by Rod Ellis
NS: Um, hmm, una taza roja.
(Leeman 2003)
implicit;
input-providing
Repetition an utterance that repeats the learner’s erroneous utterance highlighting the error S: Le … le girafe? T: Le girafe? (Lyster and Ranta 1997) implicit;
output-prompting
Metalinguistic feedback an utterance that provides comments, information, or questions related to the well-formedness of the learner’s utterance S: Euhm, le, le éléphant, Le éléphant gronde.
T: Est-ce-qu’on dit le éléphant?
explicit;
output-prompting
Elicitation a question aimed at eliciting the correct form after a learner has produced an erroneous utterance S: The chien peut court. T: The chien peut court? Le chien peut …
(Lyster and Ranta 1997)
explicit;
output-prompting
Explicit correction an utterance that provides the learner with the correct form while at the same time indicating an error was committed S: La note pour le shot. T: Oh, pour la, oh, pour ça. Tu veux dire pour la piqûre. Oui? (Lyster and Ranta 1997) explicit;
output-prompting
Table 7.1 Strategies used in the negotiation of meaning and form (from Ellis 2008)
The Input and Noticing Hypotheses
We turn now to consider what kind of input is needed for acquisition to take place. We will start by examining two hypotheses that have been very influential in SLA.
Krashen’s (1985) Input Hypothesis claims that ‘if input is understood and there is enough of it, the necessary grammar is automatically provided’ (p. 2). Krashen considered comprehensible input the essential environmental ingredient. He argued that simplified registers are beneficial for acquisition because they constitute one of the main ways of making input comprehensible. However, as Sharwood-Smith (1986) pointed out, there are two ways of processing input—for comprehension, and for acquisition. If learners use top-down strategies to process input (for example, by inferring the meaning from the situational or linguistic context) they will be able to comprehend without needing to process the input linguistically. For example, if learners hear this sentence:
Tomoko bought a new smart phone yesterday.
they can work out that the action occurred in the past without processing ‘bought’. Linguistic-processing is important for acquisition.
The importance of conscious attention to linguistic forms in the input has been demonstrated most clearly in the work of Schmidt (1994, 2001). His Noticing Hypothesis claims that ‘people learn about the things they attend to and do not learn much from the things they do not attend to’ (2001: 30). We will take a closer look at the Noticing Hypothesis in Chapter 8, but we will draw extensively on the general idea of ‘noticing’ in this chapter.
Pre-modified input and noticing
Given the importance of ‘noticing’, the key question becomes ‘How can learners be helped to notice linguistic forms in the input?’ One way is by pre-modifying the input to draw learners’ attention to specific linguistic features. This can be achieved by input-enhancement either through (1) ‘flooding’ the input with exemplars of a specific feature, or (2) highlighting a specific form through intonation in oral input or by putting it in bold type, in italics, or paraphrasing in written input. In both cases, the aim is to induce selective attention to pre-determined linguistic forms.
Several studies have investigated the effect of input enhancement on noticing. A good example is Shook (1999). In this study, learners of L2 Spanish were exposed to written input under three conditions: (1) textually enhanced, (2) textually enhanced plus explicit instruction, and (3) textually unenhanced. The highlighted grammatical features were the Spanish present perfect verb tense and relative pronouns. The learners were then asked to write down what they could recall from reading the passages. The textually unenhanced input (3) resulted in more idea units being reported suggesting better comprehension, whilst the textually enhanced model (1) led to greater noticing of the grammatical features, especially the more-meaningful present perfect. This study therefore suggests a trade-off between comprehension and noticing: if learners focus on the top-down processing required for effective comprehension, less noticing occurs and the opposite is true if learners engage in bottom-up processing and attend to the enhanced items in the text.
Not all enhanced grammatical forms are equally salient to learners. Some forms are more likely to be attended to than others. Learners are likely to pay greater attention to those enhanced linguistic features whose meanings are more transparent. Also, learners are more likely to notice forms that they have already partially acquired than those that are entirely new to them.
Overall, the research shows that text enhancement does facilitate noticing, but it has only a limited effect (Lee and Huang 2008). To date, it is not possible to comment on the relative effectiveness of different kinds of input enhancement (for example, input flooding versus highlighting). As Han, Park, and Combs (2008: 600) noted, ‘there are numerous methodological idiosyncrasies characterizing the individual studies’—such as the number of times a specific feature was highlighted, the number of texts involved, and whether learners received explicit instruction on the targeted feature(s) prior to exposure to the input.
Interactionally-modified input and noticing
Interactionally modified input also provides opportunities for noticing—arguably in ways more likely to be successful. A number of studies have used a research technique called stimulated recall to investigate whether noticing occurs as a result of the negotiation of meaning or form. This technique involves replaying extracts of interactions involving negotiation and asking learners to comment on their perceptions of what took place.
Egi (2007) is a good example of this kind of study. She investigated whether learners paid attention to specific linguistic forms in the recasts they received during negotiation. She distinguished (1) whether they showed awareness that an error had been made, but no awareness of the target-like form in the recast; (2) awareness of the target-like model, but no awareness that their original utterance was problematic; and (3) awareness of an error, and also that the target-like model had been provided. No noticing was reported most of the time for morphosyntactical features. However, Egi reported that 18.7 per cent of the learners’ comments demonstrated awareness in terms of (2) and a further 26.05 per cent awareness in terms of (3).
This and other studies suggest that interaction induces noticing but variably so. Morphosyntactical features are often not attended to probably because learners are primarily focused on meaning and thus engaged in top-down rather than bottom-up processing of the input. The extent to which noticing occurs also depends on the nature of the negotiation (for example, whether a recast is implicit or more explicit) and whether the learner is developmentally ready to notice a feature. However, one thing is clear: noticing occurs more frequently in interactionally modified input than in interaction where there is no negotiation.
Pre-modified input and acquisition
Noticing may be necessary for acquisition, but it does not guarantee it. Thus, it is important to ask whether the noticing of specific linguistic forms in the input leads to their acquisition. We will consider two ways in which researchers have investigated the effects of pre-modified input on acquisition—by simplifying the input to make it comprehensible, and by input enhancement.
Comprehensible input and acquisition
In line with the Input Hypothesis, Krashen (1985) sought to demonstrate that acquisition takes place when the input is comprehensible. One way in which this can occur is through extensive reading involving graded readers. Rodrigo, Krashen, and Gibbons (2004) reported that a group of learners—taught by a Reading Method consisting of an extensive reading programme of graded books—and another group—taught by a Reading-Discussion Method where students read graded books and then participated in debates and discussions about them—outperformed a third group taught by a traditio
nal Grammar and Composition Method in both vocabulary and grammar learning. There is now clear evidence that massive exposure to comprehensible input through extensive reading is beneficial for acquisition, especially where vocabulary is concerned. However, there are limits to how much can be acquired in this way. Low-ability readers may learn very few words incidentally.
Text-enhancement and acquisition
Comprehensible input alone may not suffice for acquisition to take place unless learners also pay conscious attention to the words and grammatical features in the input—as claimed by the Noticing Hypothesis. Perhaps, then, just as enhanced input can induce noticing, it can also facilitate acquisition.
Lee and Huang (2008) examined 20 studies that had investigated the effects of typographical enhancement and input flooding on L2 acquisition. They reported an overall positive effect in tests administered shortly after the learners had completed reading the enhanced texts, but acknowledged that this was quite small. There were considerable differences in the results of individual studies. For example, in some of the studies there was no effect at all. Lee and Huang also reported that the benefits of the exposure wore off over time.
In the case of grammar acquisition, text enhancement is limited in another way. While it may help learners to see what is grammatically possible in the target language, it may not help them to eradicate an erroneous rule, especially if this rule corresponds to an L1 rule. In an interesting study, Trahey and White (1993) investigated the effects of input-flooding on French-speaking learners’ acquisition of adverb placement in English. English permits adverb placement between the subject and the verb—for example, ‘Mary hurriedly hid her book’—but French does not; while English does not permit placement between the verb and object—for example, *Mary hid hurriedly her book’—and French does. Exposure to input containing adverbs was extensive in this study: one hour a day for ten days. The learners succeeded in learning the grammatical position for adverbs, but failed to ‘unlearn’ the ungrammatical position.
Sometimes text enhancement can lead to overuse of the target forms. Han et al. (2008) suggested that this may be due to over-enhancement—i.e. a combination of typographical enhancement and flooding. They cite studies where overuse occurred and where in each case there was double-enhancement. Han et al. concluded that for text enhancement to be effective there needs to be the right balance between frequency and saliency of target forms.
Overall, then, the evidence is quite mixed. Clearly, input enhancement does not always have an effect. As Lyddon (2011) pointed out, ‘even the most deliberate attempts to modify a stimulus are no guarantee of its perception’ (p. 116) and—even if noticing does occur—acquisition may not. At best, input enhancement only increases the likelihood of acquisition and—in the case of over-enhancement—it can have a deleterious effect.
Interactionally-modified input and acquisition
Studies that have investigated the effects of interactionally-modified input on acquisition have drawn on Long’s (1983b, 1996) Interaction Hypothesis. I will begin by explaining this hypothesis and then move on to consider studies based on it.
The Interaction Hypothesis (IH)
There is an early and later version of the IH. In the early version, Long (1983) drew on the Input Hypothesis to suggest that the negotiation of meaning assists acquisition by making input comprehensible. Long (1983) was in agreement with Krashen (1981) that simplified input and context help to make input comprehensible. However, he argued that interactive input is more important because it supplies learners with information relating to those linguistic forms that are problematic for them. In the later version, Long (1996) incorporated Schmidt’s (1994) views about the importance of noticing to claim that ‘negotiation of meaning, and especially negotiation work that triggers interactional adjustments by the native speaker or more competent interlocutor, facilitates acquisition because it connects input, internal learner capacities, particularly selective attention, and output in productive ways’ (pp. 451–52.). However, Long only claimed that negotiation facilitates acquisition—i.e. it makes it possible—not that it guarantees it. Also, the IH does not claim that meaning negotiation is the only type of interaction that fosters learning; ‘uninterrupted communication’—i.e. communication where there is no problem of understanding—can also contribute to acquisition.
The later version of the IH afforded a much richer view of how negotiation can assist language learning. It provides learners with positive evidence (i.e. ‘models of what is grammatical and acceptable’); negative evidence (i.e. ‘direct or indirect evidence of what is ungrammatical’ (Long 1996: 413)); and opportunities for modified output. Pica (1992) suggested a fourth way. She proposed that interactional modifications can help learners to see how input is composed of constituent parts. For example, it can show learners how what they first perceive as a chunk can be broken down into separate words. Closely associated with the updated IH are Long’s (1991) views about focus on form. As explained earlier, this refers to the incidental attention to linguistic form that arises when learners are primarily focused on meaning. Long viewed the negotiation of meaning as one of the main ways of achieving a focus on form and fostering acquisition. However, other researchers (e.g. Aston 1986) have questioned whether the negotiation of meaning does in fact play a major role in L2 acquisition, pointing out that learners may be resistant to nit-picking negotiation.
We will now turn to examine whether the research supports Long’s claims about the importance of negotiation and focus on form. I will first focus on two of the benefits that Long proposed for negotiation—comprehensible input and the provision of positive evidence—reserving consideration of the third—modified output—till later when I discuss the role of output.
Interactionally modified input and comprehension
Drawing on the earlier version of the IH, a number of studies investigated whether meaning negotiation assists comprehension by making input comprehensible. These studies compared the effects of three types of input on comprehension—baseline (unmodified) input, pre-modified input, and interactionally modified input.
Pica, Young, and Doughty (1987) used a task that required learners to listen to directions for choosing and placing objects on a small board illustrating an outdoor scene. The extent to which the participants were able to locate the objects correctly provided a measure of their comprehension. An analysis of the input in the three conditions showed that the interactionally-modified directions were longer and more redundant than the baseline directions with the premodified directions intermediate. The main finding was that interactionally modified input resulted in better comprehension than pre-modified input, which in turn was comprehended better than the baseline input. Other studies (for example, Loschky 1994) also found that interactionally-modified input leads to superior comprehension. These studies all involved adult learners. Young children, however, may be less ready to engage in negotiation of meaning although when they do so, their comprehension also benefits (Ellis and Heimbach 1997).
Overall, then, there is clear evidence that negotiation results in more input and more comprehensible input than unmodified input. However, it is not entirely clear whether it is the modifications themselves that facilitate comprehension or the additional time for processing input that results when negotiation occurs. Ellis and He (1999) found that when the amount of time allocated to the premodified and interactionally modified input was the same, no advantage for negotiation was seen.
Negotiated interaction and L2 acquisition
As I noted earlier in the chapter, the processes involved in acquiring from input are not the same as those involved in comprehending input. Thus it does not follow that interactionally-modified input that assists comprehension also facilitates acquisition. Increasingly, then, researchers have turned their attention to examining whether and how interaction facilitates acquisition. This vein of research has proved one of the richest in SLA to date. Mackey (2007) identified more than 40 published
studies that have investigated the relationship between interaction and L2 learning.
A number of these studies asked ‘Do learners acquire those features that learners claim to have noticed?’ In Mackey (2006), learners took part in a game show activity and received recasts or clarification requests when they made errors in questions, plurals, and past tense. Mackey found that the level of noticing varied according to target structure, with higher levels evident for questions forms, much lower levels for past tense, and intermediate levels for plurals. 83 per cent of the learners who reported noticing question forms also improved their ability to form questions. However, the relationship between noticing and the other two target features was not established. This study suggests that when learners do notice a grammatical form, acquisition occurs. However, as noted earlier, noticing itself depends on the grammatical feature.
Most studies have investigated the effects of interaction on acquisition without obtaining measures of noticing. Mackey and Goo (2007) conducted a meta-analysis of these studies. They found that interaction had a stronger effect on vocabulary than on grammar. This is not surprising as lexical problems are more likely to result in both negotiation and noticing. Just about all the studies that have investigated the impact of negotiated interaction on vocabulary have reported positive effects. Importantly, negotiation does not just benefit those learners who participate actively in it, but also learners who are auditors of negotiation involving others. This is more likely to occur if the task motivates them to attend to closely to the input as in Ellis, Tanaka, and Yamazaki (1994).