Riassunto esame Linguistica inglese, prof. Pavesi, libro consigliato Second Language Acquisition, Gass, Selinker

Chapter 8: Looking at interlanguage processing

The field of psychology has significantly influenced the study of SLA.

Connectionist/emergentist model (or constructivist approaches)

In this approach to language learning, emphasis is on usage. Learning does not rely on an innate module; rather, it takes place through the extraction of regularities from the input. As these regularities (or patterns) are used over and over, they are strengthened.

It is a frequency account of second language acquisition and relies on the assumption that "humans are sensitive to the frequencies of events". The examples present in the input form the basis from which regularities emerge. Learning proceeds based on input alone; the resultant knowledge is seen as a network of interconnected exemplars and patterns, rather than abstract rules (conception of grammar).

Connectionism is a cover term that includes a number of approaches. One is Parallel Distributed Processing (PDP), at the heart of which we find a neural network that consists of nodes that are connected by pathways. Pathways are strengthened or weakened through activation or use (influenced by neurobiology). Associations come through exposure to repeated patterns. The more often the association is made, the stronger the association becomes. Ellis & Schmidt (’97) supported Larsen-Freeman’s suggestion, finding frequency effects for the acquisition of L2 morphology.

In the case of L2 acquisition, associations may be already present, i.e., the L1 is already in place, and its associations can possibly interfere with the establishment of an L2 network. Sokolik (’90) suggested that, as a function of age, learners are less able to establish association patterns (experiment on French noun gender).

The Competition Model (Bates & MacWhinney ’82) was developed to account for the ways monolingual speakers interpret sentences. It is based on the assumption that form and function cannot be separated, and it is more concerned with performance rather than competence. A major concept of the theory is that speakers must have a way to determine relationships among elements in a sentence. Language processing involves competition among cues, each of which contributes to a different sentence interpretation. Although the range of cues is universal, they have language-specific characteristics, such as the strength assigned to each.

Native speakers of English use various cues to determine that the cows (in the cows eat the grass) is the subject of the sentence and that the grass is the object. A major cue is word order. Then, knowledge of the meaning of lexical items and animacy criteria are used to establish grammatical relationships. Finally, morphology (subject-verb agreement) contributes to interpretation. Other languages such as Italian resolve the problem of interpretation by using different cues. Word order assumes a lesser role, and morphological agreement, semantics, and pragmatics assume greater importance in interpretation.

A number of studies have been conducted using this paradigm. One of the findings is that under certain circumstances (semantics is in contradiction with grammatical cues), a meaning-based comprehension strategy takes precedence over a grammar-based one. Although the tendency of learners to adopt a meaning-based strategy is strong, there is also ample evidence that learners first look for those cues that are dominant in their L1. Only when that appears to fail (learners realize the mismatch between L1 and L2 strategies), do they adopt what might be viewed as a universal prepotency: using meaning to interpret sentences.

The research conducted within the Competition Model suggests that learners are faced with conflicts between L1 and L2 concerning the kind of cues that are relevant and their strength. They resort to a universal selection of meaning-based cues as opposed to syntax-based cues before gradually adopting the appropriate L2 interpretation strategy as their proficiency increases. The more reliable a cue (word order in English), the easier it is to be learnt. Effects of proficiency: there is greater or lesser dependence on case-marking cues depending on proficiency levels (Sasaki 1997).

Processing approaches

Characterized by a concern with processing mechanisms and capacities of the human brain that deal with the context of L2 acquisition.

Processability Theory

The main concept is that of the linguistic processor. In its simplest form, PT proposes that production and comprehension of the L2 forms can only take place when the linguistic processor is ready to handle them. This model makes strong predictions of word order development; indeed, Clahsen (’84) proposes three processing mechanisms:

• Canonical order strategy: Early learners generally use a single basic word order (e.g., SVO in Eng). Elements do not interrupt this sequence.
• Initialization/finalization strategy: Movement takes place, but elements will be moved to the initial and/or final position of the sentence. First and last positions are more salient, helping both processing and memorization.
• Subordinate clause strategy: When movement is learned, it happens in main clauses before it does in subordinate clauses. Subordinate clauses are processed differently (involves holding material in memory without a complete semantic analysis).

These processing strategies are claimed to account for the acquisition order of English questions, which require movement.

Developmental stages of English question formation:

• Stage 1: Single units (What?)
• Stage 2: Canonical word order with question intonation (Your cat is black?)
• Stage 3: Wh-fronting (Where the cats are?)
• Stage 4: Pseudo inversion – y/n questions, SV inversion in wh-questions (Have you got a dog? Where is the cat?)
• Stage 5: Auxiliary and modal verbs in II position (Where do you live?)
• Stage 6: Can inversion, negative questions, tag questions (Can you see? Doesn’t your cat look black? Isn’t it?)

Processability Hierarchy (Pienemann ’99): ordering of procedures developed by learners to process sentences.

• No procedure (single word utterances)
• Category procedure (e.g., adding plural morpheme to nouns)
• Noun phrase procedure (e.g., marking gender on both article and noun)
• Verb phrase procedure (e.g., movement of elements within a VP)
• Sentence procedure (e.g., SV agreement)
• Subordinate clause procedure (e.g., using tenses according to tenses in the main clause)

The hierarchy is implicational (one procedure is a prerequisite for the next). Even though these are universal procedures, learners can create individual solutions to processing limitations. For example, they can produce a range of forms to avoid movement of elements.

Information processing: automaticity, restructuring, and U-shaped learning

McLaughlin (’90) noted two concepts that are fundamental in L2 learning: automaticity and restructuring. Automaticity is the control over one’s linguistic knowledge. In language performance, one must bring together skills from perceptual, cognitive, and social domains. The more each of these skills is routinized, the easier becomes their use. Restructuring refers to changes made to internalized representations as a result of new learning. Integrating new information into one’s developing L2 system requires changes to parts of the existing system, thereby restructuring, or reorganizing, the current system.

An underlying assumption of this theory is that humans have a limited capacity for processing. Central to the ability to process information is the ability to attend to, deal with, and organize new information. The more we handle automatically, the more attentional resources are available to process new information.

Acquisition is seen as involving complex skills, much like learning to play tennis or the violin. From this perspective, the role of automaticity assumes great importance. When there has been a consistent and regular association between some kind of input and some output pattern, automatization may result, i.e., an associative connection is activated. This processing becomes fast, unconscious, and effortless. (The conversational routine is so automatic in a language one knows well that most people have had the experience of responding ‘fine’ before the question is even asked)

Crookes (’91) discussed the significance of planning and monitoring one’s speech. At the level of planning, the learner makes a ‘decision’ about what to say and what structures to use. That is, a learner has some choice over the structures that will be practiced, and the decision about what to practice is crucial in determining ultimate automatization. There is empirical evidence to support the benefits of planning in affecting the complexity of the discourse (Crookes, Ellis). Planning an utterance leads to the ability to use more complex language, which in turn can lead to the automatization of complex language and to the ability to plan even more complex language.

Monitoring is the activity of paying attention to one’s speech. In monitoring, learners note the successful use of a form, enabling them to use it in the following conversation. The learned system serves as a Monitor for the output of the acquired system. The Monitor cannot be used at all times. There are three conditions that must be met: learners need:

• Time to think consciously about the rules of grammar.
• Focus on form
• Know the rule

The Monitor is claimed to be used in production only, since its task is to edit utterances, but there is no absolute criterion for determining when it is in use and when it is not. Controlled processing is another mechanism of language use. Here, associations are not built up by repeated use but thanks to attentional control. Thus, one would expect a slower response. The distinction between controlled and automatic processing is the lack/presence of routinization and the creation of associations in long-term memory.

In this view, SLA takes place by the initial use of controlled processes. With experience in specific linguistic situations, learners begin to use language automatically, letting them concentrate attentional resources on new information that requires more control. The role of attention is intimately connected to automaticity in that when information use is automatic, less attention is paid to each action.

Restructuring takes place when qualitative changes occur in a learner’s internal representation of the L2. For restructuring to occur, the insertion of a new element into the system is not enough. The new element has to disturb the existing system to make reorganization necessary and let restructuring take place.

U-shaped learning: U-shaped patterns reflect three stages of linguistic use. In the earliest stage, a learner produces some linguistic form that conforms to target-like norms (error-free). At stage 2, the learner appears to lose what he or she knew at stage 1 (deviation from TL norms). Stage 3 looks like stage one in that there is again correct TL usage. Example: Lightbown 1983, French learners of English – use of the ing form; stage 1 He is taking a cake; stage 2 He take a cake; stage 3 He is taking a cake.

Input processing

[Section to be added]

Knowledge types

Description in terms of general cognition of how linguistic knowledge is acquired and organized in the brain.

Acquisition-learning

The Monitor Model (Krashen 70s) is characterized by five basic hypotheses:

• The acquisition-learning hypothesis
• The natural order hypothesis
• The monitor hypothesis
• The input hypothesis
• The affective filter hypothesis

Krashen assumed that learners have two independent means of developing knowledge of a SL: acquisition and learning. Acquisition is a process similar, if not identical, to the way children develop the ability in their L1. It is a subconscious process: learners are not aware of the fact that they are acquiring language but are only aware of the fact that they are using the language for communication. The resulting acquired competence is also subconscious because we are not aware of the rules of the language we have acquired, but we have a ‘feel’ for correctness. Acquisition is ‘picking up’ a language.

Learning is a conscious knowledge of an L2. Learners are aware of the rules and are able to talk about them. The two systems are used for different purposes: the acquisition system generates utterances because learners focus on meaning, not on form. The learned system serves as an ‘inspector’ of the acquired system. It checks to ensure the correctness of the utterance.

Declarative/Procedural

Declarative knowledge is knowledge about something, and declarative memory underlies the learning and storage of facts. With regard to language, declarative knowledge relates to word knowledge, such as collocation, pronunciation, meaning. This information is generally accessible to conscious awareness. Procedural knowledge or procedural memory relates to motor and cognitive skills that involve sequencing information (e.g., playing tennis). Using language is thought to involve procedural knowledge, especially for syntax. PK is relatively inaccessible.

The difference between the two has important consequences for learning because it is thought that, with age, the ability to use procedural knowledge and memory decreases, and older L2 learners need to rely more on declarative knowledge in learning.

Representation and control

Knowledge representation is the level of analysis and mental organization of linguistic information. Control over that knowledge is the speed and efficiency with which that information can be accessed. In many instances, learners use prefabricated patterns or language chunks with no internal analysis. They enable learners to express functions they are yet unable to produce through their linguistic system. They are simply stored. Determining the components of a chunked phrase allows the learner to use those parts in other linguistic contexts.

Interfaces

[Section to be added]

Psycholinguistic constructs

Attention: the concentration of the mental powers upon an object (the American Heritage Dictionary).

There are a number of approaches to attention and its relationship to awareness. One of the early treatments of attention in SLA comes from Tomlin & Villa (’94), who propose three components to attention:

• Alertness – readiness to receive stimuli
• Orientation – direction of resources to stimuli
• Detection – registration of stimulus

Detection is the major component that drives learning, while the other two are supports. Schmidt (’90), Noticing Hypothesis: awareness (through attention) is necessary for noticing, which in turn is essential for learning. The learner will begin to acquire language features that they have consciously noticed in input.