Input Processing: A Study of Ab Initio Learners with Multilingual ...

14 downloads 0 Views 280KB Size Report
Teachers College, Columbia University, New York, USA. Research on ... rested largely on the assumption that learners use a meaning-based approach as the.
Input Processing: A Study of Ab Initio Learners with Multilingual Backgrounds ZhaoHong Han and Stephen T. Peverly Teachers College, Columbia University, New York, USA Research on input processing in the acquisition of a non-primary language has rested largely on the assumption that learners use a meaning-based approach as the ‘default’ when processing input (VanPatten, 1996). The study reported here poses a challenge to this assumption: findings show that participants who were absolute beginners used a primarily form-based approach when processing Norwegian, a language they had not been exposed to previously. We argue that when positing principles of input processing, there is a need to differentiate between learners who have and who have not developed intermediate grammars of the target language, and that input which is linguistically incomprehensible, as well as devoid of extralinguistic clues, induces form-based processing. The paper concludes with two hypotheses: (1) learners who have acquired some knowledge of the target language will adopt a meaning-based approach to input processing; and (2) learners who have no existing knowledge of the target language will adopt a form-based approach. doi: 10.2167/ijm052.0 Keywords: input processing, Norwegian, meaning primacy, beginning learners, multilingual acquisition, attention

Introduction Much of our understanding of input processing in the acquisition of a nonprimary language is driven by the assumption that human beings have a limited and finite capacity for information processing (Baddeley, 2001; Engle, 2002; Ericsson & Kintsch, 1995; Kintsch, 1998). This assumption, in turn, makes three predictions: (1) information processing is selective; (2) individuals can process two different types of information simultaneously and effectively if the processing of one of the information types is automatised and requires little, if any, conscious attention; and (3) simultaneous processing of two different types of information that are not automatised can lead to inadequate processing of either or both types of information, and to a ‘trade-off’ effect. One issue, therefore, that has attracted the attention of researchers is whether learners are able simultaneously to process meaning and form in meaning-bearing input. Many researchers have argued, following Sharwood Smith (1986), that form and meaning are two distinct aspects of input which are processed differently via different cognitive mechanisms (e.g. Doughty, 1991; Faerch & Kasper, 1986; Gass, 1988; Klein, 1986). This proposition appears to have gained a great deal of empirical support from three major findings: first, learners do not process meaning and form at the same time; they process input for meaning, not form (e.g. Lee, 1998); second, learners process input for form and meaning dichotomously (e.g. Hulstijn, 1989; Lee, 2002; VanPatten, 1479-0718/07/01 017-21 $20.00/0 International Journal of Multilingualism

– 2007 Z-H. Han & S.T. Peverly Vol. 4, No. 1, 2007

17

18

International Journal of Multilingualism

1990); and third, learners can simultaneously process meaning and form only when form contributes to meaning (e.g. VanPatten, 1990, 1996). This general scenario, it should be noted, has emanated almost exclusively from studies of learners who had some knowledge of the target language. That is, learners in these studies were able to draw on their existing (albeit limited) linguistic and non-linguistic knowledge (e.g. Hudson, 1982) when processing input. Thus, one can argue that the emphasis on the primacy of meaning has, at least in part, been based on learners who have already developed some knowledge of the target language. If so, one would, naturally, like to know if this approach would also hold for learners who have no relevant linguistic knowledge. We ask the following question: Do beginning learners use a meaning-based approach as the default when processing input of a new language? The study reported here was an attempt to evaluate this issue. Through a sequential versus simultaneous design, subjects were exposed, for the first time, to written input in Norwegian, and were probed for the content of their processing. The findings do not support meaning primacy. In the following sections, we first review the relevant theoretical and empirical research that served as the background for the study. We then report on the method of the study and results. Next follows a discussion of the findings. The paper concludes with hypotheses for further exploration of input processing by ab initio versus en route learners.

Theoretical Background The present study derived its theoretical impetus mainly from the input processing models proposed by VanPatten (1996) and Sharwood Smith (1986). VanPatten (1996): Meaning primacy and limited capacity Predicated on the theoretical assumption that human beings have a limited information processing capacity, VanPatten’s model posits a number of principles, displayed in Table 1, which characterise the natural processing tendencies of (beginning) learners (VanPatten, 1998) of a non-primary language, such as a second or third language. The first principle focuses on meaning primacy, which has six subcomponents (1af), such as the precedence of content words over functors (1a) and the reliance on lexical items rather than grammatical form to infer meaning (1b). The first four sub-principles (1ad) are interrelated and are linked to a preference for meaning over form, while 1(e) and 1(f) speak to the constraints on processing. 1(e) stipulates the need for the availability of sufficient resources before any type of processing, meaning-oriented or formoriented, can occur; 1(f) indicates a preference for perceptually salient forms over less salient forms. The second principle suggests that learners have a general tendency to associate the first noun with the grammatical function of subject and the semantic role of agent. The principle has three corollaries (2ac), viz., (a) a reliance on lexical means of expression rather than syntactic structure,

Input Processing

19

Table 1 Principles of input processing Principle 1. The Primacy of Meaning Principle Learners process input for meaning before they process it for form Principle 1a. The Primacy of Content Words Principle Learners process content words in the input before anything else Principle 1b. The Lexical Preference Principle Learners will tend to rely on lexical items as opposed to grammatical form to get meaning when both encode the same semantic information Principle 1c. The Preference for Non-redundancy Principle Learners are more likely to process non-redundant meaningful forms before they process redundant meaningful forms Principle 1d. The Meaning-Before-Non-meaning Principle Learners are more likely to process meaningful grammatical forms before nonmeaningful forms irrespective of redundancy Principle 1e. The Availability of Resources Principle For learners to process either redundant meaningful grammatical forms or nonmeaningful forms, the processing of overall sentential meaning must not drain available processing resources Principle 1f. The Sentence Location Principle Learners tend to process items in sentence initial position before those in final position and those in medial position Principle 2. The First Noun Principle Learners tend to process the first noun or pronoun they encounter in a sentence as the subject or agent Principle 2a. The Lexical Semantics Principle Learners may rely on lexical semantics, where possible, instead of word order to interpret sentences Principle 2b. The Event Probability Principle Learners may rely on event probabilities, where possible, instead of word order to interpret sentences Principle 2c. The Contextual Constraint Principle Learners may rely less on the First Noun Principle if preceding context constrains the possible interpretation of a clause or sentence Source: VanPatten (2004)

20

International Journal of Multilingualism

(b) a reliance on world knowledge rather than syntactic structure, and (c) a reliance on linguistic context rather than syntactic structure. In summary, lexical and non-linguistic knowledge are much more important than morphosyntactic knowledge to the processing of input. The assumption and notion that learners have a limited capacity information processing system suggest that the processing of input for meaning and form can be separate and sequential; ‘learners can do only so much in their working memory before attentional resources are depleted and working memory is forced to dump information to make room for more (incoming) information’ (VanPatten, 2000: 45). This has, indeed, been confirmed by findings from empirical studies. By orienting learners to (1) form, (2) meaning, and (3) form and meaning, respectively, researchers have been able to demonstrate that learners’ processing of input is largely determined by what they attend to. In other words, if they pay attention to form while processing input, form is retained; if they pay attention to meaning, meaning is retained, and so forth. By way of illustration, we describe briefly the often cited studies of Hulstijn (1989) and VanPatten (1990). Modular processing In a study consisting of two experiments, Hulstijn (1989) investigated whether or not attention to meaning is sufficient for the implicit and incidental acquisition of the formal properties of a second language. In the first experiment, 145 intermediate learners of Dutch were randomly assigned to the Form Group, the Meaning Group, or the Form and Meaning Group. Using a pretest-treatment-post-test design, the subjects were pre-tested, via a sentence copying test, on their knowledge of a complicated structure in Dutch, put through a learning task in which they were oriented, respectively, to form and/or meaning, and finally, post-tested via a cued recall task and another copying task. The hypotheses tested were: (1) In the cued recall test, Form Group Ss will perform better than Meaning Group Ss on the formal aspects of the target sentences (function words and word order). Meaning Group Ss, however, will perform better than Form Group Ss on the semantic aspects of the target sentences. (2) On the sentence copying test (the second one), Form Group Ss will perform better than Meaning Group Ss, since retaining a long sentence in memory requires encoding its grammatical structure. Results supported the first hypothesis, but not the second. The researcher attributed this to a lack of control over subjects’ prior linguistic knowledge. To control for the latter, a second experiment (n /80) was conducted, with the target linguistic forms replaced by artificial forms. This time, the pattern of results was quite similar to that of the cued recall test in the first experiment in that ‘the responses of the Meaning Group were better than those of the Form Group, when scored for content, and poorer when scored for grammatical structure’ (p. 63), thereby confirming the first hypothesis. However, different results were obtained from the second copying test: the Form Group Ss, as predicted by the second hypothesis, did indeed perform better than the

Input Processing

21

Meaning Group Ss. In summary, the study provides evidence that (a) for implicit and incidental learning of form to take place, attention to form is sufficient, and (b) exclusive attention to meaning may inhibit the acquisition of form. Taken together, the evidence leads to the interpretation that when learners’ attention is directed to form, they can attend to both form and meaning, but not vice versa. For our purposes, we should also note that in both experiments, the Form and Meaning Group outperformed the Form Group when encoding form as well as meaning. This latter result, at first sight, seems to suggest that simultaneous processing of form and meaning is not only possible, but superior to sequential processing. However, as the review of VanPatten (1990) below will show, this conclusion does not hold up when linguistic characteristics are treated as an independent variable. Sequential as opposed to simultaneous processing VanPatten (1990) examined whether learners can consciously attend to both meaning and form when processing input. Two hundred and two (n /202) university students who were learning Spanish, representing three levels of proficiency, were randomly assigned to one of four experimental conditions: (a) attention to meaning alone, (b) simultaneous attention to meaning and an important lexical item, (c) simultaneous attention to meaning and a grammatical functor, or (d) simultaneous attention to meaning and a verb form. Results, inter alia, revealed: [C]onscious attention to form in the input competes with conscious attention to meaning, and, by extension, that only when input is easily understood can learners attend to form as part of the intake process. (VanPatten, 1990: 296) This appears to suggest that processing meaning-bearing input for form presupposes that the meaning of the input has been comprehended (see Principle 1 (e), Table 1) and that the former, when occurring, is sequential to the latter. VanPatten further differentiates between forms of high communicative value and low communicative value, noticing that simultaneous conscious attention to informational content and low communicative form in the input is difficult for the early and the intermediate stage learner. Communicative value in this context refers to ‘the relative contribution a form makes to the referential meaning of an utterance and is based on the presence or absence of two features: inherent semantic value and redundancy within the sentencestructure’ (VanPatten, 1996: 24). The two variables (i.e. semantic value and formal redundancy) may then enter into different combinations with each other to determine the communicative value of a given linguistic form as follows: a form that has inherent semantic value (i.e. /semantic) and that is not redundant (i.e. /redundant) has high communicative value; conversely, a form that has semantic value (i.e. /semantic), but is formally redundant (i.e. /redundant), has low communicative value; and so forth. An example of the latter would be the English third person singular  s. This form undoubtedly has semantic content in that it encodes the semantic notion of third person

International Journal of Multilingualism

22

singular, as well as the temporal frame within which the action occurs. Yet it is made redundant by the co-occurrence of lexical items (e.g. he, everyday) that express the same meanings. Moreover, syntactically, the English canonical subject-verb word order renders the  s redundant, for the notion of ‘third person singular’ is already carried by the subject. Hence, the form is of low communicative value, and as such, has low perceptual salience. Linguistic forms of this type are late developing. As VanPatten explains: Grammatical form conveying semantic information that also is encoded lexically will tend to not be detected; the learner instead relies on the lexical items for the semantic information. The learners’ internal mechanisms will detect grammatical form early on only if it is relatively high in communicative value. Otherwise, grammatical form is detected over time only as the learner’s ability to get meaning from the input is increasingly automatised (becomes more effort-free). This increasing automatisation of comprehension releases attentional resources for the processing of form that was previously skipped (undetected). (VanPatten 1996: 30) Notice that, according to VanPatten (1996), processing input for forms of low communicative value is predicated on attentional resources being released as a result of the automatisation of comprehension. This supposition, though intuitively appealing, can nevertheless be factually mistaken. Indeed, as the study of VanPatten (1990: 295) has shown, ‘as input becomes comprehensible, available attention and effort are not necessarily released for focusing on form’, suggesting that even when learners have attentional resources available to perform form-based processing of input, they will not automatically do so. This reluctance to perform morphosyntactic processing, for some researchers, does not result from lack of attentional resources, but from adult learners’ drive to give priority to meaning, as stated previously, and more importantly, from the nature of communication. A salient characteristic of communication, as Skehan and Foster (2001) have aptly noted when assessing the contribution of negotiation for meaning to non-primary language acquisition, is that communication is effective even when morphosyntax is far from perfect, because many grammatical inflections are redundant and context can often substitute for syntax. Applied to input processing, this means that meaning can be decoded by using contextual information and current linguistic knowledge. This fact alone is sufficient to not motivate learners to process input for form, in particular, form of limited or no communicative value. Thus, as an example, it makes no difference if the input is ‘Jack is a college student’ or ‘Jack a college student’. The trade-off phenomenon The trade-off or competition between a focus on meaning and form during input processing has not only been supported by the two studies reviewed above, but by other research as well (Lee, 1998, 2002). Building on previous findings, Lee (1998) tried to tease apart the structural characteristics of the trade-off, and subsequently offered important new insights:

Input Processing

23

When naturally attending to meaning, learners: (a) comprehend better when the forms in the input are less complex than when they are more complex, (b) detect the forms in the input, and (c) comprehend informational content and process the forms in the input via different processes. (Lee, 1998: 42) The suggestion here is that comprehension is affected when the forms are complex. Only when the forms are not complex is simultaneous processing of meaning and form possible. In the latter, Lee further notes that even among functors (e.g. morphological forms), some naturally lend themselves to being noticed while the focus is on meaning. Processing input for form and for meaning, he nevertheless contends, are based in different processes. As he puts it, ‘the good comprehenders are not necessarily the good input processors. Likewise, the poor comprehenders are not necessarily the poor input processors’ (p. 41). One remaining question, however, is: what processes underlie input processing for meaning versus form, a question to which we now turn. Sharwood Smith (1986): The dual relevance of input According to Sharwood Smith (1986), language input has dual relevance such that: At moment X, the learner’s main aim may be to extract meaning and survive or succeed in a given interchange of messages. In this case, only those aspects of input which will aid the learner in this communicative endeavor (linguistic or otherwise) will be relevant in this first sense: the learner will interpret for meaning. At the same time, there will be linguistic input which is relevant to the current state of the learner’s competence. It may contribute to the substantiation or reflection of some current hypothesis about the target language system: the acquisition device will ‘interpret’ for acquisition. (Sharwood Smith, 1986: 243; emphasis original) In this view, language input may potentially serve two purposes: extraction of meaning for a communicative purpose (i.e. processing input for meaning) and extraction of morphosyntactic information for the purpose of building a linguistic system (i.e. processing input for form). The learner, as implied in the above model, can decide what purpose s/he wants the input to fulfill, and as such, s/he is in control not only of what knowledge and attentional resources to invoke but where to deploy them. Figure 1 illustrates the processes that underlie the dual relevance of input. In Figure 1 the double arrows represent the processing of linguistic input, and the single arrows the learner’s cognitive contributions to that input. As illustrated, the learner can derive meaning in two ways, by (1) processing nonlinguistic input, which presumably draws on prior world knowledge (cf. Ellis, 1994; Faerch & Kasper, 1986; Gass, 1988); and (2) performing surface structure analyses, which draws on prior linguistic knowledge. In the latter case, Sharwood Smith explains, memory for surface structure can be short-lived; that is, ‘surface structure is just briefly registered, recoded for meaning alone

International Journal of Multilingualism

24 LINGUISTIC INPUT

Surface structure

Semantic representation

LANGUAGE ACQUISITION MECHANISMS NON-LINGUISTIC INPUT

TOTAL MEANING REPRESENTATION

Meaning representation

Figure 1 Input processing and dual relevance (reproduced from Sharwood Smith, 1986: 250)

and then discarded forthwith, [and] the acquisitional mechanism would have little or no access to crucial information for the further development of [grammatical] competence’ (Sharwood Smith, 1986: 250). For acquisition to occur, on the other hand, ‘the surface structure must first be held in an acquisition-specific memory store and not simply discarded after semantic processing has taken place’(p. 251). Only when such a cognitive condition obtains will the acquisitional mechanisms perform the series of processes listed below: (1) compare the semantic representation (derived purely from current competence) with the total meaning representation (semantic representation plus meaning derived via other means like real-world knowledge . . .) and note any discrepancy; (2) adjust the semantic representation to fit the facts where a discrepancy is noted, i.e. where current competence has apparently generated a semantic representation that is in violation of the facts of the situation; (3) generate a surface structure from the adjusted semantic representation according to the rules of the current grammar; (4) compare the original surface structure with the new surface structure (in 3) and note any discrepancy; and (5) restructure the current competence system (grammar) so that the adjusted semantic representation may be derived from the original surface structure, if there was indeed a discrepancy (in 4) (Sharwood Smith, 1986: 251). One thing that is clear from this model is that input processing for either purpose is an attention-demanding process. It is, therefore, not surprising that when the two co-occur, they lead to trade-offs and compromises, as mentioned above. Selective attention Long ago research on attention established that human beings have a limited attentional capacity when engaged in conscious, effortful processing (Broadbent, 1975). As a result, individuals’ attention is selective, i.e. they can attend to only one form or source of information at a time (for recent review, see Robinson, 2003). If they must attend to two or more sources, they do it sequentially, by switching back and forth between sources of information.1

Input Processing

25

There are two primary views on why some individuals use their limited capacity systems more effectively than others. The cognitive resources point of view suggests that those with more cognitive resources (e.g. language and content knowledge) can attend to and process more information in a limited capacity processing system than those with fewer resources (Ericcson & Kintsch, 1995; Kintsch, 1998). The central executive view suggests that the critical feature of effective processing in a limited capacity system, regardless of resources, is the ability to control attention and to avoid distraction (Baddeley, 2001; Engle, 2002). These theories are directly relevant to input processing by absolute beginners. As they are novices, they have neither the level of cognitive resources nor the skills necessary to avoid distraction that more advanced individuals possess. The question we sought to address in this study is precisely: what do learners (novices) select to attend to on their own? The issues reviewed above underpin the motivation for the present study. The questions and hypotheses we investigated are: (1) What approach, meaning-based or form-based, do absolute beginners use when processing the target language written input for the very first time? We hypothesised that subjects would adopt a meaning-based approach to processing a reading passage, as has been argued by VanPatten (1990; 1996; cf. Skehan, 1998). A meaning-based approach, operationalised in line with VanPatten’s principles 1(ae) (see Table 1), is one whereby learners selectively rely on meaningful forms. A form-based approach, on the other hand, is one whereby learners rely on meaningful and non-meaningful forms, indiscriminately and irrespective of redundancy, in processing input. (2) Is sequential processing of meaning and form superior to simultaneous processing? Based on our previous discussion, we hypothesised that sequential processing of meaning and form would be superior to simultaneous processing in terms of comprehension and intake.

Method Subjects Twelve graduate students (age /25) at a university in the United States participated in the study. Eleven were female and one male. Each participant knew at least two languages. None, however, had had any prior experience with Norwegian, the target language. Table 2 presents a biographic sketch of the participants. The information was gathered by a questionnaire. Design and materials The 12 participants were equally divided into two groups: sequential versus simultaneous.2 The sequential group (henceforth the SQ group) was exposed

International Journal of Multilingualism

26

Table 2 Participants’ biographic information Participants

S1

Years of instruction Age of exposure to the first received in additional additional languages language

First language

Additional languages

Polish

English

28 years

Russian

8 years

German

2 years

S2

Korean

English

6 years

S3

English

French

10 years

Spanish

6 years

Japanese

5 months

English

6 years

French

2 years

S4

Turkish

S5

English

Chinese

1 year

S6

English/ Dutch

Spanish

4 years

German

A few weeks

French

A few weeks

Italian

A few weeks

11 13

10

11 From birth

From early childhood

S7

Chinese

English

20 years

8

S8

English

Italian

5 years

18

S9

Turkish

English

9 years

French

13 years

English

13 years

Chinese

6 months

French

6 years

Japanese

1 year

Thai

6 months

Spanish

3 months

American Sign Language (ASL)

1 year

English

6 years

S10

S11

S12

Korean

English

Japanese

10

5

15

13

Input Processing

27

to a reading passage twice, non-consecutively, with each exposure followed by one task to complete, while the simultaneous group (henceforth the SM group) was exposed to the reading passage twice, consecutively, and were then given two tasks to complete. The reading passage, totalling 125 words, was taken from Bo I Norge, an upper-beginner textbook written by Manne (1995) for learners of Norwegian as a second or foreign language. The text was a letter written by a college student for a ‘partner-seeking’ column in a local newspaper (Appendix 1). An idea unit analysis was carried out, resulting in the division of the text into 18 units. A more complete description of an idea unit is given in the next section. In this study, input processing was operationalised as the ability to mentally register the meaning (i.e. comprehension) and form (i.e. intake) of the passage, and was, accordingly, measured by two tasks. The first was a free written recall task (cf. Lee, 1986) which was used to tap learners’ comprehension of the text (see Appendix 2). For this task, the subjects were given the following instructions: ‘What is the reading passage about? In five minutes,3 recall as many facts as you can from the passage. You can write in English.’ Note that the subjects were encouraged to write in English, a language which they were all highly proficient in, rather than in Norwegian, the target language, of which they knew nothing. This, following Lee (1986, 2002), was a way to ensure that the subjects’ understanding of what they read would not be obfuscated by their lack of knowledge of the target language. The second measure of input processing was a fill-in-the-blank task (see Appendix 3). The purpose of this task was to probe the subjects’ intake, i.e. recognition of forms in the text. In this task, five sentences were taken out of the passage, each containing one missing word (e.g. a preposition, an adverb). The subjects were asked to fill in the missing word by choosing one of four options.4 At the end of the experiment, subjects were given a questionnaire to complete which had five questions, including one directed at the information that had been drawn on when processing the text. Table 3 summarises the design of the study and procedures. As shown in Table 3, the time for the experiment was kept constant for both the SQ and the SM group so that the data elicited would be comparable. At the onset of the experiment, the subjects were each given a paper folder which contained (1) an instruction sheet which provided a general orientation to the experiment,5 (2) the two tasks, and (3) the questionnaire. The reading Table 3 Design and procedures SQ group

SM group

Time

First reading of the text

First reading of the text

5 minutes

Task 1: Free written recall

Second reading of the text

5 minutes

Second reading of the text

Task 1: Free written recall

5 minutes

Task 2: Fill in the blanks

Task 2: Fill in the blanks

5 minutes

Questionnaire

Questionnaire

5 minutes

International Journal of Multilingualism

28

passage was made into a PowerPoint file and subsequently projected onto a large screen via an LCD. One of the researchers administered the readings and the test tasks to both groups concurrently, while controlling for the time for each task. Coding, scoring, and analysis The data from the two test tasks were coded by one of the researchers twice, and the intra-rater agreement was 0.97.6 Specifically, the recall protocols were coded for (1) correct idea units; (2) TL words identified without understanding of their meaning (henceforth Word [ /M], and (3) TL words identified with understanding of their meaning (henceforth Word [/M]). A correct idea unit, after Bransford and Johnson (1973), is a phrase or sentence, rather than a single word, which transmits information that matches part of the overall message of the reading passage. Examples are given in [1] and [2]. Each correct idea unit was scored 1 point. In addition, words that were singularly recalled were coded if they matched the surface form of any words in the reading passage. Each correctly identified word, with or without understanding of its meaning, was scored 1 point. Examples of Word [/M] and Word [/M] are given in [2] and [3]. Data from the fill-in-the-blank task were, on the other hand, scored dichotomously: each answer was scored 1 (correct) or 0 (incorrect). [1] Idea unit It’s about a 25 year old man. [2] Word [/M] antirasist /antiracist [3] Word [ /M] skull should SPSS 12.0 was used to compute the descriptive statistics and to run the analysis of variance (ANOVA). Protocol data from the questionnaire were analysed qualitatively.

Results Quantitative Table 4 displays the means and standard deviations for the participants’ performance on the free written recall and the fill-in-the-blank tasks. Table 4 shows that, overall, the mean scores were rather low for both the SQ and the SM groups, suggesting that they had considerable difficulty comprehending the text. This is best reflected in the mean total for idea units across groups, 2.92 out of 18, suggesting that only 16% of the text was understood and remembered. In comparison, subjects did better overall on the fill-in-the-blank task. This can be seen in the mean total of 2.34 out of 5, suggesting that, on average, they could recognise 46% of the forms tested. This

Input Processing

29

Table 4 Means and standard deviations for SQ versus SM groups Group

Idea unitsa SQ (n/6)

SM (n/6)

Word [/M] Word [/M]

Mean

2.0

2.5

1.16

2.5

SD

2.53

1.38

1.47

1.05

Mean

3.83

1.67

2.17

2.17

SD

1.6

.52

1.72

1.47

2.1

1.67

2.34

Mean total 2.92 a

Gap fillingb

Free written recalls

100%/18. 100%/5.

b

Table 5 SQ versus SM differences on the free-written-recall and the fill-in-the-blank task Groups

df

f

Sig.

Idea units

SQ versus SM

1

2.25

0.17

Word [ /M]

SQ versus SM

1

1.92

0.20

Word [/M]

SQ versus SM

1

1.17

0.31

Gap filling

SQ versus SM

1

0.2

0.66

result, however, should be interpreted with caution, for there were far fewer answers expected of the participants in the gap-filling task than in the free recall task. An ANOVA was used to evaluate the effect of the independent variable of group (sequential versus simultaneous) on the dependent variables: the ability to recall idea units and individual words, with (/M) and without (/M) an understanding of the meaning of the words. The results are displayed in Table 5. As shown in Table 5, there were no significant differences (p ]/0.05) between the two groups on any of the four dependent measures. However, the lack of differences does not mean an absence of difference. Closer inspection of the last two columns of Table 5 reveals that the two groups differed from each other on all four measures. The differences from greatest to smallest were: free recall of idea units, recall of individual words [ /M], recall of individual words [/M], and gap filling. Figure 2 gives a better visual display of the between-group differences. As seen in Figure 2, the SM group outperformed the SQ group in recalling idea units; the SQ group, on the other hand, did better than the SM group in their recall of individual words without understanding (Word [/M]); the SM group did better than the SQ group in their recall of individual words with understanding (Word [/M]); and the SQ group outperformed the SM group

International Journal of Multilingualism

30 4.5 4

Mean score

3.5 3 2.5

SQ

2

SM

1.5 1 0.5 0 Idea units Word [-M] Word [+M] Gap filling

Figure 2 Between-group differences

on the fill-in-the-blank task. Our qualitative analysis of participants’ protocols is presented next. Qualitative The retrospective protocols reveal use of rather similar strategies across the two groups; the participants all seemed to have tried to ‘crack the code’ (Cook, 1997) by (1) making connections between the new language input and a language they already knew to identify similarities, and (2) analysing the grammatical structure of a sentence. What is also worth noting is that much of the code-cracking process happened at the word level, as can be seen in the sample recall displayed in Table 6. S11’s recall, which was consistent with her protocol as solicited by the questionnaire, shows form-based processing predominantly; that is, the subject relied on familiar forms (e.g. numbers and cognates) to infer meaning, and analysed the structure in terms of word order and words’ part of speech. She even performed phonological rehearsal (i.e. ‘reading to self in head’) to aid her processing of form. A qualitative analysis of the recall data  focusing on words (types) that triggered comprehension and tokens of the individual words that were recalled  indicated: (1) The idea units were based almost exclusively on words that looked familiar (e.g. student, diskriminere, antiracist , 25, 2540), as were recalls of individual words [/M]. (2) Recalls of individual words [/M] comprised mostly of words that appeared at least four times in the reading passage (e.g. pa˚, og, jeg, er ). That frequency and saliency had a role to play in the ability to recall words was also supported by data from the gap-filling task, where words that appeared frequently in the reading passage were better recognised than those that were less frequent. For example, 11 of the 12 participants chose the correct word (i.e. er ) for the first blank (see Appendix 3). That word appeared seven times in the reading passage. Conversely, only three of the 12 participants

Input Processing

31 10

Table 6 Written recall by S11  SQ

What is this passage about? In five minutes, recall as many facts as you can from the passage. You can write in English. (1) 25 years old (2) A student 0/this is only English cognate (3) From ‘Norge’ (Norway) (4) Name kraef(?) (5) Something 25-40 (6) Going to school (skool) (7) Jag /I (8) Er/am Jag er . . . (9) Discrimination (10) Pa˚ /prep? (11) Og/prep? (12) Three blocks/sections (13) Header footer/attribution- /bolded (14) It’s in Norwegian? or Danish? (15) Yellow background, black type (16) I was reading to self in head

chose the correct word (i.e. ha ), which appeared only once in the reading passage, for the second blank.

Discussion The present study differed from previous studies (Hulstijn, 1989; VanPatten, 1990) in two key respects: first, while previous research investigated input processing with learners who had varying exposure to the target language, and, therefore, had developed partial knowledge of that language, this study examined input processing with learners who had absolutely no prior exposure to the target language, and hence no existing knowledge of it. Second, while previous research targeted specific linguistic structures, the present study looked, generically, at the content of input processing. Small-scaled as it was, which necessarily limits its generalisability, the present study did, nevertheless, generate some findings meaningful enough to raise concerns about the assumption that (beginning) learners’ default approach to input processing is meaning-based (Table 1). First and foremost, this group of ab initio learners adopted primarily a form-based approach, even when they had been oriented to focus on meaning (i.e. the SQ group). As one

32

International Journal of Multilingualism

of the participants in the SQ group reported, she did not even pay attention to the meaning of the passage when processing the input, but rather, tried to understand the structure of sentences. Hence, our first hypothesis was disconfirmed.7 The use of a form-based approach, as we analyse it retrospectively, appeared to be based on two main factors. First, the learners had no prior knowledge of the target language, and hence they had little to resort to in order to aid their comprehension. Using Krashen’s term, they did not even have i , to begin with. Second, the stimuli (i.e. the reading passages) were devoid of any extralinguistic clues, such as pictures and background information, which would have given the learners some contextual information to make sense of the input. These two  one posing a linguistic constraint and the other functional  collectively made the input outright incomprehensible. Under such circumstances, the learners had no recourse but to perform bottom-up processing, primarily at the word level. A second finding concerns the hypothesised effects of sequential versus simultaneous processing (i.e. Hypothesis 2). The fact, as reported in Results, that the subjects in the SQ and SM conditions performed at similar levels on the recall task disconfirms our second hypothesis. But more importantly, it suggests that directing their attention to either form or meaning did not have much of an impact on the way the subjects processed the input. One can, of course, argue that the lack of differences might have resulted from the small sample of subjects. True as that may be, we refuse to give in to the notion that genuine differences ought to derive from large samples as opposed to from the data per se . Thus, in the case at hand, if genuine differences did indeed exist, they should surface in the recall data. A qualitative examination of individual recalls, however, reveals overwhelming similarity in the way of processing, and this was later corroborated by the protocol data. For us, the lack of differences was another manifestation of the function of the two constraints mentioned above. That is, driven by a lack of knowledge resources for processing the input, the subjects in both groups relied on the only information available, the given target language input, and performed surface structure analyses thereof (Sharwood Smith, 1986).8 The overall lack of a significant difference between the two groups, however, does not, and should not, blind us to the fact that on separate measures of the subjects’ performance on the recall task, the SQ and the SM group did show differences of small magnitude (see Tables 4 and 5, and Figure 2). The SM group performed slightly better than the SQ group in comprehension (i.e. Idea Units and Word [/M]), and slightly worse in registering form (i.e. Word [/M]) and word recognition (i.e. filling in the blanks). Interestingly enough, these differences point up a weak contrastive pattern: while the SM group did somewhat better in decoding input, the SQ group outperformed the SM group in encoding input in short-term memory. Thus, there does seem to be some modest support for the second hypothesis in that consecutive exposure to the target language input, as experienced by the SM group, helped somewhat with input processing for meaning, whereas non-consecutive exposure, as experienced by the SQ group, helped with processing for form. As pointed out above, the impact of sequential versus simultaneous

Input Processing

33

processing was probably cancelled out by the participants’ lack of existing knowledge of the target language.

Conclusions The study reported here provides evidence challenging the notion that the default approach to non-primary language input processing is meaning-based, as implied in the processing principles of VanPatten (1996, 2004) (i.e. 1(ae); see Table 1), a notion that has been adopted by many researchers as a conceptual basis for understanding a multiplicity of learning issues (passim the second language acquisition literature). The subjects in this study, due to the absence of existing knowledge of the target language, performed form-based processing of the given input; they did not show a preference for content words, nor for grammatical functors. Rather, their processing mechanism located forms  content words and functors  indiscriminately. The Meaning Primacy Principle, therefore, should not be assumed to hold true for all learners, and we would argue for the need to differentiate between input processing by learners with and without existing knowledge of the target language. At the very least, the present study raises the question  if lexical items are not readily identifiable, how do learners search for them? The Meaning Primacy Principle presupposes some existing knowledge of the target language, with which learners can then take the input processing route, delineated by Sharwood Smith (1986; see the Linguistic Input route in Figure 1), for comprehension and acquisition. When no such knowledge is available, however, as in the case in question, that path can only be minimally traversed, leading to little comprehension and intake. Findings from the study suggest that the quality of processing of input for meaning and form depends on the interplay of a number of factors: existing linguistic knowledge, processing capacity, and, last but not least, input frequency. However, we must concede that the measures we invoked are limited, and, hence, our data are only suggestive. Future studies are needed to verify our findings by replicating the design of the present study, but including additional measures of comprehension and intake, with beginning learners of different target languages. Furthermore, future research comparing the results from the beginning learners with those from more advanced learners may illuminate the issue of how much knowledge of the target language is necessary for learners to ‘switch’ from a form-based to a meaning-based input processing approach. Albeit tentative, findings from the present study appear to lend some support to the observations of White (1987) on the usefulness of incomprehensible input for acquisition. White specifically notes that ‘[S]trictly speaking, the driving factor for grammar change is that the input is incomprehensible, rather than comprehensible.’ (White, 1987: 98) In the present study, the input was incomprehensible to the learners, who were thus forced into formoriented processing. The implication of the study for language teaching could, therefore, be that providing input without contextual support may promote bottom-up, form-based processing.9 The lack of such processing in adult learners has been widely noted (see, e.g. Long, 1996), and has given the

34

International Journal of Multilingualism

impetus to much research in search of pedagogic strategies that may promote it. VanPatten has, for years, advocated processing instruction as a strategy to improve the quality of input processing (see, e.g. VanPatten, 2004; VanPatten & Cardierno, 1993). Similarly, Sharwood Smith (1993) has recommended input enhancement, and Long (1991) focus on form. What these strategies have in common is that they are external to the learner, who may or may not respond to them. Input-driven form-based processing, on the other hand, comes from within, and may, therefore, be a more effective strategy for promoting acquisition. This, of course, is merely speculation that warrants empirical validation. To that end, we should like to conclude with two hypotheses for future research: Assuming that there is no contextual support: (1) Learners who have developed intermediary grammars will adopt a meaning-based approach to processing input. (2) Beginning learners will adopt a form-based approach to processing input. Correspondence Any correspondence should be directed to Dr ZhaoHong Han, Teachers College, TESOL/Applied Linguistics, Columbia University, Box 66, 252 West 120th Street, New York, NY 10027-6696, USA ([email protected]) Notes 1. More recent research, however, treats selective attention not as a response to capacity limitations, but rather, as a means of action control (see, e.g. Neuman, 1996), whereby information relevant to the domain of action is selected, and perception of less relevant stimuli inhibited and suppressed. 2. One reviewer asked why we divided the subjects into two groups instead of putting them all in one group. The answer to this lies in our research questions; that is, we wanted to see differential effects, if any, on learners performing input processing under simultaneous versus sequential conditions. We were, of course, fully aware of the statistical consequences, and hence the limitation on the generalisability of the results, due to the small size of the experimental cells (see the Discussion and Conclusions sections). 3. Five minutes was set based on the length of the reading passage. As it turned out, it was more than required for completing the recall task. 4. Only five sentences were included in this task, because the original input text was short, and because we wanted to control for frequency effects by targeting both frequently appearing and less frequently appearing items. The task was designed to tap into recognition. Hence, the options were given. In terms of difficulty, definitely the first task was more challenging, as one reviewer aptly pointed out, but difficulty was not much of an issue for us, not in the present study; rather, our main concern was that the two tasks each tapped into a different aspect of input processing, comprehension versus recognition of forms. The two tasks we employed are typical in studies on input processing. Moreover, it was precisely our intention to use commonly used tasks in order to allow future comparisons of findings from our study with those from other studies. 5. Before reading, the participants were told the following: ‘You will read a passage twice on the screen. Each reading will take no more than five minutes. Upon finishing reading, you will be given a comprehension task and a fill-in-the-blank task to complete.’ Based on findings from Lee (1986), pre-reading orienting directions can aid the comprehension of low-level learners.

Input Processing

35

6. Ideally, the data should be scored independently by two or more raters, but this was not possible due to difficulty finding researchers who had a working knowledge of Norwegian. 7. One might argue that the study was designed in such a way that it was quite obvious what its outcome would be. We hasten to point out, however, that, natural as it now seems, this outcome had not been established through empirical research prior to the present study. 8. The fact that the participants were multilingual speakers may have predisposed some of them to better processing forms, due to perceived lexical or structural similarities between the prior languages and the target language. We recognise that to minimise such prior linguistic bias, it might have been better to expose them to an artificial language. Future research controlling for learners’ prior knowledge should shed light on this issue. 9. A relevant finding from Oxford et al . (2004) is that high-proficient learners used top-down strategies during reading for comprehension, whereas low-proficient learners used bottom-up strategies. 10. The written recall by S11  SQ is chosen because it best represents the general scope of the recalls made by the participants from both groups.

References Baddeley, A. (2001) Is working memory still working? American Psychologist 56, 849864. Bransford, J. and Johnson, M. (1973) Consideration of some problems of comprehension. In W. Chase (ed.) Visual Information Processing (pp. 383438). New York: Academic Press. Broadbent, D.E. (1975) The magic number seven after 15 years. In A. Kennedy and A. Wilkes (eds) Studies in Long-term Memory (pp. 318). London: Wiley. Cook, V.J. (1997) Second Language Learning and Language Teaching . London: Arnold. Doughty, C. (1991) Second language instruction does make a difference. Studies in Second Language Acquisition 13, 431469. Ellis, R. (1994) Implicit and explicit language learning. In N. Ellis (ed.) Implicit and Explicit Learning of Languages (pp. 79114). San Diego: Academic Press. Engle, R.W. (2002) Working memory capacity as executive attention. Current Directions in Psychological Science 11, 1923. Ericsson, K.A. and Kintsch, W. (1995) Long-term working memory. Psychological Review 102, 211245. Faerch, C. and Kasper, G. (1986) The role of comprehensible input in second-language acquisition. Applied Linguistics 7 (3), 257273. Gass, S. (1988) Integrating research areas: A framework for second language studies. Applied Linguistics 9, 198217. Hudson, T. (1982) The effects of induced schemata on the ‘short circuit’ in L2 reading: Nondecoding factors in L2 reading performance. Language Learning 32, 129. Hulstijn, J. (1989) Implicit and incidental second language learning: Experiments in the processing of natural and partly artificial input. In H. Dechert and M. Raupach (eds) Interlingual Processes (pp. 4973). Tubingen: Narr. Kintsch, W. (1998) Comprehension: A Paradigm for Cognition . Cambridge: Cambridge University Press. Klein, W. (1986) Second Language Acquisition. Cambridge: Cambridge University Press. Lee, J. (1986) On the use of the recall task to measure L2 reading comprehension. Studies in Second Language Acquisition 8, 201212. Lee, J. (1998) The relationship of verb morphology to second language reading comprehension and input processing. The Modern Language Journal 82 (i), 3348. Lee, J. (2002) The initial impact of reading as input for the acquisition of future tense morphology in Spanish. In S. Gass, K. Bardovi-Harlig, S. Magnan and J. Walz (eds) Pedagogical Norms for Second and Foreign Language Learning and Teaching (pp. 119140). Amsterdam: John Benjamins.

36

International Journal of Multilingualism

Long, M. (1991) Focus on form: A design feature in language teaching methodology. In K. de Bot, D. Coste, C. Kramsch and R. Ginsberg (eds) Foreign Language Research in a Cross-cultural Perspective (pp. 3952). Amsterdam: John Benjamins. Manne, G. (1990) Ny i Norge . Oslo: Fag og Kultur. Oxford, R., Cho, Y. and Leung, S. (2004) Effect of the presence and difficulty of task on strategy use: An exploratory study. International Review of Applied Linguistics 42 (1), 147. Robinson, P. (2003) Attention and memory during SLA. In C. Doughty and M. Long (eds) The Handbook of SLA (pp. 630678). Oxford: Blackwell Publishers. Sharwood Smith, M. (1986) Comprehension versus acquisition: Two ways of processing input. Applied Linguistics 7 (3), 239256. Sharwood Smith, M. (1993) Input enhancement in instructed SLA. Studies in Second Language Acquisition 15, 165179. Skehan, P. (1998) A Cognitive Approach to Language Learning. Oxford: Oxford University Press. Skehan, P. and Foster, P. (2001) Cognition and tasks. In P. Robinson (ed) Cognition and Second Language Instruction (pp. 183205). New York: Cambridge University Press. VanPatten, B. (1990) Attending to form and content in the input. Studies in Second Language Acquisition 12, 287310. VanPatten, B. (1996) Input Processing and Grammar Instruction: Theory and Research . Norwood, NJ: Ablex. VanPatten, B. (1998) Cognitive characteristics of adult second language learners. In H. Bryrnes (ed.) Learning Foreign and Second Languages: Perspectives in Research and Scholarship (pp. 105127). New York: The Modern Language Association of America. VanPatten, B. (2004) Processing Instruction . Mahwah: Erlbaum. VanPatten, B. and Cardierno, T. (1993) From input to output: Processing instruction and communicative tasks. In F. Eckman, D. Highland, P.W. Lee, J. Mileham and R.R. Weber (eds) Second Language Acquisition Theory and Pedagogy (pp. 169185). Mahwah, NJ: Lawrence Erlbaum. White, L. (1987) Against comprehensible input: The input hypothesis and the development of second-language competence. Applied Linguistics 8, 95110.

Appendix 1 The reading passage Kjære (1) Jeg er en gutt pa˚ 25 a˚r. (2) Jeg er utenlending. (3) Jeg er sa˚ redd for a˚ møte ensomhet som sa˚ mange ganger før. (4) Jeg har vært i Norge i flere ma˚neder, og (5) trenger ei jente. (6) Hun jeg ha˚per (7) pa˚ bør være 25-40, og (8) hun vil ikke diskriminere noen pa˚ grunn av farge. (9) Jeg er svart og (10) er student. (11) Du som kanskje vil svare pa˚ dette bør ikke tenke pa˚ utseendet ditt, (12) det er ikke bare det som teller, (13) mener jeg. (14) Jeg liker a˚ ga˚ ut pa˚ byen av og til. (15) Jeg trenger kjærlighet som ikke er redd. (16) Hvis du skulle ha lyst til a˚ vite mer om meg, ja, (17) sa˚ fort deg a˚ skrive. (18) Jeg venter spent pa˚ a˚ høre fra deg. Hilsen. Antirasist (125 words) English Translation: Dear

Input Processing

37

I am a 25-year old boy. I am a foreigner. I am so afraid to undergo loneliness as I have for so many times already. I have been in Norway for many months, and I need a girl. I hope she is between 25 and 40, and she will not discriminate against anyone on grounds of color. I am black and am a student. You who perhaps will answer this should not think about your appearance. That’s not the only thing that matters, in my opinion. I like going downtown once in a while. I am seeking love that is not timid. If you are interested in knowing more about me, yes, go ahead and write me. I look forward to hearing from you. Regards Antiracist

Appendix 2 Free recall What is the reading passage about? In five minutes, recall as many facts as you can from the passage. You can write in English.

Appendix 3 Gap filling Choose the appropriate words to fill in the blanks. (1) Jeg _____ ( har, er, var, hadde) en gutt pa˚ 25 a˚r. (2) Hvis du skulle ______ (var, har, ha, er) lyst til a˚ vite mer om meg, ja, sa˚ fort deg a˚ skrive. (3) Jeg er sa˚ redd for a˚ møte ensomhet som pa˚ mange ganger _____ (sa˚, før, tidlig, pa˚). (4) Jeg har vært i Norge _____ (i, for, pa˚, da) flere ma˚neder, og trenger ei jente. (5) Hun vil ikke diskriminere noen _____ (for, en, pa˚, i) grunn av farge.

Suggest Documents