by Rod Ellis
Table 4.1 Learner varieties
The ESF project found that development is similar for both learners of the same and for different second languages. All the learners showed the same progression through the learner varieties. However, the transition from one variety to the next is not sudden; rather, the utterance structure typical of one variety persists even when a learner has moved on to the next variety. Also, the source language of the learners was found to play a role. Klein and Perdue (1997) noted that when the target language offers alternative word orders (for example, Dutch and German allow both NP +V + NP and NP + NP + V word orders) the learners opted for the word order that corresponded most closely to their first language (for example, Turkish learners of these languages opted for verb-end in accordance with the word order in Turkish).
The project was directed at describing the path of development of adult, naturalistic learners. Klein and Perdue were careful to make no claims about instructed L2 acquisition. However, my own case study (Ellis 1984) suggests that some classroom learners may manifest similar development. In Table 4.2, I provide examples of the utterances that one of the learners I investigated (the Portuguese boy) produced inside an ESL classroom. These utterances are listed chronologically in the table. They show that he too seemed to proceed from the pre-basic, to the basic, and then to the post-basic variety in his communicative speech. It is possible, then, that irrespective of setting, learners follow a similar pattern of development when they attempt to use the L2 for communicative purposes.
Learner variety Utterances of an L2 classroom learner
Pre-basic variety ‘Me no ruler.’ (= I do not have a ruler.)
‘Phoc no good.’ (= Phoc is not a good boy.)
‘We no school.’ (= We don’t come to school on Monday.)
‘A door no downstairs.’ (= There is no door in the downstairs part of the house.)
Basic variety ‘Mariana no coming.’ (= Mariana is not coming to school today.)
‘Drawing the picture?’ (= Do I have to draw a picture?)
‘Here writing Friday?’ (= Do I have to write ‘Friday’ here?)
‘Playing now bingo?’ (= Are we going to play bingo now?)’
Post-basic variety ‘You did no read properly.’ (criticizing another student)
‘In this one the man is not shouting.’ (describing a picture)
‘This man can’t read because the light is green.’ (describing a picture)
Table 4.2 Examples of the learner varieties in the speech of an L2 classroom learner
What explanation is there for these learner varieties? Klein (1998) suggested that they reflect the nature of our innate capacity for learning a language. He argued that the basic variety cannot be explained by the surface properties of the input that the learners are exposed to and proposed instead that the constraints that govern it are part of our genetic endowment. This amounts to a nativist account of L2 acquisition, similar in some respects to that proposed by Chomsky (1986) for L1 acquisition. However, as we saw in Chapter 1, the claim that the human capacity for language is biologically specified and distinct from other cognitive abilities is highly contentious. N. Ellis (2002), for example, argued that both L1 and L2 acquisition are driven by input frequency and general cognitive (rather than language-specific) strategies.
The research on learner varieties provides a general picture of how L2 development in naturalistic learners takes place. It was limited, however, in that it only looked at the relatively early stages of development and did not examine how specific grammatical features are acquired.
Order of acquisition
Order of acquisition is determined by investigating either the accuracy of use of specific grammatical features by means of obligatory occasion analysis or the order of emergence of different features in learners’ production. The studies that have investigated order of acquisition have been both cross sectional and longitudinal. The early studies conducted during the 1970s (and some later ones, such as Jia and Fuse 2007) examined a miscellaneous set of English morphemes (for example, verb + -ing, plural -s, articles, and regular and irregular past tense). Later studies focused on specific grammatical systems, such as verb tenses.
The morpheme studies
The L2 morpheme studies borrowed the methodology used to study how children acquire the grammar of their first language. These studies showed that L1 acquisition involved a relatively clearly defined order of acquisition. Brown (1973), for example, reported a longitudinal study of three children learning English as their mother tongue. He showed that grammatical morphemes were mastered by all three children in the same fixed order. De Villiers and de Villiers (1973) found a very similar order for the same morphemes in a cross-sectional study of a large number of children. They examined the accuracy with which the children used the different morphemes at one time and then claimed that the accuracy order they found reflected the order of acquisition.
The key finding of the early L2 cross-sectional morpheme studies was that learners demonstrated a very similar accuracy order, irrespective of whether they were children or adults and irrespective of their first language. Dulay and Burt (1973, 1974) investigated Spanish and Chinese children learning L2 English. They found that the accuracy order for a mixed group of English morphemes was the same for both groups of learners. Bailey, Madden, and Krashen (1974) carried out a similar study on adults and reported that the order they found correlated significantly with that in Dulay and Burt. On the assumption that accuracy order reflects the acquisition order, it was proposed that there was a natural order of acquisition which all learners followed.
The existence of this ‘natural order’ has assumed an almost mythical status in SLA. It provided empirical support for Corder’s (1967) claim that L2 learners, like children acquiring their first language, have a ‘built-in syllabus’. There is, however, considerable evidence to suggest that the order is not as fixed as it was once assumed. Larsen-Freeman (1976), for example, found that different elicitation tasks produced somewhat different accuracy orders. In particular, she noted that some morphemes (for example, plural -s and third-person -s) were used more accurately in writing than in speech. Pica’s (1983) study of naturalistic, instructed, and mixed learners of English found that although the accuracy orders in all three groups of learners were the same, there were differences among the groups in specific morphemes. The instructed group used plural -s more accurately than the naturalistic group, while the naturalistic group was more accurate than the instructed group in using verb + -ing, suggesting that the linguistic environment had some influence on how and when these features were acquired. Also, longitudinal studies of the same set of morphemes failed to find the same acquisition order as the accuracy order reported in the cross-sectional studies. Jia and Fuse’s (2007) study did provide some support for the natural order. However, Hakuta’s (1976) two-year longitudinal study of a five-year-old Japanese girl indicated that plural -s, which was ‘acquired’ early according to Dulay and Burt’s accuracy order, was actually acquired much later. Hakuta suggested that this was because this morpheme had no equivalent in the learner’s first language. Shin and Milroy’s (1999) study of young Korean-American children’s acquisition also suggested that the learners’ first language influenced the order of acquisition. Thus, to some extent, both the learning environment and the learners’ first language influence the order of acquisition.
Nevertheless, it is clear that some English grammatical morphemes are inherently more difficult for L2 learners to acquire than others. Larsen-Freeman and Long (1991) argued that ‘there are … too many studies … showing sufficiently consistent general findings for the commonalities to be ignored’ (p. 92). The question that arises, then, is how these commonalities can be explained. Goldschneider and DeKeyser (2001) examined some 20 morpheme studies in order to investigate whether a number of different factors could account for the order. They were able to show that the phonological salience of a morpheme, its syntactic category, and its frequency
in input all had an influence on when a particular morpheme was acquired. They concluded that there was a single general factor that could explain the order of acquisition—’salience’. In other words, learners first learn those morphemes whose meanings are transparent and whose form is readily discernible in the input. For example, verb + -ing is more easily perceived in the input than third-person -s and is acquired earlier. Plural -s conveys a specific meaning (‘more than one’) and is acquired before third-person -s which is redundant.
Appealing to saliency as the explanation can also help to explain the differences in the morpheme order that some studies have reported. For example, it is reasonable to suppose that the learning environment and the learner’s first language will have some impact on which features are salient to learners. Verb + -ing may be overused by classroom learners because of its high frequency in language instruction. Plural -s may lack saliency to Japanese learners because there is no equivalent feature in their first language.
There is, however, a more fundamental problem with the morpheme studies. They compare apples and oranges. Noun morphemes (for example, plural -s and possessive -s) constitute very different kinds of learning problems to verb morphemes (for example, irregular and regular past tense). Also, even those morphemes that belong to same sub-system (for example, verbs) may not be learned in the same way. Pinker (1999) claimed that irregular and regular morphemes involve different types of learning—item-based versus rule-based. Learners learn and store English irregular past tense forms as items, but learn a rule for the regular past tense. Kempe, Brooks, and Kharkhurin (2010) proposed that rule-based learning can override item-based learning when the input cues for the former are transparent and salient. This can lead to overgeneralization of a rule. A good example of this is when learners extend the rule for English regular past tense to irregular verbs. This results in the U-shaped pattern of development that has been observed for irregular past tense—i.e. learners initially say ‘ate’ only to later say ‘eated’ before reverting once more to the target language form. In fact, learners’ use of grammatical morphemes manifests enormous variability from one time to the next (Huebner 1983).
Much of the later order-of-acquisition research avoided the rag-bag approach of the morpheme studies. It investigated particular grammatical sub-systems and also adopted a different way of determining whether a grammatical feature had been acquired—examining ‘emergence’ instead of ‘accuracy’. We will briefly consider two examples of this research, one of which looked at word-order rules and the other at verb tenses.
Order of acquisition of German word-order rules
Whereas the morpheme studies were carried out by US-based researchers, the study of the acquisition of word-order rules originated in the Zweitspracherwerb Italienischer und Spanischer Arbeiter (ZISA) project conducted by a group of German researchers (for example, Meisel, Clahsen, and Pienemann 1981). Initially, the research focused on German word-order rules, but was subsequently extended to the study of English word order in Johnston and Pienemann (1986) and other languages (Pienemann and Kessler 2011). It investigated learners with a variety of L1s and consisted of both cross-sectional and longitudinal studies.
Pienemann, Johnston, and Brindley (1988) claimed that the order of acquisition of German word-order rules shown in Table 4.3 is ‘probably one of the most robust empirical findings in SLA research’ (p. 222). A clear order emerged from the various studies. Learners began with a canonical word order (which varied according to the basic word order of the learner’s L1) and then progressed through a series of other rules before arriving at the last acquired verb-end rule.
Stage Name Description Examples
1 Canonical order Romance learners begin with SVO as their initial hypothesis about German word order. Adverbials appear in sentence-final position ‘Die Kinder spielen mit Ball.’ (= The children play with the ball.)
2 Adverb preposing Learners able to move an adverbial to sentence-initial position, but do not yet invert subject and verb as required when a sentence begins with an adverbial. This is not acquired until stage 4. ‘Da Kinder spielen.’ (= There children play.)
3 Verb separation Learners move non-finite verbal elements into the required clause-final position. ‘Alle Kinder muss die Pause machen.’ (= All the children must the pause make.)
4 Inversion Learners learn that in certain contexts such as sentence-initial adverbials and interrogatives, the verb must precede the subject. ‘Dann hat sie wieder die Knocht gebringt.’ (= Then has she again the bone bringed.)
5 Verb end Learners learn that the finite verb in subordinate clauses goes in clause-final position. ‘Er sagte dass er nach Hause kommt.’ (= He said that he to home comes.)
Table 4.3 Sequence of acquisition of German word-order rules (based on Pienemann, Johnston, and Brindley 1988)
Initially, learners operate in accordance with a fixed word order which functions as a default processing strategy. Over time, they develop the ability to perform a series of increasingly complex processing operations that involve, in the first place, just adding a linguistic unit to the beginning or end of an utterance (as in adverb pre-posing) and then, incrementally, moving linguistic units from within an utterance to the beginning or end (as in verb separation), re-ordering the internal structure of linguistic units (as in inversion), and finally re-organizing the basic word order for subordinate constructions (as in verb end). These operations constitute ways of handling different word orders (for example, the question ‘Is he coming?’ involves a re-arrangement of elements from the default ‘subject + verb’ word order).
This account of L2 development was later incorporated into Processability Theory (Pienemann 1998). This claims that learners have to ‘create language specific routines’ (Pienemann 2011: 33)—which they do in stages—gradually extending their ‘grammatical memory store’ to enable them to handle the increasing complex ways in which grammatical information is exchanged across constituents. The details of this theory are presented in Chapter 8.
Although Pienemann considered the language-specific routines that underlie L2 acquisition to be universal, he also acknowledged that individual variation occurs. First, the emergence of a new structure may or may not guarantee its systematic use. Some learners continue to use it alongside an earlier acquired structure. For example, when learners of L2 English reach the inversion stage—and thus are capable of producing sentences like ‘Have you seen him?’—they do not always do so consistently, sometimes producing forms belonging to an earlier period (for example, ‘Have see him?’ and ‘You have seen him?’). Learners vary in how they apply a new processing operation. Some learners appear to move slowly, consolidating each new stage before they move on to the next, whereas others move on to a new stage very rapidly, not bothering if they have not achieved a high level of accuracy in the structures belonging to the prior stage. Other learners take time to acquire the rule in all its contexts before they progress to a new stage: many of the learners in the ZISA project failed to reach stages four and five. Finally, not all grammatical features are governed by these processing operations. Some features, such as copulas (for example, the verb ‘be’) appear to be ‘variational’. That is, they are not dependent on the processing routines that govern developmental features and thus can be acquired at any timeNOTE 4.
Acquisition of the tense-aspect system
Other researchers (for example, Dietrich, Klein, and Noyau 1995; Bardovi-Harlig 2000) have focused more narrowly on the emergence of verbal morphology. They adopted a different approach, focusing not just on the order in which verb forms emerged, but also on form-function mappings. The investigation of a number of different languages made it possible to arrive at cross-linguistic generalizations about the order of acquisition.
Dietrich et al.’s (1995) study was part of the ESF project we considered earlier. They investigated 23 learners of Dutch, French, German, Swedish, and English, identifying three broad stages in learners’ expression of past time:
/>
In the pre-basic variety, the learners relied entirely on pragmatic means. They produced utterances that contained no explicit marker of past time, but could be interpreted as such in the context of a previously produced utterance. They also relied on the situational context to make meanings clear. Other pragmatic strategies involved the use of contrasting events and chronological order. At this stage, in effect, the learners made no attempt to encode pastness linguistically.
In the basic variety, the learners used various lexical expressions (for example, locative adverbials such as ‘in the morning’ and ‘yesterday’ and connectives such as ‘and’ and ‘then’) to express pastness. Verbs were used either in the base form (for example, ‘go’ in English) or some idiosyncratic form (for example, ‘going’).
Finally, in the post-basic variety, morphological markers of pastness appeared. These were used non-systematically as the learners continued to also rely on pragmatic and lexical means to convey pastness. Over time, however, past-tense morphology stabilized with a corresponding decrease in the use of adverbials.
Bardovi-Harlig (2000) reported the following order for English verb tenses in her longitudinal study of 16 learners of L2 English from four different language backgrounds:
past > past progressive > present perfect > past perfect
Drawing together the results of a range of studies that had investigated the acquisition of verb morphology, she identified four general principles that govern the acquisition of verb morphology:
Acquisition is slow and gradual.
Form often precedes function. That is, when a given morpheme first appears it is overgeneralized and thus lacks a clear contrast with existing forms. Base forms tend to be used alongside other forms even at later stages of acquisition.