Variation in Interlanguage

Tarone, Elaine (1988) Perspectives on Systematicity in Language Ethnographic Approach Rationalist Approach Systematicity is to be found in the minds of speakers (i.e. in their competence) and performance data are viewed as an imperfect reflection of competence (slips, false starts, memory overloads). Knowledge of language (competence) is accessible through introspection. Any discrepancy between the model based upon the speakers introspections and that speakers utterances are dismissed as performance errors. Function-Form Approach It tries to describe and explain how form and function are systematically related when language is used in meaningful communication in social context (i.e. how language form functions to signal social class, gender, social identity, gender, etc.). The Language of Second Language Learners Systematicity A transfer view of systematicity In 1940s through 1960s, the L2 learners language was thought to be primarily influenced by their NL. Lado (1957) proposed CA between NL and TL to find areas of difference, assumed to be also the areas of difficulty for the learners. Stockwell and Bowen (1965) emphasized the role of transfer and proposed a hierarchy of difficulty (whereby differences between languages could be assessed in terms of their degree of difficulty for learners). Anecdotal approach: Scholars hardly engaged in systematic observation and study of language produced by L2 learners and instead cited anecdotes i.e. sample sentences overheard in the classroom or remembered from student writing. Systematicity is essentially that of the learners NL, which is then gradually substituted by the TL system. A Chomskyian view of interlanguage systematicity Richards (1971) presented long lists of errors observed in learner language and classified them into types (thereby showing systematicity). He proposed several causes for these errors, only one of which was NL transfer. Corder (1971) proposed the term idiosyncratic dialect for learner language to capture the idea that it is systematic because it has its own rules. He proposed the use of both learners judgments of grammaticality and observation of learners communicative performance. Selinker (1972), who first proposed the term interlanguage for learner language, opposed the use of grammaticality judgments as data on IL, arguing that learner judgments tap a different linguistic system from the one underlying IL. A Labovian approach to interlanguage systematicity A function-form view of interlanguage systematicity In a longitudinal study of a Japanese ESL learner, Hakuta (1974, 1976) demonstrated that decreases in the use of one form were often accompanied by increases in the use of another form which took over some of the first forms functions in the IL. Models of Explaining Interlanguage Variability Inner Processing Theories (Psycholinguistic Approach) They seek to explain interlanguage variation primarily as a result of psycholinguistic processes.

Chomskyian Model Variation is seen to result from performance error: during performance in situations where the learner lacks the necessary rules, features, or items to express a particular meaning, the learners competence is thought to become permeable, i.e. the IL is either penetrated by rules or items from the NL, or by overgeneralization or distortion of an IL rule or form. This position is contradicted with empirical findings: Tarone (1983) found more influence from NL (and thus more permeability) in learners introspective data than in their casual speech performance. In like manner, Liceras (1987) data obtained from a translation task were more target-like than the same learners intuitional data. If performance (as seen in the translation task or the casual speech) is only a poor reflection of competence (as represented best in intuitions), we might expect to find just the opposite. Monitor Model Variability is considered to be due to the use or non-use of the Monitor (the learned metalinguistic knowledge system as opposed to the unconsciously acquired implicit knowledge system). Psychological Processing Models This approach makes a distinction between language knowledge and the processes used to implement that knowledge in communicative performance. For example, McLaughlin (1978, 1987) suggests that when a L2 learner can use a TL form in simple tasks, but not in communicative tasks, the learner is producing that form by means of controlled processes. At the point where the form appears in complex communicative tasks as well, McLaughlin suggests that the form is now being produced automatically. Labovian Models These are based upon William Labovs work on the style-shifting patterns of native speakers. He lists five methodological axioms relating to the study of speech styles: 1. Style-shifting. There are no single style speakers. All speakers vary their language to some extent as the social situation or topic changes. 2. Attention. Speech styles of a speaker can be arranged along a single continuous dimension defined by the amount of attention paid to speech. 3. Vernacular. The style in which minimum attention is given to monitoring speech is called the vernacular. It is the style associated with informal everyday speech and it is in this style where the most regular and systematic data can be found for linguistic study. 4. Formality. When a speaker is systematically observed, a formal context is thereby defined, and the speaker pays more than the minimum amount of attention to speech. 5. Good data. The best way to obtain good data on the speech of language users is through systematic observation as in an individual interview: a formal context. The conflict between the fourth and the fifth axioms leads to what Labov called the Observers Paradox. Good data requires systematic observation but this prevents access to the users vernacular style. Implicit in these guidelines, is a theory on the cause of variation: attention is seen as the mechanism through which other factors can affect style. In an attempt to relate that theory to IL data, variationist researchers working in a Labovian paradigm have considered attention as the cause of task-related variation (i.e. synchronic variation, as opposed to diachronic variation, which is the variation due to change in the learners knowledge of language over time). Attention causes the learner to produce a style which is close to their formal norm; this norm may contain more TL forms, but may also contain elements from the learners NL formal norm. As less attention is paid to language form, more vernacular IL forms will be produced; these forms are more systematic, in

the sense that they are more consistent with the interlanguage norm and less influenced by the TL or NL norms. A grammaticality judgment task provides information about only the more formal, or careful style; there is no sense in which this style underlies other styles. Other tasks, like the reading of word lists or connected texts, the combining of sentences, the description of entities, or narration of a story, must be assumed to tap different styles of the IL continuum. Variability in IL is caused by style-shifting along this IL continuum, which in turn is caused by variable shifts in the degree of attention which the learner pays to language form. (pp. 40-41) One problem with the Labovian model is that the psychological process of attention is very difficult to verify independently. Individuals attention may waver from one minute to the next during one single task. Labov has suggested that one possible solution to this problem of operationalizing attention to speech is to watch for channel cues: when the vernacular style is being used (i.e. when least attention is being paid to language form), speech rate increases, loudness increases, laughter and joking comments increase, and pausing and hesitation are reduced. Researchers have observed a correlation between such channel cues and the occurrence of an increased number of vernacular variants. But of course, attention to form is still inferred, not observed, here. Another problem is that attention can only be an intermediary, not an explanatory, factor. This is because even if we postulate a psychological process called attention, we are still left with establishing what it is in the task and in the situation that causes learners to pay attention to speech and therefore to style-shift. (Planning)

Sociolinguistic and Discourse Theories Social-Psychological Models Speech Accommodation Theory (SAT) Interlanguage variation is explained as style-shifting due to two accommodation strategies: adjustments or modifications which make ones speech become either more similar or less similar to the speech patterns of the interlocutor (convergence vs. divergence). Speakers may converge toward speakers who are not physically present (e.g. an absent reference group), rather than toward their actual interlocutor. For example, they may shift toward the patterns typical of their own reference group, or towards a prestige norm the interlocutor is believed to value (though he/she may not actually use it). Accommodation Theory is preferable to the Labovian Model in several respects: 1. It focuses upon determinable social-psychological factors like multiple group membership and identity assertion as explanatory, end causes of IL variation, rather than postulating an unobservable intermediary process of attention. 2. It allows the researcher to identify the origin of the variants used. For example, variants from the NL may be used by a learner to signal ethnic identity and group solidarity. 3. The Labovian Model focuses only upon shifts in the degree of standardness (i.e. whether a speaker uses a standard or a non-standard linguistic form). SAT allows the researcher to also study shifts in amount of talk, speech rate, duration, pauses and utterance length, stress, pitch, intonation, and even in the content expressed. One criticism against SAT is that an interlocutor effect cannot adequately explain all facts of IL variation such as variation due to linguistic environment (i.e. the effect of one form on another in the same utterance), learners limited repertoire, learners previous language experience, communicative demands of different genres.

Function-Form Models The problem with all inner processing theories is that they fail to propose empirically verifiable causes of IL variation. For example, the Monitor model and McLaughlins model used amount of time allowed for task as an indirect measure of Monitoring or automaticity, and the Labovian model used channel cues as an indirect measure of degree of attention. While amount of time allowed for task can indeed be shown to influence variation, there is no empirical evidence allowing us to determine whether the underlying inner process involved is Monitoring, automaticity, or degree of attention paid to form. Sociolinguistic models focus upon external observable phenomena as causes of variation, but they fail to provide any comprehensive account for all known facts of IL variation.

Causes of systematic variation Linguistic environment Ellis (1988) provides evidence that learners acquire forms first in one, most favorable, linguistic context, and then gradually extend this form into progressively less favorable contexts. Psychological processing factors Krashen (1979) believed that time and a focus on form are sufficient conditions for Monitoring to occur. The use of Monitor, in Krashens view, is considered to be the cause of variability in grammatical accuracy. Krashens position is criticized for being a post hoc analysis: there is nothing in the task itself or in the observable elicitation situation which can independently be shown to force subjects to focus on from. Dickerson (1974) found evidence suggesting that attention to speech causes the use of more target-like variants. In contrast, Beebes (1980) results showed that attention to speech may not always result in more TL variants; it may at times result in more of the prestige NL variants as well. Interlocutor As a result of studying eight Spanish-speaking child ESL learners in a bilingual classroom, Cathcart (1983) found that age and authority relationships of the subjects interlocutors influenced the relative frequency of control behavior and information-sharing behavior. So, for example, if the interlocutor was an adult, there was more information-sharing (announcing, expressing, requesting), whereas if the interlocutor was a child, there were more control acts (initiating, supporting, responding). In adultcontrolled situations, as in ESL classes, where the learners were not in a position of authority or social power, their language consisted of words and short phrases with formulaic chunks. In child-controlled situations, such as recess, there was a wide variety of communicative acts and syntactic structures, and there were more control acts. Topic Shifts of topic in conversation, for example, from a discussion of the weather to a description of a situation in which the speakers life was in danger, can cause a learner to style-shift. Also a learner who is forced to talk about an unfamiliar or complex topic in an L2 may produce disjointed language and more inaccurate grammatical forms than when conversing about more familiar and simpler topics. (p. 119)