Unit 18 Learner strategy
Pre-questions
1. What are the two types of knowledge regarding input? 2. What does procedural knowledge include? 3. What is Ellis’s framework of strategies like? 4. What are the two types of speech?
5. What are the hypotheses of language learning strategies? 6. What are the three types of inferencing? 7. How to test hypothesis?
8. What are the production strategies? 9. What are the planning strategies? 10. How to correct strategies? Introduction
This unit considers the internal processes which account for how the learner handles input data and how the learner utilizes L2 resources in the production of messages in the L2. It looks at the internal mechanisms, or the 'black box'. A complete account of SLA involves both showing how the input is shaped to make it learnable (an inter-organism perspective), and how the learner works on the input to turn it into intake (an intra-organism perspective). 18.1 Two types of knowledge
The learner has two types of L2 knowledge: declarative and procedural (Faerch and Kasper 1983b).
1
Declarative knowledge (陈明性知识) is 'knowing that'; it consists of
internalized L2 rules and memorized chunks of language. Procedural knowledge
(程序性知识) is
knowing how; it consists of the
strategies and procedures employed by the learner to process L2 data for acquisition and for use. When we talk about acquiring a L2, we normally mean declarative knowledge, as the learner is considered to have access already to a set of procedures for learning the L2. Declarative knowledge has been examined in previous units.
Procedural knowledge can be subdivided initially into social and cognitive components(社会知识和认知知识). The social component comprises the behavioral strategies (行为策略)used by the learner to manage interactional opportunities (i.e. the use of the L2 in face-to-face contact or in contact with L2 texts). Fillmore (1979) describes a number of general social strategies used by five Spanish-speaking children learning English in play situations with native-speaking children. At the beginning, the children adopted the strategy of joining a group and acting as if they understood what was going on, even if they did not. Later, they sought to give the impression that they could speak the language by utilizing a few carefully chosen words. They also relied on their friends to help them out when they were in communicative difficulty.
The cognitive component of procedural knowledge comprises the various mental processes involved in internalizing and automatizing new L2 knowledge and in using L2 knowledge in conjunction with (结合) other knowledge sources to
2
communicate in the L2. These processes, therefore, involve both learning and using the L2. Learning processes account for how the learner accumulates new L2 rules and automatizes existing ones by attending to input and by simplifying through the use of existing knowledge. They can account for the 'natural' sequence of development. The processes involved in using L2 knowledge consist of production and reception strategies and also communication strategies. Production strategies are defined by Tarone (1981) as attempts to use existing L2 knowledge efficiently and clearly with a minimum of effort. Communication strategies occur when the speaker is not able to communicate his original communicative goal in the way he planned to, and so is forced to reduce the goal or to locate alternative means to express it. Communication strategies are the result of an initial failure to implement a production plan. Language use is characterized by both production and reception strategies, which operate when the learner utilizes available resources easily and subconsciously. It is also characterized by communication strategies, which operate when the learner needs to compensate for inadequate means and which, as a result, are likely to involve greater effort and to be closer to consciousness. 18.2 Framework of strategies
A framework of the different learner strategies is provided in Figure 18.1. It should be recognized that these strategies are not special to the learner. Native language speakers must be considered to use the same strategy types.
3
What distinguishes L2 learners and native speakers is the frequency with which the same strategies are called upon. Learners will manifest more strategy tokens.
declarative knowledge (i,e. L2 rules and chunks of speech) {examined in Chapters 3 a nd 4)
social processes/strategies
(i.e. devices for managing
interaction in L2)
procedural knowledge for learning L2
(examined in this chapter} (i.e. devices for internalizing or
automatizing L2 knowledge)
cognitive
strategies/ production/reception processes processes and strategies
{i.e. devices for using existing resources automatically)
for using L2
communication strategies (i.e. devices for compensating for inadequate resources)
Figure 18.1 Types of L2 knowledge (Ellis, 1999,165)
Before examining the different types of procedural knowledge in detail, it is important to consider the metalanguage (元语言) used to describe this kind of
4
mental phenomenon. It is perhaps inevitable, given the abstractness of the concepts involved, that the metalanguage is often confusing and vague. In particular, researchers do not use terms like 'process', 'strategy', or 'principle' consistently. Sometimes they use them as synonyms for general mental operations, but sometimes they use them to differentiate operations involved in language processing. For example, Faerch and Kasper (1980) make a clear distinction between 'strategy' and 'process'. They define the former as plans for controlling the order in which a sequence of operations is to be performed, and the latter as the operations involved either in the development of a plan (the planning process) or in the realization of a plan (the realization process)--see Figure 18.2 below.
goal
plan
action
realization process planning process Figure 18.2 The planning and realization of intellectual behavior (Faerch and Kasper 1980)
A problem with such a distinction is that there is little consensus
识)concerning
(共
which behaviors belong to 'processes', as opposed to
'strategies'. It is reasonable to distinguish the idea of a sequence of operations (as in 'the production/reception process') (一系列操作)and the idea of a single operation as a feature of a process (as in 'a strategy of simplification')
5
(单次操作).
18.3 Specific learning strategies
The proliferation of terms and concepts, so characteristic of accounts of every aspect of procedural knowledge, is perhaps most evident in discussions of
learning
strategies.
Strategies
as
varied
as
memorization,
overgeneralization, inferencing, and prefabricated patterns (预模) have all been treated under the general heading of 'learning strategies'. In addition there is constant reference to the process of hypothesis-testing, which underlies in a rather ill-defined way the operation of the more specific strategies.
18.4 Two types of speech
Some order is imposed by Ellis (1999) on this heterogeneity (异质性) represented by framework. The basis of the framework is a distinction between two types of linguistic product: formulaic speech and creative speech.
18.4.1 Formulaic speech
Formulaic speech consists of 'expressions which are learned as unanalyzable wholes and employed on particular occasions…’(Lyons 1968: 177). It can be observed in the speech of native speakers as well as of learners. Krashen and Scarcella(1978) distinguish between routines (常规) and patterns. Routines refer to whole utterances learnt as memorized chunks, and
6
patterns refer to utterances that are only partially unanalyzed and have one or more open slots
e.g. Can I have __? Can I have a ___? I’d love to but____. Do you mind my V–ing? According to Ellis (1984c), formulaic speech can consist of entire scripts, such as greeting sequences, which the learner can memorize because they are more or less fixed and predictable.
e.g.
A: How are you?
B: I’m fine. Thank you. And you? A: I’m fine, too.
Formulaic speech has been observed to be very common in SLA, particularly in the early stages of development. It figures frequently in the speech both of child learners (cf. Huang and Hatch 1978) and of adult learners (cf. Hellwig 1983) in naturalistic SLA, and also in some classroom learners (cf. Ellis 1984c). The particular unanalyzed chunks which are learnt are likely to vary from learner to learner, but the following appear to be typical:
e.g. I don't know. Can I have a_____?
7
There is no _____. What's this? I wanna __. This is a __. How do you do? I can't speak English.
Each formula is closely tied to a particular communicative goal. Ellis (1984a), for instance, notes that three classroom learners rapidly developed a number of formulas to meet the basic communicative needs in an ESL classroom where English functioned as the medium of communication. It has been suggested by Krashen and Scarcella (1978) that learners develop formulas as a response to communicative pressure. That is, they memorize a number of ready-made expressions to compensate for lack of sufficient L2 rules to construct creative speech. Krashen (1982) argues that formulaic speech occurs when the learner is forced to speak before he is ready. Left to his own devices, he will engage in a 'silent period' while he builds up sufficient L2 rules to speak creatively. The important point, however, is that formulaic speech is closely tied to the performance of specific meanings, and that it is common in early SLA because it reduces the learning burden while maximizing communicative ability. Pattern memorization
8
What are the learning strategies involved in acquiring formulaic speech? It is very unlikely that these are the same as those involved in the acquisition of rules responsible for creative speech. One suggestion is that formulaic speech may involve the right hemisphere of the brain rather than the left hemisphere, which is responsible for the creative language function in most people. The right hemisphere is generally considered to involve holistic processing (Seliger 1982). That is, the learner subjugates (抑制) the individual parts that make up an entity to the whole. The learner perceives the whole pattern as a gestalt(--Psychological point of view that says it is necessary to consider the whole of something, since the
whole has a meaning apart from its individual elements完全形态)rather than the elements that
constitute it. In the case of formulaic speech the learner may operate a strategy of pattern memorization. He attends to the input and, using mechanisms located in the right hemisphere, identifies a number of commonly occurring whole utterances in terms of the contexts in which they are used. He is aided in this by the facts that (1) the patterns are highly frequent, and (2) each pattern is linked to a communicative function which the learner is motivated to perform. The patterns are stored in the right hemisphere and are available for immediate processing in production and reception. The extent to which the strategy of pattern memorization is invoked is a product of the type of input the learner receives (i.e. how predictable and context-dependent it is) and the strength of the learner's need to communicate in the L2.
Pattern memorization is a psycholinguistic strategy. It occurs without
9
the learner needing to activate it consciously and it has no overt manifestation. We cannot see or hear pattern memorization taking place. Some researchers refer to the use of prefabricated patterns as a learning strategy, but this is to confuse product and process. The patterns are the products of the process of pattern memorization.
Pattern imitation
The product--the patterns themselves--can also result from another, related strategy. Pattern imitation is the behavioral equivalent of pattern memorization. It is a behavioral strategy because it is usually consciously activated and because it can be observed when it takes place. Pattern imitation involves the deliberate and methodical(有条理的)copying of whole utterances or parts of utterances used in the speech of an interlocutor. It is common in classrooms where the audiolingual techniques of pattern practice are used. It can also occur in naturalistic SLA, when the learner imitates the previous utterance of a native speaker irrespective of its communicative appropriateness. This is very common, particularly with child learners (Itoh and Hatch 1978). Seliger(1982) considers that pattern imitation also involves right hemisphere abilities.
It has been suggested in both first and second language acquisition research that formulaic speech serves as the basis for creative speech (格式化言语
是创新性言语的基础。). That is, the learner comes to realize that utterances initially
10
understood and used as wholes consist of discrete constituents which can be combined with other constituents in a variety of rule-bound ways. (学生逐渐认识到：起
初当作整体理解和使用的言语是由分离成分组成的，这些成分可以以各种各样受规则约定的方法用来跟其他成分结合。)
Clark(1974) illustrates how in first language acquisition, new structures can result from the juxtaposition (并置) of two routines or from the embedding(嵌入) of one within another. Fillmore (1976) suggests that in SLA, formulas are slowly submitted to an analytical process that releases constituent elements for use in 'slots' other than those they initially occupied. Seliger (1982) proposes that patterns initially learnt through right hemisphere abilities are brought to the attention of left hemisphere abilities, which work on them in order to analyze out their parts. Ellis (1984c) shows how the 'I don't know' formula is built on, by combining it with other formulas: e.g. That one I don't know. I don't know what's this.
and also broken down, so that ―don't‖ is used in similar but different expressions:
e.g. I don't understand. I don't like.
'Know' is used without 'don't':
e.g. I know this.
and subjects other than 'I' begin to occur: e.g. You don't know where it is.
11
It is possible that formulas are slowly unpackaged so that valuable information can be fed into the creative rule system.
The basis for this analysis must lie in the learner comparing utterances in order to identify which parts recur (重现) and which parts remain the same. The learner gradually notices variation in the formulaic structures according to the situation and also detects similarities in the parts of different formulas (Fillmore 1979). Thus, earlier utterances, learnt and used as formulas, are often more grammatical in appearance than later utterances, which are constructed from rules. To account for this process, a strategy of pattern analysis can be posited(假定). This works by comparing formulas and looking for similarities and differences. It ascribes to the learner something of the skills of a linguist seeking to identify the constituent structure of utterances.
There is not total agreement about pattern analysis. Krashen and Scarcella (1978) argue that formulaic speech and rule-created speech are unrelated. They claim that the learner does not unpackage the linguistic information contained in formulas, but internalizes L2 rules by attending to the input in accordance with the 'natural' sequence of development. They see the process of 'creative construction' of L2 rules as entirely separate.
Formulaic speech is an important factor in SLA, but is probably only a major factor in early SLA. The strategies of pattern memorization, pattern imitation, and (more controversially) pattern analysis are to be seen as minor learning strategies in comparison with those contributing directly to the creative
12
rule system.
【Stop for a break at 17:08, Mar 07, 2012】
18.4.2 Creative speech
Creative speech is the product of L2 rules. These are 'creative' in the Chomskian sense that they permit the L2 learner to produce entirely novel sentences. They are the rules that constitute the learner's interlanguage system and which account for the 'natural' sequence of development. They are variable in that they permit the learner to vary his performance according to both the linguistic and the situational context.
A plethora (过多) of strategies have been proposed to account for the creative rule system. Faerch and Kasper (1980; 1983b) provide a framework which can be used to consider these strategies systematically. They distinguish strategies involved in establishing interlanguage rules, and strategies involved in automatizing interlanguage knowledge. In the former they also distinguish two basic and related processes: hypothesis formation and hypothesis testing. Ellis (1999) uses this framework in the following discussion, but does not limit himself to the specific strategies identified by Faerch and Kasper.
18.5 Hypothesis formation
Faerch and Kasper (1983b) suggest that hypotheses about interlanguage rules are formed in three ways:
1. by using prior linguistic knowledge (i.e. first language knowledge, existing
13
L2 knowledge, or knowledge of other languages); 2. by inducing new rules from the input data; 3. by a combination of (1) and (2).
Underlying these general processes it is possible to identify two general strategies, each with a number of more specific strategies associated with them. The two general strategies are simplification and inferencing. 18.5.1 Simplification
Many of the early interlanguage publications recognized that learners seek to ease the burden of learning in various ways. Richards (1974) defines 'strategies of assimilation' as attempts to reduce the learning burden. As an example he quotes his own use of the equivalent of the 'going to' form in French to express future meaning, simply because he found it easier to learn. Simplification consists of attempts by the learner to control the range of hypotheses he attempts to build at any single stage in his development by restricting hypothesis formation to those hypotheses which are relatively easy to form and will facilitate communication.
Simplification is evident in the use of a number of strategies. Widdowson (1975b) has suggested that Selinker's five interlanguage processes are tactical variations of the same underlying simplification strategy. For instance, transfer involves the use of the learner's L1 as a basis for forming hypotheses about the L2, while overgeneralization involves the use of existing L2 knowledge by
14
extending it to new interlanguage forms. Both strategies can be seen as manifestations of the same basic strategy of relying on prior knowledge to facilitate new learning (Taylor 1975; McLaughlin 1978a). Which strategy is used may be partly a matter of individual preference, but is also likely to be governed by such factors as the learner's stage of development, the linguistic properties of the L2, and the learning context--Ervin-Tripp (1974), for example, proposes that transfer will be more common in a classroom than in a natural learning environment. Irrespective of which strategy is used, simplification takes place, in that the learner seeks to override (推翻) the evidence of the input by positing(假
定) interim rules(临时规则) which he finds easier to construct on the basis of
existing knowledge. In this sense simplification serves as an alternative to inferencing by attending closely to input data.
Not all researchers agree that simplification is a learning strategy. Faerch and Kasper (1980), for example, argue that strategies such as simplifying, regularizing, overgeneralizing, and redundancy reducing are, in fact, strategies of non-learning, as they prevent the formation of correct hypotheses. If it is accepted that the formation of non-correct hypotheses is essential in the overall process of SLA, this argument appears tenuous(脆弱的,无力的). Learning does not only involve the formation of correct hypotheses; it also involves interim hypotheses (过渡期间的假设)which are systematically amended (改进) until the final correct hypothesis is arrived at.
Simplification plays a positive role in that it delimits the proportion of
15
hypotheses which are formed by attending to input (i.e. through inferencing) at any one time.
Another objection to the notion of simplification is that it does not make sense to refer to the learner as simplifying what he does not possess. Corder (1981) has argued that SLA should be viewed as a process of 'complexification'. This confuses product and process again. While Corder is right to claim that the learner cannot simplify L2 rules he has not acquired, it is perfectly feasible to argue that he simplifies the burden of learning by restricting hypothesis formation dependent on attending to input.(L2学习者不可能简化他尚未习得的二
语规则，而他以限制赖以注意输入的假设形成的方式来简化学习负担) Even if we accept that the product
cannot be simplified(--and this will be challenged in the next section), it is still viable (可实行的) to argue that the process is simplified (即使表达不能简化，但过程可以简化) . This is the position adopted by Meisel (1983). He uses the term 'elaborative simplification' (煞费苦心的简化)to refer to the learner process of adjusting what has just been acquired to what has been acquired before, by forming approximations of the rule represented in the input.(通过构成以输入所代表的近似规则，把习得
的语言规则调整到以前习得的语言过则。)
Simplification is sometimes listed as a production strategy rather than a learning strategy (Tarone 1981). That is, the learner is equipped with L2 rules but is unable to utilize them because of language processing difficulties. He therefore simplifies them (e.g. by using some but omitting others). Meisel refers to this as 'restrictive simplification'. Simplification is clearly an important aspect
16
of production. So, at least, simplification is both a learning strategy and a production strategy.
18.5.2 Inferencing(推断)
Inferencing is the means by which the learner forms hypotheses by attending to input(学生通过专注于输入而形成假设的手段). That is, in cases where the appropriate L2 rules cannot be successfully derived by means of transfer or overgeneralization of existing interlanguage knowledge, the learner will need to induce the rule from the input. (适当的二语规则不能从通过对现有中介语知识迁移或过度概括的手段获
得，在这种情况下，学生需要二语输入中归纳出规则。)For example a Spanish learner of English
will not succeed in acquiring the rule for negative sentences on the basis of simplification. A transfer strategy might lead to the 'no + V' rule (e.g. 'No like beer'). Such a rule might also be derived by overgeneralization (i.e. 'no' is overgeneralized as the negator of verbs as well as nouns)( 过度概括). But in order to arrive at the correct rule for generating sentences like I don't like beer.
18.5.2.1 Three types of inferencing
Existing treatments of inferencing do not entirely conform with the one Ellis has just given. Carton (1971), for instance, discusses inferencing in terms of three types of cues:
(1) intralingual(语言内部的)(i.e. cues derived from the morphological and
17
syntactic regularity of the L2);
(2) interlingual (中介语(的))(i.e. cues derived from loans between languages where similar forms are hypothesized); and
(3) extralingual (语外的) or contextual（情境的） (i.e. cues based on regularities in the objective world which make predictions possible).
18.5.2. 2 Byalystock’s three types of inferencing
Bialystok (1983a) also identifies three types of inferencing, but they are not the same as Carton's:
(1) inferencing from implicit knowledge (i.e. making explicit L2 knowledge that is only intuitive to begin with);
(2) inferencing from other knowledge (i.e. using knowledge of other languages, in particular the L1, and/or knowledge of the world), and (3) inferencing from context.
In both Carton's and Bialystok's frameworks, however, inferencing is treated as a general process that incorporates (包含) both what Ellis has called inferencing and simplification.
18.5.2. 3 Ellis’s two types of inferencing
Ellis (1999) restricts inferencing to two types: intralingual inferencing
(语内推断)and extralingual inferencing(语外推断).
Intralingual inferencing involves a process similar to that of pattern
18
analysis of formulaic speech(格式化语言的模式分析), the difference being that in this case the learner operates on external L2 data rather than internal L2 data (i.e. stored formulas). Intralingual inferencing is the result of intake analysis(吸纳分析). The use of the strategy may be governed by innate linguistic or cognitive predispositions (倾向) to attend to specific features of the input. If they are linguistic, they will resemble the 'universal grammar' which Chomsky believes the first language learner is endowed with. If they are cognitive, they will involve the use of general perceptual strategies(一般的知觉策略) such as those listed by Slobin as 'operating principles'. The detailed workings of intake analysis, however, have not yet been described, and our understanding of them is limited.
Extralingual inferencing is one of the most powerful devices available to the learner for building hypotheses from external input. It consists of paying attention to features of the physical environment and using these to make L2 input comprehensible. By observing the non-linguistic correlates (关联项) of utterances, the learner can convert input that is beyond his competence into intake(吸纳). MacNamara (1972) talks of the L1 learner using 'meaning as a clue to language rather than language as a clue to meaning'(把意思当做线索了解语言，而不是把
语言当做线索了解意思). The same must also hold true for the L2 learner, including the
adult learner. The situational context of a speech act enables the learner to work out meanings right from the start of SLA. Therefore，extralingual inferencing serves as the principal means that the beginner uses to formulate
19
hypotheses from the external input. It continues to be important throughout SLA(语外推断作为初学者使用的主要手段以便从外部输入中制定假设).
【Stop at 17:58 for a break, Mar 07, Wednesday, March 07, 2012.】
18.6 Hypothesis testing
Language learners may make errors in order to test out hypotheses about the L2 rule system. It remains to examine in some detail how this takes place and to consider some of the objections about hypothesis testing which have been raised.
Once the learner has developed a hypothesis, he can test it out in a variety of ways. There are four ways put forward by Faerch and Kasper (1983b) to test hypothesis.
1 Receptively(接收式地) (i.e. the learner attends to L2 input and compares his hypotheses with the data provided--by means of intake analysis);
2 Productively (表达式地) (i.e. the learner produces L2 utterances containing rules representing the hypotheses he has formed and assesses their correctness in terms of the feedback received);
3 Metalingually (元语言式地)(i.e. the learner consults a native speaker, teacher, grammar, or dictionary to establish the validity of a hypothesis);
4 Interactionally interlocutor).
As a result of hypothesis testing carried out in one or more of these ways,
20
(互动式地)
(i.e. the learner elicits a repair from his
the learner is in a position either to confirm or reject an initial hypothesis(初始假设). The constant revision of interlanguage rules is the result of the learner responding to evidence that requires modification of hypotheses. SLA ceases either when the learner no longer receives contrary evidence, or when he stops testing out hypotheses (i.e. because he is satisfied with his existing competence).
There have been several criticisms of the hypothesis testing view of SLA. Most of these have to do with the role of feedback. It has been observed that the provision of negative feedback (i.e. corrections) does not appear to lead to more accurate performance, at least not immediately. Even when the negative feedback is provided in the course of ordinary conversation (e.g. in the form of expansions and paraphrases serving as confirmation checks and requests for clarification), there is still no evidence to suggest that the learner amends his hypothesis immediately. Moreover, as Long (1977) points out, it makes little sense to talk about learners actively seeking out negative feedback. The' falsification idea' is not feasible.
Nevertheless, the hypothesis testing model is still viable. These criticisms relate to how hypotheses are rejected. The important process may be hypothesis confirmation. The learner often builds two or more hypotheses relating to the form of a single rule. The role offered back may be to enable him to decide which hypothesis to accept finally. He is unlikely to come to a conclusion until he has received substantial feedback to test each of the
21
hypotheses fully. As a result, the alternative hypotheses are maintained for some time, with perhaps a gradual tendency to favor one rather than the other(s). Eventually one of the competing hypotheses wins through and becomes a permanent L2 rule. This view of hypothesis testing meets the objections described above and also gives due recognition to interlanguage as a variable phenomenon. The entire process is a subconscious one; that is, the learner does not carry out hypothesis testing in order to learn the L2, but as part of the process of communication. Thus different types of communication can hinder or facilitate the process, as suggested.
The role of hypothesis testing is considered when the 'universal hypothesis' in SLA is examined. The argument is that inductive hypothesis formation cannot adequately explain how acquisition takes place. 18.7 Automatization processes
The variability of interlanguage phenomena is also a reflection of rules which have been differentially automatized. Competing hypotheses are held with varying degrees of certainty. The less certain a hypothesis is, the less accessible it is for use in L2 production. Part of the learning process involves consolidating hypotheses by accumulating confirmatory evidence. Faerch and Kasper (1983b) suggest that this can be brought about by practising the L2 productively and receptively. They distinguish formal and functional practice, depending on whether the focus is on formal features of the L2 or communicative endeavor. Automatization involves both the practising of L2
22
rules which enter interlanguage at the formal end of the stylistic continuum and the practising of rules which are already in use in the 'vernacular'.
summary
The processes and strategies are responsible for creative speech. They are typically subconscious procedures (i.e. they are spontaneously activated by the learner while he is focused on some communicative purpose), but they can also be conscious (i.e. deliberately activated by the learner with the intention of increasing his L2 knowledge). Some procedures—metalingual
(元语言的)
hypothesis testing and formal practice--are invariably conscious, however.
Process Strategy Hypothesis formation Simplification
(1) overgeneralization (2) transfer Inferencing
(1) intralingual (via intake analysis) (2) extralingual
Hypothesis testing Receptive (via intake analysis) Productive Metalingual Interactional Automatization Formal practice Functional practice
Table 18.3 Processes and strategies in creative language learning 18.8 Production strategies
23
It will be recalled that utilizing interlanguage rules involves both production-reception strategies and, when initial plans prove problematic, also communication strategies. This section considers the unproblematic use of L2 knowledge. It will present a basic model of production (reception is not considered, for reasons already stated), and then consider specific strategies associated with the planning and articulating components of this model.
A model of L2 production
It is hypothesized that L2 production follows the same pattern as native-speaker production. The model which is summarized below is that described by Clark and Clark (1977).
The starting point is the speaker's communicative goal. This might be to tell a story, to issue an instruction, or to answer a question. The communicative goal will determine the type of discourse plan the speaker needs to form. This plan will reflect whether the communicative goal needs to be realized through dialogue or through monologue. It will also reflect to what extent the discourse is conventional and therefore 'scriptal'(脚本的), or requires unique planning. Once the speaker has formed an appropriate discourse plan, he begins to construct sentence plans. These involve outlining the constituent structure of each utterance, after determining the general propositional content and illocutionary meaning. The discourse and sentence plans together constitute a 'skeleton'. The next stage leads into the execution component of the model. It consists of
24
building plans for the structure of each constituent. However, the language user does not necessarily build plans for all the constituents of the proposed utterance. He is likely to alternate between the planning of individual constituents and their execution. Thus the constituent planning stage and the articulatory program stages do not follow in linear fashion. Evidence that the user moves backwards and forwards from planning to execution can be found in the pauses which occur in speech. These are typically found at constituent boundaries and seem to be motivated by the learner's need for time to plan the next constituent. Clark and Clark suggest that the articulatory program itself has five stages:
(1) the selection of the meaning that each constituent is to have; (2) the selection of a syntactic outline for the constituent (i.e. specifying the actual word slots);
(3) content word selection (i.e. choosing nouns, verbs, adjectives, and adverbs to fit into the appropriate slots);
(4) affix and function word formation (i.e. slotting in form words and grammatical inflections); and
(5) specification of the phonetic segments. The final component of the model is the motor program，i.e. the actual production of the utterance. Figure 18.4 provides a diagrammatical(图解的)representation of Clark and Clark's 'skeleton and constituent' model.
25
Planning program 1 Communicative goal 2 Discourse plan 3 Sentence plan 4 Constituent plans Articulatory program 1 Meaning selection 2 Syntactic outline 3 Content word selection 4 Affix and function word format 5 Specification of phonetics segments Motor program
Figure 18.4 The 'skeleton and constituent' model of language production (Based on Clark and Clark 1977)
Littlewood (1979) proposes a production model specifically to account for L2 use. His model closely follows that of Clark and Clark. Littlewood distinguishes
26
two sets of strategies based on the model: minimal strategies and maximal strategies. The former characterize production in the early stages of SLA. They revolve the learner simplifying both the constituent planning and articulatory stages in a number of ways. The learner may not develop constituent plans for all the major constituents of the sentence. Instead he may plan for only the minimal number of unshared meaning elements, allowing the situational context to stand in for those missing. Ellis (1982a) describes this as a strategy of semantic simplification. Another type of simplification can occur in the articulatory program when the learner chooses to encode only content words and to omit affix and function word formation. This may occur either because the learner has not yet constructed hypotheses for those aspects of the L2, or because of processing pressures which do not permit him time to make a thorough search of his linguistic resources. Thirdly, simplification may take place in the motor program (i.e. sounds may be elided 27 reflected in the complexification of the inter-language system. The skeleton and constituent model also provides a framework for interpreting Seliger's (1980) interesting distinction between planners and correctors. The planners are those who plan each constituent carefully before starting the articulatory program. As a result, their performance is likely to be hesitant but correct, at least in terms of their current interlanguage system. In contrast, the correctors only partially plan each constituent before embarking on the articulatory program, with the result that they produce fewer pauses, but are more prone to monitoring or to the use of communication strategies when they find they have not conformed to their interlanguage system or run into problems. The use of planning and correcting strategies is considered in greater detail below. 18.9 Planning strategies Two basic planning strategies can be identified. These are semantic simplification and linguistic simplification. They are the procedures used by the learner in the early stages of SLA when he has minimal L2 resources and needs to employ them in a maximally easy and efficient manner. Semantic simplification Semantic simplification occurs when the learner simplifies the sentence plan by reducing the propositional elements that are linguistically coded. The 28 learner selects specific constituents for encoding and leaves the others to be inferred by the listener from extralinguistic cues. The constituents deleted are not syntactic but semantic; that is, they consist of cases such as agent, object, dative, etc.(Fillmore 1968) which can be thought of as general conceptual categories, used in both the perception of real-life events and their linguistic representation. Imagine a situation in which Person A is hitting Person B. A full linguistic representation of this (from the viewpoint of Person B) might be: He is hitting me. which involves the case categories: (Agent) (Action process) (Patient) The learner may produce any one of the following abridged versions: Hitting (= Action process) He hitting (= Agent + Action process) Hitting me (-- Action Process + Patient) He me (= Agent + Patient) Which version the learner chooses will reflect (1) the linguistic resources he has available (e.g. he may not know the verb 'hit'), and (2) which constituents he feels will be maximally informative in terms of his communicative goal and the context of situation. Ellis (1982a; 1984a) has suggested that semantic simplification provides a powerful explanation of processes involved in both first and second language acquisition. Stop here for the first break this morning, March 14, 2012 29 Linguistic simplification Linguistic simplification involves the omission of form words and affixes. Unlike semantic simplification, it runs up against the logical objection that the learner cannot simplify what he does not possess. Thus if the learner produces an utterance like: He hitting me. where aux-be is omitted, this can only constitute linguistic simplification if the learner can be shown to possess an aux-be rule in his interlanguage. In fact there is plenty of evidence to suggest that learners do in fact omit functors such as aux-be, even though they have acquired the necessary forms. It is a common characteristic of the learner's variable use of his L2 knowledge. As Meisel (1983: 127) puts it: '...learners do produce utterances that are structurally simplified ...although it can be assumed that their 'transitional competence at this point contains rules that generate the corresponding nonsimplified structures'. Meisel argues that 'restrictive simplification' accounts not only for L2 production, but also for simplified registers such as motherese and foreigner talk. Linguistic simplification, therefore, is a very general language production strategy. 18.10 Correcting strategies: monitoring The principal strategy responsible for correcting is monitoring. This has 30 received considerable prominence in the form of Krashen's (1981) Monitor Model of L2 performance. Krashen argues that the learner possesses two kinds of knowledge, which are implicit or intuitive knowledge (referred to as 'acquisition' by Krashen), and explicit or metalinguistic knowledge (referred to as 'learning'). According to Krashen, the learner initiates utterances using only implicit knowledge, but is able to monitor his performance using explicit knowledge either before or after articulation. Krashen's theory has come under considerable attack. His notion of monitoring is seen as too narrow, in the sense that the learner is clearly able to edit his performance using implicit as well as explicit knowledge. Krashen, in fact, allows for this in his discussion of correction by 'feel', but this is allocated only a minor place in the overall theory. The kind of monitoring Ellis wishes to consider here is the broader kind, involving the use of any form of knowledge to correct incipient or actual output. Morrison and Low (1983) propose a very similar production model to Clark and Clark's in order to discuss the role of monitoring. They distinguish post-articulatory monitoring, which occurs after the articulatory program has been implemented, and pre-articulatory monitoring, which can occur at any stage during the articulatory program. After the learner has begun the process of filling out constituent plans with linguistic forms, he can operate on the emerging utterance by substituting an initially selected form with another preferred form. This kind of monitoring can be carried out on lexis, syntax, 31 morphology, and the phonetic realization. In other words, it can occur at any of the five stages in the articulation program. It can also occur at higher levels, when it leads to an adjustment in communicative goal or discourse and sentence plans. But because monitoring can occur when no communicative or linguistic difficulty is experienced, it should be treated as a means of maximizing existing resources in an easy and efficient manner; it is, therefore, a production strategy rather than a communication strategy. Summary Production comprises a hierarchical process involving a planning, an articulation, and a motor program. This process is common both to native speakers and to L2 users. Production strategies can be divided into those strategies involved in the planning of utterances, and those involved in the correction of utterances. Semantic simplification and linguistic simplification are 'minimal strategies' used to facilitate easy and effective use of the L2 system. They are planning strategies. Monitoring is a correcting strategy. It can occur before or after articulation and can utilize both implicit and explicit knowledge. 18.11 Strategies of communication The term 'communication strategy' was coined by Selinker (1972) in his account of the processes responsible for interlanguage. There has been a steady increase of interest in the learner's communication strategies since then. Much of this interest, however, has been taken up with the problems of 32 definition. Following this, a typology (类型学)of communication strategies will be provided and a survey of empirical studies of their use by L2 learners undertaken. This section will conclude with some comments about the role of communication strategies in SLA. Defining communication strategies There are two key concepts in most discussions of communication strategies: conscious, and problem-oriented. Communication strategies are conscious. Varadi (1973) points out that L2 errors may arise either inadvertently or deliberately. Inadvertent errors are the result of production strategies and reflect the transitional state of the learner's L2 knowledge. Deliberate errors are the result of communication strategies that are consciously employed by the learner in order to reduce or replace some element of meaning or form in the initial plan. Faerch and Kasper (1980) also consider consciousness to be a defining characteristic of communication strategies, but they recognize the difficulty of deciding empirically whether a strategy is conscious or otherwise. As learners may not always be aware of their use of communicative strategies, they suggest that a better definition is to refer to them as 'potentially conscious'. Communication strategies are problem-oriented. Communication strategies are employed by the learner because he lacks or cannot gain access to the linguistic resources required to express an intended meaning. As 33 Corder (1978c) puts it, there is a lack of balance between means and ends. Faerch and Kasper (1980) classify communication strategies as part of a particular kind of plan which is activated when the initial plan cannot be carried out. The learner is forced into substituting a 'strategic plan' for his original production plan, because he discovers he has insufficient means to implement the production plan. In terms of the model of ―speech production‖, this involves adding a loop, which starts from the source of the problem, which might lie at any point in the articulation process, and then leads into an alternative plan (which can be at discourse, sentence or constituent level) before feeding back into the articulation program. However, communication strategies are not alone in being problem-oriented. Learner strategies can also be motivated by the learner's recognition that existing means are insufficient. But communication strategies differ from learning strategies in that the problem arises as a result of attempts to perform in the L2, and the strategies are needed to meet a pressing communicative need. If learning strategies are the long-term solution to a problem, communication strategies provide the short-term answer. To begin with, communication strategies were discussed in psycholinguistic terms. That is, they were treated as the mental phenomena which underlay actual language behavior. In some later discussions (e.g. Tarone 1981), an interactional perspective has been taken. In this, communication strategies are seen as attempts to bridge the gap between the linguistic knowledge of the L2 learner and the linguistic knowledge of the 34 learner's interlocutor in real communication situations. They are characterized by the 'negotiation of an agreement on meaning' between interlocutors (Tarone 1981: 288). Tarone, therefore, sees communication strategies as the learner's contribution to the interactional work required to overcome a communication problem. Limitations of the interactional definition However, as Faerch and Kasper (1983c; 1984) point out, there are several difficulties with this interactional definition. First, it is difficult to apply to monologue (e.g. writing), when the L2 learner's interlocutor is not present, and there is no overt negotiation of meaning. Communicative problems, however, occur in monologue just as much as in dialogue. Second, the application of a communicative strategy can take place without this becoming manifest in interaction. The learner may realize the inoperability of his initial production plan before he begins to execute it. The substitution of an alternative plan, therefore, can take place with no other signal than a pause, perhaps a slightly longer one than those characteristic of normal production. In Tarone's interactional definition, however, only those communicative strategies that are marked in performance by some form of appeal on the part of the learner are considered. In general, therefore, the psycholinguistic definition is to be preferred. The interactional perspective is best tackled by discourse analysis, which considers the joint contribution of learner and interlocutor, rather than singling out the learner's activity for separate analysis. 35 Communication strategies are employed by native speakers as well as by L2 learners. Most of the communicative strategies listed in the typology in the following section are common to both. They are to be seen as a part of communicative competence. Canale and Swain (1980:25) identify 'strategic competence', defined as 'how to cope in an authentic communicative situation and how to keep the communicative channel open', as an integral part of the language user's overall communicative competence. In the light of the foregoing discussion, communication strategies can be defined as follows: Communication strategies are psycholinguistic plans which exist as part of the language user's communicative competence. They are potentially conscious and serve as substitutes for production plans which the learner is unable to implement. 18.12 A typology of communication strategies Perhaps because of the problems of definition, there is no generally agreed typology of communication strategies. Various typologies have been proposed by Varadi (1973), Tarone et al. (1976), Corder (1978c), and Faerch and Kasper (1980). In addition, typologies relating specifically to lexical problems are provided by Blum-Kulka and Levenston (1978), and Paribakht (1982). The typology provided in Table 18.2 is a summary of that given in Faerch and Kasper (1984). It should be noted that they refer to production and not reception. 36 Many of these strategies will be difficult to identify in actual data. Some (e.g. the retrieval strategies) will not be manifest in actual performance at all. Ideally, therefore, the identification of the use of communication strategies needs to make use of introspective research techniques as well as the analysis of speech data. Empirical research Theoretical discussion of communication strategies has predominated (支配) over empirical research into their use. This is a reflection of the uncertainties of their definition and the consequent problems of identification. Some empirical research has taken place, however, and is rapidly growing. A number of rather different approaches have been followed. Early research (e.g. Varadi 1973; Tarone 1977) consisted of a comparison of learners' performance on story-telling tasks in their first and second languages. The comparison was motivated by the belief that L2 communication strategies can be identified only if base-line first language data are available. For example, avoidance cannot be said to have taken place if it is also evident in L1 performance. Another rather similar approach involves comparing the performance of a group of native speakers with that of L2 learners on an identical task (Ellis, 1984d). A third approach consists of focusing on the use of specific lexical items. This can be done by embedding them in a picture story reconstruction task (by asking subjects to label pictures or translate from the 37 L1). Yet another approach involves the analysis of video-taped conversations between L2 and native speakers (Haastrup and Phillipson 1983). The results of the available research are suggestive rather than definitive (决 定性的). They can be summarized in terms of the effects of different variables on the use of communication strategies. 18.12. 1 Effects of proficiency level The proficiency level of the learner influences his choice of strategy. Tarone (1977) notes that the less able students whom she investigated preferred reduction to achievement strategies. Ellis (1983) also found that one of the learners in his longitudinal study opted for reduction-type behavior in the earlier stages, but increasingly turned to achievement-type behavior as he progressed. Ellis (1984d), not surprisingly, found quantitative but not qualitative differences between the strategy use of ESL children and native-speaking English children. The former relied more on avoidance, and the latter more on paraphrase. Bialystok (1983b) found that advanced learners used significantly more L2-based strategies and significantly fewer L1-based strategies than less advanced learners. In general L2 learners of limited proficiency prefer either reduction strategies or L1-based achievement strategies while the more advanced learners prefer L2-based achievement strategies such as paraphrase. 2 Effects of the problem-source There is less evidence to demonstrate that strategy choice is influenced 38 by the specific nature of the problem, but this would seem likely. Tarone (1977) notes that code-switching is more likely when the first and second languages have close cognates(同系语言）. Hamayan and Tucker (1980) found that the extent to which L2 child learners displayed avoidance depended on the grammatical structures involved. 3 Effects of personality(个性) Tarone (1977) observes definite differences in her learners' overall approach to story-telling. One learner spoke quickly and provided little detail in either L1 or L2 performance, whereas another elaborated (详细述说) and frequently appealed for assistance. She suggests that personality factors may correlate highly with strategy preference. 4 Effects of the learning situation It would seem probable that learners' use of communication strategies is affected by the situation of use. For instance, learners may use fewer strategies in a classroom environment than in a natural environment, particularly if the pedagogic focus is on correct L2 use, rather than on fluent communication. The situation may also influence the type of strategy used. Piranian (1979) found that American university students learning Russian relied more on avoidance, whereas learners with natural exposure used paraphrases too. Of central importance in the study of communication strategies is their effectiveness in promoting L2 communication. This has received little attention. 39 Bialystok (1983b) suggests that the best strategy users are those with adequate formal proficiency who modify the strategy to suit the specific concept to be conveyed. Haastrup and Phillipson (1983) argue that L1-based strategies are the least effective and L2-based strategies the most effective. They found that L1-based strategies nearly always led to partial or absolute incomprehension and that non-linguistic strategies did not fare much better. They suggest that paraphrase is the strategy most likely to be successful. However, it may not be appropriate to argue about the relative merits of alternative strategies, as learners often use several communication strategies together, first trying one (e.g. a L1-based strategy) and then turning to another (e.g. a L2-based strategy) to supplement the first choice or to try again if it failed. 18.13 The role of communication strategies in SLA Communication strategies are concerned with L2 production. An important issue is to what extent and in what ways they contribute to L2 learning. Communication strategies have been allocated a constitutive role in SLA. Corder (1978c), for instance, characterizes reduction strategies as 'risk-avoiding' and achievement strategies as 'risk-taking'. Faerch and Kasper (1980) argue that a basic condition for communication strategies to have a potential learning effect is that they belong to achievement behavior rather than reduction behavior. They base this view on the grounds that only achievement 40 behavior encourages hypothesis formation, and that risk is essential for automatization. Tarone (1980) challenges this view. She suggests that the conversational effect of communication strategies in general is to enable the native speaker to help the L2 learner use the right form to say what he wants. Thus all strategies can help to expand resources. Another argument to reinforce this point of view is that the main contribution of communication strategies is to keep the channel open. Thus even if the learner is not provided with the particular structure he needs, he will be exposed to a number of other structures, some of which may constitute a suitable intake for his learning strategies to operate on. As Hatch(1978c: 434) argues, the 'most important thing of all has to be \"don't 'give up\"'. Communication strategies are one of the main ways of keeping going. It might also be argued that the successful use of communication strategies will prevent acquisition. A learner may become so skillful in making up for lack of linguistic knowledge by the use of various communication strategies that the need for hypothesis formation or testing is obviated (消除). Such learners are not unfamiliar in anecdotal accounts of SLA, but have not been investigated rigorously. Another issue to do with the role of communication strategies in SLA is what aspect of interlanguage development is affected. In particular it is important to know whether the use of communication strategies facilitates the acquisition of lexis and/or grammatical rules. An argument can be presented for 41 restricting their influence just to lexis. It has been observed that the strategies are common to both learner and native speaker performance. In the case of native speakers, it is feasible to argue that the strategies provide a means by which lexical knowledge can be expanded, but it would be counter-intuitive to argue that they lead to the acquisition of new grammatical rules. If it were to be argued that L2 learners enhance both lexical and grammatical development through strategy deployment, it would be necessary to claim that communicative strategies have a different effect for learners than for native speakers. It may be that communication strategies aid the acquisition of lexis rather than grammar. This discussion of the role of communication strategies in SLA has necessarily been speculative, as there is little hard research. At the moment, it is difficult to come to any firm conclusions. Summary Communication strategies are used by L2 learners (and native speakers) when they are faced by a production problem. They consist of substitute plans and are potentially conscious. A typology of communication strategies distinguishes reduction strategies, which are used to avoid the problem altogether, and achievement strategies, which are used to overcome the problem. The latter can be further divided into compensatory strategies (including both L1 and L2 based strategies) and retrieval strategies. There 42 has been only limited empirical study of communication strategies, but there is evidence to suggest that their use is influenced by the learner's proficiency level, the nature of the problem-source, the learner's personality, and the learning situation. It is not yet clear what effect, if any, communicative strategies have on linguistic development. The issues are (1) which strategies are facilitative (e.g. L2- or Ll-based strategies) and (2) which aspects of SLA (e.g. lexis or grammar) are affected. Conclusion Peering into the 'black box' to identify the different learner strategies at work in SLA is rather like stumbling blindfold around a room to find a hidden object. There are, perhaps, two main consequences of the attempt. First, it is perhaps inevitable that the focus of attention has been on production rather than reception, as the researcher needs the support of the learner's actual utterances to guide his inquiry. The processes and strategies involved in reception have been neglected, therefore. Second, the mapping of strategies into a tight conceptual framework is bound to be arbitrary to some extent. It is doubtful whether learner strategies can be divided into learning, production, and communication as neatly as suggested. This is evident in the discussion of simplification, which has been classified as both a general learning strategy and a strategy of production. It might also be counted as a communication strategy--semantic simplification, for instance, may occur when the learner has 43 experienced a problem with his initial production plan. Researchers differ in the frameworks they provide, because of the problems of identifying and classifying the psycholinguistic events that underlie learning and use. They devote considerable effort to the problems of definition, for the same reason. Increasingly they are turning to introspective methods for studying SLA as one way of gaining greater insight. Understanding language-learner language, however, is not complete without an explanation of the 'black box'. This unit has attempted to review the various psychological explanations of acquisition and use in order to provide an account of the learner's procedural knowledge. Summary of Unit 18 The two types of knowledge—declarative and procedural—lead us to think more about language learning strategies. Further categorization of strategies makes a clear framework of strategies of language learning. A close look finds specific strategies. Some strategies like simplification strategy are two-sided, strategy of learning or strategy of production. Other strategies like inferencing are further classifiable. Communication strategy is regarded as part of language learning strategy. Learner’s choice of strategies is dependable on the level of his L2, and hypothesis making and testing, and personal preferences. Questions 1. What is the relationship between declarative and procedural knowledge? 2. What knowledge does procedural knowledge belong to? 44 3. How does Ellis classify language learning strategies? 4. What do you know about the formulaic and creative speech in regard to strategies of language learning? 5. What are the types of inferencing? 6. What position does communication strategy have in L2 learning strategies? 7. What influences the learner’s choice of L2 learning strategies? 45 因篇幅问题不能全部显示，请点此查看更多更全内容