereas vocabulary growth is slow in first language acquisition up to the first fifty words and then rapidly accelerates, it was initially much more rapid, but then decelerated in these children. Both of these studies investigated learners in a naturalistic setting. Palmberg (1987) reported a longitudinal study of vocabulary growth in Swedish learners of L2 English in a classroom setting. Unsurprisingly, most of the words the learners were able to produce were traceable to the textbook vocabulary. Laufer (1998) examined the development of passive and active vocabulary in adult foreign language learners over one year, reporting that their passive knowledge progressed considerably, but their controlled active knowledge much less so, and their free active vocabulary not at all.
These studies mainly focused on development in terms of vocabulary size. Schmitt (1998) in a longitudinal study of four advanced learners of L2 English investigated changes in depth of knowledge. He focused on four components of the learners’ knowledge of 11 polysemous English words: form (i.e. spelling); association (i.e. the extent to which the learners’ word associations corresponded to those of native speakers); grammatical characteristics (i.e. word class and word derivations); and meaning. The main finding of this study was that there was no evidence that these types of word knowledge were acquired in a sequence. Rather, development occurred in the different areas unevenly, but concurrently.
L2 pragmatic development
Pragmatics is the study of how we put language to use in real-life contexts. That is, it is concerned with what we do with language in communication rather than what language is. It entails both how people use language to produce continuous discourse (for example, how they are able to stay on topic and manage the organization of conversations) and also how they perform specific speech
acts—such as requests and apologies—in ways that are socially appropriate. For reasons of space, I will only consider how learners acquire the ability to perform speech acts.
Becoming pragmatically competent in a language is the process of developing sociopragmatic and pragmalinguistic competence. Sociopragmatic competence refers to the ability to adjust the way you speak to take account of the extent of the imposition on the other person, how familiar you are with that person and the person’s social status. Thus, it concerns ‘the social conditions placed on language in use’ (Thomas 1983: 99). These conditions are cultural in nature and thus differences will be found in speakers of different languages. For example, Japanese people are likely to emphasize status differences over speaker familiarity, whereas Americans tend to do the opposite. Pragmalinguistic competence is knowledge of the linguistic forms required to encode specific sociopragmatic meanings. For example, making a polite request that places a high level of imposition on the addressee requires the use of complex request formulas such as ‘I wonder if it would be possible for you to …’.
I will focus here on pragmalinguistic L2 development. This is investigated by identifying the linguistic strategies that a learner utilizes at particular stages of development. Drawing on a number of longitudinal studies of L2 requests, Kasper and Rose (2002) suggested that development involves five stages. Table 4.4 describes the stages and provides examples of the pragmalinguistic strategies found in each. The examples are taken from two different sources—from my case study summarized earlier in this chapter and Achiba’s (2003) study of the acquisition of English requests by one seven-year-old Japanese girl living in Australia. What is interesting is that a very similar pattern of development was evident in the communicative speech of both the classroom and naturalistic learners.
Stage Description Examples
1 The pre-basic stage
Requests are context-dependent and lacking in verbs. ‘Sir, sir pencil.’
(L asking for his pencil back).
2 The formulaic stage
The learner relies on unanalysed formulaic sequences and imperatives. ‘Give me paper.’
(L asking for a piece of paper)
‘Can I have pen please?’
(L asking for a pen)
‘Have you got sellotape?’
(L requesting another L to give him the sellotape).
3 The unpacking stage
Formulaic sequences are analysed allowing for more productive use and the learner also relies less on very direct requesting strategies. ‘Can you pass on your paper?’ (L telling another L to pass over her paper for collection)
‘Can you make for me please?’ (L asking T to staple his papers)
4 The pragmatic expansion
The learners’ pragmalinguistic repertoire is extended. They use a greater variety of strategies and make more attempts to mitigate the force of a request using more complex syntax. ‘Could you go over there?’
(L asking T to move away) ‘Could you tell me how to play this?’
(L to an family adult)
‘I think you’d better put this …’ (L to an adult family friend)
5 The fine-tuning stage
Learners are able to fine-tune the force of their strengths to take account of participants, goals, and contexts. ‘Maybe we could paint it in this colour.’
(L to an adult family friend) ‘Don’t eat too much because you might get tummy ache.’
(L to another L)
‘Maybe you could make it if you want to.’
(L to another L)
Key
L = Learner; T = Teacher
Table 4.4 Stages in the pragmalinguistic development of requests (based on Kasper and Rose 2002)
A similar pattern of development is evident for other speech acts (for example, apologies and refusals). Three observations are in order. Not all learners reach the final, fine-tuning stage. The classroom learner in my study only reached stage four. There was no evidence of this learner modifying his requests to take account of different addressees, for example the teacher or other students. In contrast, Achiba’s learner manifested a higher level of development and had clearly entered stage five, although she was still somewhat limited in her ability to mitigate the force of her requests at the end of the study. Both studies covered a similar period of time so the difference may reflect the different contexts in which the data were collected (i.e. the classroom versus play situations involving other children and adults). In both cases, development was quite slow.
The second observation is that the pattern of development reflects progression through the learner varieties described by Klein and Perdue (1997). Thus, requests in stage one are representative of the pre-basic variety, those in stages two of the basic variety, while those in stages four and five the post-basic variety. This raises the important question of the relationship between pragmalinguistic development and grammatical development. Do learners acquire grammar and then put this to use to convey pragmatic meanings or do they acquire pragmalinguistic devices and then derive grammar from them, or do pragmalinguistic and grammatical development feed off each other?
Schmidt’s (1983) case study of Wes suggests that pragmatic and grammatical development proceed separately. Schmidt commented that Wes ‘developed considerable control of the formulaic language that acts as social grease in interaction’ (p. 154), but demonstrated little grammatical development. This study suggests that the early stage of acquisition is essentially pragmatic rather than grammatical. However, the development of more advanced stages of pragmalinguistic ability is only possible if learners have acquired the necessary grammatical resources. For example, Japanese university students’ failure to make use of bi-clausal request formulas (Takahashi 1996) is best explained by the fact that they had not yet acquired bi-clausal constructions. The setting may also determine whether pragmatic need drives the acquisition of grammar or vice versa. Bardovi-Harlig and Dörnyei (1998), for example, found that—whereas ESL learners in the USA identified pragmatic errors more consistently than grammatical errors—the reverse was the case for EFL learners in Hungary.
The third observation is that the performance of speech acts like requests and apologies is clearly dependent on the use of formulas. Most of the examples in Table 4.4 are formulaic routines with a single open slot (for example, ‘Give me …’ and ‘Can I have a …?). Thus—to a considerable extent—the development of pragmatic competence rests on learners acquiring an increasing range of formulaic expressions. However, as Bardovi-Harlig (2006) pointed out—and as noted earlier in this chapter—formulas are themselves developmental. Thus, pragmatic development depends not just on learning new formulas but on the analysis of formulas already acquired. She proposed that formulas emerge in stages and illustrates this with data that show how a very simple formula (‘yeah but’ used as a marker of disagreement) evolved over time from an initial stage consisting of ‘but ah …’, through an interim stage involving the use of ‘yeah so’ and ‘yeah no’ to the final stage where the target formula (‘yeah but’) finally appears. As formulas develop, so pragmalinguistic ability also develops.
Summing up
We have now considered a range of descriptive research about how learners’ grammatical, phonological, and lexical systems develop and also how their L2 pragmalinguistic abilities evolve. This research has used a variety of methods: (1) case studies of individual learners, (2) descriptions of the learner varieties that evolve over time, (3) obligatory occasion analysis, (4) emergence, (5) frequency analysis, and (6) usage-based analyses. These different approaches are based on different theoretical assumptions about how L2 acquisition proceeds. For example: obligatory occasion analysis is predicated on the assumption that acquisition entails conformity to target-language norms; frequency analysis on the assumption that acquisition is a stage-like process involving transitional constructions; emergence on the assumption that the appearance of a new form demonstrates its acquisition; and usage-based analyses view
acquisition as the unpacking of formulaic sequences.
These differences in research methods and their underlying assumptions make it difficult to arrive at a simple summing up of how L2 development takes place. However, I will attempt a number of generalizations about what the research shows about L2 development, reserving a consideration of how they can be explained for later chapters.
The focus of the research has been on incidental acquisition
The bulk of the research reviewed in this chapter has investigated the communicative speech of learners in naturalistic settings. In effect, then, it has examined how languages are learned incidentally and how implicit L2 knowledge develops. Of course, learners can also learn intentionally through studying a language formally and, as a result, develop metalinguistic understanding of L2 features (i.e. explicit knowledge). However, there are limits on learners’ ability to make use of their explicit knowledge in their communicative speech. Differences in the order of acquisition were only found when data were collected by means of tests or unspeeded writing tasks (Larsen-Freeman 1976), which allow time for learners to access their explicit knowledge. We have also seen that classroom learners manifest similar acquisitional tendencies to naturalistic learners when their communicative speech is examined (Ellis 1984; Eskildsen 2012).
L2 development has been investigated primarily in terms of production
The regularities in L2 development that we have observed are regularities in the acquisition of productive abilities. However, a full account of how learners acquire an second language needs to account for their ability to process and comprehend the meaning of different grammatical forms in the input they are exposed to as well as their ability to produce them. Learning a morphological feature such as English plural -s does not commence with the production of this form, but with attending to it in the input and understanding the meaning it conveys. By and large, receptive knowledge precedes productive knowledge for grammar, just as it does for vocabulary.