You are on page 1of 210


Is There an Academic Vocabulary? Ken Hyland and Polly Tse 235

The Effect of Focused Written Corrective Feedback and Language Aptitude on ESL Learners Acquisition of Articles 255 Younghee Sheen Task Familiarity and Interactional Feedback in Child ESL Classrooms Alison Mackey, Alec Peter Kanganas, and Rhonda Oliver 285

Development of Speed and Accuracy in Pragmatic Comprehension in English as a Second Language 313 Naoko Taguchi Pointing Out Frequent Phrasal Verbs: A Corpus-Based Analysis Dee Gardner and Mark Davies Spoken Grammar and ELT Course Materials: A Missing Link? Richard Cullen and I-Chun (Vicky) Kuo 339 361


Pedagogical Implications: Genuine or Pretentious? ZhaoHong Han 387 394

The Research/Pedagogy Interface in a 21st-Century Publication Context Margaret Cargill A Bridge Too Far? Diane Belcher 396 400

Gauging the Scholarly Value of Connecting Research to Teaching Sally Sieloff Magnan Pedagogical Implications in TESOL Quarterly? Yes, Please! Carol Chapelle 404

The Author Replies to Z. H. Hans Pedagogical Implications: Genuine or Pretentious? 407 Youngkyu Kim A Reader Responds to J. Jenkinss Current Perspectives on Teaching World Englishes and English as a Lingua Franca 409 Luke Prodromou The Author Replies Jennifer Jenkins 414

Volume 41, Number 2

June 2007

Theorizing Language Teacher Education Constructing Pedagogical Awareness with Brazilian Language Educators Marlia Dos Santos Lima and Beatriz Fontana English as Cultural Capital in the Oaxacan Community of Mexico Angeles Clemente 421 417


Teaching English to the World: History, Curriculum, and Practice George Braine (Ed.) Reviewed by Dennis Bricault Language Planning in Higher Education: A Case Study of Pakistan Sabiha Mansoor Reviewed by Tariq Rahman Theory Construction in Second Language Acquisition Geoff Jordan Reviewed by Natalie Hess Bilingualism and Language Pedagogy 438 Janina Brutt-Griffler and Manka M. Varghese (Eds.) Reviewed by Josephine OBrien 436 431


Is There an Academic Vocabulary?

University of London London, England

Hong Kong University of Science and Technology Hong Kong SAR, China

This article considers the notion of academic vocabulary: the assumption that students of English for academic purposes (EAP) should study a core of high frequency words because they are common in an English academic register. We examine the value of the term by using Coxheads (2000) Academic Word List (AWL) to explore the distribution of its 570 word families in a corpus of 3.3 million words from a range of academic disciplines and genres. The ndings suggest that although the AWL covers 10.6% of the corpus, individual lexical items on the list often occur and behave in different ways across disciplines in terms of range, frequency, collocation, and meaning. This result suggests that the AWL might not be as general as it was intended to be and, more importantly, questions the widely held assumption that students need a single core vocabulary for academic study. We argue that the different practices and discourses of disciplinary communities undermine the usefulness of such lists and recommend that teachers help students develop a more restricted, discipline-based lexical repertoire.


he idea of an academic vocabulary has a long history in teaching English for academic or specic purposes (EAP and ESP). Variously known as subtechnical vocabulary (Anderson, 1980; Yang, 1986), semitechnical vocabulary (Farrell, 1990), or specialized nontechnical lexis (Cohen, Glasman, Rosenbaum-Cohen, Ferrara, & Fine, 1988), the term is used to refer to items which are reasonably frequent in a wide range of academic genres but are relatively uncommon in other kinds of texts (Coxhead & Nation, 2001). This vocabulary is seen as a key element of essayist literacy (Lillis, 2001) and an academic style of writing and is considered to be more advanced (Jordan, 1997) than the core 2,0003,000 words that typically make up around 80% of the words students are likely to encounter in reading English at university (Carter, 1998).

TESOL QUARTERLY Vol. 41, No. 2, June 2007


Many teachers regard helping undergraduates develop control over such a specialist vocabulary as an important part of their role, and attempts have been made to assemble lists of key terms to guide materials writers and help students plan their learning more efficiently. Early ESP materials, for example, sought to identify and present forms with a high frequency in scientic and technical writing (e.g., Barber, 1988; Herbert, 1965) and considerable effort has been devoted to investigating the vocabulary needed for academic study (e.g., Campion & Elley, 1971; Coxhead, 2000; Nation, 1990). Such research is usually based on the assumption that learners are seeking to build a repertoire of specialized academic words in addition to their existing basic or general service vocabulary, and this repertoire building is often seen as the purpose of developing university vocabulary. Consequently, vocabulary is typically seen as falling into three main groups (Nation, 2001): 1. High frequency words such as those included in Wests (1953) General Service List (GSL) of the most widely useful 2,000-word families in English, covering about 80% of most texts. 2. An academic vocabulary of words which are reasonably frequent in academic writing and comprise some 8%10% of running words in academic texts. 3. A technical vocabulary which differs by subject area and covers up to 5% of texts. First year undergraduate students are said to nd an academic vocabulary (2) a particularly challenging aspect of their learning (Li & Pemberton, 1994). This aspect of their learning is challenging because, although technical vocabulary is central to students specialized areas, general academic vocabulary serves a largely supportive role and the words are not likely to be glossed by the content teacher (Flowerdew, 1993, p. 236). Many of these words also occur too infrequently to allow incidental learning (Worthington & Nation, 1996), encouraging researchers and teachers to develop vocabulary lists for directly teaching these terms. The notion that some words occur more frequently in academic texts than in other domains is generally accepted. It also appears to correspond with EAPs distinctive approach to language teaching, based on the identication of the specic language features and communicative skills of target groups, and devoted to learners particular subject-matter needs. However, whether it is useful for learners to possess a general academic vocabulary is more contentious because it may involve considerable learning effort with little return. It is by no means certain that there is a single literacy which university students need to acquire to participate in academic environments, and we believe that a perspective

which seeks to identify and teach such a vocabulary fails to engage with current conceptions of literacy and EAP, ignores important differences in the collocational and semantic behavior of words, and does not correspond with the ways language is actually used in academic writing. It is, in other words, an assumption which could seriously mislead students. In this article we therefore set out to question some of the assumptions underlying the idea of a general academic vocabulary by analyzing the distribution of items in the widely used Academic Word List (AWL) (Coxhead, 2000) in a corpus of 3.3 million words from a variety of genres and disciplines.


To identify the most valuable words in academic contexts, a variety of vocabulary lists have been compiled from corpora, or collections, of academic texts. Corpus studies have shown that the most frequent words in English cover a large percentage of word occurrences in any text. The top three words (the, of, to) make up some 10% of uses in the 400-millionword Bank of English corpus, for instance, and the rst 100 comprise about one half of all written and spoken texts (e.g., Hunston, 2002; Kennedy, 1998). Extending the value of such frequency counts to academic uses, researchers have sought to compile lists of frequently occurring words found in different disciplines and different kinds of texts (e.g., Coxhead, 2000; Farrell, 1990; Praninskas, 1972; Xue & Nation, 1984). Common to these lists is the focus on word families, that is, the base word plus its inected forms and transparent derivations (Bauer & Nation, 1993). This approach overcomes the difficulty of specifying what counts as a word by including all closely related affixed forms as well as the stems most frequent, productive, and regular prexes and affixes. It is also supported by the view that knowledge of a base word can facilitate the understanding of its derived forms and evidence which suggests that members of the same word family are stored together in the mental lexicon (Nagy, Anderson, Schommer, Scott, & Stallman, 1989, p. 262; Nation, 2001). The most recent compilation is the AWL (Coxhead, 2000), which contains 570 word families believed to be essential for students pursuing higher education irrespective of their chosen eld of specialization.1 The 3,112 individual items in this inventory do not occur in Wests (1953) GSL and were included in the AWL if they met certain frequency and

The headwords of families in the AWL together with the items grouped by sublist are available from the Massey University (2004) Web site.



range criteria. Words were selected on the basis of occurring at least 100 times in an academic corpus of 3.5 million words of varied genres and in at least 15 of the 28 disciplines within the four broad subject groupings of the corpus: arts, commerce, law, and science (Coxhead, p. 221). Coxhead found that the AWL covered 10% of the words in her corpus and only 1.4% of a similar-sized corpus of ction, suggesting that the items are more relevant for learners with academic purposes. Taken together with items on Wests GSL, the AWL accounted for 86% of the academic corpus. There is no doubt that the AWL is an impressive undertaking, representing the most extensive investigation to date into core academic vocabulary. By stressing students target goals and the need to prioritize items in their lexical development, the AWL helps distinguish EAP from general English and sets an agenda for focused language learning. But although it has become a benchmark for developing teaching materials for EAP (e.g., Schmitt & Schmitt, 2005), the list has not been independently evaluated. It remains unclear how well the AWL can be said to represent the lexical composition of academic writing in English, and we have little idea of its coverage in particular disciplines and genres. In fact, a major difficulty of such lists, and not just the AWL, is the assumption that a single inventory can represent the vocabulary of academic discourse and so be valuable to all students irrespective of their eld of study. We explore this hypothesis by examining the frequency, range, preferred meanings and forms, and the collocational patterns of items in the AWL.


Our study employs both qualitative and quantitative analyses of a medium-sized academic corpus organized by discipline, genre, and writer expertise. Our academic corpus offers a broad cross section of writing in the disciplines and includes a range of professional and learner texts representing key academic genres across a broad span of disciplines (Table 1). The disciplines are biology, physics, and computer science (sciences); mechanical and electronic engineering (engineering), and sociology, business studies, and applied linguistics (social sciences). Within each of these elds we collected 30 research articles, seven textbook chapters, and 20 academic book reviews in each of seven disciplines; and 45 scientic letters in physics and biology. This corpus represents the range of sources students are often asked to read at university and so include the kinds of lexical items they will frequently encounter. In addition to these professionally written texts we added eight masters theses, six doctoral dissertations, and eight nal-year undergraduate project theses by L2 students in each of six disciplines to represent students productive uses of vocabulary.

TABLE 1 Academic Corpus Sciences Research articles Textbooks Book reviews Scientic letters Expert writers (subtotal) Masters theses doctoral dissertations Final year project theses Student writers (subtotal) Overall 189,800 106,100 31,600 122,000 449,500 139,300 191,800 87,100 418,200 867,700 Engineering 178,900 108,000 15,900 0 302,800 96,200 93,900 77,000 267,100 569,900 Social Sciences 633,400 176,800 77,000 0 887,200 205,700 457,000 305,100 967,800 1,855,000 Totals 1,002,100 390,900 124,500 122,000 1,639,500 441,200 742,700 469,200 1,653,100 3,292,600

The academic corpus thus comprises contemporary examples of the main academic discourse genres and includes both long and short texts, published peer-refereed articles, pedagogic texts, and student writing. Unlike Coxhead, whose corpus was opportunistic, including unequal numbers of texts in each eld, a number of 2,000 word text fragments from the now dated Lancaster-Oslo/Bergen (LOB) Corpus and Brown Corpus as well as the Wellington Corpus, and no examples of student writing, we attempted to more systematically represent a range of key genres in several elds, with equal numbers of entire texts in each genre. Although our procedure resulted in subcorpora of different sizes, we compensated by comparing item frequencies per 10,000 words. Having compiled our corpus, we then combined the texts into disciplines and elds using Corpus Builder (Cobb, 2004) and explored these using RANGE, a program developed by Nation (2002) and used to create the AWL (Coxhead, 2002). This software is preloaded with Wests (1953) GSL of the most frequent 2,000 English words and the AWL, and it shows the frequency of items from each list in any corpus together with its range, or the number of different subcorpora they occurred in. To determine the frequency and spread of AWL items, we examined these gures for the entire academic corpus, for the three subcorpora of engineering, science, and social science, and then for the disciplines within these elds. Finally, we ran concordances on selected items to see if they were used in the same ways and with the same meanings.

RESULTS: IN SEARCH OF AN ACADEMIC VOCABULARY Overall Frequencies and Distributions

Coxhead and Nation (2001, p. 254) claim that a word should be included in a general academic vocabulary if it is common to a range of

academic texts and accounts for a substantial number of words in an academic corpus. In our study, all 570 of the AWL word families occurred in the academic corpus, with 541 occurring in all three subcorpora. As Table 2 shows, the AWL covers an impressive 10.6% of the words in the corpus, and together with the 2,000 words of the GSL it provides an accumulative coverage of 85%, representing roughly one unknown word in every seven words of text. Although the list offers a good overall coverage of our academic corpus, we can see that this coverage is not evenly distributed. Students in the sciences are far less well served because the combined AWL and GSL failed to account for 22% of the words in our science corpus, meaning that students would stumble over an unknown item about every ve words, making the text incomprehensible. This variation may suggest that writing in the sciences demands a more specialized and technical vocabulary, but as we discuss later, the fact that all disciplines shape words for their own uses seriously undermines attempts to describe a core academic vocabulary. To explore the issue of range further, we examined the frequency of individual items in the three subcorpora themselves. Coxheads criteria for uniformity of frequency was 100 occurrences overall with at least 10 in each of four elds (Coxhead, 2000, p. 221), but this seemed a remarkably low threshold given the size of her corpus and the fact that the word list contains more than 3,000 individual words. We therefore used a more rigorous and systematic standard, identifying items as frequent if they occurred above the mean for all AWL items in the corpus (i.e., 597). Using this measure, only 192 families, or about a third of the AWL items, could be regarded as frequent, with the research terms process, analyze, research, data, and method being the most common. Interestingly, the top 60 items in our corpus provided a similar coverage (3.9%) to Coxheads most frequent items, but only 35 items were common to both lists. The most frequent 60 families form an important sublist in Coxheads compilation because they represent a coverage more than twice as numerous as the next most frequent 60 items, about
TABLE 2 Coverage of Academic Word List (AWL) in Academic Corpus Frequency words Engineering Social Sciences Sciences Overall 551,891 1,822,660 838,926 3,213,477 Coverage % AWL items 61,408 200,393 78,234 340,035 Mean 108 352 137 597 AWL 11.1 11.0 9.3 10.6 GSL 73.3 77.0 69.0 74.0 Overall 84.4 88.0 78.3 84.7

Note. GSL = General Service List (West, 1953).



one occurrence every four pages. At the other end of the scale, commence, concurrent, levy, and forthcoming were among 23 families we judged to be extremely infrequent because they occurred less than 60 times in the corpus (below 10% of the overall mean).

Distributions Across Fields

Looking more closely at the distributions across subcorpora, some items are frequent overall because of their concentration in one or two subcorpora, 15 of our top 50 items, for example, had more than 70% of their occurrences in one eld. Taking the means of individual elds as a benchmark, we found that of the 192 families which were frequent overall, only 82 were frequent in all three subcorpora and 50 in just one. Nor were the same items the most frequent in all subcorpora. Table 3 shows that only analyze and process in the overall most frequent list also occurred in the top 10 most frequent families in each subcorpora. We can also see here that the AWL items differed considerably in their
TABLE 3 Most Frequent Items by Field With Percentages of Families in That Field Overall (all three elds) Family Process Analyze Research Data Method Vary Strategy Culture Function Signicant Freq 4,501 4,498 3,841 3,789 3,214 3,156 3001 2962 2909 2742 Sciences Family Data Method Process Analyze Concentrate Require Function Obtain Extract Similar Freq 1395 1271 1118 1029 865 848 759 750 739 726 Item % 1.8 1.6 1.4 1.3 1.1 1.1 1.0 1.0 0.9 0.9 Cum % 1.8 3.4 4.8 6.2 7.3 8.3 9.3 10.3 11.2 12.1 Family Research Strategy Culture Analyze Process Consume Response Individual Participate Signicant Item % 1.3 1.3 1.1 1.1 0.9 0.9 0.9 0.9 0.9 0.8 Cum % 1.3 2.6 3.8 4.9 5.8 6.8 7.6 8.5 9.4 10.2 Family Equate Process Design Method Data Analyze Function Require Output Input Engineering Freq 1,418 1,143 999 920 913 895 847 844 839 818 Item % 2.3 1.9 1.6 1.5 1.5 1.5 1.4 1.4 1.4 1.3 Cum % 2.3 4.2 5.8 7.3 8.8 10.2 11.6 13.0 14.4 15.7

Social Sciences Freq 3261 2795 2583 2574 2240 1947 1910 1894 1800 1762 Item % 1.6 1.4 1.3 1.3 1.1 1.0 1.0 0.9 0.9 0.9 Cum % 1.6 3.0 4.3 5.6 6.7 7.7 8.6 9.6 10.5 11.4

Note. Freq = frequency. Cum = cumulative.



spread in each subcorpora, with the top 10 comprising nearly 16% of families in engineering and only 11.4% in social sciences. Distributions are also unequal when we consider the least frequent words. Using 10% of the mean in each subcorpora as a reference, we found 78 families to be extremely infrequent in one subcorpora, 63 in two subcorpora and 6 in all three. In other words, 27% of all the AWL families have a very low occurrence in at least one subcorpora and so have an extremely low chance of being encountered by students. Although comparing the words frequency of occurrence relative to the mean helps to determine the relative signicance of particular words in different subcorpora, variations in the sizes of the subcorpora mean that this procedure gives us only a general impression of distributions. A more accurate picture is obtained by norming frequencies to give occurrences per 10,000 words and then looking at the percentage spread of each item across the subcorpora. Because we considered three elds, an even distribution would be roughly 33% of an item in each subcorpora, but no family met this criteria and more than half of all items occurred mainly in one subcorpora only. Table 4 shows that of the 570 AWL families, 534 (94%) have irregular distributions across the three subcorpora. Of these, 227 (40% of items) have at least 60% of all instances concentrated in just one subcorpora, and 13% have at least 80% in a single subcorpora. Among the most frequent items, more than 90% of all cases of participate, communicate, output, attitude, conict, authority, perspective, and simulate occurred in one subcorpora. Overall, only 36 word families were relatively evenly distributed across the science, engineering, and social science subcorpora, and so might therefore be regarded as candidates for an academic word list, albeit a very limited one. Of these, however, only 22 were frequent by our criterion of being above the overall mean, and only seven were in the top 60 items. Just six families appeared in the top 60 of both Coxheads list and our own: analyze, consist, factor, indicate, period, and structure, which together comprised about 0.5% of the corpus. In other words, although the AWL may describe certain high frequency words in the register, the
TABLE 4 Concentration of Items in One Field (Adjusting for Text Size) Number of families Concentration of items 40%59% of occurrences in one eld 60%79% of occurrences in one eld Over 80% of occurrences in one eld Total families with uneven distribution All items 307 (53.9%) 154 (27.0%) 73 (12.8%) 534 (93.7%) Most frequent items (above mean) 103 (53.6%) 48 (25.0%) 19 (9.9%) 170 (88.5%)



distributions in our data suggest that most items have a limited range across subcorpora. The concentration of items is no less pronounced when we move to a ner level of delicacy and examine how words are distributed within subcorpora. Table 5 shows the extent of these concentrations, with 283 items in engineering (52% of all families) having more than 65% of all cases in just one discipline, 244 items in the sciences (43%) with more than 65% in just one discipline, and 128 (22.5%) of items in the social sciences with more than 65% in one discipline. Overall, only one family occurred roughly equally across the three disciplines in the sciences, seven families spread evenly across the social sciences, and 47 across the two engineering elds. Once again then, the patterns suggest a more complex picture of language use in the disciplines than notions of a general academic vocabulary allow, pointing to more specialized language uses.

Word Meanings and Uses

There is a further difficulty with compiling a so-called common core of academic vocabulary in that not only should it include items that meet frequency and range criteria, but also items which behave in roughly similar ways across disciplines. When reading academic texts, students need to be condent that they are understanding words in the right way, which means a vocabulary list must either avoid items with clearly different meanings and dissimilar co-occurrence patterns, or these items must be taught separately rather than as parts of families. We need, then, to be cautious about claiming generality for families whose meanings and collocational environments may differ across each inected and derived word form (Oakey, 2003). Most words have more than one sense and Wang and Nation (2004) have recently examined the potential monosemic bias of the AWL to discover if cases of homography, or unrelated meanings of the same written form, misrepresent the composition of word families and so affect the
TABLE 5 Concentration of Items in Disciplines (% Adjusted for Corpus Size) Total Families 542 568 570 570 Total of all items occurring in one discipline 40%64% 259 (47.8%) 322 (56.7%) 409 (71.8%) 336 (59.0%) 65%79% 133 (24.5%) 116 (20.5%) 74 (13.0%) 110 (19.4%) Over 80% 150 (27.7%) 128 (22.5%) 54 (9.5%) 114 (20.0%)

Disciplines Engineering Sciences Social Sciences Overall



learning burden for students. Although Wang and Nation found only a small number of families containing homographs (e.g., major which can mean both important and military rank), they further suggest that words have essentially similar meanings across elds. In our corpus, however, different disciplines showed clear preferences for particular meanings and collocations. This nding has implications for the notion of an academic vocabulary and attempts to base word lists on it. As brief examples, we might take the two most frequent AWL items in the corpus, process and analyze, both of which occur far more often in academic discourse than in other registers. Despite its high frequency in all three subcorpora, the word process is far more likely to be encountered as a noun by science and engineering students than by social scientists. This likelihood is the result of the well known practice of nominalization, or grammatical metaphor (Halliday, 1998), which refers to the way that writers in the sciences regularly transform experiences into abstractions to create new conceptual objects. Embedding an item such as process into complex abstract nominal groups produces terms such as A constant volume combustion process . . . the trouble call handling process . . . processing dependent saturation junction factors . . . the graphical process conguration editor . . .

Though nominalization enables writers in the sciences to give new objects stable names and discuss their properties without further investigation, it does not help novice users unpack specialized meanings from the individual lexical item. We believe this practice is therefore likely to present difficulties to both native and nonnative-English-speaking students. Similarly, analyze appears to be used differently across elds. In the social sciences, it tends to occur more regularly as a noun, while in engineering, students are six times more likely to come across the form analytical. There are also semantic differences. The word analysis, for instance, is often associated with particular types of approach, so that it appears in discipline-specic compound nouns such as genre analysis or neutron activation analysis. The verb form also has eld-specic meanings. In engineering, it tends to refer to methods of determining the constituent parts or composition of a substance (see Examples 1 and 2), and in the social sciences it tends to mean simply considering something carefully (see Examples 3 and 4):
Example 1. The proteins were separated by electrophoresis on 10% polyacrylamide SDS gels and analyzed by autoradiography. (Physics doctoral dissertation)

Example 2. We used a variety of methods to analyse fungal spore load, volatiles and toxins. (Biology article) Example 3. The major objective of this report is to analyze developments in political sociology over the last half century. (Sociology article) Example 4. Following this, the results were analyzed to determine buyer behaviour. (Business masters thesis) A random analysis of AWL families with potential homographs reveals a considerable amount of semantic variation across elds. Table 6 shows the main meanings for selected words with different overall frequencies in the AWL together with their distributions. The rst four are from our high frequency list, with occurrences above the overall mean, and show that even where uses are very frequent, preferred uses still vary widely, with social science students far more likely to meet consist as meaning to stay the same and science and engineering students very unlikely to come across volume meaning a book or journal series unless they are reading book reviews. With less frequent words the preferred meanings differ dramatically, and we have to question the value of teaching the secondary meanings at all. These preferred uses become more apparent when we consider patterns at the disciplinary level. In Table 7, for instance, we can see minimal uses of issue to mean ow out in all disciplines but business, an overwhelming preference for attribute to mean a feature in business, sociology, and computer science and as the verb to accredit in applied linguistics, biology, and physics. Similar disciplinary reversals arise in the preferred uses of volume, credit, abstract, and offset among these examples, and no doubt there are more in the corpus that we have not analyzed. In addition, all disciplines adapt words to their own ends, displaying considerable creativity in both shaping words and combining them with
TABLE 6 Distribution of Meanings of Selected Academic Word List Word Families Across Fields (%) Family Consist (rank 41) Issue (46) Attribute (93) Volume (148) Generation (245) Credit (320) Abstract (461) offset (547) Meaning stay the same made up of ow out topic feature ascribe to book quantity growth stage create acknowledge payment prcis/extract theoretical counter out of line Science 34 66 7 93 83 17 1 99 2 98 0 100 76 14 0 100 Engineering 15 75 6 94 35 65 7 93 2 98 60 40 100 0 14 86 Social Science 55 45 18 82 60 40 50 50 36 64 52 48 13 87 100 0



TABLE 7 Distribution of Meanings of Selected Academic Word List Word Families Across Disciplines (%) Families Consist (41) Issue (46) Attribute (93) Volume (148) Generation (245) Credit (320) Abstract (461) offset (547) Meanings stay the same made up of ow out topic feature ascribe to book quantity growth stage create acknowledge payment prcis/extract theoretical counter out of line AL 54 46 3 97 21 79 96 4 38 62 65 35 12 88 100 0 BS 74 26 28 72 90 10 1 99 18 82 9 91 28 72 100 0 Soc 63 37 4 96 71 29 80 20 71 29 71 29 20 80 100 0 EE 30 70 14 86 36 64 5 95 1 99 100 0 75 25 4 96 ME 20 80 0 100 56 44 41 59 6 94 44 66 80 20 0 100 Bio 38 62 0 100 12 88 4 96 3 97 0 0 100 0 25 75 CS 64 36 0 100 99 1 0 100 8 92 0 100 92 8 0 100 Phy 35 65 0 100 0 100 14 86 1 99 0 0 0 100 0 100

Note. AL = applied linguistics, BS = business sciences, Soc = sociology, EE = electronic engineering, ME = mechanical engineering, Bio = biology, CS = computer science, Phy = physics.

others to convey specic, theory-laden meanings associated with disciplinary models and concepts. Thus, only texts in business studies contain the form issuer, for example, and 56% of instances of convert are related to convertible security (42% of all cases) and convertible bond (14% of all cases). Also creating difficulties for learners is the fact that words take on additional meanings as a result of their regular co-occurrence with other items (e.g., Arnaud & Bejoint, 1992), so that the term value in computer science, for instance, is often found as value stream (21% of all cases) and multiplevalue attribute mapping (7% of all cases). Even high frequency items such as strategy have preferred associations with marketing strategy forming 11% of all cases in business, learning strategy making up 9% of cases in applied linguistics, and coping strategy constituting 31% of cases in sociology. Nor are word choices themselves simply arbitrary selections from an equally appropriate pool of semantically equivalent candidates. Frequency patterns reveal clear disciplinary preferences as routine uses take on the constancy of convention. The word phase, for example, is favored in biology, where it co-occurs with a limited range of items such as stationary and mobile; in mechanical engineering, stage is preferred, and in applied linguistics, the preferred term is period. In sum, these different word choices, collocates, and xed phrases colour the everyday uses of words with more particular discipline-specic meanings, reecting how writers need to represent themselves and their ideas through a locally appropriate theoretical and methodological framework. These patterns reinforce the view that particular lexical bundles, or strings of words which commonly go together in natural discourse (Biber, Johansson, Leech, Conrad, & Finegan, 1999, p. 990), also contribute to meaningmaking in academic contexts. Such stable word combinations are an impor246 TESOL QUARTERLY

tant part of a disciplines discoursal resources but enormously complicate the business of constructing general word lists. By breaking into single words items which may be better learnt as wholes, vocabulary lists simultaneously misrepresent discipline-specic meanings and mislead students.

Despite the pedagogic attraction of a universal academic vocabulary, there appear to be good reasons to approach the concept with caution. A growing body of research suggests that the discourses of the academy do not form an undifferentiated, unitary mass, as might be inferred from such general lists as the AWL, but constitute a variety of subject-specic literacies (e.g., Hyland, 2000, 2002). Such ideas as communicative competence in applied linguistics, situated learning in education, and social constructivism in the social sciences have contributed to a view which places the notion of community at the heart of writing. Each subject discipline represents a way of making sense of human experience that has evolved over generations, and each is dependent on its own particular rhetorical practices. Words are often associated with different meanings and uses across registers (e.g., Biber et al., 1999) and we tend to nd similar variations in the practices in different disciplines. As Trimble (1985) cautions, academic vocabulary can take on extended meanings in technical contexts, and in different disciplinary environments words may have quite different meanings. We noted earlier, for instance, that the AWL families were underrepresented in the sciences, suggesting the need to recognise a distinct scientic vocabulary. The resources which have developed in the sciences for construing reality as a world of logical relations and abstract entities are far removed from our routine ways of describing the world and so represent a more precise disciplinary lexical arsenal. But all elds, not only the sciences, draw on a specialized lexis. All academic representations shape and manipulate language for disciplinary purposes, often refashioning everyday terms so that words take on more specic meanings. In fact, it would be surprising if specialised meanings had not evolved to discuss eld-specic topics more precisely. Persuasion in academic writing involves using language to relate individual beliefs to shared experience, tting observations and data into patterns which are meaningful to disciplinary insiders (Bazerman, 1994; Geertz, 1988). Writers must encode ideas and frame arguments in ways that their particular audience will nd most convincing, drawing on conventional ways of producing agreement between members and frequently moulding everyday words to the distinctive meanings of the disciplines (e.g., Hyland, 2000; Myers, 1990). In other words, different views of knowledge, different research practices, and different ways of

seeing the world are associated with different forms of argument, preferred forms of expression, and, most relevantly, specialised uses of lexis. Corpus evidence shows that words do not occur randomly in language use and that choices are governed by both rule-based systems of categories (e.g., Sinclair, 1991) and community-based conventional practices. The ndings of this study show that these practices do not just operate at the level of register, as assumed in the notion of academic vocabulary, but indicate greater specicity which undermine claims that there is a general vocabulary of value to all students preparing for, or engaged in, university study. We have found that various parameters of variation support this assertion, and analysis of our corpus using the most recent list reveals uneven word frequencies, restricted item range, disciplinary preferences for particular items over semantic equivalents, and additional meanings lent to items by disciplinary convention and associations in lexical bundles. Many items are considerably underrepresented in particular elds or disciplines, and this clustering in one eld suggests a considerable degree of disciplinary specicity in their use. The AWL seems to be most useful to students in computer sciences, where 16% of the words are covered by the list, and least useful to students in biology, with only 6.2% coverage. Coxhead acknowledged a bias toward the commerce subcorpus in her study but investigated no further. It seems likely, however, that this bias resulted from the inclusion of disciplines which shared greater similarities in her commerce corpus (e.g., accounting, economics, and nance), and dissimilar disciplines in arts (e.g., psychology, history, and linguistics) and sciences (e.g., geography, mathematics, and biology). This procedure seems to have produced remarkably high frequencies of words in the AWL common to nance-oriented disciplines, such as consume, corporate, invest, partner, purchase, and nance. These words all occur overwhelmingly in business studies and, because it tends to develop business-related applications, in the computer studies texts in our corpus. Similarly, a law subcorpus comprises a full quarter of Coxheads database, and a glance at the AWL shows a considerable number of lawrelated headwords (legal, legislate, regulate, compensate, etc.) which are unlikely to gure signicantly in the texts of most undergraduates. We are also concerned that the AWL corpus is partly made up of 2,000-word text fragments rather than whole texts and that some of these, namely the Brown and LOB Corpora, date back almost 30 years. In short, we believe that the selection of Coxheads corpus may have skewed the AWL and squeezed other potential candidates out of the corpus altogether. These ndings therefore serve to undermine the value of relying on decontextualized lists of vocabulary as a source of generally available and equally valid items for student writers across the disciplines. Within each discipline or course, students need to acquire the specialized discourse

competencies that will allow them to succeed in their studies and participate as group members. This means seeing literacy as something we do, an activity located in the interactions between people (Barton and Hamilton, 1998, p. 3), rather than as a thing distanced from both teacher and learner and imposing on them external rules and requirements (Street, 1995, p. 114). Because literacy is a practice integral to particular social and institutional contexts, we are forced to acknowledge that lists which claim to represent an academic vocabulary are likely to have a limited usefulness. The words they contain are unlikely to be of equal value to all students, and many words will be of almost no use to them at all. As teachers, we have to recognize that students in different elds will require different ways of using language and so we cannot depend on a list of academic vocabulary. We are therefore unable to support the division between academic and technical vocabulary assumed by list compilers. The fact is that although some words appear to be more generally used in the register, ranging across several disciplines, these items are not used in the same way and do not mean exactly the same thing in different disciplinary contexts. Instead, it might be more accurate to regard academic vocabulary as a cline of technically loaded or specialized words ranging from terms which are only used in a particular discipline to those which share some features of meaning and use with words in other elds. The general point is, however, that we need to identify students target language needs as clearly as possible and to address these needs as well as we can, and part of this work will involve introducing, making salient, and practicing the specialized vocabulary of their elds or disciplines. Nor do we nd reasons to be sanguine about the unexamined view that students rst acquire a multipurpose general service vocabulary which they only then top up with an academic repertoire. Clearly, many students learn English for purposes other than higher education and may not encounter large numbers of academic words in their studies. But often students do not follow this kind of trajectory and, instead, build their vocabularies less systematically, as they come across words from a variety of sources. Second language acquisition research indicates that students are likely to acquire items as they need them rather than in a taught sequence, and students will meet many of these academic words before gaining control of a general service vocabulary. This sort of acquisition is particularly likely given the fact that the AWL assumes knowledge of the words in the rather archaic GSL, which is now more than 50 years old and does not reect current usage. It is not hard to imagine, for example, a student encountering so-called academic items (from the AWL) such as available, percent, achieve, similar, and assist before GSL items such as headdress, shilling, redden, cultivator, and beak. In sum, because academic knowledge is embedded in processes of

argument and consensus-making, it will always be particular to specic disciplines and the socially agreed on ways of discussing problems in those disciplines. The fact that writing actually helps to create disciplines, rather than being just another aspect of what goes on in them, is a serious challenge to identifying uniformities in academic language use that apply in the same ways to all disciplines. If this view is correct, then it is also a serious challenge for students crossing disciplinary boundaries in modular and joint degree courses. As Bhatia (2002) observes: Students interacting with different disciplines need to develop communication skills that may not be an extension of general literacy to handle academic discourse, but a range of literacies to handle disciplinary variation in academic discourse (p. 27). It is difficult to see how students equipped with a general academic vocabulary are in a position to make these transitions. Although this article may tend toward the negative, we have not set out to discredit the AWL. On the contrary, we feel it offers a useful characterization of register-level vocabulary choices which may provide learners with a basis for challenging stereotypes or examining specic practices in their own elds. We also agree with the pedagogical principles that lay behind it: that teachers should seek to teach the most relevant and useful vocabulary to their students and that corpus analyses are the best way of ascertaining which vocabulary to teach (Coxhead, 2002). However, the evidence from this study urges caution in seeing the surfaces of texts as assemblies of discrete words with similar uses and meanings. By considering context, cotext, and use, academic vocabulary becomes a chimera. It gives a misleading impression of uniform practices and offers an inadequate foundation for understanding disciplinary conventions or developing academic writing skills.


Designing EAP materials is a complex task and, by arguing for more specic approaches to vocabulary teaching, it may seem that we are making it even harder. There are, however, important theoretical and pedagogic issues at stake which make it important to examine the assumptions that inform the idea of a general academic vocabulary. Modern conceptions of EAP, and language teaching more generally, stress the role of communication rather than language and the processes by which texts are created and used as much as they stress the texts themselves. Thus, teaching looks beyond lists of common core features and the autonomous views of literacy that such lists assume, recognizing that contextual factors are crucial to language choices because we communicate as members of social groups.

Vocabulary lists, such as the AWL, may provide some guidelines for teaching purposes, but if we nd that individual items occur and behave in dissimilar ways in different disciplines then we are forced to acknowledge the importance of contextual environments which reect different disciplinary practices and norms. Obviously vocabulary is more than individual words acting separately in a discourse, and compilers of vocabulary lists rightly insist that items should not be learnt out of context (e.g., Coxhead, 2000, 2002; Coxhead & Nation, 2001; Nation, 2001). Acquisition clearly needs to be part of a well-planned and sequenced program, with a mix of explicit teaching and incidental learning, a range of activities which focus on elaboration and consolidation, and sufficient information about contexts and denitions. We would argue, however, that the most appropriate starting point for such a program, offering the best return for learning effort, is the students specic target context. Like the AWL itself, corpus-informed lists and concordances can be used to help establish vocabulary learning goals for EAP courses, design relevant teaching materials, and generally target instruction more carefully. We believe it is important, however, that these lists and concordances are derived from the genres students will need to write and the texts they will need to read. Learners should be encouraged, for example, to notice these high frequency items and multiword units through repeated exposure and through temporary decontextualisation activities such as matching and item identication activities. Consciousness-raising tasks which offer opportunities to retrieve, use, and manipulate items can be productive, as can activities which require learners to produce the items in their extended writing. Students can, in fact, be encouraged to explore specialised corpora themselves and to try to identify meanings inductively from repeated examples or to familiarise themselves with common collocates. In sum, although the generic label academic vocabulary may be a convenient shorthand for describing a general variety, it conceals a wealth of discursive variability which can misrepresent academic literacy as a uniform practice and mislead learners into believing that there is a single collection of words which they can learn and transfer across elds. As we have learnt more about the different contexts in which students nd themselves at university, the complexity of the particular communicative demands their studies placed on them has become increasingly clear. We have began to see that many language features, including vocabulary, are specic to particular disciplines, and that the best way to prepare students for their studies is not to search for overarching, universally appropriate teaching items, but to provide them with an understanding of the features of the discourses they will encounter in their particular courses.

Ken Hyland is Professor of Education and Head of the Centre for Academic and Professional Literacies at the Institute of Education, University of London, London, England. He has taught in numerous countries and published more than 120 articles and 11 books on writing and applied linguistics. Polly Tse is an instructor at the Hong Kong University of Science & Technology, Hong Kong SAR, China, where she teaches English for specic purposes to business and science students. Her research interests include academic writing, interaction in text, and systemic functional linguistics.

Anderson, J. (1980). The lexical difficulties of English medical discourse for Egyptian students. English for Specic Purposes (Oregon State University), 37, 4. Arnaud, P., & Bejoint, H. (Eds.). (1992) Vocabulary and applied linguistics. London: Macmillan. Barber, C. (1988). Some measurable characteristics of modern scientic prose. In J. Swales (Ed.), Episodes in ESP (pp. 116). Cambridge: Cambridge University Press. (Reprinted from Contributions to English syntax in and philology, pp. 2143, by F. Behre, Ed., 1962, Gothenburg, Sweden: Almqvist & Wiksell.) Barton, D., & Hamilton, M. (1998). Local literacies. London: Routledge. Bauer, L., & Nation, P. (1993). Word families. International Journal of Lexicography, 6(4), 253279. Bazerman, C. (1994). Constructing experience. Carbondale: Southern Illinois University Press. Bhatia, V. K. (2002). A generic view of academic discourse. In J. Flowerdew (Ed.), Academic discourse (pp. 2139). London: Longman. Biber, D., Johansson, S., Leech, G., Conrad, S., & Finegan, E. (Eds.). (1999). Longman grammar of spoken and written English. London: Longman. Campion, M. E., & Elley, W. B. (1971). An academic vocabulary list. Wellington, New Zealand: New Zealand Council for Educational Research. Carter, R. (1998). Vocabulary: Applied linguistics perspectives. London: Routledge. Cobb, T. (2004). The online corpus builder. Available from nobel/r21270/cgi-bin/tools/corpus_builder Cohen, A., Glasman, H., Rosenbaum-Cohen, P. R., Ferrara, J., & Fine, J. (1988). Reading English for specialized purposes: Discourse analysis and the use of standard informants. In P. Carrell, J. Devine, D. Eskey (Eds.), Interactive approaches to second language reading (pp. 152167). Cambridge: Cambridge University Press. Coxhead, A. (2000). A new academic word list. TESOL Quarterly, 34, 213238. Coxhead, A. (2002). The academic word list: A corpus-based word list for academic purposes. In B. Ketterman & G. Marks (Eds.), Teaching and language corpora (TALC) conference proceedings (pp. 7389). Atlanta, GA: Rodopi. Coxhead, A., & Nation, I. S. P. (2001). The specialized vocabulary of English for academic purposes. In J. Flowerdew & M. Peacock (Eds.), Research perspectives on English for academic purposes (pp. 252267). Cambridge: Cambridge University Press. Farrell, P. (1990). Vocabulary in ESP: A lexical analysis of the English of electronics and a study of semi-technical vocabulary (CLCS Occasional Paper No. 25). Dublin, Ireland: Trinity College, Centre for Language and Communication Studies.

Flowerdew, J. (1993). Concordancing as a tool in course design. System, 21, 231244. Geertz, C. (1988). Words and lives: The anthropologist as author. Palo Alto, CA: Stanford University Press. Halliday, M. A. K. (1998). Things and relations: Regrammaticising experience as technical knowledge. In J. Martin & R. Veel (Eds.), Reading science (pp. 185235). London: Routledge.. Herbert, A. (1965). The structure of technical English. London: Longman. Hunston, S. (2002). Corpora in applied linguistics. Cambridge: Cambridge University Press. Hyland, K. (2000). Disciplinary discourses: Social interactions in academic writing. London: Longman. Hyland, K. (2002). Specicity revisited: How far should we go? English for Specic Purposes, 21, 385395. Jordan, B. (1997). English for academic purposes. Cambridge: Cambridge University Press. Kennedy, G. (1998). An introduction to corpus linguistics. London: Longman. Li, S. L., & Pemberton, R. (1994). An investigation of students knowledge of academic and subtechnical vocabulary. Proceedings of the Joint Seminar on Corpus Linguistics and Lexicology (pp. 183196). Hong Kong SAR, China: Hong Kong University of Science and Technology. Lillis, T. (2001). Student writing: Access, regulation, desire. London: Routledge. Massey University. (2004). The headwords of the Academic Word List. Retrieved March 13, 2007, from Myers, G. (1990). Writing biology: Texts in the social construction of scientic knowledge. Madison: University of Wisconsin Press. Nagy, W. E., Anderson, R., Schommer, M., Scott, J., & Stallman, A. (1989). Morphological families in the internal lexicon. Reading Research Quarterly, 24, 263282. Nation, I. S. P. (1990). Teaching and learning vocabulary. New York: Newbury House. Nation, I. S. P. (2001). Learning vocabulary in another language. New York: Cambridge University Press. Nation, I. S. P. (2002). RANGE [computer software]. Available from http:// Oakey, D. (2003, June 1618). Academic vocabulary in academic discourse: The phraseological behaviour of EVALUATION in the discourse of economics. Paper presented at Evaluation in Academic Discourse, University of Siena, Italy. Praninskas, J. (1972). American university word list. London: Longman. Schmitt, D., & Schmitt, N. (2005). Focus on vocabulary: Mastering the academic word list. London: Longman. Sinclair, J. (1991). Corpus, concordance and collocation. Oxford: Oxford University Press. Street, B. (1995). Social literacies. London: Longman. Trimble, L. (1985). English for science and technology: A discourse approach. Cambridge: Cambridge University Press. Wang, K. M.-T., & Nation, P. (2004). Word meaning in academic English: Homography in the academic word list. Applied Linguistics, 25, 291314. West, M. (1953). A general service list of English words. London: Longman. Worthington, D., & Nation, I. S. P. (1996). Using texts to sequence the introduction of new vocabulary in an EAP course. RELC Journal, 27, 111. Xue, G., & Nation, I. S. P. (1984). A university word list. Language Learning and Communication, 3, 215299. Yang, H. (1986). A new technique for identifying scientic/technical terms and describing science texts. Literacy and Linguistic Computing, 1, 93103.



The Effect of Focused Written Corrective Feedback and Language Aptitude on ESL Learners Acquisition of Articles
American University Washington, DC, United States

This study examines the differential effect of two types of written corrective feedback (CF) and the extent to which language analytic ability mediates the effects of CF on the acquisition of articles by adult intermediate ESL learners of various L1 backgrounds (N = 91). Three groups were formed: a direct-only correction group, a direct metalinguistic correction group, and a control group. The study found that both treatment groups performed much better than the control group on the immediate posttests, but the direct metalinguistic group performed better than the direct-only correction group in the delayed posttests. It also found a signicantly positive association between students gains and their aptitude for language analysis. Moreover, language analytic ability was more strongly related to acquisition in the direct metalinguistic group than in the direct-only group. The results showed that written CF targeting a single linguistic feature improved learners accuracy, especially when metalinguistic feedback was provided and the learners had high language analytic ability.

his article examines the role of corrective feedback (CF) in L2 acquisition by addressing written CF and the role of one individual difference factor: language analytic ability. In addition, this research is relevant to writing pedagogy, given that such pedagogy aims to improve students written grammatical accuracy. However, writing is a complex activity, and writing teachers view CF more broadly than second language acquisition (SLA) researchers (i.e., as encompassing correction of content, organization, and mechanics as well as linguistic form). Thus, this article does not intend to present a model for correcting students writing but rather to investigate one important aspect of the model, linguistic accuracy. Truscotts (1996) review of written CF studies and his controversial conclusion that written CF is ineffective and even harmful in promoting L2 acquisition constituted a challenge to researchers. Subsequently, the CF literature has offered an array of theoretical counterarguments and

TESOL QUARTERLY Vol. 41, No. 2, June 2007

empirical studies demonstrating that CF can lead to acquisition (e.g., Chandler, 2004; Ferris, 1999, 2004). The study reported in this article provides further evidence refuting Truscotts position. I begin by examining how corrective feedback has been addressed in the SLA and second language (L2) writing literatures, then I discuss a number of methodological issues relating to research on written CF, the relative effectiveness of different types of CF and the role played by language aptitude in mediating the effects of CF. I then report the results of a quasi-experimental study. I conclude by considering the implications of the results of the study for both SLA theory and L2 writing pedagogy.


SLA research into CF has been primarily concerned with oral CF in relation to theoretical claims about the role of input and interaction (e.g., Gass, 1997; Mackey, 2007) and focus on form (Long, 1996).1 There is now growing evidence that oral CF, as a focus-on-form technique, facilitates interlanguage development, although there is less consensus about the effects of different types of oral CF (e.g., explicit versus implicit, input-providing versus prompts; see Ellis, 2006). L2 writing researchers can benet from examining the methodology of oral CF research in SLA. For example, written CF studies (e.g., Polio, Fleck, & Leder, 1998; Robb, Ross, & Shortreed, 1986) have typically compared outcomes in terms of overall improvement across a number of different grammatical structures, whereas studies of oral CF in SLA (e.g., Han, 2002; Lyster, 2004) have focused on a specic grammatical feature and measured acquisition in terms of that feature. The SLA research suggests that intensive CF that repeatedly targets a single linguistic feature can have a benecial effect on interlanguage development (e.g., Doughty & Varela, 1998; Han, 2002; Iwashita, 2003; Long, Inagaki, & Ortega, 1998; Mackey & Philp, 1998). There are, however, some obvious differences between written and oral CF. Written CF is delayed whereas oral CF occurs immediately after an error has been committed. Written CF imposes less cognitive load on memory than oral CF, which typically demands an immediate cognitive comparison, thus requiring learners to heavily rely on their short-term memory. Written CF is also different pedagogically. Writing teachers often try to improve content and organization while focusing on the

Focus on form (Long, 1996) constitutes a type of pedagogical intervention. It refers to attempts to draw learners attention to form in the context of communication. One way in which this occurs is reactively through CF directed at learners linguistic errors.



overall quality of students writing, in which case accuracy is often a secondary issue. On the other hand, a teachers provision of oral CF typically draws learners attention to their erroneous utterances as they arise in communicative activities. These differences may explain why, in contrast to the SLA research that in general has shown that oral CF is effective, L2 writing researchers have not been able to convincingly demonstrate that written CF leads to improvement in grammatical accuracy in new pieces of writing. However, another reason that the results of SLA research into CF contrast with the results of L2 writing research into CF may lie in the methodologies used by writing researchers.

Methodological Issues in Written CF Research

A key issue concerning the efficacy of written CF was identied by Ferris (2004): Does written CF help students to improve in written accuracy over time? (p. 56). At present, this question has not been denitively answered because past research has failed to provide unambiguous evidence of the effectiveness of written CF (for a comprehensive and inuential discussion of this issue, see Ferris, 2004). The reasons for this are primarily methodological. Many written CF studies, for example, lacked a control group that did not receive CF (e.g., Kepner, 1991; Robb et al., 1986). In studies that did have a control or comparison group, the group differences were in fact not statistically signicant (e.g., Lalande, 1982; Semke, 1984). Also, the control or comparison group in the written CF studies typically received some kind of feedback (e.g., comments on content and/or organization) in addition to CF directed at linguistic accuracy (e.g., Fazio, 2001), which makes the effects of CF difficult to isolate. Researchers adopted this procedure for ethical reasons. When using intact writing classes in the study, researchers felt they had to provide some kind of correction because students had a right to it and probably expected it. The absence of a true control group, however, constitutes a major limitation in these studies. The studies also varied in how they measured the effectiveness of CF. Some measured improvement in terms of whether learners incorporated the corrections in a revision of their rst draft (e.g., Fathman & Whalley, 1990; Ferris & Roberts, 2001); others measured improvement in homework essay assignments or journal entries over a long period of time (e.g., Chandler, 2003; Kepner, 1991). A few studies measured improvement in writing in terms of gains in linguistic accuracy as well as uency (Ashwell, 2000; Robb et al., 1986). This variation in how the effectiveness of CF was measured makes reaching any denite conclusion very difficult. Other studies have measured improvement only by examining learnFOCUSED WRITTEN CORRECTIVE FEEDBACK AND LANGUAGE APTITUDE 257

ers revised texts. For example, in an examination of intermediate ESL college students writing, Fathman and Whalley (1990) found that feedback on form and feedback on content were equally effective in producing improvement in the students revisions. Ashwell (2000) also found that providing adult learners with grammar correction improved their grammatical accuracy in written compositions. One could argue, however, as Truscott (1999) did, that improvement in revisions alone does not constitute evidence that learning has occurred. To claim that error correction results in learning, one must examine whether the improvement in revisions carries over to a new piece of writing or if the improvement is manifested on posttest or delayed posttest measures. In sum, the methodological limitations of past studies on CF and the inconclusiveness of their ndings have resulted in a somewhat confusing picture, leading CF critics to argue against the effectiveness of CF and even to claim that CF can be harmful (Krashen, 1982; Truscott, 1996, 1999, 2004). Criticisms in this vein include practical concerns about the quality of the CF that teachers provide and also the danger of raising the students affective lter (i.e., provoking anxiety in the learner which inhibits learning). However, Ferris (1999, 2004) points out that most students do want their writing errors to be corrected and that it is the job of L2 writing teachers to attend to their needs.

Relative Efficacy of Different Types of Written CF

Although the absolute efficacy of written CF remains controversial, most written CF research has compared different types of written feedback to investigate whether certain types of CF produce a more positive effect than others (e.g., Fazio, 2001; Lalande, 1982; Robb et al., 1986). In one of the earliest studies, Robb et al. examined the writing of 134 EFL students in Japan over one academic year to see if four different CF types produced differential effects on the improvement in their written essays. The four methods were (a) direct correction, indicating the errors and providing the correct form; (b) coded feedback, indicating the type of error based on an abbreviated code system; (c) uncoded feedback, indicating errors in the text with a yellow highlighting pen without specifying their type; and (c) marginal, indicating the total number of errors in the margins of the students article. In all treatments, revisions were required and the instructor checked for accuracy. Robb et al. found no major differences across the four treatment types. Ferris (2002) contends that indirect error correction (i.e., identifying errors without providing the correct form) is more benecial than direct correction because it pushes learners to engage in hypothesis testing, thereby inducing deeper internal processing, which helps them inter258 TESOL QUARTERLY

nalize the correct forms. However, empirical evidence to date suggests that there is no advantage for indirect CF over direct CF (Chandler, 2003; Lalande, 1982; Robb et al., 1986). In fact, Chandler found that direct correction was superior to other types of indirect correction in producing more accurate writing. Chandler hypothesized that a teachers direct correction helps ESL students internalize the correct form in a more productive way because indirect feedback, though it demands greater cognitive processing, delays conrmation of students hypotheses. She also reported that her ESL students favored direct correction. These ndings suggests that contrary to pedagogical suggestions in the L2 writing literature (e.g., Ferris, 2002; Ferris & Hedgcock, 2005), indirect written CF may not be superior to direct CF. This conclusion is also supported by SLA research on oral feedback, which provides evidence that explicit feedback facilitates acquisition better than implicit feedback (Carroll & Swain, 1993; Ellis, Loewen & Erlam, 2006). However, the research on written and oral CF to date has not explored the possibility that the effectiveness of different types of CF will vary depending on the individual learner. I now turn to consider one major individual difference factor that may mediate the effect of CF.

Language Aptitude
Aptitude is a complex construct, distinct from general intelligence and achievement. Carroll (1981) denes language aptitude as having the following four components: (a) phonemic coding ability, (b) grammatical sensitivity, (c) inductive language learning ability, and (d) rote learning ability. Skehan (1998) combines (b) and (c) in a component he labels language analytic ability. This study is concerned only with this component, dened as the ability to analyze language by creating and applying rules to new sentences (Sawyer & Ranta, 2001). It is reasonable to think that learners with high language analytic ability will be better able to engage in the kind of cognitive comparison that is required if CF is to result in learning. It can be further argued that direct metalinguistic correction will benet analytically strong language learners to a greater extent than analytically weak learners because the former will nd it easier to use the metalinguistic information. To date, only two studies have explored the role of aptitude in the efficacy of CF. DeKeyser (1993) found that learners with high previous achievement, high language aptitude, and low anxiety beneted the most from error correction. Similarly, Havranek and Cesnik (2001) reported that corrective feedback is likely to benet learners who have a positive attitude toward error correction and high language ability. However, these two studies did not isolate different types of CF. Nor did they

investigate the interrelated effects of language aptitude and CF on specic target linguistic structures. The complex nature of CF (e.g., implicit or explicit, input providing or input prompting) raises the question whether different aspects of CF are mediated differently by language analytic ability. Moreover, the previous studies only examined language aptitude as a whole. The current study, therefore, chose language analytic ability as the component to investigate because it seemed to be potentially the most likely to interact with two direct types of CF in determining learning outcomes in the context of a written task. The relationship between language aptitude and learning outcome has often been discussed in terms of the cognitive load imposed on the learner (e.g., DeKeyser, 1993; Snow, 1987). In the two types of CF compared in the current study, the difference between the direct correction and direct metalinguistic correction was whether the learners received metalinguistic comments. Schmidt (2001) distinguishes noticing from metalinguistic awareness and argues that metalinguistic awareness requires a deeper level of learning. Thus, it may be hypothesized that whereas written direct feedback increases noticing, direct metalinguistic feedback increases not only noticing but also encourages awareness-asunderstanding (i.e., a deeper level of cognitive processing).

Research Questions
This study constitutes an attempt to address some of the perceived problems in written CF research by drawing on the theory and methodology of oral CF research in SLA and by acknowledging that individual difference factors such as language analytic ability may mediate the effect of CF. The study considers the following research questions: 1. Does focused written corrective feedback have an effect on intermediate ESL learners acquisition of English articles? 2. Is there any difference in the effect of direct correction with and without metalinguistic feedback on ESL learners acquisition of English articles? 3. To what extent does the learners language analytic ability mediate the effectiveness of CF?

This study used a quasi-experimental research design with a pretest treatmentposttestdelayed posttest structure, using intact ESL class260 TESOL QUARTERLY

rooms. The research also involved a correlational analysis of the relationship between one individual learner variable (i.e., language analytic ability) and criterion test scores in the posttests and delayed posttests with a view to examining the moderating effect of language aptitude on the effect of the error correction treatment. Two weeks prior to the start of the corrective feedback treatments, the participating students took a language analytic ability test. In the following week, they completed the pretests. The immediate posttests were completed following the two CF sessions and the delayed posttests 3 to 4 weeks later. During each testing session, three subtests were administered: a speeded dictation test, a writing test, and an error correction test.

The current study was conducted in six intact classrooms in the American Language Program (ALP) of a community college in the United States. The ALP offers intensive and extensive English language courses for those who wish to speak and write English accurately and uently. Classes are offered each semester, and each class lasts from 1 hour and 20 minutes to 3 hours each session. Teachers have either native or nativelike English language prociency. The program has four levels: Foundation and Levels I, II, and III (most advanced level) with an average class size of 1522 students. Prior to the current study, the researcher visited the site many times and observed and piloted a number of instruments in several classes at different levels. The Level II classes consisting of intermediate-level learners chosen for the study came from all areas of the program, not specically writing classes.

The participants were ve native-English-speaking American teachers and their 111 intermediate-level students. The students were drawn from both international and immigrant ESL populations and represented various language and ethnic backgrounds. The three major groups were Korean, Hispanic, and Polish. Their ages varied from 2156 years and they came from different educational backgrounds (e.g., high school diploma to doctoral degree). Class sizes ranged from 1522 students. In the end, 91 students completed the language analytic ability test, pretests, posttests, and delayed posttests; students with incomplete datasets were excluded from the sample. Out of six intact classrooms, three groups were formed: the direct-only correction group (n = 31), the direct metalinguistic group (n = 32), and the control group (n = 28).

Direct-only correction constitutes a traditional error correction strategy that consists of indicating the location of an error on the students text and the provision of the correct form by deleting and/or replacing the error or by adding a linguistic element. Direct metalinguistic correction is operationalized as indicating the location of an error, providing the correct form, and including metalinguistic comments that explain the correct form.

Target Structure
Articles were chosen as the target structure for the current study with a view to isolating the effect of error correction from any potential effect of grammar instruction in general. This decision was made after a series of discussions with the participating faculty members at the college, which revealed that (a) participating students are not explicitly taught articles during the semester, and (b) articles, though constituting a structure where students commonly make errors, are infrequently corrected because they are nonsalient and they require complicated rule explanations. The and a belong to the top ve most frequently occurring words in English according to the COBUILD corpus (Sinclair, 1991). However, the SLA literature has clearly shown that English articles are considered to be a nonsalient feature. They are not phonologically salient, and misuse of articles rarely leads to communication breakdown (Master, 2002). Learners have been observed to experience difficulty in learning articles because of their complex nature; that is, the choice of an article is determined by both linguistic and pragmatic factors (Butler, 2002; Liu & Gleason, 2002). For this reason, care was taken in the current study to focus the correction on errors involving just two major functions of indenite and denite articles: a as rst mention and the as anaphoric reference, as in the following example:
He hid a snake in his bag, but the snake escaped.

Instruments and Procedures

The two experimental groups completed the treatments and tests. The control group completed the tests only. It did not perform the narrative tasks and did not receive any feedback but instead followed normal classes.

Narrative Task Instruments

There were two treatment sessions. Each session involved a narrative stimulus to elicit article errors from the learners. It should be noted that because the context of this research was not an L2 writing class, it was not considered necessary to ensure that the narrative task constituted an ecologically valid writing task. The rst narrative task involved an adapted Aesops fable, The Fox and the Crow.
There was once a crow who stole a piece of cheese from a kitchen window. She ew off with the cheese to a nearby tree. A fox saw what the crow had done, and he walked over to the tree. Oh, Mistress Crow, you have such lovely black feathers, such little feet, such a beautiful yellow beak, and such ne black eyes! You must have a beautiful voice. Would you please sing for me? The crow felt very proud. She opened her beak and sang CAW-CAW-CAW-CAW. Of course the cheese fell down, and the fox ate the piece of cheese.

To minimize the processing load on the learners when reproducing the narrative, the researcher replaced some difficult words in the original fable with easier words and simplied a few sentence structures. The narrative contained seven indenite articles and seven denite articles. For the second task, the researcher constructed an interesting yet simple story titled The Pet Snake.
A boy bought a snake from a pet shop. He took the snake home. His mother screamed when she saw the snake. She told him to take the snake back to the pet shop but the owner refused to take the snake back. The boy put the snake in a box and left it on a seat in the park near his house. An old woman found the box. When she saw the snake she had a heart attack.

This narrative contained easy vocabulary with seven indenite and ten denite articles. The ALP faculty considered the two tasks suitable for their intermediate-level students, yet expected that the students would often make article errors.

Corrective Feedback Treatment Procedures

Each session was audio-recorded using a clip-on microphone attached to the teacher. There was a one-week interval between the rst and second narrative tasks. The specic procedures of the treatment session were as follows: 1. First, the teacher handed out the story with an empty writing sheet attached to it and told the students that they were going to read the story and then rewrite the story.

2. Students were asked to read the story silently. 3. The teacher explained key words and discussed the moral of the story with the class. 4. The teacher then collected the stories by asking the students to tear off the story part and keep the writing sheet only. 5. Before asking the students to rewrite the story, the teacher read the story aloud once while the students noted down key words. 6. Then the students were asked to rewrite the story as closely as they could remember. 7. The teacher collected the students written narratives which were then handed to the researcher. 8. The researcher corrected the narratives focusing mainly on article errors based on the correction guidelines. 9. In the following class (typically 2 or 4 days later), the students took part in a corrective feedback session during which they received their narratives with corrections. The students were asked to look over their errors and the corrections carefully for at least 5 minutes. However, the teacher did not comment further on their errors or give any additional explanation nor were students asked to revise their writing.

Correction Guidelines
Given the amount of time and labour involved, it was decided that the researcher would serve as the corrector. The researcher corrected all the article errors in the learners narratives, which contained between one and seven errors. To mask the focus of the study, the researcher corrected a few errors other than those involving articles. This CF treatment used two different types of written correction. For the direct-only correction group, each correction consisted of indicating the error and writing the correction above it. For the direct metalinguistic correction group, each error was rst indicated with a number. Notes for each numbered error were given at the bottom of a learners sheet. The notes indicated what was wrong using metalinguistic information and also provided the correct form.

Testing Instruments and Scoring Guidelines

Three tests were used to measure acquisition in this study: (a) a speeded dictation test, (b) a narrative writing test, and (c) an error correction test. In the case of (a) and (c), the same test was used for the

pretest, posttest, and delayed posttest, but the order of the items was randomly changed from one testing session to another. In the case of (b), the same test was used for the pretest and posttest, but the delayed posttest was based on a different narrative. This change was in response to students negative reactions in a preliminary study to being asked to repeat the same story in the posttest. The results of a pilot study involving a different group of students from a similar intermediate-level class showed that the narrative tasks used for the two different posttests were comparable (i.e., a t test showed no statistically signicant difference in the scores obtained from the two tasks). The content of the writing tests was different from the content of the CF treatment tasks and thus provided a measure of the extent to which the effects of the CF transferred to new pieces of writing. Students were also administered a language analytic ability test.

Speeded Dictation Test (8 Minutes)

This test consisted of 14 items, each of which contained one or two sentences involving the use of indenite and denite articles (see Examples 1 and 2). It was time-pressured to limit learners ability to draw on their explicit grammatical knowledge (Ellis, 2005). Each item in the test had one or two stimuli involving article-obligatory contexts (see Appendix A). The total number of article stimuli in the test was 9 indenite and 12 denite articles. Example 1 measures learners receptive and productive knowledge of the indenite article a with the referential-rst mention article function (a very kind doctor is the stimulus in the item). Example 2 measures knowledge of the indenite article a and the denite article the with the referential-second mention article function (the two stimuli in the item are a movie and the movie).
Example 1: There was a very kind doctor in my home town. Example 2: I saw a movie last night. The movie made me sad.

In administering this test, each student was provided with a small notebook. The researcher rst explained the procedures to the students. Then the teacher read two sample sentences so that the students could familiarize themselves with the procedure. Each item was read at a normal speed and students were directed to write down one item per page as fast as they could and exactly as they heard it. Once the students turned to the next page for the next item, they were not allowed to return to the previous page. This procedure prevented the students from consciously reworking what they had written. Target-like use (TLU) scores were calculated (Pica, 1991). The TLU analysis was used to measure learners knowledge of articles by taking overuse of the target form into consideration. Articles were rst scored

for correct use in obligatory contexts. This score then became the numerator of a ratio whose denominator was the sum of the number of obligatory contexts for articles and the number of nonobligatory contexts in which articles were supplied inappropriately. The scoring formula is shown in the following equation: n correct suppliance in contexts 100 = percent accuracy n obligatory contexts + n suppliance in nonobligatory contexts

Writing Test (12 Minutes)

This test was adapted from one of Muranois (2000) test instruments for English articles. It consisted of four sequential pictures; the students were asked to write one coherent story based on them. Word prompts next to each picture were included to elicit noun phrases involving article usage. For example, next to the rst picture the word prompts were old man, paint, and picture, thereby encouraging the students to construct a sentence such as An old man likes to paint a picture. It should be noted that a revised version of the writing test was administered for the delayed posttest. Care was taken to make the revised writing task equivalent to the original in terms of difficulty and the likelihood of eliciting noun phrases containing articles. The results of a pilot test with 25 ALP students who completed two versions of the writing test indicated that students tended to score higher on the original test, but the difference between the two sets of scores was not signicant, t(24) = 0.89, ns. The writing data were coded using TLU (Pica, 1991) scores. Using the same TLU formula as was used for the dictation tests, students scores were calculated as percentages. However, whereas the dictation tests afforded clear obligatory contexts based on the stimuli, the constrained free-production writing test did not. It required the following additional scoring guidelines: 1. When it was not clear whether a noun phrase (NP) constituted an obligatory context for a or the based on the students writing, the NP was not coded. Only suppliance/nonsuppliance in unambiguous contexts was coded (i.e., the contexts where the researchers could denitely determine that a or the was needed). This meant that some possible errors were ignored. Exceptions are noted in the points that follow. 2. In the case of the word prompt park, both in the park or in a park were possible, so NPs containing this word were excluded from coding. However, when neither article was present in the NP, it was coded as nonsuppliance.

3. In the case of the word prompts boy and girl, when the student wrote, A boy and girl, only the rst NP (i.e., a boy) was coded because it was not clear whether the student used the elision rule correctly. In a similar vein, when the student used the boy and girl as second mention, only the rst NP was coded. However, if the rst NP was erroneous, either as rst mention or second mention (e.g., boy and a girl or boy and girl), each NP was included in the coding. 4. Coding excluded any NP where a determiner and an article were copresent, as in the following example: A boy and girl are look at the his picture. 5. Articles in idiomatic phrasesfor example, all of a sudden, a few minutes, at the momentwere also excluded from coding.

Error Correction Test (15 Minutes)

This test consisted of 17 items (see Appendix B). Each item contained two related statements, one of which was underlined and contained an error, that the learners were asked to correct in writing. The items were adapted from test instruments used in Liu and Gleason (2002) and Muranoi (2000). Four distracter items were included, involving the use of past tense, modal choice, and subjectverb agreement. The two examples are taken from the test, followed by the correct answers.
Example 1: I saw an interesting movie last night. I forgot the name of movie. Answer: I forgot the name of the movie. Example 2: Is your uncle car salesman? Im looking to buy a car Answer: Is your uncle a car salesman?

The error correction test was scored on a discrete item basis. One point was given for each correct suppliance of an article in the 14 obligatory contexts in the underlined sentences in the test. Excluding the distracters, 14 points was the perfect score for the test, and students nal scores were calculated as percentages.

Language Analytic Ability Test (20 Minutes)

The instrument was based on a language analysis test developed by Ott and used previously by Schmitt, Drnyei, Adolphs, and Durow (2003). The test consisted of 14 multiple choice items. The learners were given a glossary consisting of words and sentences from an articial language and their English translations (see Figure 1). They were then given 14 English sentences and for each sentence were asked to choose

FIGURE 1 Language Analysis Test Examples Imaginary language Kau Meu kau meud bi kau meud bo So Ciu Example 1 The dog is watching the cat. a. kau meud so c. meu kaud so English translation Dog Cat The dog was chasing the cat. The dog is chasing the cat. Watch Mouse

b. kau meud si d. meu kaud si

the correct translation from the four choices provided. To make the correct choice, the learners needed to analyze grammatical markers supplied in the glossary and apply these to the multiple-choice translations. Figure 1 contains an example question. To choose the correct translation in Example 1, the student had to rst deduce the rule that i is a past (progressive) marker and o is a present (progressive) marker and then apply that rule to the translated choices. Because this test was challenging to the students, the researcher guided them through the rst item in the test. This guidance reduced their anxiety and ensured that they were familiar with the procedure they needed to follow. The language analysis test was scored on a discrete item basis with 14 points being the perfect score. Students nal scores were calculated as percentages.

Test Reliability
In the dictation and writing tests, a second researcher coded a sample of 25% of the total data. The sample came equally from the pretests, posttests, and delayed posttests. In the dictation test, the percentage agreement scores were 89.3%, 87.2%, 91.4%, respectively. In the writing test, the percentage agreement scores were 78.4%, 83.3%, 79.2%, respectively. Although higher levels of reliability for the writing test are desirable, the levels achieved all exceeded 75%, which is satisfactory. It should also be noted that the complexity of the English article system makes a high level of consistency difficult to achieve. As for the error correction and language analysis tests, internal consistency reliability was estimated using Cronbachs alpha. The reliability coefficient for the 14 items in the error correction pretest was = 0.84 (M = 5.17, SD = 3.59, n = 111), and for the language analysis test, = 0.92 (M = 51.0, SD = 20.0, n = 111).

TABLE 1 Group Means and Standard Deviations for Total Test Scores Pretest Correction type Direct-only group (n = 31) Direct metalingustic group (n = 32) Control group (n = 28) M 44.1 49.6 48.3 SD 11.9 16.9 14.2 Posttest 1 M 58.3 65.4 52.1 SD 15.3 16.3 15.6 Posttest 2 M 57.5 69.4 51.2 SD 14.4 15.3 16.2

All scores were entered into SPSS (2002) and a range of descriptive and inferential statistics were computed. The following analyses were used to answer the three research questions. First, descriptive statistics for the language analysis test, speeded dictation test, writing test, and error correction test were computed. Then one-way ANOVAs with posthoc multiple comparison tests using Tukey, repeated measures ANOVAs, and ANCOVAs were performed, followed by Pearson product moment correlation.

Tables 1 through 4 present the descriptive statistics for total scores for the tests taken together as well as for the scores on the dictation test, writing test, and error correction test separately over the three testing periods: pretest, Posttest 1 (i.e., immediate posttest) and Posttest 2 (i.e., delayed posttest). A one-way ANOVA showed no statistically signicant group differences in the pretest total scores among the three groups, F(2, 88) = 1.23, ns. Figure 2 shows each groups mean speeded dictation test scores for the three testing periods. In the speeded dictation test, all three groups produced gains at Posttest 1. However, the graph indicates that only the
TABLE 2 Group Means and Standard Deviations for the Speeded Dictation Test Pretest Correction type Direct-only group (n = 31) Direct metalinguistic group (n = 32) Control group (n = 28) M 50.8 54.6 58.1 SD 15.7 18.6 12.9 Posttest 1 M 67.6 70.7 64.1 SD 14.1 17.2 13.9 Posttest 2 M 65.2 72.6 62.4 SD 15.6 15.6 13.7



TABLE 3 Group Means and Standard Deviations for the Writing Test Pretest Correction type Direct-only group (n = 31) Direct metalinguistic group (n = 32) Control group (n = 28) M 53.1 57.6 57.3 SD 19.8 21.0 21.4 Posttest 1 M 66.3 73.9 63.9 SD 21.4 19.3 24.4 Posttest 2 M 64.9 78.8 63.9 SD 19.8 18.4 22.8

direct metalinguistic (henceforth direct meta) group shows a consistent increase over time whereas the control group and the direct-only correction (henceforth direct-only) group show a slight decrease from Posttest 1 to Posttest 2. One-way ANOVAs with multiple comparisons revealed that whereas the group differences in Posttest 1 were not signicant, the differences between the direct meta group and control group in Posttest 2 were signicant, F(2, 88) = 3.74, p < 0.05 Figure 3 shows the mean scores for the writing test over time. The three groups had considerably different scores, especially the direct meta group and the control group, who started at almost the same pretest point. At Posttest 1, the mean score of the direct meta group is considerably higher than that of the control group. The gure also shows that the direct-only group had the same pattern as in the dictation test in that the direct-only group barely sustained its gains indicated on Posttest 1 into Posttest 2, whereas the direct meta group continued to gain in Posttest 2. The posthoc comparisons revealed a signicant difference in Posttest 2 among the three groups, F(2, 88) = 5.12, p < 0.01, indicating that the meta group outscored both the direct-only and control group. Figure 4 shows the results for the error correction test. The pattern of the three groups in this graph reveals that the two treatment groups gains over time were substantial but the control group showed no improvement. Moreover, the gains in the meta group appear to be greater than those of the direct-only group. Also the difference between the direct meta group and the control groups is shown to be greater in this test than in the other tests. One-way ANOVAs with the posthoc tests
TABLE 4 Group Means and Standard Deviations for the Error Correction Test Pretest Correction type Direct-only group(n = 31) Direct metalinguistic group (n = 32) Control group(n = 28) M 28.3 36.6 29.5 SD 17.0 19.7 19.6 Posttest 1 M 41.2 51.7 28.4 SD 23.1 24.5 21.4 Posttest 2 M 42.4 56.9 27.4 SD 24.1 21.9 19.3



FIGURE 2 Group Means on Speeded Dictation Test

showed that these differences were signicant in both Posttest 1, F(2, 88) = 7.60, p = .001, and Posttest 2, F(2, 88) = 13.5, p < .001. More specically, in Posttest 1 both direct-only and direct meta groups outperformed the control, in Posttest 2 the direct-only group manifested signicantly
FIGURE 3 Group Means on Writing Test



FIGURE 4 Group Means on Error Correction Test

greater gains than the control group, and the meta group outperformed both the direct-only and control groups. It should be noted that one-way repeated measures ANOVAs with total scores showed that even the relatively moderate gains observed in the control group between Pretest and Posttest 1 as well as between Pretest and Posttest 2 were signicant: F(1, 27) = 9.57, p = <.01, F(2, 27) = 4.93, p = < .05, respectively. Given that all the groups showed signicant longitudinal gains, the crucial analysis concerns whether there was a time treatment interaction. Table 5 shows the results of a two-way repeated measures ANOVA with total scores as the dependent variable and with time (pretests, posttests, delayed posttests) and CF treatment (three levels) as independent variTABLE 5 Repeated Measures ANOVA of the Total Scores Across the Three Treatment Conditions and the Three Testing Sessions Source Between students CF treatment Error Within students Time Time CF treatment Error Df 2 88 1.80 3.60 157.9 F 4.79 (617.8) 114.7 16.9 (40.2) p .01 <.001 <.001



TABLE 6 Summary of Signicant Contrasts Detected by Posthoc Comparisons Total Posttest 1 Posttest 2 **D > C **M > C M > D, C Dictation Writing Error Correction **D > C M > C D > C M > D, C

*M > C

**M > D, C

Note. D = Direct-only correction group, M = Direct metalinguistic group, C = Control group; *p < .05, **p < .01, p < .001.

ables. As can be seen in Table 5, the three groups performed differently on total test scores, indicating a signicant effect for CF. Also there was a signicant interaction between time and CF treatment. To statistically examine the differences between pairs of groups, posthoc multiple comparison tests were performed. The results showed that the direct meta group produced higher total scores than the direct-only group and the control group. One-way ANOVAs with the posthoc comparisons revealed that the differences in the scores were signicant in both Posttest 1, F (2, 88) = 5.40, p < 0.01, and Posttest 2, F (2, 88) = 11.1, p < 0.001. More specically, both CF treatment groups outperformed the control group in Posttest 1 and the direct meta group performed better than the directonly group in Posttest 2. Table 6 summarizes where the signicant differences among the three groups exist for the three separate test scores as well as total scores. In total test scores, Posttest 1 results favored both treatment groups, but Posttest 2 results favored the direct meta group. Looking at the tests separately, the signicant contrasts in Posttest 1 mainly came from the error correction test. However, in Posttest 2, all three tests favored the direct meta group. Table 7 displays the descriptive statistics for the language analysis test. The direct meta group had the highest mean score and the control group the lowest. However, a one-way ANOVA revealed that these differences were not signicant, F(2, 88) = 1.09, ns. The results of the repeated measures ANCOVA showed that there was a signicant effect for aptitude as the covariate, F(2, 88) = 4.95, p < 0.05. In other words, the
TABLE 7 Descriptive Statistics for the Language Analysis Test Group Direct only (n = 31) Direct meta (n = 32) Control (n = 28) M 48.39 52.23 45.66 SD 21.25 20.80 18.68 Min 14 14 21 Max 93 93 86



students total scores on the tests over time were mediated by their language aptitude, as measured by the language analysis test. However, the ANCOVA also showed that there was still a signicant effect for CF after controlling for the effect of aptitude, F(2, 88) = 2.45, p < 0.001. That is, the learners test performance changed over time after removing the effect of their language analysis ability. Table 8 presents correlations between the language analysis scores and the test gain scores. Short-term gains were calculated by subtracting the pretest scores from the Posttest 1 scores. Longer-term gains were computed by subtracting the pretest scores from the Posttest 2 scores. In both treatment groups, the correlation analysis revealed a signicantly positive association between students short-term and longer-term total score gains on the one hand and their aptitude for language analysis on the other. When examining the correlations involving the score gains for the three separate tests, the relationship is stronger in the direct meta group. The direct meta group also yielded the highest correlation coefcient (r = 0.63) with the language analysis score accounting for approximately 39% of the variance in the longer-term total gains. In sum, both treatment groups in the immediate error correction test outperformed the control group, though the direct meta group performed better in all three delayed posttests. Students score gains were positively correlated with their aptitude for language analysis, and this correlation was greater for the direct meta group.

The rst research question asked whether written CF focusing on article errors produced a signicant positive effect on acquisition. The results of the error correction test in Posttest 1 and all three tests in
TABLE 8 Correlations Between the Score Gains and Language Analysis Scores Language Analysis Score (r) Test Total Dictation Writing Error correction Note. *p < .05 **p < .01. Gain Short term Longer term Short term Longer term Short term Longer term Short term Longer term Direct only (n = 31) .46** .54** .11 .39* .30 .32* .38* .35* Direct meta (n = 32) .57** .63** .09 .28 .61** .59** .44** .49**



Posttest 2 indicate that the written CF had a positive effect on the learning of English articles. In particular, direct metalinguistic feedback proved to be effective in improving students accuracy in all three tests (and total scores) in the second administration of these tests. The current study is different from previous written CF studies in that only one linguistic feature was targeted for the provision of CF and the tests developed measured students written accuracy alone. During the 2-month period of this study, the teacher did not explicitly teach or correct articles outside the treatment. The students in all groups were of the same level of prociency and received the same amount and type of instruction involving identical writing and reading materials. Thus, CF alone could be seen as responsible for the acquisition of English articles. Therefore, the ndings of this study provide evidence that the focused written CF resulted in improved accuracy. SLA research involving oral CF has shown that CF facilitates learning when it focuses on a single linguistic feature and makes the error salient (Han, 2002; Nicholas, Lightbown, & Spada, 2001). Also, the provision of CF targeting a specic error increases its effectiveness (Doughty & Varela, 1998; Muranoi, 2000). The current study treated only two simple, rule-based functions of articles, and students were likely to have been familiar with the form of the articles at the beginning of the study. In this context, written CF proved effective in enabling the learners to improve their accuracy in article use in both the short and longer term. It should be noted that the control group as well as the experimental groups improved over time, which suggests that there was a test practice effect. However, the CF treatment groups outperformed the control group, indicating that the CF treatment had an effect over and above the test practice effect. The second research question examined the relative effects of two direct CF strategies on learning: direct correction with and without metalinguistic information. The ndings indicate that the two CF types had differential effects: direct correction with metalinguistic comments was superior to direct correction without metalinguistic comments. This result can be explained by Schmidts account of the role of awareness in L2 acquisition. Schmidt (1995, 2001) distinguished awareness at the level of noticing and at the level of understanding, which is a higher level of awareness. Noticing involves simply attending to exemplars of specic forms in the input (e.g., English has a and the in sentences); understanding entails knowing a rule or principle that governs that aspect of language (e.g., English uses a before the rst mention of a noun and the before the second mention). Thus, it can be argued that whereas both direct CF with and without metalinguistic comments are likely to promote awareness as noticing, only direct CF with metalinguistic comments promotes awareness with understanding. Schmidt further contends that

such conscious rule awareness arising from understanding strongly facilitates later learning. This is borne out by the current study, which found that longer-term gains favoured the direct metalinguistic group. This view is supported by studies of oral CF. Carroll and Swain (1993) found that a group who received more explicit and informative CF (i.e., direct metalinguistic CF) outperformed groups who received other types of CF in a study investigating the acquisition of English dative verbs. Ellis et al. (2006) also found that explicit feedback (in the form of metalinguistic comments) was superior to implicit feedback (in the form of recasts) in a study that investigated English regular past tense. A recent study of written correction points to the importance of providing metalinguistic information. Bitchener, Young, & Cameron (2005) found that written correction in conjunction with conferencing that provided learners with metalinguistic comments on their errors resulted in statistically signicant gains in accuracy in two grammatical structures, one of which was the denite article. It is interesting that, in contrast to the study reported in this article, Bitchener et al. did not nd any statistically signicant effect for direct corrective feedback alone (i.e., without metalinguistic comments). Taken together, these ndings suggest that focused, metalinguistic feedback serves to improve learners grammatical accuracy. The third research question concerned the extent to which any effect for written CF is mediated by language analytic ability. The ndings showed that learners with a high level of language analytic ability beneted more from both types of CF. This result suggests that CF treatments are more likely to increase awareness (both at the level of noticing and understanding) when learners have higher aptitude for language analysis. However, this advantage was found to be more evident when the CF included metalinguistic information. Again, the greater enhancing effect of language analytic ability for the metalinguistic group can be explained by Schmidts (1995) notion of awareness as understanding, which is triggered by metalinguistic CF. Robinson (2001) found a strong positive correlation between aptitude and awareness. In other words, metalinguistic CF affords a higher level of awareness that facilitates learning and this awareness is triggered by a process that takes place more readily in learners with a greater capacity to engage in language analysis. The greater benecial effect of analytic ability evident for the metalinguistic CF group is not surprising. If language analysis is seen as a measure of learners capacity to acquire explicit knowledge2 (in particu2

Explicit knowledge refers to knowledge that is conscious, verbalizable, and declarative in nature (i.e., involves knowledge of stated grammatical rules). It contrasts with implicit knowledge, which is intuitive and automatic and more readily available for use in unplanned communication.



lar), it follows that such analytic skills will be more strongly related to gains in the group that receives metalinguistic CF, especially in a test such as an error correction test, which invites the use of metalinguistic knowledge. On the other hand, it is interesting that, in the case of the direct metalinguistic group, the writing test gains were more strongly correlated with the language analysis scores than the error correction test gains. A possible explanation is that the writing test most closely resembled the treatment task where learners initially needed to apply their language analytic abilities to understand the metalinguistic comments. Another possibility is simply that learners used their explicit knowledge to a greater extent in the writing test. The current study is limited in several ways. First, the study was not carried out in the context of L2 writing classes. Investigating the effect of written CF in that context would have afforded stronger ecological validity. Second, the writing task treatment was very short. A more substantial CF treatment might have produced even stronger and more robust effects. Third, the study examined the effects of CF on just two relatively simple functions of English articles and clearly the results cannot be generalized to other areas of grammatical accuracy, or even to other aspects of articles. Fourth, the students in the treatment groups were not required to revise their writing. This might be considered a limitation. However, one advantage of excluding revision is that it allows for the effect of the CF treatment by itself to be investigated. This study also examined a highly focused type of written CF. It was focused in two senses. First, it addressed only linguistic accuracy. Second, it was directed at a single grammatical area. Further research is needed to examine the effectiveness of the selective approach to CF recommended earlier. What benets accrue if CF is directed at specic writing skills? Does focused linguistic CF help students achieve long term gains in accuracy to a greater extent than unfocused linguistic CF? How focused does the corrective feedback need to be? One structure or two or three? L2 writing researchers need to conduct studies (especially longitudinal) that isolate the effects of specic types of correction. Future studies should also examine the effects of CF on more open-ended and extensive writing tasks.

As Chandler (2004) noted in her rebuttal of Truscotts (1996, 2004) criticism of written CF, the controversy surrounding the effectiveness of written CF can only be resolved through carefully designed studies:
I accept [his] argument that the efficacy of error correction for accuracy of subsequent writing can only be demonstrated by studies containing a

control group which receives no correction and experimental groups which correct their errors after either receiving direct correction or having the location of their errors pointed out. So I hope someone will do such a well-designed study. (p. 348)

The current study was designed with these points in mind to address the central question, does error feedback help L2 students written accuracy? The answer to this question is a denite yes. Thus, the current ndings do not support Truscotts claim that written CF is ineffective. Truscott based this claim on the fact that no studies had demonstrated that written CF had a positive effect on acquisition. Truscotts critique of written feedback research presented a challenge to researchers to develop methodologically sound studies. In this respect, the research reported in this article is a start in that direction. Written CF is complex. It addresses different aspects of writing content, organization, rhetoric, and mechanics, as well as linguistic accuracy. The question arises, however, whether written CF should deal with all these aspects at the same time or address different aspects selectively when correcting different pieces of writing. L2 learners have limited processing capacity and asking them to attend to corrections that address a range of issues at the same time may tax their ability to process the feedback. One reason that previous studies of written CF have failed to demonstrate any effect on students accuracy in subsequent writing may simply be that the linguistic feedback was not sufficiently focused and intensive. The study reported in this article points to the importance of a selective approach to correcting students written work. Teachers can be selective in two senses. First, they can elect to focus variably on different aspects of a students writing at different times, sometimes focusing on content or organization and on other occasions on linguistic correctness. Second, when the focus is on linguistic correctness, teachers may achieve better results if they select a specic grammatical problem that they have observed in their students writing rather than a whole range of linguistic errors. The results of this study suggest that focused linguistic CF is more effective when it incorporates both provision of the correct form and metalinguistic explanation.

This research would not have been possible without the support of the faculty in the American Language Program at Bergen Community College, Paramus, New Jersey, United States, and the willing participation of their students. I am also deeply grateful for the valuable feedback and support by Zoltn Drnyei on earlier versions of this article. Additional thanks go to George Ganat for his assistance with data coding and the anonymous TESOL Quarterly reviewers for their constructive criticism and helpful suggestions. A version of this paper was presented at the 2005 Second Lan278 TESOL QUARTERLY

guage Research Forum held October 79 at Teachers College, Columbia University, New York, New York, United States.

Younghee Sheen has recently completed her doctorate at the University of Nottingham, Nottingham, England. In fall 2007, she will be an assistant professor in TESOL at American University, Washington, DC, United States. Her research interests include instructed second language acquisition and, in particular, oral and written corrective feedback.

Ashwell, T. (2000). Patterns of teacher response to student writing in a multiple-draft composition classroom: Is content feedback followed by form feedback the best method? Journal of Second Language Writing, 9, 227258. Bitchener, J., Young, S., & Cameron, D. (2005). The effect of different types of corrective feedback on ESL student writing. Journal of Second Language Writing, 14, 191205. Butler, Y. (2002). Second language learners theories on the use of English articles. Studies in Second Language Acquisition, 24, 451480. Carroll, J. B. (1981). Twenty-ve years of research in foreign language aptitude. In K. Diller (Ed.), Individual differences and universals in language learning aptitude (pp. 83118). Rowley, MA: Newbury House. Carroll, S., & Swain, M. (1993). Explicit and implicit negative feedback: An empirical study of the learning of linguistic generalizations. Studies in Second Language Acquisition, 15, 357386. Chandler, J. (2003). The efficacy of various kinds of error feedback for improvement in the accuracy and uency of L2 student writing. Journal of Second Language Writing, 12, 267296. Chandler, J. (2004). A response to Truscott. Journal of Second Language Writing, 13, 345348. DeKeyser, R. (1993). The effect of error correction on L2 grammar knowledge and oral prociency. Modern Language Journal, 77, 501514. Doughty, C., & Varela, E. (1998). Communicative focus on form. In C. Doughty & J. Williams (Eds.), Focus on form in classroom second language acquisition (pp. 114138). Cambridge: Cambridge University Press. Ellis, R. (2005). Measuring implicit and explicit knowledge of a second language: A psychometric study. Studies in Second Language Acquisition, 27, 141172. Ellis, R. (2006). Researching the effects of form-focused instruction on L2 acquisition. In K. Bartovi-Harlig & Z Drnyei (Eds.), Themes in SLA Research (AILA Review, Vol. 19, pp. 1841). Philadelphia: John Benjamins. Ellis, R., Loewen, S., & Erlam, R. (2006). Implicit and explicit corrective feedback and the acquisition of L2 grammar. Studies in Second Language Acquisition, 28, 339368. Fathman, A., & Whalley, E. (1990). Teacher response to student writing: Focus on form versus content. In B. Kroll (Ed.), Second language writing: Research insights for the classroom (pp. 178190). Cambridge: Cambridge University Press. Fazio, L. (2001). The effect of corrections and commentaries on the journal writing

accuracy of minority and majority language students. Journal of Second Language Writing, 10, 235249. Ferris, D. (1999). The case for grammar correction in L2 writing classes: A response to Truscott (1996). Journal of Second Language Writing, 8, 110. Ferris, D. (2002). Treatment of error in second language writing classes. Ann Arbor: University of Michigan Press. Ferris, D. (2004). The grammar correction debate in L2 writing: Where are we, and where do we go from here? (And what do we do in the meantime . . . ?). Journal of Second Language Writing, 13, 4962. Ferris, D., & Hedgcock, J. (2005). Teaching ESL composition: Purpose, process, & practice (2nd ed.). Mahwah, NJ: Lawrence Erlbaum. Ferris, D., & Roberts, B. (2001). Error feedback in L2 writing classes: How explicit does it need to be? Journal of Second Language Writing, 10, 161184. Gass, S. (1997). Input, interaction, and the second language learner. Mahwah, NJ: Lawrence Erlbaum. Han, Z.-H. (2002). A study of the impact of recasts on tense consistency in L2 output. TESOL Quarterly, 36, 543572. Havranek, G., & Cesnik, H. (2001). Factors affecting the success of corrective feedback. EUROSLA Yearbook, 1, 99122. Iwashita, N. (2003). Negative feedback and positive evidence in task-based interaction. Studies in Second Language Acquisition, 25, 136. Kepner, C. G. (1991). An experiment in the relationship of types of written feedback to the development of second-language writing skills. Modern Language Journal, 75, 305313. Krashen, S. D. (1982). Principles and practice in second language acquisition. New York: Pergamon Institute of English. Lalande, J. (1982). Reducing composition error: An experiment. Modern Language Journal, 66, 140149. Liu, D., & Gleason, J. (2002). Acquisition of the article the by nonnative speakers of English. Studies in Second Language Acquisition, 24, 126. Long, M. (1996). The role of the linguistic environment in second language acquisition. In W. Ritchie & T. Bhatia (Eds.), Handbook of second language acquisition (pp. 413468). San Diego, CA: Academic Press. Long, M., Inagaki, S., & Ortega, L. (1998). The role of implicit negative feedback in SLA: Models and recasts in Japanese and Spanish. Modern Language Journal, 82, 357371. Lyster, R. (2004). Differential effects of prompts and recasts in form-focused instruction. Studies in Second Language Acquisition, 26, 399432. Mackey, A. (2007). Interaction as practice. In R. DeKeyser (Ed.), Practice in second language learning: Perspectives from linguistics and psychology (pp. 85110). Cambridge: Cambridge University Press. Mackey, A., & Philp, J. (1998). Conversational interaction and second language development: Recasts, responses, and red herrings? Modern Language Journal, 82, 338356. Master, P. (2002). Information structure and English article pedagogy. System, 30, 331348. Muranoi, H. (2000). Focus on form through interaction enhancement: Integrating formal instruction into a communicative task in EFL classrooms. Language Learning, 50, 617673. Nicholas, H., Lightbown, P., & Spada, N. (2001). Recasts as feedback to language learners. Language Learning, 51, 719758. Pica, T. (1991). Foreign language classrooms: Making them research-ready and re280 TESOL QUARTERLY

searchable. In B. Freed (Ed.), Foreign language acquisition research and the classroom (pp. 393412). Lexington, MA: D.C. Heath. Polio, C., Fleck, N., & Leder, N. (1998). If only I had more time: ESL learners changes in linguistic accuracy on essay revisions. Journal of Second Language Writing, 7, 4368. Robb, T., Ross, S., & Shortreed, I. (1986). Salience of feedback on error and its effect on EFL writing quality. TESOL Quarterly, 20, 8393. Robinson, P. (2001). Individual differences, cognitive abilities, aptitude complexes and learning conditions in second language acquisition. Second Language Research, 17, 368392. Sawyer, M., & Ranta, L. (2001). Aptitude, individual differences, and instructional design. In P. Robinson (Ed.), Cognition and second language acquisition (pp. 319 353). New York: Cambridge University Press. Schmidt, R. (1995). Attention and awareness in foreign language learning. Honolulu: University of Hawaii Press. Schmidt, R. (2001). Attention. In P. Robinson (Ed.), Cognition and second language instruction (pp. 332). Cambridge: Cambridge University Press. Schmitt, N., Drnyei, Z., Adolphs, S., & Durow, V. (2003). Knowledge and acquisition of formulaic sequences: A longitudinal study. In N. Schmitt (Ed.), The acquisition, processing, and use of formulaic sequences (pp. 5586). Amsterdam: John Benjamins. Semke, H. (1984). The effects of the red pen. Foreign Language Annals, 17, 195202. Sinclair, M. (1991). Corpus, concordance, collocation. Oxford: Oxford University Press. Skehan, P. (1998). A cognitive approach to language learning. Oxford: Oxford University Press. Snow, R. E. (1987). Aptitude complexes. In R. E. Snow & M. J. Farr (Eds.), Aptitude, learning and instruction (pp. 1359). Hillsdale, NJ: Lawrence Erlbaum. SPSS. (2002). SPSS, version 11.5 [software]. Chicago, IL: Author. Available from Truscott, J. (1996). The case against grammar correction in L2 writing classes. Language Learning, 46, 327369. Truscott, J. (1999). The case for the case for grammar correction in L2 writing classes: A response to Ferris. Journal of Second Language Writing, 8, 111122. Truscott, J. (2004). Evidence and conjecture on the effects of correction: A response to Chandler. Journal of Second Language Writing, 13, 337343.

Speeded Dictation Test
Name: Professor: Date:

There are 15 sentences in this test. The professor will read each sentence only once, so please listen carefully. After listening to each sentence, write down the sentence in the small notebook provided. Write the sentence as fast as you can. Try to write the sentence exactly as you hear it. After you have nished, turn to the next page and get ready for the next sentence. Once you nish each sentence, you must NOT return to the previous page. (Note: Do not worry about exact spelling. This is not a spelling test.)



Example 1
You will hear: Money cannot buy love. Then write, Money cannot buy love. And then, TURN TO THE NEXT PAGE.

For the Professor

Total 15 items (9 indenite articles, 13 denite articles) Example 1: I feel good when I speak English. Example 2: Tom speaks many languages. Hes very talented. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. I know the man who runs this college. The red car across the road looks suspicious. Do you know the pilot who ies this airplane? I saw a movie last night. The movie made me sad. Johns uncle was killed in a plane crash in New York. The man I met in New York became my husband. There was a temple near my house. The temple burned down yesterday. Can you move the car blocking my driveway? Please tell me who the leader of your club is. I know a lawyer. The lawyer wants to marry me. Yesterday, I saw a police officer chasing your dog. Jenny has a dog. The dog bit her boyfriend. John is a student of biology at BCC. Tom bought a car. He crashed the car the next day. There was a very kind doctor in my home town.

Error Correction Test
Please read each statement. Each statement has two sentences that are related. One of the sentences is underlined. The underlined sentence contains at least one error. There may be more than one error in each underlined sentence. Write out the underlined sentence correcting all the errors. (Note: There are no punctuation or spelling errors.) Example 1: Gloria have lived in New York during 2001. She really enjoys living in New York. Answer: Gloria has lived in New York since 2001. Example 2: John got a cold. He couldnt went to school yesterday. Answer: He couldnt go to school yesterday. 1. 2. 3. 4. 5. 6. 7. Mary used to living in Chicago. She lives in New York now. I look after a little girl and a little boy on Saturday. A little girl was smart but the boy isnt. I took three tests yesterday. Tests was so difficult. Tom quits smoking last week. He started smoking again because he is too stressed out. There might be easy way to get to Johns house. Can you show me his house on the map? I saw a man in a car across the street. I realized that the man driving car was my brother. Jen and Brad used to being so happy together. I couldnt believe that they broke up.



8. I saw a very interesting movie last night. I forgot the name of movie. 9. Last night I read a magazine and a news article. I dont know where a news article is today. 10. A young woman and a tall man were talking outside my house. Ten minutes later, a young woman was shouting at tall man. 11. I read book about New York. The author, however, was from California. 12. We rented a boat last summer. Unfortunately, boat hit another boat and sank. 13. We went to basketball game on Saturday. The players at the game were all very tall. 14. When you turn onto Paramus Road, you will see two houses: a blue one and a yellow one. I live in a blue house. 15. Is your uncle car salesman? Im looking to buy a car. 16. Bill was so drunk last night. He couldnt even recognized his girlfriend. 17. My mother was red yesterday. She will have to nd new job.



Task Familiarity and Interactional Feedback in Child ESL Classrooms

Georgetown University Washington, DC, United States


Highgate Primary School Perth, Western Australia

Edith Cowan University Bunbury, Western Australia

In this article we report on a study undertaken with ESL children between 7 and 8 years old. They carried out communicative tasks in pairs in second language classrooms. We examined patterns of their taskbased conversational interactions while we manipulated their familiarity with the procedure and content of the tasks. We found that learners working through unfamiliar tasks (in terms of both content and procedure) produced more clarication requests and conrmation checks and provided more corrective feedback on nontargetlike utterances to each other. However, learners engaged in procedurally familiar tasks had more opportunities to use feedback, and learners engaged in tasks that were familiar in both content and procedure showed more actual use of feedback.


t is increasingly accepted that conversational interaction in a second language plays an important facilitative role in second language learning (Gass, 2003; Long, 1996; Mackey, in press). Researchers are now asking how interaction works to bring about these positive developmental effects. Because most classroom interaction is task or activity oriented, one way to address that question is to examine how the structure of interaction varies across different types of tasks and task conditions. In this study, we examine learners familiarity with the procedure and content of a task, exploring the relationship between task familiarity and the

TESOL QUARTERLY Vol. 41, No. 2, June 2007


amount, nature, and immediate use of feedback learners receive during the interaction. Interaction is believed to facilitate the process of acquiring a second language by exposing learners to meaningful input, giving them feedback on their linguistic efforts (Long, 1996) and allowing them to produce and modify their output (Swain, 1985, 1995, 2005). Interaction is claimed to encourage learners to attend to linguistic form as they strive to understand and to make themselves comprehensible to their interlocutors. When learners pay attention to corrective feedback they receive on nontargetlike forms that they produce during interaction, it is claimed that they might notice mismatches between their current knowledge of the forms of the target language and the system of the target language. This feedback they receive during interaction may take a number of forms, including clarication requests, which are requests during interaction for more information to clarify an interlocutors previous utterance (see Example 1) and recasts, which are targetlike reformulations of interlocutors previous incorrect utterances (see Example 2). All examples in this article come from data collected for the current study.

Example 1. Clarication Request

(1) Learner A: Where do I put the girl balancing? (2) Learner B: What? Balancing? Whats that? (3) Learner A: You know . . . standing on one leg and you not fall down but still standing up so balancing. (4) Learner B: Oh! Like here standing on one leg on a horse, like this here on the horse. (5) Learner A: Yeahthats where I going now put it.

By requesting that Learner A clarify the word balancing (line 2), Learner B is able to obtain input that is more comprehensible and thus more suited to his particular developmental needs. In Example 2, Learner B recasts Learner As initial nontargetlike utterance. Researchers have argued that this juxtaposition of a learners nontargetlike utterance with a targetlike reformulation can help the learner notice the mismatch between the two, and that this attention to linguistic form is crucial for second language acquisition (e.g., Long, 1996; Schmidt, 1995). In Example 2, the learner appears to have noticed the mismatch; she reformulates her original utterance, correcting her error in line 3.

Example 2. Recast
(1) Learner A: The sun is top of page. (2) Learner B: Is at the top? (3) Learner A: Yes, is at the top.

A number of recent studies have focused on the outcomes of a range of different kinds of interaction (e.g., Ayoun, 2001; Braidi, 2002; Doughty & Varela, 1998; Ellis, Tanaka, & Yamazaki, 1994; Leeman, 2003; Long, Inagaki, & Ortega, 1998; Lyster & Ranta, 1997; Mackey, 1999; Mackey, Gass, & McDonough, 2000; Mackey & Philp, 1998; Oliver, 1995, 1998, 2000 and the summary in Mackey & Gass, 2006), with most nding that interaction has a positive impact on the second language learning process. However, the majority of these studies have been carried out with adults and adolescents and relatively little work has been done with children.

Research on Childrens Interaction

The few studies that exist on child interaction (Ellis & Heimbach, 1997; Harley, 1998; Lyster, 1998a, 1998b, 2001, 2004; Lyster & Ranta, 1997; Mackey & Oliver, 2002; Mackey, Oliver, & Leeman, 2003; Oliver, 1995, 1998, 2000, 2002; Van den Branden, 1997) have produced rather mixed results. For example, the literature suggests that children, with their supposedly more limited or immature attention spans or cognitive capacities, may not be able to recognize the corrections inherent in recasts and thus may benet more from feedback eliciting formal and metalinguistic information that directly focuses their attention on linguistic errors (Lyster, 1998a, 1998b, 2001, 2004; Lyster & Ranta, 1997). Lyster (2004), for instance, used a pretestposttestdelayed posttest design to compare the effect of three types of feedback on the ability of fth-grade students enrolled in French immersion classrooms to assign grammatical gender correctly. The types of feedback were prompts (including clarication requests, repetitions, metalinguistic clues, and elicitations), recasts, and no feedback. The group receiving prompts after their errors in grammatical gender obtained signicantly higher scores on the posttests than those students who had received recasts or no feedback, leading Lyster to conclude that recasts, when compared to other feedback options, are not necessarily the most effective type of feedback in communicatively oriented classrooms (p. 428), echoing his previous conclusion(Lyster, 2001) that in communicatively oriented classrooms where interactional exchanges are motivated by a variety of purposes and foci . . . young L2 learners may not readily notice targetnontarget mismatches in the interactional input (p. 268). Research by Lyster and others has shown mixed results in terms of the relationship between interactional features such as recasts and language benets in children (cf. Ellis & Heimbach, 1997, who did not nd a relationship between negotiation and vocabulary acquisition, and Van den Branden, 1997, who found relationships between negotiation and

subsequent output production and vocabulary range, but not improvements in grammatical accuracy or syntactic complexity). Researchers who have reported unequivocally positive ndings include Oliver (1998), who grouped 192 children into age- and gender-matched dyads and asked them to complete a picture-description task and an objectplacement task. She found that the children negotiated for meaning and provided recasts to their peers and that they beneted from these interactions in terms of receiving comprehensible input and feedback as well as opportunities for producing modied output. Another study that found a positive relationship between interactional feedback and child language development involved 22 children learning ESL who were interacting in dyads with adult nonnative English speakers (NESs) (Mackey & Oliver, 2002). Half of the children (the experimental group) received corrective feedback in response to nontargetlike question forms through interactional modications, and the other half (the control group) did not receive any corrective feedback in the interaction. The results of immediate and delayed posttests showed that signicantly more children in the experimental group improved in their formation of English questions than did children in the control group. Mackey and Oliver concluded that interactional feedback had facilitated second language development for the child learners and that this development had occurred more rapidly for children than previous research had shown for adult learners. Mackey and Silver (2005) found a similar positive outcome. The variation in the literature concerning the utility of interactional feedback for younger learners may be due in part to the different contexts in which the studies were conducted (laboratory and intact classrooms) and the types of tasks that were used, as well as to differences in individual learner variablesfor example, the childrens rst languages and the different age ranges studied. Oliver and Mackey (2003) have also suggested that the primary focus of the interactional context (i.e., whether it is oriented explicitly toward language, or toward content, communication, or classroom management) may be related to the learning outcomes of recasts. Nevertheless, the increasing evidence that interaction does play a role in childrens second language learning cannot be ignored.

Tasks and Task Familiarity

Given the growing recognition that interactional modications can facilitate both adult and child second language learning, researchers and practitioners have sought to identify those learning conditions which best promote helpful interactional patterns. One particularly productive

approach has been the use of communicative tasks. Tasks, generally speaking, refer to bounded classroom activities in which learners use language communicatively to achieve an outcome, with the overall purpose of learning language (Bygate, 1999a, p. 186). The crucial elements of this denition are that the activity is goal oriented (i.e., meaningful), that the learners interact with one another to achieve the goal at hand, and that some focus on form (i.e., attention to the linguistic features of the target language) is involved. Tasks vary, however, along a number of different dimensions, including task type (e.g., whether the information ow is one way or two way), cognitive complexity (e.g., what type of reasoning demands are involved), task implementation (e.g., whether learners are provided with planning time, how they are grouped, and whether they are familiar with the task), and perceived difficulty (Ellis, 2003; Robinson, 2001a, 2001b, 2003). One recent line of research in the eld of second language acquisition has been the investigation of these sorts of task conditions on learner interactional patterns and language learning outcomes (e.g., Bygate, 1996, 2001; Gass, Mackey, Alvarez-Torres, & Fernndez-Garca, 1999; Lynch & Maclean, 2000, 2001; Mehnert, 1999; Plough & Gass, 1993; Robinson, 1995, 2001a, 2001b, 2003; Salaberry & Lopez-Ortega, 1998; Skehan, 1998; Skehan & Foster, 1997, 1999; Swain & Lapkin, 2001; Yuan & Ellis, 2003; Yule, Powers, & MacDonald, 1992; for a comprehensive review, see Ellis, 2003). The focus of the current study is one task condition in particular, that of task familiarity.

Task Familiarity
Different types of task familiarity have been explored in previous research, including (a) familiarity with ones interlocutor, (b) experience with carrying out a particular task or type of task before (also known as task repetition or rehearsal), and (c) knowledge of the subject matter or content that is the focus of the task (e.g., Bygate, 1996, 2001; Gass et al., 1999; Lynch & Maclean, 2000, 2001; Plough & Gass, 1993; Skehan & Foster, 1997, 1999). In general, being familiar with a task is believed to be benecial in terms of learners ability to focus on form, and thus to have a positive effect on second language acquisition. As mentioned earlier, it has been argued that attention to form facilitates interlanguage development and that for second language learners to develop linguistic skills in the target language, they must focus not only on meaning (semantic processing) but also on form (syntactic processing). However, some researchers claim that learners, because of their limited working memory capacities (usually operationalized as the abilities to process and store information

simultaneously), are not always able to engage in processing both form and meaning simultaneously (see, e.g., VanPatten, 1990). These researchers maintain that learners tend to prioritize a focus on meaning and, in doing so, may overlook certain linguistic featuresespecially those that are nonsalient, redundant, or of low communicative value. However, if learners are familiar with a task (e.g., if they remember having described a particular picture to their partner before), attention that otherwise would be devoted to learning the task procedure or expressing new meaning can be redirected to form. Thus, familiar tasks may mitigate the limitations of learners working memories. As Bygate (1999b) explains:
Familiarity gives us the time and awareness to shift attention from message content to the selection and monitoring of appropriate language. By enabling a shift of attention, learners may be helped to integrate the competing demands of uency, accuracy and complexity. (p. 41)

In a study investigating the effects of task familiarity on language production, Bygate (2001) compared 48 ESL learners performances on two types of task. At the beginning of the 10-week period of the study, all learners performed two tasks: an interview and a retelling of a video clip. Over the following 10 weeks, one group practiced the interview task type and the second practiced the narrative task type. At the end of the 10-week period, Bygate examined the groups performance on both task types, using one repeated version and one new version of each. When learners were asked to perform exactly the same task that they had performed 10 weeks earlier, their speech was more accurate and complex than it had been on the rst performance. However, these linguistic improvements did not carry over to new exemplars of the two types of task. Gass et al. (1999) found similar results in their study of how content familiarity affects the language produced by 103 students learning Spanish as a second language. Gass et al. investigated how repeating a narrative-retelling task affected the learners overall prociency, morphosyntactic accuracy (operationalized as use of the verbs ser and estar), and lexical sophistication (i.e., the use of low-frequency words). For the retelling task, they used videotaped episodes of Mr. Bean, a sketch comedy in which the title character, played by Rowan Atkinson, suffers a series of unfortunate events, such as getting stuck as he tries to get dressed in a very small car on his way to an appointment. After watching the video, learners were asked to provide a narrative retelling in Spanish of the events in the episode. The Same Content Group viewed the same Mr. Bean episode three times, followed by a new episode of Mr. Bean. The Different Content Group performed the same task, but watched a different Mr. Bean episode on each of the four viewing days, and the Con290 TESOL QUARTERLY

trol Group saw the rst episode on Day 1 and the fourth episode on Day 4, but no episodes on the other days. Gass et al. found that when learners repeated exactly the same task, their use of the Spanish verb estar improved overall, and they showed greater lexical sophistication. However, like Bygate (2001), Gass et al. found that these improvements did not generalize to the new exemplar of the same type of task. The literature contains some evidence that content familiarity may also assist learners performances when they complete similar but not identical tasks. Lynch and Maclean (2000, 2001), for example, investigated the effects of task repetition in a poster-carousel session, nding a variety of improvements. In the task, each pair of participants was given a different research article and instructed to make a poster based on it. These posters were then displayed in a large room, and one person from each pair (the host) stood by the poster to answer questions about it, while the other person (the visitor) circled the room asking questions. The roles were then switched, and the cycle was repeated six times. As the researchers explain, the tasks were not identical to each other:
[I]t is not a second (or third, etc.) performance by one speaker on an identical task with the same listener; repetition in the case of the carousel means something more like recycling, or retrial . . . where the basic communication goal remains the same, but with variations of content and emphasis depending on the visitors questions. (Lynch & Maclean, 2000, p. 227, emphasis in the original)

Lynch and Maclean (2000, 2001) found that in the later iterations of the task, participants paid greater attention to language through, for example, more self-corrections of vocabulary and pronunciation; greater attention to content (e.g., speaking more slowly at conceptually difficult points and incorporating content introduced by previous interlocutors); and linguistic improvements, such as increased grammatical accuracy and lexical precision. They speculate that these gains were possible in part because the learners may have felt more relaxed as the task progressed, and also because recycling the task allowed the learners to shift their attention from what they wanted to say (i.e., the content or meaning) to how they were going to say it (i.e., the form). In a similar vein, certain kinds of task familiarity may also foster an environment in which learners are more likely to indicate a lack of understanding. Two studies by Plough and Gass (1993) support this assertion. Plough and Gass investigated the relationship between task familiarity and the types of discourse features present in the conversations. In the rst study, investigating interlocutor familiarity, pairs of nonnative speakers (NNSs) who were familiar with each other and pairs who were unacquainted with each other completed two tasks. The researchers found that acquainted dyads used a greater number of conrTASK FAMILIARITY AND INTERACTIONAL FEEDBACK IN ESL CLASSROOMS 291

mation checks (i.e., moves made by listeners during interaction to conrm that they have understood the message correctly) and clarication requests (requests made for more information) during the tasks than the unacquainted dyads. In the second study, examining task familiarity, the researchers found similar results. That is, the pairs of NNSs who had experience with similar tasks used more conrmation checks and clarication requests than did those pairs who did not have this experience. Given the importance of these interactional features for the development of second language skills, such ndings on interlocutor and task familiarity may have implications for second language development.

Summary and Research Questions

In summary, previous studies of task familiarity have provided evidence that repeating an identical task positively impacts learners linguistic performance. Although less support exists for the benecial effects of familiarity on new exemplars of the same type of task, there is some evidence nonetheless that task familiarity may help learners to redistribute their limited attentional resources so that they may pay attention to linguistic form and content and engage in interactional processes that have been associated with second language development. What we have to keep in mind, though, is that all of these studies have been carried out with adults. Given the differences in the ways adults and children approach classroom language learning as well as hypothesized differences in their memory and processing capacities, it remains an empirical question whether being familiar with a task is benecial for younger learners in the same way as it is for older learners. For example, considering that younger and older learners may not have the same priorities with respect to focusing on form and meaning, does task repetition help younger learners focus on form? And in view of childrens more limited attention span and perhaps greater desire for novelty, do the linguistic benets of repeating a task outweigh the possibility that children will become bored? (Plough & Gass, 1993, showed that learners familiar with a task seemed less interested; unfamiliar learners seemed more involved.) Although the second language literature has suggested that child second language learners benet from being in classes where they have familiar routines and expectations and know what to do procedurally (Wong-Fillmore, 1985), the lack of studies on the interactional structure of child second language learners conversation makes further investigation necessary. The current study contribute to this line of research by examining the relationship between task familiarity and the interactional feedback provided to younger second language learners. The specic research questions addressed were as follows: (a) Are there

differences in the amount, nature, and immediate use of interactional feedback depending on learners familiarity with the procedures involved in carrying out a task? And (b) are there differences in the amount, nature, and immediate use of interactional feedback depending on learners familiarity with the content involved in carrying out a task?

METHOD Participants
Forty children aged 7.08.5 took part in the study. All children were classied as being of intermediate English prociency by their language programs, were in ESL classes at the same level, and were familiar with each other from their classes and activities in the semester of the study. They came from a variety of rst language backgrounds, as shown in Appendix A. Each child had completed between 10 and 14 months of English instruction in Australia. They were selected as a stratied random sample on the basis of gender and were paired to form 10 male male and 10 femalefemale learner dyads.

Materials and Design

The communicative tasks were designed so that specic types of task familiarity could be controlled. Procedural familiarity was operationalized as the participants knowledge and prior experience of procedural aspects of a task, including the different roles participants take to effectively complete a particular task type. At the outset of the study, none of the learners had any prior experience carrying out tasks of the type used (as reported by their teachers and observed in the classroom). To manipulate the variable of procedural familiarity, learners in the class were divided so that half of them carried out practice tasks, while the other half of the class continued their regular lessons in another room so that they remained unfamiliar with the procedure for that particular task. Then the learners who had remained in the classroom were taken out and familiarized with the procedure of a different task. Practice tasks were done twice, one week before the experimental tasks were carried out. At the conclusion of the study, the learners were asked whether they had ever done an activity of that type before to ascertain whether the tasks were indeed familiar or unfamiliar as intended. Learners in all familiar dyads reported that they had carried out similar tasks before, and learners in all unfamiliar dyads reported that they had not.

Content familiarity was operationalized as knowledge of the subject or topic and the discourse domain inherent in the task. The subject matter selected for the rst two tasks was familiar to all of the learners. The topics chosen (The Farmyard and The Park) were based on the classroom themes operating at the time, so that both groups had been exposed equally to the content. For the second two tasks (Assemble a Circus and Secret Island), the topics had not been covered in class in the 18 months prior to the experiment. Three steps were taken to make sure participants received equal treatment exposure: (a) researcherteacher interviews to ascertain what topics the teachers and students had covered prior to the study, (b) carefully selected topics in researcher teacher collaborations, (c) group interviews, in which all the participants and their teachers were asked about their prior exposure to the topic and what they knew about the topic. This interview process with teachers and learners ensured that all participants understood the same basic content. As with the procedure, task content was made familiar to half of the class one week before data collection through three 30-minute talks using visual aids based on the task topics. In this way, essential vocabulary and aspects of language inherent in the topic were provided to the learners. Again, familiarity and unfamiliarity with content were conrmed through nal interviews with teachers and learners. Each dyad completed four tasks. First they completed two tasks with familiar content. One of these tasks followed a familiar procedure and the other followed an unfamiliar procedure. Two weeks later each dyad completed two tasks that followed a familiar procedure. One of these tasks had familiar content, and the other had unfamiliar content. This design (and the counterbalanced order of presentation for the tasks) is illustrated in Figure 1. The participants engaged in the tasks at designated learning centers situated in the classroom where they did their daily work. The dyads sat at opposite sides of a desk, facing each other, with a screen 30 centimeters high in the middle of the desk to keep them from seeing each others work. They were, however, still able to see each others faces. The researcher sat at the side of the desk to supervise, observe, and introduce the tasks. The researcher read the task instructions, which involved explaining the drawing or picture placement aspects of the task, from an instruction sheet (see Appendix B). The learners were also informed that they would have an opportunity to compare their results after they had nished. They were then given 5 minutes of silent planning time to gather their thoughts about how they would approach their part of the task. No discussions took place during this time. Learners were given 20 minutes to complete the task. When they nished, their sharing of the results amounted to approximately another 10 minutes.

FIGURE 1 Procedural and Content Familiarity



The researchers made transcriptions of the rst 100 utterances in each task. (In the few cases where there were fewer than 100 utterances, all utterances were transcribed.) This method ensures comparability of data where the tasks consist of a range of utterances, and is based on second language interaction research (Braidi, 2002; Oliver 1995, 1998, 2000, 2002; Oliver & Mackey, 2003), as well as rst language research (e.g., Wong-Fillmore, 1985). Alternative methods of ensuring comparability of data often involve ratio percentages or some measure of equivalence for data density. Because interaction research with children has traditionally used the rst 100 utterances, this measure also ensured that the current study was comparable with relevant prior work. All transcriptions were checked by a trained observer to ensure consistency of utterance segmentation and utterance content. The entire corpus consisted of 7,915 utterances.

Coding and Analysis

The coding process is illustrated in Figure 2 (from Mackey, Oliver, & Leeman, 2003, reprinted with permission). The coding process consisted of rst categorizing initial NNES utterances as either targetlike or nontargetlike. Because the current study sought to explore the provision and nature of feedback, only exchanges that began with a nontargetlike utterance were included in the analyses. Learner responses to nontargetlike utterances were then coded according to whether they provided feedback. Each of these features was explained and examples were provided. Recasts, conrmation checks, comprehension checks, and clarication requests were all classied as interactional feedback, given that they all imply unacceptability of the original utterance. Next, interlocutor feedback was classied according to whether learners were provided with the opportunity to use that feedback in producing modied output. When an interlocutor provided negative feedback but then continued her turn without affording the learner an opportunity to produce modied output, this was classied as no opportunity for modied output. In contrast, incidences of feedback that provided learners with the opportunity to reformulate their original utterances were coded as opportunity for modied output. Learner utterances immediately following feedback with opportunity for modied output were coded according to whether the learner corrected the original grammatical error. Although immediate incorporation of feedback or the production of modied output may not be a reliable indicator of

FIGURE 2 Coding Categories

the long-term effects of negative feedback (Mackey & Philp, 1998), various researchers have argued that learners ability to modify their output in response to feedback suggests that they have made some use of that feedback (see, e.g., McDonough, 2005). Intercoder agreement, measured as simple percentage agreement on 30% of the whole dataset, was 96%. Next, we turn to some examples from the data in the current study. Example 3 shows the learners initial nontargetlike utterance in line 1, followed by feedback in line 2 that provides no opportunity for modied output.

Example 3. No Opportunity for Modied Output

(1) Learner A: The magician is hold rabbit. (2) Learner B: Oh! Holding the rabbit. I see birds too. Where can I put? (3) Learner A: On the roof.

In Example 4, in contrast, the learners initial nontargetlike utterance in line 1 is followed by feedback in line 2 which does provide an opportunity for negative feedback.

Example 4. Opportunity for Negative Feedback

(1) Learner A: Next to fence is dog. (2) Learner B: Next to the fence is a dog? (3) Learner A: Yes, is a dog.

The different types of interactional feedback that were used in familiar/unfamiliar procedural and content tasks were compared using twotailed paired t tests. The provision of feedback (negotiation strategies and recasts), and the opportunity for and production of modied output according to familiarity with procedure and content were submitted to hierarchical log linear analysis along with 2 Habermans (173) residual tests for the source of signicant differences. This analysis follows commonly used procedures for frequency data analysis where more than two-way associations are involved and was based on previous research involving the same type of analyses (Braidi, 2002; Oliver, 1995, 2000).

The research questions asked whether there were any differences in the amount, nature, and immediate use of interactional feedback depending on whether participants were familiar or unfamiliar with (a) the typical procedures for carrying out a task and (2) the content of the task. Dyad types were balanced for gender following prior research showing gender might impact childrens interactional processes (Oliver, 1998).1 In general, signicant, albeit complex, relationships were found between familiarity with task procedure and content, and the patterns of interaction that occurred while the task was being carried out. For example, as Table 1 shows, dyads working through an unfamiliar task procedure produced signicantly more clarication requests (M = 7.31) and conrmation checks (M = 7.88) than those working through a familiar task procedure (M = 4.22, M = 3.74, respectively). The results were the same when the content was unfamiliar: Dyads produced signicantly more clarication requests (M = 9.12) and conrmation checks (M = 8.18). On the other hand, dyads working through familiar tasks used signicantly more comprehension checks (procedural, M = 1.36; content, M = 1.76) than dyads working through unfamiliar tasks (procedural, M = 0.27; content, M = 0.46). Similarly, as Table 2 shows, dyads engaged in unfamiliar tasks provided signicantly more feedback on nontargetlike forms (procedural,

In this study, on all tests, gender was nonsignicant.



TABLE 1 Mean Percentage of Negotiation Strategies Used by Dyads Working Through Familiar and Unfamiliar Tasks Procedure Familiar Negotiation Clarication Conrmation Comprehension Self-repetition Other repetition M 4.22 3.74 1.36 7.03 7.77 SD (1.88) (1.59) (1.82) (3.97) (2.94) Unfamiliar M 7.31 7.88 0.27 8.15 8.70 SD (2.45) (2.67) (0.83) (3.5) (3.54) Sig. (t) 4.46* 5.94* 2.43* 0.94 0.079 Familiar M 4.48 3.83 1.76 5.49 7.94 SD (2.05) (1.88) (1.83) (2.69) (2.64) Content Unfamiliar M 9.12 8.18 0.46 8.40 9.55 SD (3.39) (3.49) (1.34) (3.11) (2.88) Sig. (t) 5.23* 4.89* 2.56* 2.25* 1.83

Note. *p < 0.05. Sig. = signicance.

46.1; content, 48.6), than dyads engaged in familiar tasks (procedural, 32.5; content, 35.4). However, when we look at opportunity to use that negative feedback, in data shown in Table 3, we see that dyads working through procedurally familiar tasks had signicantly more opportunity to use that feedback (87.0) than dyads engaged in unfamiliar tasks (71.8). However, dyads using familiar (91.3) and unfamiliar (88.0) content showed no signicant differences in terms of opportunity to use negative feedback. Likewise, when we examine the data on actual use of corrective feedback shown in Table 4, we see once again that dyads working through familiar tasks produced signicantly more modied outputthat is, they used the feedback (procedural, 74.7; content, 80.0)than dyads working through unfamiliar tasks (procedural, 20.2; content, 17.5). These results are summarized and shown in Table 5.

The results of this research show that learners working through unfamiliar tasks, both in terms of content and procedure, produced (a)
TABLE 2 Provision of Response to Nontargetlike Turns by Dyads Working Through Familiar and Unfamiliar Tasks Procedure Familiar Negative feedback Ignore 32.5* 67.5* Unfamiliar 46.1* 53.9*

Content Familiar 35.4* 64.6* Unfamiliar 48.6* 51.4*

Note. Procedure: 2 (1, n = 574) = 8.090, p < 0.05; Content: * = Haberman (1973) adjusted residuals > 2 or < 2.

(1, n = 541) = 7.992, p < 0.05;



TABLE 3 Opportunity to Use Feedback by Dyads Working Through Familiar and Unfamiliar Tasks Procedure Familiar Opportunity existed No opportunity Note. Procedure: < 2.

Content Familiar 91.3 8.7 Unfamiliar 88.0 12.0

Unfamiliar 71.8* 28.2*

87.0* 13.0*

(1, n = 231) = 7.770, p < 0.05; * = Haberman (1973) adjusted residuals > 2 or

more clarication requests and conrmation checks and (b) more negative feedback on initial nontargetlike utterances than learners working through familiar tasks. However, there were both more opportunities to use feedback by learners doing procedurally familiar tasks and more actual use of feedback by learners engaged in familiar tasks (in terms of content and procedure). A close examination of the data sheds more light on these patterns. For example, if learners are unfamiliar with the content of a task, they are likely to work collaboratively to develop some mutual understanding about the topic. In contrast, if the procedure of a task is unfamiliar, learners often need to rst discuss and develop a common understanding of how to undertake the task. We turn now to a more detailed discussion of the ndings.

Content Familiarity
It may be that familiarity with the content of a task is helpful for learners because they do not need to think about conveying meaning; instead, they can focus on the linguistic forms needed to convey that meaning, as Van Patten (1990) and others have argued. For example, Bygate and Samuda (2005) explain,
An initial encounter with a task can be seen as creating a holistic representation of the task, along with the experience of handling it in real time. This representation and the accompanying experience can be
TABLE 4 Use of Feedback by Dyads Working Through Familiar and Unfamiliar Tasks Procedure Familiar Use Ignore 74.7* 25.3* Unfamiliar 20.2* 79.8*

Content Familiar 80.0* 20.0* Unfamiliar 17.5* 82.5*

Note. Procedure: 2 (1, n = 181) = 53.959, p = <0.05; Content: * = Haberman (1973) adjusted residuals > 2 or < 2.

(1, n = 198) = 77.475, p < 0.05;



TABLE 5 Summary of Results for Dyads Working Through Familiar and Unfamiliar Tasks Task procedure Familiar Clarication requests Conrmation checks Comprehension checks Self repetition Other repetition Provision of NF Opportunity to use NF Use of NF Unfamiliar + + + + + + + + + + Task content Familiar Unfamiliar + +

Note. NF = Negative feedback; + = signicant results, each indicating where the higher amount occurred.

stored, creating a kind of plan which can be drawn on on a second occasion, enabling the learner to integrate a broader range of their resources into their performance. (p. 38)

Learners may also take more linguistic risks when the content of a task is familiar. It seems plausible that learners might be more likely to rely on automated aspects of their interlanguage when carrying out a task for the rst time, but that they may be ready to take more risks when they encounter the same task again, stretching their interlanguage further. However, this is not to say that unfamiliar tasks have no place in the child ESL classroom. For example, when the content of the task was unfamiliar to the children in this study, their interactions included signicantly more clarication requests and conrmation checks. It is possible that when a task is unfamiliar, children, like adults, may need to collaborate and modify their interactions to develop an understanding of the topic, as shown in Example 5. These modications are supposedly helpful for their second learning process.

Example 5. Learners Unfamiliar With Content

Learner Learner Learner Learner Learner Learner Learner Learner Learner Learner Learner A: B: A: B: A: B: A: B: A: B: A: Am there is . . . There is something like . . . What? Like a where boats go. Boats go? Yeh in the water. What you saying? I dont know I forgot what how to say but . . . Boats can stay there in the water. In the water? Yeh for boats to stay there all the time. Like a place like a jetty? Yeh for boats.


As shown in this example, the potential ambiguity that results from being unfamiliar with the content can create a situation that requires learners to negotiate for meaning and to provide and use interactional feedback. Given the importance ascribed to interactional modications such as these for second language development, it cannot be concluded that unfamiliar tasks are less benecial for second language learning than more familiar ones. A need to negotiate during unfamiliar tasks also arose from the fact that learners who were unfamiliar with the content tended to produce lengthy descriptions that caused confusion, as shown in Example 6.

Example 6. Learners Unfamiliar With Content

Learner A: You have to draw like a long snake going its there going in a hole. Well its going into the hole and then out. Its got to be long like a snake going all around. Learner B: A snake? Learner A: Yes but like where trains go on. Learner B: Train lines? Learner A: Yeh like that. Learner B: I draw train lines going in a hole in the ground? Learner A: Yeh But it goes in a mountain and out on other side too. Learner B: Is train lines going through that mountain here. Learner A: Yeh, a mountain.

In contrast, learners familiar with the content were better able to express exactly what they wanted to say. Conversations proceeded with more concise and correct terminology. However, the conversation between these dyads was at times elliptical and incomplete in surface form. When this occurred, learners seemed to be able to successfully rely on their background knowledge to interpret the intended meaning (see Example 7), reducing the need to negotiate for meaning, as pointed out initially by Sato (1986).

Example 7. Learners Familiar With Content

Learner A: Draw a train track going all around. And through the mountain. Learner B: A tunnel. Learner A: Yep.

Procedural Familiarity
As with the children who were unfamiliar with the content of a task, children who were unfamiliar with the procedure for carrying out a task

also negotiated more than children who were familiar with the procedure, producing signicantly more clarication requests and conrmation checks. For instance, in Example 8, the children use several clarication requests to negotiate a common understanding of how to work through the particular task.

Example 8. Learners Unfamiliar With Procedure: Clarication Requests

Learner Learner Learner Learner Learner Learner Learner Learner Learner Learner Learner Learner Learner Learner Learner Learner A: B: A: B: A: B: A: B: A: B: A: B: A: B: A: B: You go rst. What I do? What? I dont know. Well, I see this you have it the horses. The horses? Yeh. I give you it. Now put on the picture. What? I dont know. I think yeh. I tell you where it goes then you it on the picture the picture. The picture? Yeh. Near the truck it goes. The truck? I put it near the truck? Now? Yeh see I tell you and you put it there. On the spot on the picture. Yeh.

Age might play a particular role in the importance of familiarity with task procedure. Children, with their less extensive experience with all forms of tasks and less developed skills in carrying out collaborative work, may nd familiarity with a task particularly important in their ability to carry out the task successfully, in terms of both the linguistic and nonlinguistic goals of the task. Without procedural familiarity, the children often seemed to seek reassurance from their partner (who was equally unfamiliar with the task) that they were doing what they were supposed to, in other words, that they were working through their particular role in a task correctly. Learners often sought reassurance through conrmation checks, as shown in Example 9.

Example 9. Learners Unfamiliar With Procedure: Conrmation Checks

Learner Learner Learner Learner Learner Learner A: B: A: B: A: B: Now draw some ducks in a pond. Yeh. Draw now. Now? All the time you tell me now what to draw? Yeh I have the picture and you need to draw it. No looking. Oh, I draw ducks then all over here.


Children familiar with a tasks procedure, on the other hand, produced more comprehension checks than children unfamiliar with the procedure (see Example 10). However, in this case, although the nding was signicant, the absolute differences were smaller, and no condition had many comprehension checks. Nevertheless, when the child learners were familiar with the procedure of a task, they became more sensitive to the kinds of potential problems that their partners might experience and worked to resolve them before they arosea nding that has been reported with adult second language learners as well (e.g., Yule, Powers, & MacDonald, 1992). Our nding brings into question claims about childrens egocentric natures. In these data, the child second language learners (even those aged 78 years) are cognizant, like adults, of their partners conversational needs.

Example 10. Learners Familiar With Procedure: Comprehension Check

Learner A: Draw a tree big tree on the left. You understand on the left? You know the left? Learner B: Yep, the left.

In addition, learners who were familiar with the procedure of a task were able to express ideas and concepts more uently than learners who were unfamiliar with procedure. It seems that learners familiar with the procedure were able to rely on language structures that they had previously acquired. In other words, because they had completed similar tasks before, the children were better equipped to successfully communicate particular concepts and ideas, as shown in Example 11.

Example 11. Learners Familiar With Procedure: Communicating Ideas and Concepts
Where do I put the pigs? In the fence. Where do I put the boy? Near the chickens. Because he giving food to the chickens. What about the truck? Where does it go? Learner A: Oh yeh, on the hill. Learner Learner Learner Learner A: B: A: B:

In contrast, learners unfamiliar with a task paused more frequently and expressed uncertainty about what they were doing, as evidenced through their greater use of hedges such as maybe, could be, I think, and so on. Also, the learners who were unfamiliar with procedure had a greater incidence of false starts and errors in lexis when instructing each other in what they had to do next. As a result, they often needed to negotiate to understand what their interlocutors were saying to them, as shown in Example 12.

Example 12. Learners Unfamiliar With Procedure: Negotiating Understanding

Learner A: I pig have a pig? You got it? Learner B: Yeh. Learner A: Well what can I do . . . No where can I seeyou see . . . Oh it go where? Learner B: The pig? Learner A: Yeh. Learner B: Could be . . . somewhere . . . near the dog . . . maybe. Learner A: What?

Further examination of the data highlights another difference in the interactional structure associated with familiarity. The data indicate that dyads familiar with a tasks procedure often provided a setting for the task by describing the content in some detail. For example, they provided a description of some of the pictures in each task before they gave instructions as to what they had to do to complete the task correctly, as shown in Example 13. This setting provided a focus that helped their partner interpret and infer what was being talked about. In contrast, most of the unfamiliar dyads did not provide a clear setting. Therefore, exchanges were not interpreted accurately at times (see Example 14), creating a need to negotiate for meaning.

Example 13. Learners Familiar With Procedure: Creating a Setting

Learner A: I have a farm. And some farm animals. The farm animals everywhere. There some ducks in a pond. The pond on the left side. And a truck on the right. Four ducks near the pond. The ducks are talking. The middle I see two horses together. Ha! Ha! and pigs talking too much in the bottom. Your turn.

Example 14. Learners Unfamiliar With Procedure: Misinterpreted Exchanges

Learner Learner Learner Learner Learner Learner Learner A: B: A: B: A: B: A: Boy in mid. What? Boy in mid. I dont know what you saying? Draw a boy. Yeh but where? Not top not bottom. But the in the mid.


Learner Learner Learner Learner

B: A: B: A:

In the mid? Mid oh middle! Yes, there he ying a kite. Oh what now? Stop so I can draw. A boy in the middle.

Pedagogical Comments
This study shows that teachers might benet from being aware of the different kinds of interactional modications that can result from different types of familiarity, although as Edwards and Willis (2005) point out, further research into task-based teaching and learning are necessary. In this study, clarication and conrmation, and the resulting adjustments to form, occurred more often in unfamiliar tasks, but feedback was actually used more often in familiar tasks, which led to modications in learner output. So both types of tasks had benets, which leads to the conclusion that the more varied a teachers tools are in terms of tasks, the better for the learners. What do we mean by this? Increasing child learners familiarity with the content and procedure of tasks may help create contexts for learners to focus on the linguistic form of their utterances. As Wong-Fillmore (1985) suggested, and as some of the children in the current study expressed, when tasks are familiar to younger learners, they sometimes enjoy them more. As the learner in Example 15 says, Its great doing this. I done before (line 1).

Example 15. Learners Familiar With Task

Learner A: Its great doing this. I done before. Learner B: Yeh! You properly do so I know how do it right. Learner A: Of course I will do it properly! Im very good! This I know how to do it properly! Learner B: I do it properly too. Now I start. This is fun.

However, as the results of this study also indicated, learners engaged in unfamiliar tasks more often use certain types of interactional modications, such as clarication requests and conrmation checks; thus, unfamiliar tasks also support the language learning process. As many researchers have commented, it is when childrens linguistic skills are stretched, for whatever reason, that they develop (e.g., Tarone & Liu, 1995). Unfamiliar tasks might help with this stretching process. Like all children, those in second language classrooms can also become bored if they are asked to repeat identical tasks, or even the same task type too many times. Not all tasks were received like the task in Example 15. For this reason, it may be most benecial for teachers to include a mix of old and new tasks in the classroom. Another alternative

would be to present variations of new tasks over time, modifying the tasks slightly in some respects. In making these changes, the teacher may be able to isolate and retain some element that holds childrens interest, so that some aspects of the process and content are familiar to the children. This background experience may enable more active participation, which again, can provide opportunities for language learning.


In considering recommendations, however, the limitations of the current study must be noted. With only 40 learners, the study was relatively small scale. In addition, because this was fundamentally a classroom study, the lack of randomization is a factor. Future research will need to examine larger populations of learners before any generalizations can be made. In addition, because of the small number and constrained type of tasks used in this study, it remains to be seen how other task characteristics (e.g., task type and complexity) interact with the various types of familiarity. Clearly, a great deal more research is required to conrm the extent to which these ndings might apply to children of different age groups carrying out different tasks in a variety of settings. It would also be interesting to examine the interactional patterns of dyads and small groups where one interlocutor is more or less familiar with the content or procedure than the other. In a migrant ESL setting where children come and go, this is often the case. Notwithstanding these limitations, this study showed that characteristics of learners task-based conversational interactions are related to their familiarity with the procedure and content of tasks. Learners who were working through unfamiliar tasks (in terms of both content and procedure) produced more clarication requests and conrmation checks, as well as more corrective feedback on nontargetlike utterances. Learners who engaged in procedurally familiar tasks had more opportunities to use feedback, and learners who engaged in tasks that were familiar in both content and procedure were more likely to use the feedback. Given the importance of these interactional features to the development of second language skills, the results suggest that teachers should consider task familiarity as one of the factors important in the design and sequencing of language teaching materials.

Alison Mackey has studied the effects of input, interaction, and feedback and the role of cognitive factors such as attention and memory in second language learning. She

has been an English language teacher, given workshops for teachers, and taught applied linguistics in the United Kingdom, Japan, Australia, and the United States. Alec Peter Kanganas is a teacher and deputy principal at Highgate Primary School. He has taught ESL for a number of years and recently completed his masters degree in education at Edith Cowan University, Western Australia. He is the Western Australian president of the Australian Literacy Educators of Australia. Rhonda Oliver was an English language teacher in schools working with migrant children in Australia before commencing her academic career. She has taught TESOL and applied linguistics at universities for several years and currently holds the chair in education at Bunbury Campus, Edith Cowan University, Western Australia.

Ayoun, D. (2001). The role of negative and positive feedback in the second language acquisition of the pass compos and imparfait. Modern Language Journal, 85, 226243. Braidi, S. M. (2002). Re-examining the role of recasts in native-speaker/nonnativespeaker interactions. Language Learning, 52, 142. Bygate, M. (1996). Effects of task repetition: Appraising the developing language of learners. In J. Willis & D. Willis (Eds.), Challenge and change in language teaching (pp. 134146). London: Heinemann. Bygate, M. (1999a). Quality of language and purpose of tasks: Patterns of learners interaction on two oral communication tasks. Language Teaching Research, 3(3), 185214. Bygate, M. (1999b). Task as context for the framing, reframing and unframing of language. System, 27, 3348. Bygate, M. (2001). Effects of task repetition on the structure and control of oral language. In M. Bygate, P. Skehan, & M. Swain (Eds.), Researching pedagogic tasks: Second language learning, teaching, and testing (pp. 2348). Harlow, England: Longman. Bygate, M., & Samuda, V. (2005). Integrative planning through the use of task repetition. In R. Ellis (Ed.), Planning and task performance in a second language. (pp. 3774). Amsterdam: John Benjamins. Doughty, C., & Varela, E. (1998). Communicative focus on form. In C. Doughty & J. Williams (Eds.), Focus on form in classroom second language acquisition (pp. 114138). Cambridge: Cambridge University Press. Edwards, C., & Willis, J. (Eds.). (2005). Teachers exploring tasks in English language teaching. London: Palgrave Macmillan. Ellis, R. (2003). Task-based language learning and teaching. Oxford: Oxford University Press. Ellis, R., & Heimbach, R. (1997). Bugs and birds: Childrens acquisition of second language vocabulary through interaction. System, 25, 247259. Ellis, R., Tanaka, Y., & Yamazaki, A. (1994). Classroom interaction, comprehension and the acquisition of L2 word meanings. Language Learning, 44, 449491. Gass, S. M. (2003). Input and interaction. In C. Doughty & M. H. Long (Eds.), Handbook of second language research (pp. 224255). London: Blackwell. Gass, S. M., Mackey, A., Alvarez-Torres, M. J., & Fernndez-Garca, M. (1999). The effects of task repetition on linguistic output. Language Learning, 49, 549581.

Haberman, S. J. (1973). The analysis of residuals in cross-classied tables. Biometrics, 29, 205220. Harley, B. (1998). The role of form-focused tasks in promoting child L2 acquisition. In C. Doughty & J. Williams (Eds.), Focus on form in classroom second language acquisition (pp. 156174). New York: Cambridge University Press. Leeman, J. (2003). Recasts and second language development: Beyond negative evidence. Studies in Second Language Acquisition, 25, 3763. Long, M. H. (1996). The role of the linguistic environment in second language acquisition. In W. C. Ritchie & T. K. Bhatia (Eds.), Handbook of research on language acquisition. Vol. 2: Second language acquisition (pp. 413468). New York: Academic Press. Long, M. H., Inagaki, S., & Ortega, L. (1998). The role of implicit negative feedback in SLA: Models and recasts in Japanese and Spanish. Modern Language Journal, 82, 357371. Lynch, T., & Maclean, J. (2000). Exploring the benets of task repetition and recycling for classroom language learning. Language Teaching Research, 4(3), 221250. Lynch, T., & Maclean, J. (2001). A case of exercising: Effects of immediate task repetition on learners performance. In M. Bygate, P. Skehan, & M. Swain (Eds.), Researching pedagogic tasks: Second language learning, teaching and testing (pp. 141 165). Harlow, England: Longman. Lyster, R. (1998a). Negotiation of form, recasts, and explicit correction in relation to error types and learner repair in immersion classrooms. Language Learning, 48, 183218. Lyster, R. (1998b). Recasts, repetition and ambiguity in L2 classroom discourse. Studies in Second Language Acquisition, 20, 5181. Lyster, R. (2001). Negotiation of form, recasts, and explicit correction in relation to error types and learner repair in immersion classrooms. Language Learning, 15(Suppl. 1), 265301. Lyster, R. (2004). Differential effects of prompts and recasts in form-focused instruction. Studies in Second Language Acquisition, 26, 399432. Lyster, R., & Ranta, L. (1997). Corrective feedback and learner uptake: Negotiation of form in communicative classrooms. Studies in Second Language Acquisition, 19, 3766. Mackey, A. (1999). Input, interaction and second language development: An empirical study of question formation in ESL. Studies in Second Language Acquisition, 21, 557587. Mackey, A. (in press). Interaction and second language development: Perspectives from SLA research. In R. DeKeyser (Ed.), Practice in second language learning: Perspectives from linguistics and psychology. Cambridge: Cambridge University Press. Mackey, A., & Gass, S. (2006). Pushing the methodological boundaries in interaction research: An Introduction to the special issue. Studies in Second Language Acquisition, 28, 169178. Mackey, A., Gass, S. M., & McDonough, K. (2000). How do learners perceive interactional feedback? Studies in Second Language Acquisition, 22, 471497. Mackey, A., & Oliver, R. (2002). Interactional feedback and childrens L2 development. System, 30, 459477. Mackey, A., Oliver, R., & Leeman, J. (2003). Interactional input and the incorporation of feedback: An exploration of NS-NNS and NNS-NNS adult and child dyads. Language Learning, 53, 3566. Mackey, A., & Philp, J. (1998). Conversational interaction and second language development: Recasts, responses, and red herrings? Modern Language Journal, 82, 338356.

McDonough, K. (2005). Identifying the impact of negative feedback and learners responses on ESL question development. Studies in Second Language Acquisition, 27, 79103. Mehnert, U. (1999). The effects of different lengths of time for planning on second language performance. Studies in Second Language Acquisition, 20, 83108. Oliver, R. (1995). Negative feedback in child NSNNS conversation. Studies in Second Language Acquisition, 17, 459481. Oliver, R. (1998). Negotiation of meaning in child interactions. Modern Language Journal, 82, 372386. Oliver, R. (2000). Age differences in negotiation and feedback in classroom and pairwork. Language Learning, 50, 119151. Oliver, R. (2002). The patterns of negotiation for meaning in child interactions. Modern Language Journal, 86, 97111. Oliver, R., & Mackey, A. (2003). Interactional context and feedback in child ESL classrooms. Modern Language Journal, 87, 519533. Plough, I., & Gass, S. M. (1993). Interlocutor and task familiarity: Effects on interactional structure. In G. Crookes & S. M. Gass (Eds.), Tasks and language learning: Integrating theory and practice (pp. 3556). Clevedon, England: Multilingual Matters. Robinson, P. (1995). Task complexity and second language narrative discourse. Language Learning, 45, 99140. Robinson, P. (2001a). Task complexity, cognitive resources and second language syllabus design: A triadic theory of task inuences on SLA. In P. Robinson (Ed.), Cognition and second language instruction (pp. 285317). Cambridge: Cambridge University Press. Robinson, P. (2001b). Task complexity, task difficulty and task production: Exploring interactions in a componential framework. Applied Linguistics, 22, 2757. Robinson, P. (2003). The cognition hypothesis, task design, and adult task-based language learning. Second Language Studies, 21(2), 45105. Salaberry, M. R., & Lopez-Ortega, N. (1998). Accurate L2 production across language tasks: Focus on form, focus on meaning, and communicative control. Modern Language Journal, 82, 514532. Schmidt, R. (Ed.). (1995). Attention and awareness in foreign language learning. Honolulu: University of Hawaii, Second Language Teaching & Curriculum Center. Skehan, P. (1998). A cognitive approach to second language learning. Oxford: Oxford University Press. Skehan, P., & Foster, P. (1997). Task type and task processing conditions as inuences on foreign language performance. Language Teaching Research, 1(3), 185 211. Skehan, P., & Foster, P. (1999). The inuence of task structure and processing conditions on narrative retellings. Language Learning, 49, 93120. Swain, M. (1985). Communicative competence: Some roles of comprehensible input and comprehensible output in its development. In S. M. Gass & C. Madden (Eds.), Input in second language acquisition (pp. 235253). Rowley, MA: Newbury House. Swain, M. (1995). Three functions of output in second language learning. In G. Cook & B. Seidlhofer (Eds.), Principle and practice in applied linguistics: Studies in honour of H. G. Widdowson (pp. 125144). Oxford: Oxford University Press. Swain, M. (2005). The output hypothesis: Theory and research. In E. Hinkel (Ed.), Handbook on research in second language learning and teaching. Mahwah, NJ: Lawrence Erlbaum. Swain, M., & Lapkin, S. (2001). Focus on form through collaborative dialogue:

Exploring task effects. In M. Bygate, P. Skehan, & M. Swain (Eds.), Researching pedagogic tasks: Second language learning, teaching and assessment. London: Pearson International. Tarone, E., & Liu, G. Q. (1995). Situational context, variation, and second language acquisition theory. In G. Cook & B. Seidlhofer (Eds.), Principle and practice in applied linguistics: Studies in honor of H. G. Widdowson (pp. 107124). Oxford: Oxford University Press. Van den Branden, K. (1997). Effects of negotiation on language learners output. Language Learning, 47, 589636. VanPatten, B. (1990). Attending to content and form in the input: An experiment in consciousness. Studies in Second Language Acquisition, 12, 287301. Wong-Fillmore, L. (1985). When does teacher talk work as input? In S. M. Gass & C. Madden (Eds.), Input in second language acquisition (pp. 1750). Rowley, MA: Newbury House. Yuan, F., & Ellis, R. (2003). The effects of pre-ask planning and on-line planning on uency, complexity and accuracy in L2 monologic oral production. Applied Linguistics, 24, 127. Yule, G., Powers, M., & MacDonald, D. (1992). The variable effects of some taskbased learning procedures on communicative effectiveness. Language Learning, 42, 249277.

Participants First Language Backgrounds
First language Afghan Arabic Cantonese Farsi Indonesian Bosnian Serbian Italian Thai Vietnamese Russian Polish Mandarin Romanian Number of students (N = 40) 1 3 3 4 5 3 4 1 4 3 2 2 2 3 Percentage of sample 2.5 7.5 7.5 10.0 12.5 7.5 10.0 2.5 10.0 7.5 5.0 5.0 5.0 7.5

Task Instructions
One way tasks
In this activity x has a picture. X is going to describe this picture to you so you can draw an exact copy. While doing this work you can both talk to one another. You must work together. I cannot help you. X, you will have 5 minutes to look at the picture and think about what you are going



to say. Then you will both have 20 minutes to do this work, at the end you will have 10 minutes to share and check your work with your partner.

Two way tasks

In this activity both you and your partner have a picture which is nearly the same. The reason why it is not exactly the same is because you both have missing parts to it. Your partners picture has different parts missing. The missing parts of your picture are on yellow cards, which are in front of you. What you have to do is this: you have to talk and ask questions to one another so you can give your partner information on where to put the yellow cards. When you nish, both your pictures should be the same. You must work together. I cannot help you. You will have 5 minutes to look and think about what is in your picture. After you will have 20 minutes to do this work, then you will have 10 minutes to share and check your work with your partner.



Development of Speed and Accuracy in Pragmatic Comprehension in English as a Foreign Language

Carnegie Mellon University Pittsburgh, Pennsylvania, United States

This study examined development of pragmatic comprehension ability across time. Twenty native speakers and 92 Japanese college students of English completed a computerized listening task measuring ability to comprehend two types of implied meaning in dialogues: indirect refusals (k = 24) and indirect opinions (k = 24). The participants comprehension was analyzed for accuracy (scores on the listening task) and comprehension speed (average time taken to answer each item correctly). L2 learners accuracy and comprehension speed improved signicantly over a 7-week period. However, the magnitude of effect was lower for comprehension speed than for accuracy. This study also examined the relationships among general L2 prociency (measured on the ITP TOEFL), speed of lexical judgment (measured on a word recognition task), and pragmatic comprehension ability. There was a signicant relationship between prociency and accuracy (r = 0.39), as well as between lexical access speed and comprehension speed (r = 0.40). However, L2 prociency bore no relationship to comprehension speed, and lexical access speed had no relationship with accuracy. Moreover, accuracy and comprehension speed were not related to each other. These ndings suggest that development of pragmatic knowledge and processing capacity of using the knowledge may not coincide perfectly in L2 development.


he eld of second language acquisition (SLA) has much discussed communicative competence, as marked by the emergence of several models (Bachman & Palmer, 1996; Canale & Swain, 1980; Hymes, 1972). These models, emphasizing functional and context-dependent language use, have situated pragmatic competence as a distinct, indispensable component of communicative competence. Pragmatic competence, assumed in the roles of the speaker and listener in interaction, refers to the

TESOL QUARTERLY Vol. 41, No. 2, June 2007

ability to produce meaning in a socially appropriate manner and to interpret meaning, explicitly or implicitly stated, according to contexts (Thomas, 1995). With the recognition of the role of pragmatic competence in communicative ability, abundant second language (L2) research has analyzed learners pragmatic performance in communicative contexts. In the existing L2 literature, pragmatic competence has been analyzed mainly from production skills, specically production of speech acts. Little L2 research has investigated comprehension of pragmatic functions (Kasper & Rose, 2002). Moreover, research has not substantially studied issues related to developmental aspects or relationships of pragmatic competence with general L2 prociency and other cognitive and noncognitive variables. According to Thomas (1995), meaning has two levels: utterance meaning, or assigning sense to the words uttered; and force, or speakers intention behind the words. Pragmatic comprehension entails understanding meaning at both levels. It involves the ability to understand what words and sentences mean, as well as to understand what speakers mean by them. Therefore, comprehension of implied meaning, namely, meaning that goes beyond what is given by the language form itself or what is literally said (Verschueren, 1999, p. 25), is an important aspect of comprehension ability. Grice (1975) explains the comprehension process of implied meaning by using a notion of conversation maxims, or rules of communication. The conversation maxims enable people to interpret implicit meaning, such as in Bs response:
A: Im out of petrol. B: There is a garage around the corner. (p. 51)

The listener, assuming that the speaker has followed the conversation maxims and has made an appropriate comment to A, draws the conclusion that the garage has gasoline. Implied meaning is understood based on the assumption that the speaker operates under the cooperative principle, a universal axiom, as well as the listeners ability to supply contextual information to make inferences from seemingly unrelated utterances. Sperber and Wilson (1995) used relevance theory to extend Grices insight and claimed that communication is achieved by interpreting contextual cues and using them to infer speaker intention. Contextual cues include not only such external factors as physical environment or the immediately preceding discourse, but also internal factors such as ones knowledge about the world, conventions, and experiences. Relevance theory also emphasizes the relation between context and processing effort. The theory argues that implicatures vary in their degree of strength; some implicatures are strongly conveyed, but others are weakly under314 TESOL QUARTERLY

stood because of to the number of contextual cues to be processed (Sperber & Wilson, 1991, 1995). The greater the number of contextual cues to be processed, the more extensive the search for meaning becomes, resulting in greater processing effort. One factor that could reduce processing effort is the level of conventionality encoded in utterances. When implicatures convey conventional meaning, that is, when speaker intentions are linguistically coded or embedded within predictable, xed patterns of discourse, the listener may not attend to many contextual cues, consequently reducing the processing effort.1 A relatively small number of L2 studies have examined whether learners can comprehend implied meaning accurately (Bouton, 1992, 1994, 1999; Cameron & Williams, 1997; Carrell, 1981, 1984; Cook & Liddicoat, 2002; Garcia, 2004; Kasper, 1984; Koike, 1996; Rver, 2005; Takahashi & Roitblat, 1994; Taguchi, 2002, 2003; Ying, 1996). Some studies examined the ability to comprehend implicit meaning in relation to L2 prociency and the degree of directness or conventionality encoded in utterances. Cook and Liddicoat (2002) examined L2 comprehension of requests of various types. ESL learners of two prociency levels responded to a written questionnaire that contained brief scenarios followed by a requestmaking expression. The expressions had three directness levels: direct (e.g., Pass me the salt), conventional indirect (e.g., Can you pass me the salt?), and nonconventional indirect (e.g., Are you putting salt on my meat?). The high prociency learners had difficulty in understanding nonconventional indirect forms, and the low prociency group had difculty with both indirect forms. Results suggest a difference in comprehension difficulty among different types of expressions. Nonconventional indirect requests were difficult to comprehend even for the high prociency learners, but conventional expressions were not. As L2 prociency develops, learners are able to comprehend a winder range of indirect expressions. Taguchi (2002) also documented different difficulty levels across implied meaning types. Eight Japanese ESL learners at two different prociency levels listened to dialogues that contained three types of indirect utterances: indirect refusals, indirect disclosure of information, and indirect expression of opinions. The results showed that higher prociency learners more accurately comprehended implied speaker intentions than lower prociency learners. Retrospective verbal reports revealed that comprehension of refusals required less effort; learners used fewer strategies to interpret indirect refusals than the other

Morgan (1978) dened conventionality as common knowledge of the way things are done (p. 279) that encompasses both knowledge of language and language use. Morgan called some indirect speech acts short-circuited implicatures, meaning that they are types of implicatures that do not require extensive inferencing because the listener understands the intended meaning based on conventions of language and language use, and does not have to process a great number of contextual cues.



two types of implied meaning. These differences in comprehension difculty among indirect expressions may stem from different levels of conventionality encoded in the utterances. In indirect refusals, refusal intention is often embedded within predictable, xed patterns of discourse. For instance, when refusing someones invitation, it is customary to provide a reason for not accepting the invitation (Nelson, Carson, Batal, & Bakary, 2002). In contrast, other indirect expressions such as opinion expressions are considered less conventional because they are idiosyncratic; the set of possible expressions to express opinions is open (e.g., endless qualications of liking and disliking). In summary, a few previous studies have documented that successful comprehension of implied meaning depends on the degree of processing effort required for comprehension, as well as learners general L2 prociency. What is not examined systematically relates to the developmental aspect of pragmatic comprehension. Only one study to date has examined development of pragmatic comprehension using a pre- and posttest design. Bouton (1992, 1994) investigated L2 learners comprehension of conversational implicatures. ESL learners took a test that had short written dialogues including different types of implicatures.2 The results showed that learners overall comprehension improved over time along with the length of residence in the target language country. However, learners showed much less pronounced development for formulaic implicatures that had typical structural and semantic features (e.g., showing agreement by saying Is the pope Catholic?). Building on Boutons work, research should examine whether the ability to understand implied meaning develops over time and what factors inuence the development. Because Bouton did not directly address the difference in comprehension load in relation to the differential degree of conventionality among implied meaning types, integration of conventionality could provide useful insights into the nature and development of pragmatic comprehension ability. Another gap in the previous research is that most studies on pragmatic comprehension have analyzed comprehension accuracy, namely the knowledge dimension of pragmatic ability, and only a few studies to date have addressed speed of pragmatic comprehension, an aspect of the processing dimension of pragmatic ability. As Segalowitz (2001, 2003) has claimed, uencyspeed and ease of language processingbelongs to

Implicatures tested in Boutons studies included relevance-based implicatures (violation of Grices relevance maxim), Pope implicatures (violation of Grices relevance maxim), indirect criticism (violation of Grices quality maxim), sequence-based implicatures, irony, scalar implicatures, and indirect criticism. Learners cultural knowledge and language background affected comprehension ability. Korean, Japanese, and Chinese speakers showed signicantly poorer comprehension ability than German, Spanish, and Portuguese speakers.



the study of performance. The study of uency should provide an alternative way to characterize language acquisition because rapid, effortless, and accurate skill execution is the fundamental goal of SLA. A growing number of studies have inquired into processing speed in L2 performance by examining response times in language tasks. In L2 research, response-time measures have been used mostly to examine lower order processes, such as lexical decisions (Jiang, 2002; Neumann, McCloskey, & Felio, 1999; Segalowitz, Segalowitz, & Wood, 1998), grammaticality judgments (White & Genesee, 1996; White & Juffs, 1998), and sentence decoding (Hoover & Dwivedi, 1998; Juffs, 1998; Juffs & Harrington, 1996). Results have generally conrmed that processing speed is related to automatization of lower order processes, and that the automaticity is related to prociency (Carber, 1990; Hirai, 1999; Segalowitz, 2001; Towell, 2002). However, little research has examined speed of higher order processing, such as comprehension of implied intentions. Thus, it is important to nd out whether previous ndings about the relationship between prociency and processing speed can be extended to higher level language use, such as pragmatic processing, which involves processing of multiple pragmatic cues, including contextual information, schemata, and knowledge of communication conventions. Only a few studies to date have examined speed in pragmatic comprehension (Takahashi & Roitblat, 1994; Taguchi, 2005). Takahashi and Roitblat examined L2 processing patterns of conventional indirect requestswhether L2 learners activate literal interpretation of indirect requests rst or they immediately process illocutionary force of requests. ESL learners and native speakers of American English read 12 stories: six including implied requests and six including literal interpretations of the utterances. Results showed that, although ESL learners took longer to comprehend the indirect requests than native English speakers, they showed processing patterns similar to those of native speakers. The learners read at the same speed whether the target sentence was interpreted as a conventional request or was interpreted literally. In Taguchis (2005) study, native English speakers and Japanese college students of EFL completed a listening test measuring ability to comprehend more and less conventional implicatures. More conventional implicatures had indirect requests and refusals that had conventional features (e.g., making a request by using Do you mind if followed by a subject). In less conventional implicatures, meaning was not attached to specic linguistic forms (e.g., indicating a negative opinion of a movie by saying I was glad when it was over). Comprehension was analyzed for accuracy and speed. Results showed that, for L2 learners, comprehension of more conventional implicatures took less time than comprehension of less conventional implicatures. Prociency had a signicant imDEVELOPMENT OF SPEED AND ACCURACY IN PRAGMATIC COMPREHENSION 317

pact on accuracy, but not on comprehension speed, and no signicant relationship was found between accuracy and comprehension speed. These previous studies suggest that processing speed could be a reection of learners ability to comprehend implied meaning. Taguchis study in particular suggests that speed has distinct characteristics on its own, independent from general L2 prociency or accurate demonstration of pragmatic knowledge. It seems that pragmatic knowledge and the capacity to use or process the knowledge are distinguishable underlying aspects of L2 communicative competence and thus should be examined separately. What is not yet examined to date is longitudinal development of processing speed when comprehending pragmatic meaning. In L2 research, a limited number of studies have investigated development of performance speed over time using a pre- and posttest design, mostly in the areas of oral uency (Freed, 2000; Segalowitz & Freed, 2004; Towell, 2002) and word recognition (Fukkink, Hulstijn, & Simis, 2005; Segalowitz & Segalowitz, 1993; Segalowitz, Segalowitz, & Wood, 1998; Segalowitz, Watson, & Segalowitz, 1995). Segalowitz and Freed administered an oral prociency test at the beginning and the end of a semester to college students of Spanish in a domestic environment and in a studyabroad environment. The results showed that the study-abroad students made signicant gains in oral uency. Theories of cognitive skill development claim that increased performance speed is related to increased L2 practice (Anderson, 1990; McLaughlin, 1987; McLaughlin & Heredia, 1996). According to Anderson, skill development involves a shift from declarative knowledge (knowledge that) to procedural knowledge (knowledge how). For instance, knowing a rule that English plural nouns are marked with a suffix s is declarative knowledge, but using the rule in real time reects procedural knowledge. Language acquisition is a process where declarative knowledge becomes procedural. At the initial stage of skill development, L2 learners retrieve and use language rules consciously. After repeated practice in applying rules, the use of rules becomes increasingly automatic, rapid, and unconscious. Automaticity develops through consistent associative practices between input and learners response, as shown in some L2 studies. Snellings, van Gelderen, and Glopper (2002) examined the effect of training on uent lexical retrieval among Dutch learners of L2 English. Learners who received semantic access training had higher accuracy scores and faster reaction times in recognizing word meaning. Based on the theoretical claims and empirical ndings, it is reasonable to assume that increased L2 experience over time will promote rapid skill execution. However, in L2 pragmatics, little research has examined the development of speed over time. Performance speed in pragmatics might show a different developmental course because of a number of

resources to be processed, including linguistic, contextual, and sociocultural resources. Another gap in the literature is the analysis of factors that may affect speed of pragmatic processing. Taguchi (2005) found that general L2 prociency did not inuence speed of pragmatic comprehension among EFL students. A question then remains as to what factors actually affect processing speed in pragmatic comprehension. Some candidate factors belong to the domain of cognition, including semantic access ability, pattern recognition, short-term memory, and attention control. These factors reect overall cognitive abilities, a set of components that enable us to allocate attention and select and coordinate information in language tasks (Bialystok, 1990; Robinson, 2003, 2005; Skehan, 1995, 1998; Widdowson, 1989). Among those cognitive factors, lexical access speed, one processing characteristic, has been recognized as a factor that is directly implicated in performance uency. Processing speed of lexical items (e.g., words) is considered as one of the underlying abilities in speedy skill execution (Ackerman, 1988). An increasing number of L2 studies have examined lexical access ability as an underlying component skill in language performance (DeKeyser, 2001; Schmidt, 1992; Segalowitz, 1997, 2000; 2003; Segalowitz & Segalowitz, 1993). Segalowitz and Freed (2004) examined whether lexical access speed and attention control are related to oral uency gains among L2 Spanish learners. Results revealed that oral uency signicantly correlated with lexical access speed but not with attention control ability. Their study conrmed that the ability to access meanings quickly could promote uent speech. Building on recent work, L2 pragmatics research that incorporates measures of cognitive processing abilities, such as lexical access speed, in addition to L2 prociency, could reveal factors that affect speed as well as accuracy of pragmatic comprehension. Pragmatic comprehension involves processing at multiple levels. It involves lower level processing of attending to and assigning meaning to aural stimuli (Wolvin & Coakley, 1985). It also involves higher level processingsupplementing acoustic information with nonacoustic information (e.g., prior knowledge, schemata) to derive meaning behind utterances (Smith, 1975). One can assume that lexical access speed, namely the ability to access word meaning quickly, reects lower level processing that which underlies the process of comprehending pragmatic meaning. However, because this assumption has not been tested, it merits empirical investigation. In summary, very little research has addressed the development of pragmatic comprehension ability. Studies that have addressed both accuracy and speed in development are even scarcer. As Kasper (2001) states, pragmatic competence refers to the acquisition of pragmatic knowledge and to gaining automatic control in processing it in real time. Thus, the two dimensions (i.e., pragmatic knowledge and processing capacity

for the knowledge) should be addressed together because they provide complementary insight into developmental accounts of pragmatic comprehension. In addition, with increasing attention to the interdependence between cognition and language performance, cognitive processing abilities, such as lexical access skill, are considered to affect performance. Because limited empirical evidence has demonstrated the relationship among cognitive variables, general L2 prociency, and pragmatic comprehension, future research also awaits in this direction.

Based on the gaps in the existing literature, this study investigated the ability to comprehend implied meaning among L2 learners of English as a foreign language. The purpose of the study was to examine the development of accuracy and comprehension speed across time, as well as general prociency and cognitive ability that may affect accuracy and speed of pragmatic comprehension. Two research questions guided this study: (1) Does L2 learners ability to comprehend implied meaning improve over time, in terms of accuracy and speed of comprehension? (2) To what extent are L2 general prociency and speed of cognitive processing related to accurate and speedy comprehension of implied meaning?

METHODOLOGY Participants
Ninety-two Japanese learners of English and 20 native English speakers participated in this study. The Japanese learners were freshmen enrolled in an intensive English program in a Japanese university in Japan. There were 21 males and 71 females with their average age of 18.26, ranging from 17 to 22 years old. The students received between 16 and 18 hours of content-based instruction per week in the program over a half-semester period (7 weeks). The program promoted an integrated skills approach to the acquisition of academic English abilities, and there was no specic course for pragmatic instruction. All classes were taught in English and provided academic activities, such as listening to lectures and taking notes, participating in discussion, writing essays, and using computer skills.

Because there were no international students from English-speaking countries on campus at the time of data collection, it was assumed that the students did not have extensive contact with native English speakers outside of the classroom. Among the 92 Japanese participants, three students had from 6 to 9 months experience living overseas, and three students had from 2 to 3 months experience living overseas. Twenty native English speakers (12 males and 8 females) were international students from U.S. universities who were studying Japanese in the same university. Their average age was 23.83, ranging from 2039.

Measures Pragmatic Listening Task

Participants ability to comprehend implied meaning was assessed by a computerized listening task (for sample items, see Taguchi, 2003) that had 60 items: 2 practice items, 10 ller items, and 48 experimental items (i.e., 24 refusal and 24 opinion items). The ller items that tested literal comprehension were included to divert participants attention from the purpose of the experiment, namely testing comprehension of nonliteral meaning. Each item had a short dialogue spoken by a male and female native English speaker. In the experimental dialogues, the reply that appeared at the end violated the maxim of relevance and did not provide a straightforward answer to the speakers question. Thus, the experimental items intended to test L2 learners ability to comprehend the implied speaker intention encoded in the seemingly irrelevant reply. Each dialogue was followed by a yesno question that was designed to check the participants comprehension of the speakers intention. The task was computerized using the SuperLab Pro Program for Windows (Cedrus, 2003). Table 1 displays sample items. Indirect refusal items included four types: refusals to requests (6 items), invitations (6 items), suggestions (6 items), and offers (6 items). Utterances used for indirect refusals took the form of providing a reason and were written so as to avoid explicit linguistic markers of refusals such as I cant, No, or I dont want to, which were identied as direct refusal expressions by Beebe, Takahashi, and Uliss-Weltz (1990). Indirect opinion items were written as expressions used to convey opinions in a nonliteral manner. To counteract response biases, half of the indirect opinion items were designed to exemplify negative opinions, and the other half were written to imply positive opinions. These two item types (refusals and opinions) were included in the listening task to counteract any response biases, as well as to add variety

TABLE 1 Sample Pragmatic Listening Task Items Indirect refusal item Mary: Hey, John, whatre you doing? John: Im working on my paper for the English class. Mary: Youve been working on that paper for a week. Why dont you take a break? Lets go to the movies tonight. John: I have to nish my paper by eight in the morning. Question: Is John going to the movie? Indirect opinion item John: Hey, Mary, you got a package today, didnt you? From who? Mary: Its from my cousin in Florida. He sent me a gift for my birthday. A little alarm clock. John: Oh, thats nice of him. Did you like it? Mary: The wrapping paper was nice. Question: Does Mary like the gift?

to the items degree of conventionality. Indirect refusal items in this task were considered more conventional because all the utterances used to convey refusals took the form of providing a reason, which is considered as a common and predictable pattern of refusal routines (Beebe et al., 1990; Nelson et al., 2002). In contrast, indirect opinion expressions were considered less conventional because they did not reect specic discourse routines or conventions of language use. The items used for this study were drawn from Taguchis (2005) study, which adapted partial items from previous studies of the comprehension of implicatures (Bouton, 1992, 1994; Holtgraves, 1999, Kotsonis, 1981, Rver, 2005). To control impact on short-term memory, the length of dialogues was kept approximately the same. Each conversation had between 42 and 48 words, with a mean of 44.8 and standard deviation of 1.9 words. The number of words in questions was also kept approximately the same, either six or seven words, to reduce the effect of extraneous variables (e.g., reading time) and to make response times more comparable. The number of turns in each dialogue was also kept approximately the same, between three and ve turns. In addition, following Rver (2005), to reduce the effect of construct-irrelevant variance from the learners difference in vocabulary knowledge, all vocabulary in items were drawn from Longmans 2000-word dening vocabulary list (Summers, 1995). The 2000 words on the list are identied as common, frequent, and basic English words and are thus used to write all the word denitions in the Longman dictionary. Internal consistency reliability of the listening task was estimated using Cronbachs alpha, yielding 0.85 for the full test (k = 58).

Lexical Access Test

To seek the relationship between pragmatic comprehension ability and cognitive processing skills, a computerized word recognition test

adapted from Segalowitz and Freed (2004) was used. This lexical access test (LAT) measured participants speed in semantic classications of individual words. The LAT had 80 words, and the participants made speeded judgments on whether each word that appeared on a computer screen referred to a living or nonliving object (e.g., the boy = living; a pen = nonliving). The words used for the test were all English nouns that were generally high in written frequency and were considered fairly familiar to English speakers (Segalowitz & Freed, 2004). A high school English teacher who had more than 10 years of teaching experience conrmed that the 80 words used in the instrument were familiar for average high school students and also the students enrolled at the institution. In addition, 91% of the words in the list appeared in the basic word list for high school students published in the market. The words were also found in the basic vocabulary list introduced in junior and senior high school textbooks authorized by the Japanese Ministry of Education.


To examine L2 learners pragmatic comprehension in relation to their general language ability, the institutional TOEFL (i.e., ITP TOEFL) was used to determine general L2 prociency. Educational Testing Services (ETS) provides the opportunity to administer the TOEFL locally to students at their home institutions throughout the world, through a program called the Institutional Testing Program (ITP). Because the ITP uses previous versions of the TOEFL test, the format, content, and scoring of the ITP TOEFL are identical to the actual TOEFL (ETS, 2003). The ITP TOEFL was given twice to the Japanese participants, at the beginning and at end of the 7-week session.

Data Collection Procedures and Analysis

The pragmatic listening task was given to the participants using Windows computers on campus. After sitting in front of the computer, the participants put on headphones and read instructions on the computer screen in their native language. Before starting the test items, they completed two practice items. Immediately following each dialogue, a yesno question appeared on the screen. Participants responded yes by pressing the number 1 key or no by pressing the number 2 key. Once they had chosen the answer, the computer automatically took them to the next item. Response time was measured between the moment when the question appeared on the screen and the moment when the participant pressed the number key. The computer recorded all responses and their

latencies. The same procedure was repeated for the Japanese participants at the end of the intensive 7-week course. In the second listening test, the test items were randomized. Immediately following the second listening test session, the same computer was used to administer the LAT. The participants read instructions on the screen in their native language. After completing four practice items, they completed the 80 test items. When a word appeared on the computer screen, they made a speeded judgment on whether the word referred to a living thing by pressing the number 1 key or a nonliving thing by pressing the number 2 key. Once they had made a choice, the computer automatically moved to the next item. Response time was measured between the moment when the target word appeared on the screen and the moment when the participant pressed the number key. The computer recorded all responses and their latencies. For the analyses of the rst research question (i.e., development of pragmatic comprehension ability), L2 learners comprehension ability was operationalized in terms of comprehension accuracy of implied meaning items (i.e., indirect refusals and indirect opinions combined), and comprehension speed of the items. Accuracy was measured by the pragmatic listening task, which had an interval scale between 0 and 48 (the scale for each item type was between 0 and 24). Comprehension speed, also interval data, was operationalized as response times and was calculated by averaging the number of seconds taken to answer items correctly. Development of accuracy and speed were analyzed using two matched-pair t tests comparing the differences in accuracy scores and response times between the pre- and posttest sessions. The 10 ller items in the listening task were excluded from all analyses. The second research question addressed the relationship among cognitive processing ability, general L2 prociency, and pragmatic comprehension ability. The cognitive processing ability, operationalized as the ability to make speedy lexical judgments, was determined as average time in seconds on the LAT, and it formed continuous data. General L2 prociency was determined by the ITP TOEFL scores, which were treated as a continuous variable. The learners in this study were at beginning to highintermediate prociency level; their scores ranged from the 4th to the 39th percentile according to the ETSs score scale (ETS, 2005). Relationships among the four variablesaccuracy scores and response times in pragmatic listening task, response times in LAT, and ITP TOEFL scoreswere examined using Pearson correlation. Prior to the statistical analyses, distributions of the variables were examined to check underlying assumptions. Data screening results documented that the distributions of comprehension time data did not meet the underlying assumptions of normality, linearity, and homoscedasticity. Because the distributions were positively skewed, a logarithmic trans324 TESOL QUARTERLY

formation was performed following Tabachnick and Fidell (2001). After the transformation, the distributions were found to be normal and linear, and homoscedasticity was conrmed. The Kolmogorov-Smirnov test conrmed normality at the alpha level of 0.05. Thus, it was decided that the statistical analyses should be performed on the logarithmically transformed data. Based on previous conventions, the signicance level for hypothesis testing was set at 0.05; however, because the analyses of the rst research question used two statistical comparisons, the alpha level was adjusted to 0.025 using the Bonferroni correction in order to avoid a type one error (Brown, 1990; Hatch & Lazaraton, 1991; SPSS, 1998). Similarly, for the analyses of the second research question, the signicant level was adjusted to 0.0125 because it used four separate correlation analyses.

RESULTS Development of Pragmatic Comprehension Ability

In response to the rst research question, two attributes of pragmatic comprehension, namely accuracy and response time in a pragmatic listening task, were compared between pretest and posttest for L2 learners. Tables 2 and 3 display descriptive statistics of accuracy scores and response times in the two test sessions for L2 learners, as well as descriptive statistics of native English speaker data. As expected, compared with L2 learners, native speakers were almost perfectly accurate and much faster
TABLE 2 Descriptive Statistics of Native Speakers and L2 Learners Accuracy Scores in a Pragmatic Listening Task Mean Native Speaker Data (n = 20) Implied meaning items (k = 48) Refusal items (k = 24) Opinion items (k = 24) L2 Learner Data (n = 92) Pretest Implied meaning items (k = 48) Refusal items (k = 24) Opinion items (k = 24) Posttest Implied meaning items (k = 48) Refusal items (k = 24) Opinion items (k = 24) Note. k = the number of items. 46.56 23.11 23.44 SD 1.34 0.83 0.98 Min 43.00 21.00 20.00 Max 48.00 24.00 24.00

35.21 19.38 15.84 38.15 21.09 17.07

4.84 2.98 2.67 4.12 2.17 2.56

22.00 8.00 8.00 25.00 14.00 10.00

45.00 24.00 23.00 46.00 24.00 22.00



TABLE 3 Descriptive Statistics of Native Speakers and L2 Learners Response Time in Seconds in a Pragmatic Listening Task Mean Native Speaker Data (n = 20) Implied meaning items (k = 48) Refusal items (k = 24) Opinion items (k = 24) L2 Learner Data (n = 92) Pretest Implied meaning items (k = 48) Refusal items (k = 24) Opinion items (k = 24) Posttest Implied meaning items (k = 48) Refusal items (k = 24) Opinion items (k = 24) 3.30 3.39 3.22 SD 0.97 1.00 1.02 Min 1.54 1.63 1.36 Max 4.67 5.41 4.74

5.70 5.18 6.35 5.22 4.81 5.69

1.93 1.65 1.70 1.65 1.62 2.01

2.72 2.74 2.71 2.32 2.25 2.40

13.24 10.72 16.07 11.46 8.44 14.32

Note. k = the number of items. The response time refers to the average number of seconds taken to answer each item correctly.

in comprehending implied meaning items. In both tests, L2 learners were more accurate and faster in comprehending refusals items than opinion items. Matched-pair t test results revealed a signicant score difference between pre- and posttests for implied meaning items, t = 8.15, df = 91, p = 0.00, indicating that the L2 learners achieved a signicant gain in accuracy scores over a 7-week period. The effect size of 0.42 based on 2 also suggests a large effect (Hatch & Lazaraton, 1991). A matched-pair t test was also conducted to check the response time difference between the pre- and posttest. The results revealed a signicant difference between the two test sessions, t = 3.16, df = 91, p = 0.00, suggesting that the L2 learners also made a signicant improvement in processing speed; they became faster in responding to the items over a 7-week period. However, compared with the accuracy gain, the effect size for the response time difference was much smaller, 0.10 based on 2, which shows small effect. The results thus suggest that the degree of development differed between accuracy and response speed; the gain of speed when processing pragmatic information was smaller than that of accurate understanding of pragmatic meaning.

Relationships Among Cognitive Ability, General Prociency, and Pragmatic Comprehension

The second research question addressed the extent to which cognitive processing ability and general L2 prociency were related to ability to

TABLE 4 Descriptive Statistics of Lexical Access Time for Native Speakers and L2 Learners Mean Native speaker LAT time (n= 20) L2 learner LAT time (N = 92) Notes. LAT = lexical access test. 0.98 1.18 SD 0.20 0.25 Min 0.64 0.80 Max 1.39 1.76

comprehend implied meaning. Table 4 displays descriptive statistics of lexical access speed, namely the average time in seconds in judging each word correctly. There was a small difference between the native English speakers and L2 learners lexical decision times (i.e., 0.20 seconds). Variance in response time was smaller for native speaker data than L2 learner data. Pearson correlation was conducted to seek relationships among lexical access speed, general L2 prociency, and pragmatic comprehension ability. Accuracy scores and response time in the pragmatic comprehension posttest were correlated against the response time in LAT and the ITP TOEFL scores. Table 5 displays a correlation matrix among the four variables. As shown in Table 5, lexical access speed correlated signicantly with response time for the pragmatic comprehension task, r = 0.40, N= 92, p = 0.01, but not with general L2 prociency measured by the ITP TOEFL (r = 0.03). Furthermore, accuracy scores of the pragmatic comprehension task correlated signicantly with general L2 prociency, r = 0.39, N = 92, p = 0.01, but not with lexical access speed (r = 0.01). Finally, there was no signicant correlation between accuracy scores and response time for the pragmatic comprehension ability (r = 0.11). The results indicate that the cognitive processing ability measured in this study, namely speed of semantic access, revealed a signicant relationship with the speed dimension of pragmatic comprehension ability but not with the accuracy dimension. On the other hand, general knowledge of L2 measured by the ITP-TOEFL in this study was found to be signicantly related to the accuracy dimension of pragmatic comprehension but not with the speed dimension.
TABLE 5 Correlations Among Lexical Access Speed, Prociency, and Pragmatic Comprehension Ability PLT-accuracy PLT-accuracy PLT-response time LAT-time ITP TOEFL PLT-time 0.11 LAT-time 0.01 0.40* ITP TOEFL 0.39* 0.07 0.03

Notes. *p = 0.01; PLT= pragmatic listening task; LAT= lexical access task; n = 92.



This study revealed improvement of pragmatic comprehension ability among EFL learners over a 7-week period, in terms of accuracy scores as well as response times. The ndings suggest that exposure to the target language context and culture is not the only contributing factor in pragmatic development. In this study, pragmatic development, in terms of accuracy and speed in comprehending implied meaning, was also noted among learners in a foreign language context with limited exposure to L2 pragmatic conventions and sociocultural practice. The results, then, lend support to the previous claim that learning in a foreign language context does not necessarily disadvantage pragmatic development (Ohta, 2001); it is the way in which learning is organized and fostered that enables or hinders pragmatic development. Another potential explanation is that pragmatic comprehension ability might be inuenced more strongly by L1-based inferential skills than L2-specic sociocultural knowledge. Whether in L1 or L2, the ability to seek the relevance of a message is part of innate human cognition and develops naturally as we mature through our socialization process. Lack of this ability is labeled as a type of cognitive disorder (Baron-Cohen, Leslie, & Frith, 1985; Hala & Carpendale, 1997). It is possible that the EFL learners in this study became more skilled at deriving implied speaker intentions over time because they were able to transfer their L1-based inferential practice to L2 comprehension. The transfer of inferential skills occurred as learners received more classroom practice in their general listening skill, and thus their listening ability, as well as their overall language prociency, improved over time. The increase of speed observed among foreign language learners also adds to the previous theoretical claims in the area of cognitive skill development. Information processing models developed by cognitive psychologists argue that the speed of language processing develops naturally in accordance with increased associative practices between input and response (Anderson, 1990; McLaughlin, 1987; Rumelhart & McClelland, 1986). Learners in a foreign language context are considered to receive limited incidental L2 exposure and practice outside the classroom. Nonetheless, the EFL learners in this study became signicantly faster when processing pragmatic information. The results suggest that the development of processing speed could occur in a domestic, formal classroom setting. One potential explanation is the intensive instruction that the learners received at the institution. Although they received L2 input mainly in classrooms, they received 1618 hours of intensive English instruction per week during a short period of 7 weeks. Thus, it is possible that the intensity of the English learning experience in the English for academic purposes program facilitated speed development.

It may be the function of the amount of in-class contact that caused the gain in processing speed over a short period of time. Several previous studies lend support to this interpretation (Freed, Segalowitz, & Dewey, 2004; Segalowitz & Freed, 2004). Segalowitz and Freed found that, although Spanish language learners in a second language context gained more oral uency than learners in a foreign language context, the gain did not reect greater out-of-class contact or extracurricular use of the second language. One possible explanation for the differences is the number of class hours. The domestic students were enrolled in one Spanish class, but the students in the study abroad group were enrolled in three courses. Thus, it may be the intensive format of instruction that promoted oral uency, which reects the speed of processing language information. These observations could explain the present ndings because the students, who were enrolled in the intensive English program, improved in speed in pragmatic processing, a type of language processing activity. Another notable nding in this study is that, although the development of accuracy and comprehension speed was observed in both types of indirect expressions (i.e., indirect refusals and indirect opinions), learners comprehension was more accurate and faster for refusals than opinions. These ndings suggest differential comprehension demands between the two item types because they have different degrees of conventionality encoded in them. Expressions used to convey indirect refusals are more routinized because they reect relatively xed patterns of discourse exchange (e.g., giving an excuse when refusing). In more conventionalized situations, comprehension of speech intentions becomes easier and takes less time because, when the conventions are shared between L1 and L2, learners can activate previous experience and knowledge that facilitate comprehension. Thus, although comprehension of less conventional expressions such as indirect opinions did improve over time, both in terms of accuracy and speed of comprehension, the development itself might take longer than the ability to comprehend more conventional indirect forms. Although both accuracy and speed showed signicant developmental trends over time, the respective degrees of development differed: Speed increased much less than accuracy did during the 7-week period. The effect size of the mean difference between the pretest and posttest, based on 2, was 0.42 for accuracy scores, but it was 0.10 for response times. Thus, analyses of the two attributes, accuracy and processing speed, revealed different types of information about L2 pragmatic development. Development of performance speed, namely, automatic realization of pragmatic knowledge, seems to lag behind in L2 acquisition and does not develop as quickly as accurate demonstration of pragmatic knowledge. Processing of pragmatic functions involves the coordinated action

of a number of constituent processes, including linguistic, cognitive, and sociocultural processes, which must be automatized to achieve speedy performance. The EFL learners in this study seemed to take longer to automatize these underlying components. To examine the potential factors that affect knowledge and processing capacity, this study addressed the relationships among general L2 prociency (a language-specic factor), speed of lexical judgment (a cognitive-specic factor), and pragmatic comprehension ability. The relationship between the prociency and accuracy dimensions of pragmatic comprehension was signicant (r = 0.39), as was the relationship between lexical access speed and speed dimension of pragmatic comprehension (r = 0.40). However, L2 prociency bore no relationship to the speed of comprehending implied meaning, and lexical access speed had no relationship with accuracy of comprehension. Moreover, accuracy and speed were found to be unrelated to each other. As shown in this study, development of performance speed, namely automatic realization of pragmatic knowledge, seems to lag behind in second language acquisition and does not develop as quickly as accurate demonstration of pragmatic knowledge. Processing of pragmatic functions involves the coordinated action of a number of constituent processes, including linguistic, cognitive, and sociocultural processes, which must be automatized to achieve speedy performance. For the EFL learners in this study, these underlying components seem to take longer to be automatized. The present ndings imply that analyses of the two attributes, language knowledge and processing capacity, provide different types of information about L2 pragmatic comprehension ability. Development of pragmatic comprehension was found to be at least twodimensional, including acquiring pragmatic knowledge and achieving automatic control of processing it in real time. To examine the potential factors that affect knowledge and processing capacity, this study addressed the relationships among general L2 prociency (a languagespecic factor), speed of lexical judgment (a cognitive-specic factor), and pragmatic comprehension ability. The relationship between the prociency and accuracy dimensions of pragmatic comprehension was signicant (r = 0.39), as was the relationship between lexical access speed and speed dimension of pragmatic comprehension (r = 0.40). However, L2 prociency bore no relationship to the speed of comprehending implied meaning, and lexical access speed had no relationship with accuracy of comprehension. Moreover, accuracy and speed were found to be unrelated to each other. These results further support the distinction between the language knowledge and processing capacity in language performance. Bachman and Palmer (1996) dene language knowledge as an underlying domain of information available in producing and interpreting language. It in330 TESOL QUARTERLY

cludes organizational knowledge that deals with formal aspects of language (e.g., grammar, lexis, discourse structure) and pragmatic knowledge that deals with functional and sociolinguistic aspects of language. Accuracy of pragmatic comprehension seems to reect learners organizational knowledge and pragmatic knowledge, which, in part, correspond to the underlying constructs measured in language prociency exams such as TOEFL. L2 prociency seems to play a decisive role in improving the accuracy of pragmatic comprehension. Speed, on the other hand, is an indication of processing abilities, not knowledge (White & Genesee, 1996). Speed of pragmatic comprehension represents learners ability to process and access pragmatic knowledge in real time. Because language processing subsumes the coordinated action of a number of underlying component processes, performance speed results from efficient and automatic operation of those component processes (Schneider & Fisk, 1982; Segalowitz, 2000, 2003). Lexical access speed, a cognitive processing variable, is considered to form one of the component processes and directly underlies performance. Therefore, when lexical retrieval occurs quickly, overall performance speed is promoted, as observed in the speed in pragmatic comprehension in this study. Speedy processing of lower component process was found to be associated with the speedy processing of higher order information, namely pragmatic meaning.


This study has several implications for the teaching of English as a foreign language. First, the signicant impact of prociency on accurate comprehension of implied meaning found in this study suggests that classroom instruction of pragmatic comprehension could be benecial for low prociency learners. In terms of the teaching order, because more conventional refusal items were found easier to comprehend than less conventional opinion-giving items, the more conventional items should be introduced rst. When teaching indirect refusals, teachers can highlight typical refusal patterns (e.g., giving an excuse for a refusal), emphasizing that these conventional features assist comprehension. Teachers can also identify other indirect items that include conventional features and develop listening materials based on the items. For instance, previous research found conventional expressions associated with some speech acts, such as requests, compliments, complaints, and suggestions (e.g., Blum-Kulka, House, & Kasper, 1989; Matsumura, 2001; Wolfson & Manes, 1980). As an additional activity, learners could compare the conventional expressions cross-linguistically with their L1s. Such activity

could raise learners awareness of conventionality; some conventional features can be shared universally across languages, and others can be specic to a language group. Comprehension of the former types might be easier because learners can draw on pragmatic procedures and patterns that are conventionalized in their L1s. When teaching less conventional implicatures, on the other hand, classroom teachers could develop listening materials from movies and TV dramas. Scenes from movies, dramas, or plays often serve as a rich source of pragmatic input because they contain a variety of conversational exchanges in which the speakers reply does not provide a straightforward answer to the question. The concept of adjacency pairs from the eld of conversation analysis (CA) can be applied to identify the exchanges where the relevance maxim is violated. According to CA, conversation usually consists of a sequence of turns, and specic interactional constraints determine what is relevant in the next turn (e.g., Goffman 1976). The concept of adjacency pairs in CA explains how the second pair part in a conversation is constrained by the rst pair part. A question asking for evaluative comments should be followed by good or bad. When asked for information, a person usually provides the requested information; thus, not providing the expected or straightforward answer is a case of a marked violation of the adjacency pair rule. These cases can be identied in movie and drama scripts and then used as teaching materials. When practicing comprehension of less conventional implicatures, teachers could encourage learners to use contextual information, such as paralinguistic cues (e.g., intonation, tone of voice, pause length), background information, and personal experience, to infer indirect speaker meaning. A follow-up discussion on why speakers use indirect utterances instead of literal ones would also deepen learners understanding of the strategic intention behind implicatures. Violating the relevance maxim sometimes appears as a politeness strategy to avoid explicit answers and save face in a social interaction. In other times, outing is also used to emphasize the point to be made, or to make the point more dramatic and interesting (Thomas, 1995). By understanding the logic and strategy behind using implicatures, learners will realize the functions that implicatures play in every day social interaction. Another instructional implication relates to training in language processing. It is now widely believed that the fundamental goal of language acquisition is to achieve accurate as well as uent and speedy skill execution in language tasks (Segalowitz, 2001). If performance speed takes longer time to develop or does not develop naturally along with general prociency in a foreign language context, as implied in this study, L2 learners could benet from focused classroom training in processing speed. The types of training that best enhance uency involve repetitions

with consistent associations between linguistic input and its meaning. Because the speed of lexical retrieval was found signicantly related to overall speed in processing pragmatic information, training of such a lower level component could provide a substantial amount of processing experience in L2 and consequently enhance overall performance speed. It is assumed that once the lower level processing, such as lexical processing, is automatized, learners can pay more attention to processing of higher level, semantic and pragmatic information (Segalowitz, 2003). At the initial stage of language acquisition, performance requires a great deal of attention, resulting in slower processing speed (Segalowitz, 2000). As learners become more skilled, basic processing components become routinized and automatic, freeing up space for other more complex, higher order processing components. Therefore, it could be more practical and benecial to gradually stage the training sessions of performance speed, shifting from training on lower level mechanisms such lexical and syntactic information to training on higher level mechanisms such as discourse, sociocultural, and pragmatic information. Training of processing speed should address diverse mechanisms and should be practiced in an incremental manner in a meaningful context (Gatbonton & Segalowitz, 1988, 2005). This type of classroom practice will add to the growing literature on the effect of training on uency development, by showing the types of training that best enhance speed in language performance (Fukkink, Hulstijn, & Simis, 2005; Segalowitz, Segalowitz, & Wood, 1998; Snellings, van Gelderen, & Glopper, 2002).


One major limitation in this study is the homogeneity of the participant population. Because the participants were limited to adult Japanese EFL learners, ndings cannot be generalized to different groups of English language learners. This study found pragmatic gains among EFL learners in an intensive, massed instructional format. However, without comparative groups of ESL learners or learners in a traditional English class in nonintensive format, the precise contextual factors that affect the development are not clear. Compared with EFL learners, ESL learners are potentially more often exposed to everyday out-of-class practice in inferential comprehension, and have more out-of-class associative practices between input and meaning. Therefore, when overall L2 prociency is controlled, ESL learners might perform pragmatic functions better and faster than EFL learners who have limited exposure to target input outside the classroom. Similarly, it is possible that EFL learners in

an intensive program receive rich, condensed, in-class input that promotes general language skills, and that EFL learners in a regular university class receive a limited amount of in-class input that is distributed over months. Thus, the intensive format of classroom practice may contribute to accurate and speedy comprehension of implied meaning. However, without data from ESL and EFL learners in a traditional instructional setting, this study was not able to make such comparisons. To overcome these limitations, future research should examine pragmatic comprehension over a wider range of instructional environments. Another limitation of this study relates to the limited number of data collection sessions. Because this study compared performance gains between the beginning and end of the 7-week session, the study did not capture the incremental developmental trend over an extended period of time. Such longitudinal analysis is important, particularly because the size of speed gain was smaller than that of accuracy gain. Future research should incorporate more frequent observations and monitor students progress for a longer period of time. Finally, this study examined the effects of one cognitive-specic factor (lexical access speed), and one language-specic factor (general prociency) on pragmatic comprehension. Although speedy pragmatic processing was found to be signicantly related to speedy lexical processing, the degree of relationship, as exemplied in correlation coefficient, was rather small (i.e., approximately 16% of shared variance). The results imply that there are other cognitive and noncognitive variables that affect processing speed. For instance, speed might be affected by individual processing strategies. Some learners might activate multiple interpretations of nonliteral meaning and choose the most plausible one after closely evaluating them, and others might jump to the most accessible interpretation without considering other options. Therefore, it remains for future research to identify other processing related factors, both cognitive and noncognitive, that could inuence the development of speed in pragmatic comprehension.

I thank the two anonymous TESOL Quarterly reviewers for their insightful comments on earlier versions of this paper. Thanks also go to Ingrid Johnson for proofreading this manuscript.

Naoko Taguchi is an assistant professor in the Department of Modern Languages at Carnegie Mellon University, Pittsburgh, Pennsylvania, United States, where she

teaches Japanese and second language acquisition. She holds a doctorate in applied linguistics from Northern Arizona University. Her research interests include second language pragmatics, classroom-based research, and foreign language teaching.

Ackerman, P. (1988). Determinants of individual differences during skill acquisition: Cognitive abilities and information processing. Journal of Experimental Psychology. General, 117, 288318. Anderson, J. R. (1990). Cognitive psychology and its implications (3rd ed.). New York: W. H. Freeman. Bachman, L. F., & Palmer, A. S. (1996). Language testing in practice: Designing and developing useful language tests. Oxford: Oxford University Press. Baron-Cohen, S., Leslie, A. M., & Frith, U. (1985). Does the autistic child have a theory of mind?. Cognition, 21, 3746. Beebe, L. M., Takahashi, T., & Uliss-Weltz, R. (1990). Pragmatic transfer in ESL refusals. In R. Scarcella, D. Andersen, & S. Krashen (Eds.), Developing communicative competence in a second language (pp. 5574). New York: Newbury House. Bialystok, E. (1990). Communication strategies. Oxford: Blackwell. Blum-Kulka, S., House, J., & Kasper, G. (1989). Cross-cultural pragmatics: Requests and apologies. Norwood, NJ: Ablex. Bouton, L. (1992). The interpretation of implicature in English by NNS: Does it come automatically without being explicitly taught? Pragmatics and Language Learning Monograph Series, 3, 6477. Bouton, L. (1994). Can NNS skill in interpreting implicature in American English be improved through explicit instruction?: A pilot study. Pragmatics and Language Learning Monograph Series, 5, 88108. Bouton, L. (1999). Developing nonnative speaker skills in interpreting conversational implicatures in English. In E. Hinkel (Ed.), Culture in second language teaching and learning (pp. 4770). Cambridge: Cambridge University Press. Brown, J. D. (1990). The use of multiple t tests in language research. TESOL Quarterly, 24, 770772. Cameron, R., & Williams, J. (1997). Sentence to ten cents: A case study of relevance and communicative success in nonnativenative speaker interactions in a medical setting. Applied Linguistics, 18, 415445. Canale, M., & Swain, M. (1980). Theoretical aspects of communicative approaches to second language teaching and testing. Applied Linguistics, 1, 147. Carber, R. P. (1990). Reading rate: A review of research and theory. New York: Academic Press. Carrell, P. L. (1981). Relative difficulty of request forms in L1/L2 comprehension. In M. Hines & W. Rutherford (Eds.), On TESOL 81 (pp. 141152). Alexandria, VA: TESOL. Carrell, P. L. (1984). Inferencing in ESL: Presuppositions and implications of factive and implicative predicates. Language Learning, 34, 119. Cedrus. (2003). SuperLab 3.0. San Pedro, CA: Author. Cook, M., & Liddicoat, A. J. (2002). The development of comprehension in interlanguage pragmatics: The case of request strategies in English. Australian Review of Applied Linguistics, 25, 1939. DeKeyser, R. (2001). Automaticity and automatization. In P. Robinson (Ed.), Cognition and second language instruction (pp. 124151). New York: Cambridge University Press.

ETS. (2003). TOEFL testing program. Princeton, NJ: Author. Available from http:// ETS. (2005). TOEFL test and score data summary: 20042005 test year data (Report No. TOEFL-SUM-0405-DATA). Princeton, NJ: Author. Available from http:// Freed, B. (2000). Is uency, like beauty, in the eyes (and ears) of the beholder? In H. Riggenback (Ed.), Perspectives on uency (pp. 243265). Ann Arbor: University of Michigan Press. Freed, B. F., Segalowitz, N., & Dewey, D. P. (2004). Context of learning and second language uency in French. Studies in Second Language Acquisition, 26, 277303. Fukkink, R., Hulstijn, J., & Simis, A. (2005). Does training in second-language word recognition skills affect reading comprehension? An experimental study. Modern Language Journal, 89, 5475. Garcia, P. (2004). Developmental differences in speech act recognition: A pragmatic awareness study. Language Awareness, 13, 96115. Gatbonton, E., & Segalowitz, N. (1988). Creative automatization: Principles for promoting uency within a communicative framework. TESOL Quarterly, 22, 473492. Gatbonton, E., & Segalowitz, N. (2005). Rethinking communicative language teaching: A focus on access to uency. Canadian Modern Language Review, 61, 325353. Goffman, E. (1976). Replies and responses. Language in Society, 5, 254313. Grice, P. (1975). Logic and conversation. In P. Cole & J. Morgan (Eds.), Syntax and semantics. Vol. 3: Speech acts (pp. 4158). New York: Academic Press. Hala, S., & Carpendale, J. (1997). All in the mind: Childrens understanding of mental life. In S. Hara (Ed.), The development of social cognition (pp. 189240). East Sussex, England: Psychology Press. Hatch, E., & Lazaraton, A. (1991). The research manual: Design and statistics for applied linguistics. Boston: Heinle & Heinle. Hirai, A. (1999). The relationship between listening and reading rates of Japanese EFL learners. Modern Language Journal, 83, 367384. Holtgraves, T. (1999). Comprehending indirect replies: When and how are their conveyed meanings activated? Journal of Memory and Language, 38, 519540. Hoover, M. L., & Dwivedi, V. D. (1998). Syntactic processing by skilled bilinguals. Language Learning, 48, 129. Hymes, H. D. (1972). On communicative competence. In J. B. Pride & J. Holmes (Eds.), Sociolinguistics: Selected readings (pp. 269293). Middlesex, England: Penguin. Jiang, N. (2002). Formmeaning mapping in vocabulary acquisition in a second language. Studies in Second Language Acquisition, 24, 617637. Juffs, A. (1998). Main verb vs. reduced relative clause ambiguity resolution in second language sentence processing. Language Learning, 48, 107147. Juffs, A., & Harrington, M. (1996). Garden path sentences and error data in second language sentence processing. Language Learning, 46, 283326. Kasper, G. (1984). Pragmatic comprehension in learnernative speaker discourse. Language Learning, 34, 120. Kasper, G. (2001). Four perspectives on L2 pragmatic development. Applied Linguistics, 22, 502530. Kasper, G., & Rose, K. (2003). Pragmatic development in a second language. Boston: Blackwell. Koike, D. A. (1996). Transfer of pragmatic competence and suggestions in Spanish foreign language learning. In S. M. Gass & J. Neu (Eds.), Speech acts across cultures (pp. 257281). Berlin: Mouton de Gruyter.

Kotsonis, M. E. (1981). Childrens interpretations of conversationally implied meanings. Unpublished doctoral dissertation. University of Virginia, Charlottesville. Matsumura, S. (2001). Learning the rules for offering advice: A quantitative approach to second language socialization. Language Learning, 51, 635679. McLaughlin, B. (1987). Theories of second language learning. London: Edward Arnold. McLaughlin, B., & Heredia, J. (1996). Information-processing approaches to research on second language acquisition and use. In W. C. Ritchie & T. K. Bhatia (Eds.), Handbook of second language acquisition (pp. 213228). San Diego, CA: Academic Press. Morgan, J. (1978). Two types of convention in indirect speech acts. In P. Cole (Ed.), Syntax and semantics. Vol. 9: Pragmatics (pp. 261280). New York: Academic Press. Nelson, G. L., Carson, J., Batal, M. A., & Bakary, W. E. (2002). Cross-cultural pragmatics: Strategy use in Egyptian Arabic and American English refusals. Applied Linguistics, 23, 163189. Neumann, E., McCloskey, M. S., & Felio, A. C. (1999). Cross-language positive priming disappears, negative priming does not: Evidence for two sources of selective inhibition. Memory & Cognition, 27, 10511063. Ohta, A. (2001). Second language acquisition processes in the classroom: Learning Japanese. Mahwah, NJ: Lawrence Erlbaum. Robinson, P. (2003). Cognition and second language instruction. New York: Cambridge University Press. Robinson, P. (2005). Aptitude and second language acquisition. Annual Review of Applied Linguistics, 25, 4673. Rver, C. (2005). Testing ESL pragmatics: Development and validation of a Web-based assessment battery. Frankfurt, Germany: Peter Lang. Rumelhart, D. E., & McClelland, J. (1986). Parallel distributed processing: Explorations in the microstructure of cognition. Cambridge, MA: MIT Press. Schmidt, R. (1992). Psychological mechanisms underlying second language uency. Studies in Second Language Acquisition, 14, 357385. Schneider, W., & Fisk, A. (1982). Degree of consistent training: Improvements in search performance and automatic process development. Perception & Psychophysics, 31, 160168. Segalowitz, N. (1997). Individual differences in second language acquisition. In A. de Groot & J. Kroll (Eds.), Tutorials in bilingualism (pp. 85112). Hillsdale, NJ: Lawrence Erlbaum. Segalowitz, N. (2000). Automaticity and attentional skill in uent performance. In H. Riggenbach (Ed.), Perspectives on uency (pp. 200219). Ann Arbor: University of Michigan. Segalowitz, N. (2001). On the evolving connections between psychology and linguistics. Annual Review of Applied Linguistics, 21, 322. Segalowitz, N. (2003). Automaticity and second languages. In C. Doughty & M. Long (Eds.), The handbook of second language acquisition (pp. 382408). Malden, MA: Blackwell. Segalowitz, N., & Freed, B. (2004). Context, contact, and cognition in oral uency acquisition. Studies in Second Language Acquisition, 26, 175201. Segalowitz, N., & Segalowitz, S. J. (1993). Skilled performance, practice, and the differentiation of speed-up from automatization effects: Evidence from second language word recognition. Applied Psycholinguistics, 14, 369385. Segalowitz, S. J., Segalowitz, N., & Wood, A. (1998). Assessing the development of second language word recognition. Applied Psycholinguistics, 19, 5367. Segalowitz, N., Watson, V., & Segalowitz, S. (1995). Vocabulary skill: Single case

assessment of automaticity of word recognition in a timed lexical decision task. Second Language Research, 11, 121136. Skehan, P. (1995). Analyzability, accessibility, and ability for use. In G. Cook & B. Seidlhofer (Eds.), Principle and practice in applied linguistics (pp. 91106). Oxford: Oxford University Press. Skehan, P. (1998). A cognitive approach to language learning. Oxford: Oxford University Press. Smith, F. (1975). Comprehension and learning: A conceptual framework for teachers. New York: Holt, Rinehart, & Winston. Snellings, P., van Gelderen, A., & Glopper, K. (2002). Lexical retrieval: An aspect of uent second language production that can be enhanced. Language Learning, 52, 723754. Sperber, D., & Wilson, D. (1991). Loose talk. In S. Davis (Ed.), Pragmatics: A reader (pp. 540550). Oxford: Oxford University Press. Sperber, D., & Wilson, D. (1995). Relevance: Communication and cognition (2nd ed.). Cambridge: Cambridge University Press. SPSS. (1998). SPSS base 8.0 applications guide. Chicago: Author. Summers, D. (Ed.). (1995). Longman dictionary of contemporary English (3rd ed.). New York: Longman. Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics (4th ed.). Boston: Allyn & Bacon. Taguchi, N. (2002). An application of relevance theory to the analysis of L2 interpretation processes: The comprehension of indirect replies. International Review of Applied Linguistics, 40, 151176. Taguchi, N. (2003). Pragmatic performance in comprehension and production of English as a second language. Unpublished doctoral dissertation. Northern Arizona University, Flagstaff. Taguchi, N. (2005). Comprehending implied meaning in English as a second language. Modern Language Journal, 89, 543562. Takahashi, S., & Roitblat, H. L. (1994). Comprehension process of second language indirect requests. Applied Psycholinguistics, 15, 475506. Thomas, J. (1995). Meaning in interaction: An introduction to pragmatics. London: Longman. Towell, R. (2002). Relative degrees of uency: A comparative case study of advanced learners of French. International Review of Applied Linguistics, 40, 117150. Verschueren, J. (1999). Understanding pragmatics. London: Arnold. White, L., & Genesee, F. (1996). How native is near-native? The issue of ultimate attainment in adult second language acquisition. Second Language Research, 12, 233265. White, L., & Juffs, A. (1998). Constraints on wh-movement in two different contexts of narrative language acquisition: Competence and processing. In S. Flynn, G. Martohardjono, & W. ONeil (Eds.), The generative study of second language acquisition (pp. 111129). Mahwah, NJ: Lawrence Erlbaum. Widdowson, H. G. (1989). Knowledge of language and ability for use. Applied Linguistics, 10, 128137. Wolfson, N., & Manes, J. (1980). The compliment as a social strategy. Papers in Linguistics, 13, 391410. Wolvin, A. D., & Coakley, A. G. (1985). Listening (2nd ed.). Dubuque, IA: William C. Brown. Ying, H. G. (1996). Multiple constraints on processing ambiguous sentences: Evidence from adult L2 learners. Language Learning, 46, 681711.



Pointing Out Frequent Phrasal Verbs: A Corpus-Based Analysis

Brigham Young University Provo, Utah, United States

Supposing is good, but nding out is better. Mark Twain (DeVoto, 1922)

This study attempts to shed new light on one of the most notoriously challenging aspects of English language instructionthe English phrasal verbs. The highest frequency phrasal verb constructions in the 100-million-word British National Corpus are identied and analyzed. The ndings indicate that a small subset of 20 lexical verbs combines with eight adverbial particles (160 combinations) to account for more than one half of the 518,923 phrasal verb occurrences identied in the megacorpus. A more specic analysis indicates that only 25 phrasal verbs account for nearly one third of all phrasal-verb occurrences in the British National Corpus, and 100 phrasal verbs account for more than one half of all such items. Subsequent semantic analyses show that these 100 high-frequency phrasal verb forms have potentially 559 variantmeaning senses. The authors discuss how learners, teachers, and materials developers might utilize the ndings of the study to improve instruction of phrasal verbs in English language education.

any English language teachers have noted the importance of multiword knowledge in developing their learners nativelike uency (Moon, 1997; Schmitt, 2004; Wray, 2000, 2002). Idioms (e.g., kick the bucket), phrasal verbs (e.g., chew out), stock phrases (e.g., how do you do), prefabs (e.g., the point is), and other multiword structures are crucial to English, and they add a denite richness to the language. However, there has been a general confusion regarding which multiword items to teach and the best ways to include them in language training (Condon & Kelly, 2002; Darwin & Gray, 1999; Nesselhauf, 2003) and language assessment (Read, 2000). Of particular concern is the evidence that multiword items are not learned well through ordinary language experience (Coady, 1997, p. 282).

TESOL QUARTERLY Vol. 41, No. 2, June 2007


During the past two decades, as many researchers and teachers have begun to shift attention from syntax to vocabulary in second language education (Folse, 2004; Laufer, 1997), interest in multiword vocabulary items has been growing. High-powered computers, robust software, and large electronic collections of actual language (called corpora; singular, corpus) have enabled researchers to better identify and classify these otherwise elusive structures that permeate English as well as many other languages (Hunston, 2002; Moon, 1997; Read, 2004; Stubbs, 2001). However, whereas corpus linguists have been able to provide many more insights regarding these items, it remains clear that the surface of this complex issue has scarcely been scratched (Read, 2004). Perhaps the most important area for progress in the discussion of multiword items is English phrasal verbs. The study of phrasal verbs promises to provide valuable insights into what many linguists and applied linguists have begun to recognize as a multiword middle ground between syntax and lexis that has important ramications for second language acquisition (Gass & Selinker, 2001, p. 391). Additionally, phrasal verbs are notoriously difficult for nonnative learners to acquire, a problem exacerbated by the fact that they tend to be very common and highly productive in the English language as a whole (Celce-Murcia & Larsen Freeman, 1999; Darwin & Gray, 1999; Moon, 1997). This dilemma is further complicated by the fact that many nonnative English speakers actually avoid using phrasal verbs altogether, especially those learners at the beginning and intermediate levels of prociency. (See Liao & Fukuya, 2004, for a review of this topic.) Even learners whose native language actually contains phrasal verbs (e.g., Dutch) often avoid using such forms when communicating in English (Hulstijn & Marchena, 1989). The purpose of the current study is to establish a logical rationale for narrowing the scope of phrasal verbs in English language training based on frequencies of actual occurrence in a large representative corpus of EnglishBritish National Corpus: World Edition (BNC; British National Corpus Consortium, 2000). With this aim in mind, we are more concerned with probabilities of encounter than with possibilities of acquisition. The latter will be left to the fruitful discussions of how multiword items are stored in and retrieved from the mental lexicon (e.g., Wray, 2002) and to actual testing of such items with learners of English (e.g., Read, 2000). We are also more concerned with establishing a basis for what to teach, rather than how to teacha topic better suited to accounts with that particular focus (e.g., Condon & Kelly, 2002; Cornell, 1985; McCarthy & ODell, 2004; Side, 1990; Wyss, 2003). Because our study is intended to be purely exploratory and informative in nature, no issues of causality will be tested or implied. However, it is our hope that the rich data resulting from the study can be used in

productive ways to inform English language teaching, materials development, and testing and to provide important information for future empirical studies involving language learners and their actual acquisition of English phrasal verbs.


It is rare to read an article about phrasal verbs without some discussion of denitions. Historically, linguists have focused much of their attention on characterizing and classifying phrasal verbs based on syntactic and semantic considerations such as single-word replacement, separability, literal versus gurative meanings, and so forth (e.g., Bolinger, 1971). Some attention has also been given to distinguishing between phrasal verbs, prepositional verbs, phrasal prepositional verbs, and free combinations (e.g., Biber, Johansson, Leech, Conrad, & Finegan, 1999). However, as Darwin and Gray (1999) point out, many of the traditional tests for phrasal verbs have notable exceptions, and experts often disagree on which items to include under this fuzzy grammatical category. This confusion among sources, according to the researchers, leads to confusion for students and teachers (p. 67). Interestingly, however, even their own alternative approach to phrasal verb classication has come under criticism on the grounds that their system excludes forms that might actually aid in the teaching of phrasal verbs (Sawyer, 2000, p. 151), and that their proposed classication system for phrasal verbs may not do much to advance second language curriculum development and instruction beyond what is already available (Sheen, 2000). Perhaps missing in this ongoing debate among linguists is the ecological reality of phrasal verb forms in the actual language experience of nonnative speakers of English. In other words, if even the linguists and grammarians struggle with nuances of phrasal verb denitions, of what instructional value could such distinctions be for the average second language learner? Therefore, for purposes of the current study, we will rely on a more functional and objective denition of phrasal verb that is stated as follows: all two-part verbs in the BNC consisting of a lexical verb (LV) proper (tagged as VV in the BNC) followed by an adverbial particle (tagged as AVP) that is either contiguous (adjacent) to that verb or noncontiguous (i.e., separated by one or more intervening words). No other criteria for classication will be used in this study, and we will rely on the good judgment of the readers to decide whether, and for what purpose, the data may warrant further scrutiny. In terms of examples, our denition would encompass verb units with varying degrees of semantic transparency, including literal (e.g., sit down;

sit yourself down), gurative (e.g., chew out; chew the team out), and all degrees in between. In other words, we make no distinctions in our analysis based on the semantic transparency of the phrasal verbs. Whereas some evidence suggests that second language learners struggle more with gurative phrasal verbs than literal phrasal verbs (Liao & Fukuya, 2004), it is not our task in the current study to classify or distinguish phrasal verbs in any way; rather, our aim is simply to determine which phrasal verbs have the greatest impact in the language as a whole. We also recognize, at the outset, that potential limitations are inherent in our denition of phrasal verbs. First, we are relying heavily on the BNC tagging software to accurately identify the AVP functions (e.g., You can turn in for the night) as opposed to the prepositional functions that many of these forms also have (e.g., The police told me to turn in the opposite direction). However, whereas some tagging inaccuracies may exist, we also assume that these will be very few in number, as the creators of the tagged BNC report a classication error rate of 1.58% (less than 2 in 100) for AVPs and 0.59% (less than 1 in 100) for prepositions (Leech & Smith, 2000). In addition, we acknowledge that any computerized analysis of linguistic forms tends to underestimate the number of meanings elicited by those forms in real-language contexts (cf. Widdowson, 2000).


Despite the potential limitations of the denition noted earlier, there are strong reasons for pursuing a rationale for dealing with phrasal verbs that utilizes actual frequency of occurrences in the language, rather than more traditional approaches that have often relied on isolated linguistic examples, teachers intuitions, or random groups of phrasal verbs in language training curriculaa point clearly articulated by Darwin and Gray (1999):
Another reason for the somewhat arbitrary presentation of phrasal verbs is that very little has been done to determine frequency of particular phrasal verbs. Thus, instructors, curriculum designers, and researchers are left working with what they determine by intuition to be the most common or most needed phrasal verbs. Their intuition, though, may not be correct. (p. 67; cf. Darwin & Gray, 2000)

As stated previously, one of the growing strengths of corpus linguistics is in the identication and classication of multiword units:
A major part of the patterning revealed by concordances is the extent of phraseology, which is not obvious to speakers, and has indeed been ignored by many linguists. The patterns have been discovered, but not created, by the computer. (Stubbs, 2001, p. 153; cf. Hunston, 2002)

To date, there have been several attempts to identify and classify English phrasal verbs based on corpus ndings. Chief among these are the Longman Dictionary of Phrasal Verbs (Courtney, 1983), the Collins Cobuild Dictionary of Phrasal Verbs (Sinclair & Moon, 1989), NTCs Dictionary of Phrasal Verbs and Other Idiomatic Verbal Phrases (Spears, 1993), and the Cambridge International Dictionary of Phrasal Verbs (Walter & Pye, 1997). Each of these extensive works attempts to identify a large number of English phrasal verbs and to provide a denition and contextualized example of each form. Implied in all the methodologies is that the compilers have attempted to identify and dene important phrasal verbs based on corpus data. Beyond this implication, however, there is very little information about actual frequency data that could be used by teachers, materials developers, and test designers who may wish to prioritize treatment of phrasal verbs based on their relative impact in the language as a whole. Perhaps the best treatment of relative frequency is contained in Biber et al., (1999), the Longman Grammar, where a short subsection of the text is devoted to describing the most prolic verbs that combine with AVPs to form phrasal verbs within seven semantic domains (activity intransitive, activity transitive, mental transitive, etc.). The authors also consider how the relative frequencies of the most prolic phrasal verbs vary by register (i.e., in conversation, ction, news, and academic texts). They provide similar, but separate, information regarding prepositional verbs. Their criteria for inclusion is that an item must occur over 40 times per million words in at least one register of their particular corpus (p. 410). Beyond this, however, they provide no information to allow comparison between their narrow set of 31 phrasal verbs and the next most prolic set, for example, or between their set and other phrasal verbs in general. We do not view this as an oversight, but as function of space constraints and the broader focus and purposes of the Longman Grammar text in general. However, we feel that more focused and expanded data analyses of phrasal verbs are warranted and that such analyses can benet English language teaching, materials development, and assessment. The remainder of our study is dedicated to addressing this issue. The primary aims of our study are 1. To determine the most frequent AVPs. 2. To determine how often these forms function as AVPs versus prepositions. 3. To determine the most frequent lexical verbs used in phrasal verb constructions. 4. To determine how often such verbs function as stand-alone verbs versus phrasal verbs.

5. To determine the extent to which these verbs interact with various AVPs. 6. To establish a list of the most frequent verb-plus-particle combinations based on overall frequency and coverage. 7. To determine the approximate number of word senses associated with each of the most frequent verb-plus-particle combinations.

METHOD The Corpus

The source of phrasal verbs analyzed in this study is the BNC, described as follows:
The British National Corpus [BNC] is a one hundred million word snapshot of British English, both spoken and written, at the end of the twentieth century. . . . The corpus contains about 4000 samples from the widest possible range of linguistic productions, automatically annotated with part-of-speech tags by the CLAWS system, and SGML-encoded according to the Text Encoding Initiatives Guidelines. (CD cover)

Data Gathering
The grammatically tagged version of the BNC was rst converted into a format that could be imported into Microsoft SQL Server, a powerful relational database program. In short, this process included dividing the corpus into every possible two-, three-, four-, ve-, six-, and seven-word chunk, with their accompanying grammatical tags. Table 1 shows an example of how the database appeared after the transformations took place. The second step in gathering the data consisted of software queries to identify and report every instance where a lexical verb (LV) was followed by an adverbial particle (AVP). The particle could be immediately adjacent to the verb (LV + AVP), within two words (LV + X +AVP), within three words (LV + X + X + AVP), within four words (LV + X + X + X + AVP), within ve words (LV + X + X + X + X + AVP), or within six words (LV + X + X + X + X + X + AVP). For example, taken (LV) over (AVP) on the rst data line of Table 1 is a LV + AVP phrasal verb because there are no words between the LV (taken) and the AVP (over), whereas took (LV) a (AT0) step (NNI) back (AVP) on the fth data line is a LV + X + X + AVP phrasal verb because there are two intervening words (the article a and the noun step) between the LV (took) and the AVP (back). Our subsequent analyses of all outcomes indicated the presence of many false

TABLE 1 Sample of Four-Word Chunks From BNC With Grammatical Parts of Speech Identied Chunk 1 2 3 4 5 6 7 8 9 10 11 12 13 14 W1 taken taken took taken took take taken taken take takes taken taken took taken POS 1 LV LV LV LV LV LV LV LV LV LV LV LV LV LV W2 over up on up a you over up it us up back over over POS 2 AVP AVP AVP AVP AT0 PNP AVP AVP PNP PNP AVP AVP AVP AVP W3 by by a in step back by with out back by to from from POS 3 PRP PRP AT0 PRP NN1 AVP PRP PRP AVP AVP PRP PRP PRP PRP W4 the The new the back to a the on to a the the the POS 4 AT0 AT0 AJ0 AT0 AVP PRP AT0 AT0 PRP PRP AT0 AT0 AT0 AT0

Note. W = word. POS = part of speech tag: LV = lexical verb (tagged as VV in BNC); AVP = adverbial phrase, AT0 = article, PNP = pronoun, PRP = preposition, NN1 = singular noun.

phrasal verbs beyond the four-word scenario. This fact, coupled with the relatively infrequent occurrences of phrasal verbs with longer separations (mostly a frequency of 1), caused us to limit our attention to those of the two- (take on), three- (take it on), and four-word (took the nodules back) varieties. We should note, however, that legitimate longer separations do exist in the BNC and may be worth studying in the future (e.g., send your certicate of motor insurance back). In the third step, we lemmatized the outcomes, so that all inectional forms of the same verb were counted together. For example, the forms look, looks, looking, and looked were grouped under the lemma LOOK; the forms take, takes, taking, taken, took, and taken under TAKE; and the forms give, gives, giving, gave, and given under GIVE. Hereafter, we distinguish lemmas from individual types by using all uppercase letters to indicate lemmas (cf. Stubbs, 2002).

Form-Based Data Analyses

To perform frequency and coverage analyses, we created Excel spreadsheets from the query outcomes explained in the preceding section. This procedure allowed us to form lists consisting of frequency rankings, cumulative frequencies, cumulative percentages, and so forth.

Meaning-Based Data Analyses

In recognition of the fact that phrasal verb forms, like individual word forms, have multiple, context-sensitive meanings, we also used WordNet

(Miller, 2003) to establish a more realistic picture of the benets of establishing lists of high-frequency phrasal verbs for teaching purposes. WordNet (Miller, 2003) is an electronic lexical database that, among other functions, recognizes distinctions between different senses of the same word forms (Fellbaum, 1998). For instance, WordNet contains 10 different senses for the phrasal verb put out (to cause inconvenience, to give considerable effort, to smother, to anesthetize, etc.) and 8 different senses for the phrasal verb work out (to happen in a certain way, to elaborate, to exercise, to solve, etc.). Our intention with this semantic analysis of the data is to avoid the kinds of oversimplication found in many corpus-based vocabulary studies, namely, that frequency of word forms is discussed as though it were equivalent to frequency of word meanings. We view this as one of the fundamental issues to be addressed as corpus linguists attempt to build bridges to language education.


Table 2 shows the total number of all BNC grammatical tags for each of the 16 preposition-particle forms. It also indicates the number of times each of these forms is tagged as an AVP as opposed to a preposition or
TABLE 2 Frequency of Adverbial Particles (AVPs) in BNC Form out up down back off round along over around on through about in under by across Total Total tags 149,727 180,792 91,832 97,154 67,479 30,821 18,555 128,304 43,391 705,790 81,184 190,615 1,845,077 60,049 504,969 24,053 4,219,792 # as AVP 145,706 158,064 72,709 75,233 37,751 10,895 4,925 32,526 10,384 54,956 5,797 12,587 34,411 313 371 13 656,641 % as AVP 97.3 87.4 79.2 77.4 55.9 35.3 26.5 25.4 23.9 7.8 7.1 6.6 1.9 0.5 0.1 0.1 15.6*

Note. # = token frequency. * = Average of column.



other grammatical structure, such as the noun out (in sports language) or the noun back (body part). The overall totals indicate that these forms function 15.6% of the time as AVPs. This fact, coupled with the sheer number of particle forms overall (656,641), suggests that phrasal verb (PV) constructions (comprised of verbs plus AVPs) constitute a major grammatical class. To put this in another light, learners will encounter, on average, one in every 150 words of English they are exposed to, or roughly 2 per average page of written text (i.e., assuming 300 words per page). Of course, these exposure estimates are for the corpus as a whole and could vary somewhat based on register type (informal speech, narrative ction, expositions, etc.)a point which future research should more carefully consider. It is also clear from the values in Table 2 that certain forms are more likely to act as particles than as prepositions. In particular, out (97.3%), up (87.4%), down (79.2%), and back (77.4%) occur much more often as AVPs in PV constructions (e.g., she picked out a birthday card) than they do as prepositions in prepositional phrases (e.g., she ran out the door). Learners who understand this concept would have a denite advantage in recognizing PV forms containing these highly prolic AVPs. In contrast, certain forms, such as under (0.5%), by (0.1%), and across (0.1%), will rarely appear as particles in PV constructions. Learners could also put this information to good use. From a particle approach, the most difficult forms to deal with appear to be words like on (7.8% as particles) and in (1.9% as particles) that function much more often as prepositions in general but also appear to have substantial frequencies as particles (on 54,956; in 34,411). This dilemma suggests the need for learners to have other methods for recognizing PVs besides particle identication and leads naturally to a discussion of the verbal component in PV constructions.

Frequencies of Lexical Verbs in Phrasal Verb Constructions

Table 3 provides information relative to all lexical verb (LV) forms tagged in the BNC, as well as those that function in PV constructions. Approximately 5% of all LV tokens function in PV constructions, or one in every 20. Perhaps this nding becomes more meaningful when one considers that PVs, as a grammatical class, have a higher overall frequency than the verb are, the determiners this or his, the negative not, the conjunction but, or the pronoun they. Looking at this from still another angle, there are 10,404,107 tagged LVs in the 100-million-word BNC, or approximately 1 in every 9.6 words, on average. By extrapolation, this means that a learner will encounter 1 PV, on average, in every 192 words (9.6 20) of English, or nearly 2 per page of written text. It is also important to note that only 8% (1,572 lemmas) of the 19,682

TABLE 3 Descriptive Statistics for BNC Lexical Verbs (LVs) Functioning in Phrasal Verbs (PVs) Measurement LV tokens LV lemmas Average lemma frequency PV lemmas Total # 10,404,107 19,682 529 12,508 # in PVs 518,923* 1,572 330 % of total 5.0 8.0

Note. # = token frequency. Lemma = all inectional forms of a verb (e.g., look, looks, looking, looked) considered to be the same verb (e.g., LOOK). Average lemma frequency = calculated by dividing number of tokens by number of lemmas. PV lemma = a distinct phrasal type comprised of a lexical verb lemma and an (e.g., LOOK up, LOOK out, LOOK over = three distinct phrasal verb [PV] lemmas). * = Count based on PVs that are contiguous (verb [V] + adverbial phrase [AVP]), separated by one word (V + X + AVP), and separated by two words (V + X + X + AVP).

total LV lemmas function in PV constructions. Whereas 1,572 is still a staggering gure from a teaching and learning perspective, it is certainly more manageable than trying to gure out which of the 19,683 different LV lemmas function in PV constructions. Perhaps of more pedagogical concern, however, is the fact that, on average, any given LV lemma functioning in PV constructions repeats only 330 times in 100 million words, or once in every 303,030 words. Also problematic, from a teaching and learning perspective, is that at least 12,508 distinct PV lemmas exist in the BNC alone. For instance, the LV lemma PUT combines with at least 15 different particles in the BNC to form PV lemmas (PUT out, PUT up, PUT on, etc.). It is therefore crucial to establish a frequency ranking of PVs to determine if some are noticeably more prolic than others.

Verb-Particle Frequencies Involving Top 20 Lexical Verbs

Table 4 displays a frequency ranking of the top 20 LVs found in PV constructions. Perhaps the most important statistic in the table is that these 20 verbs are found in 53.7% of all PVs in the BNC; in other words, more than half of all the PVs contain a verb from this short list. Additionally, the PVs containing these 20 LVs make up 2.7% of all the LVs that learners would encounter in the BNC (279,882 10,404,717) approximately 1 in every 39. Interestingly, several of the verbs in the table appear more often in PV constructions than as stand-alone LVs (i.e., PICK, 70.0%; POINT, 52.0%; CARRY, 51.1%). It is also noteworthy that many of the LVs in the list are among the most prolic in the BNC, with six of the verbs ranked in the top 10 (GO, 2; GET, 3; MAKE, 4; TAKE, 7; COME, 9; GIVE, 10), and four others in the top 20 (LOOK, 11; FIND, 13; PUT, 16; WORK, 19). The fact that these 20 prolic verbs function 24.2% of the time in PV constructions underscores the importance of English PVs, in general, and establishes the

TABLE 4 Descriptive Statistics of Top 20 Lexical Verb (LV) Lemmas Functioning in Phrasal Verb (PV) Forms LV lemma GO COME TAKE GET SET CARRY TURN BRING LOOK PUT PICK MAKE POINT SIT FIND GIVE WORK BREAK HOLD MOVE Total # in BNC PVs 48,016 36,878 22,970 20,223 18,569 15,617 13,040 12,514 12,226 11,970 9,997 7,368 7,159 7,112 6,934 6,174 5,985 5,428 5,403 5,197 278,780 % of all BNC PVs 9.3 7.1 4.4 3.9 3.6 3.0 2.5 2.4 2.4 2.3 1.9 1.4 1.4 1.4 1.3 1.2 1.2 1.0 1.0 1.0 53.7 Cum % of all BNC PVs 9.3 16.4 20.8 24.7 28.3 31.3 33.8 36.2 38.6 40.9 42.8 44.2 45.6 47.0 48.3 49.5 50.6 51.7 52.7 53.7 53.7 Total # in BNC 227,103 145,047 173,996 213,726 39,149 30,572 44,051 42,567 109,110 67,839 14,274 210,880 13,767 27,388 96,010 125,312 63,104 18,642 46,773 37,820 1,747,130 BNC rank 2 9 7 3 40 53 32 33 11 16 138 4 149 64 13 10 19 109 30 41 39*

% as PVs 21.1 25.4 13.2 9.5 47.4 51.1 29.6 29.4 11.2 17.6 70.0 3.5 52.0 26.0 7.2 4.9 9.5 29.1 11.6 13.7 24.2*

Note. # = token frequency. Cum % = cumulative frequency percentage. Values based on nonseparable and separable counts (i.e., verb [V] + adverbial phrase [AVP], V + X + AVP, V + X + X + AVP). Total phrasal verb (PV) tokens in the British National Corpus (BNC) = 518,923; total lexical verb (LV) tokens in BNC = 10,404,107. * = Average of column.

value of this particular list for instructional purposes. To further illustrate this point, we must rst reiterate that LV lemmas functioning in PV constructions repeat only 330 times on average in a corpus of 100 million words (see Table 3). In stark contrast, these 20 high-frequency LV lemmas repeat 13,994 times on average when functioning in PV constructions (279,882 20). Table 5 provides the raw counts of verb-plus-particle constructions for the top 20 LV lemmas. Considering cumulative frequency percentages, it is clear that these 20 LV lemmas combine with only eight particles (out, up, on, back, down, in, over, and off)a total of 160 combinationsto account for more than half (50.4%) of the PVs in the BNC. It is equally clear, however, that the actual verb + particle combinations are highly idiosyncratic. For example, the particle on combines with the verb lemma GO a total of 14,743 times but never once with the lemma POINT, and the particle over combines often with the verb TAKE (5,158) but never with the verbs SET, POINT, or FIND. Also, several combinations with higher frequencies contain particles that are infrequently used in PVs in general (e.g., BRING about, 2,083; GO round, 1,366; COME along, 1270).

TABLE 5 Verb-Particle Frequencies of Top 20 Lexical Verbs Functioning in Phrasal Verb (PV) Forms (Continued on p. 351) Verb GO COME TAKE GET SET CARRY TURN BRING LOOK PUT PICK MAKE POINT SIT FIND GIVE WORK BREAK HOLD MOVE Total % of PV Cum % Out 7,688 5,022 3,426 3,545 4,633 10,798 4,284 1,425 1,641 1,660 856 1,105 6,984 191 6,619 532 4,703 996 1,507 573 68,188 13.1 13.1 Up 3,678 5,523 4,608 3,936 10,360 36 2,710 2,507 3,871 2,835 9,037 5,469 104 1,158 33 4,186 334 1,286 1,624 477 63,772 12.3 25.4 On 14,903 4,830 4,199 2,696 11 3,869 292 390 244 1,428 35 25 0 118 9 34 411 3 908 1,419 35,824 6.9 32.3 Back 8,065 8,029 1,628 4,552 265 172 1,373 2,200 2,251 1,369 3 270 7 834 128 507 36 4 823 566 33,082 6.4 38.7 Down 4,781 3,305 775 1,538 504 84 1,051 1,022 2,221 2,873 3 65 56 4,478 34 11 98 2,199 369 306 25,773 5.0 43.7 In 1,974 4,814 509 1,127 281 32 149 2,505 250 810 1 16 0 145 57 579 182 220 34 790 14,475 2.8 46.5 Off 2,104 518 2,163 1,086 1,869 170 594 31 2 742 44 277 6 1 4 121 33 549 91 242 10,647 2.1 48.5 Over 991 1,004 5,420 293 1 131 975 129 207 76 18 75 2 3 5 198 31 2 40 201 9,802 1.9 50.4

These phenomena suggest that certain semantic constraints exist in the possible combinations of LVs and AVPs in English. Pedagogically, therefore, English language learners would have to learn more than the 320 verbparticle combinations depicted in Table 5 (i.e., 20 verbs 16 particles); they would also have to understand which combinations are less likely to occur or do not exist at all. However, given the prolic nature of this set of verbs and particles in general, such an effort is certainly justiable, especially if one considers the more random nature of typical PV selection for instructional purposes.

Most Frequent Phrasal Verbs in BNC

Appendix A contains the 100 most frequent PVs in the BNC along with their statistical information. This list is simply a reorganization of the highest verb + particle combinations from Table 5. However, this repurposing of the data offers additional insights regarding the impact of these high-frequency phrasal forms. To facilitate this discussion, we refer the reader to Table 6, which is a consolidation of the statistical information found in Appendix A. First, it is noteworthy that only 25 PV lemmas make up nearly one

TABLE 5 Verb-Particle Frequencies of Top 20 Lexical Verbs Functioning in Phrasal Verb (PV) Forms (Continued from p. 350) Round 1,366 1,107 78 365 0 10 1,146 105 694 21 0 16 0 34 3 3 23 0 2 19 4,992 1.0 51.4 About 244 741 2 102 645 29 38 2,083 45 35 0 0 0 18 3 1 0 0 0 178 4,164 0.8 52.2 Through 972 567 31 533 0 127 1 11 21 90 0 40 0 4 10 0 100 169 0 2 2,678 0.5 52.7 Around 394 139 37 241 0 107 423 18 779 16 0 2 0 126 29 2 19 0 1 340 2,673 0.5 53.2 Along 717 1,270 94 163 0 52 0 88 0 1 0 8 0 1 0 0 5 0 0 84 2,483 0.5 53.7 Under 95 2 0 3 0 0 4 0 0 9 0 0 0 1 0 0 10 0 4 0 128 0.0 53.7 By 44 7 0 42 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 93 0.0 53.7 Across 0 0 0 1 0 0 0 0 0 5 0 0 0 0 0 0 0 0 0 0 6 0.0 53.7 Total 48,016 36,878 22,970 20,223 18,569 15,617 13,040 12,514 12,226 11,970 9,997 7,368 7,159 7,112 6,934 6,174 5,985 5,428 5,403 5,197 278,780 53.7

Note. Cum % = cumulative frequency percentage.

third (30.4%) of all PV occurrences in the BNC. Fifty PV lemmas constitute 42.7% of the total (see Cum % Tot PV), and only 100 are needed to cover more than one half (51.4%) of all PV occurrences in the BNC. To put this observation into a more practical perspective, language learners familiar with this list of 100 distinct PV lemmas would, on average, be able to negotiate more language containing PV constructions

TABLE 6 Frequency and Coverage Summary of Top 100 Phrasal Verb (PV) Lemmas in BNC PV rank 125 2650 5175 76100 Subtotal 10112,508 Total # 157,921 63,437 29,485 15,781 266,624 252,299 518,923 % of total PV 30.4 12.3 5.6 3.1 51.4 48.6 100.0 Cum % total PV 30.4 42.7 48.3 51.4 51.4 100.0 100.0 % of total LV 1.52 0.61 0.28 0.15 2.56 2.44 5.00 Cum % total PV 1.52 2.13 2.41 2.56 2.56 5.00 5.00

Note. PV = phrasal verb. # = token frequency. Cum % = cumulative frequency percentage.



than if they knew the remaining 12,408 distinct PV lemmas in the BNC, which cover only 48.6% of all PV occurrences. Table 6 also shows that these same 100 PV lemmas account for 2.56% of all lexical verbs in the BNC, or roughly 1 in every 40. In practical terms, this result means that, on average, 1 of these top 100 phrasal verbs will occur in every 400 words of English (1/400)that is, 10 words of every 100 words in the BNC are lexical verbs (1/10) and 1 lexical verb of every 40 will be from this high-frequency list of PVs (1/40). Put in this light, the effort of teaching these top 100 PVs certainly appears to be justied, especially when one considers the haphazard manner in which most PV instruction takes place in second language instructional settings.

Word Senses of Frequent Phrasal Verbs

Table 7 displays the word-sense frequencies from WordNet (Miller, 2003) for the top 100 PVs in the BNC. These values underscore the need
TABLE 7 Number of WordNet Senses for Top 100 Phrasal Verbs (PVs) in BNC PV Go on Carry out Set up Pick up Go back Come back Go out Point out Find out Come up Make up Take over Come out Come on Come in Go down Work out Set out Take up Get back Sit down Turn out Take on Give up Get up Look up Senses 5 2 15 16 4 5 6 3 4 12 8 8 11 5 5 8 8 3 13 4 3 12 5 12 8 1 PV carry on go up get out take out come down put down put up turn up get on bring up bring in look back look down* bring back break down take off go off bring about go in set off put out look out take back hold up get down hold out Senses 4 7 7 14 5 7 8 5 7 8 5 2 5 2 8 9 6 5 1 7 10 2 6 7 7 5 PV put on bring out move on turn back put back go round** break up come along sit up turn round** get in come round** make out get off turn down bring down come over break out go over turn over go through hold on pick out sit back hold back put in Senses 9 9 1 4 2 5 19 2 2 3 5 1 10 11 5 6 1 5 4 9 5 5 2 2 5 7 PV move in look around take down put off come about go along look round*** set about turn off give in move out come through move back break off get through give out come off take in give back set down move up turn around Senses 3 1 4 5 1 3 0 3 3 2 2 4 1 5 5 4 3 17 1 6 2 0

Note. Total senses = 559. PV = phrasal verb. *Consulted Longman Dictionary of Phrasal Verbs (Courtney, 1993). **WordNet = around. ***See look around. See turn round.



in corpus studies to address the semantic as well as the formal characteristics of PVs as we attempt to inform language teaching. In short, the same 100 forms that make up roughly half of all PVs in the BNC have expanded to 559 potential meanings, or 5.6 meanings per PV on average. From our perspective, however, this is still a manageable number for language teachers and materials writers to deal with, especially if we consider the alternative of resorting to the more random lists of PVs with their questionable utility in addition to their own multiplicity of meaningsthat are typically used in English language education. The noted multiplicity of PV senses also conrms the need for language learners to be exposed to these structures in multiple and varied contextsa task perhaps aided by the use of concordancing software (Cobb, 1997, 1999; Gavioli & Aston, 2001; Sun & Wang, 2003). The ndings also suggest the value of constructing more semantically tagged corpora (Landes, Leacock, & Tengi, 1998) that would allow semantic frequencies to be established for instructional purposes. For instance, the list-high 19 senses of the PV break up (see Table 7) could be arranged from highest to lowest semantic frequency, thus prioritizing them for language learning. We acknowledge, however, that corpora of this nature are much easier talked about than constructed.


The primary purpose of this corpus-based study of PVs is to contribute to what Read (2004) describes, in talking about multiword research, as fresh insights for vocabulary learning and for language teaching that may yet transform our understanding of vocabulary and the way it is taught (p. 156). We hope that our high-frequency lists of PVs will partially answer the where-do-we-start question so often asked by English language learners, teachers, curriculum designers, and materials developers. We also hope that the ndings will make a useful contribution to the growing research base involving multiword language.

Suggestions for Pedagogical Applications

Whereas our concern in this article is not with the actual teaching or acquisition of PVs, we offer the following suggestions regarding how the data of our study might best be used for pedagogical purposes. 1. The concept of word lemma is critical to taking full advantage of our lists. Therefore, learners would need to be aware of inectional rePOINTING OUT FREQUENT PHRASAL VERBS: A CORPUS-BASED ANALYSIS 353





lationships in word families (e.g., TURN down includes turn down, turns down, turning down, and turned down). Learners should commit to memory those AVPs that occur much more often in phrasal verb constructions than in prepositional phrases (i.e., out, up, down, and back) and be taught to look for corresponding verbs, keeping in mind that some combinations can be separated by one or more intervening words (e.g., turn down the offer versus turn the offer down). Knowing these prolic particles will also allow learners to identify many PVs that occur less frequently in the language in general. Learners should commit to memory our list of the top 20 lexical verb lemmas functioning in PVs that cover 53.7% of all phrasal verb tokens in the BNC (see Table 4). Learners should also know the 16 s indicated in the study and have ample practice in exibly combining these particles with the top 20 lexical verb lemmas (see Table 5). A special emphasis should be given to the eight most prolic particles (out, up, on, back, down, in, off, over) that combine with the 20 lexicalverb lemmas to account for approximately one-half of all phrasal verb tokens in the BNC. Learners should have ample exposure (contextualized and decontextualized) to the top 100 phrasal verb lemmas (see Appendix A), with priority given to the top 25 lemmas (covering nearly one third of all phrasal verb constructions). Learners should be made aware of, and have ample exposure to, the multiple meaning senses that are characteristic of high-frequency PVs (see Table 7). Electronic resources such as WordNet (Miller, 2003) and VIEW (Davies, 2005) could be used to ascertain these senses, and provide example contexts for exposure and practice.

Suggestions for Future Research

We also offer the following suggestions for future research based on our experiences in this study. 1. The list of high-frequency PVs in this study must be tested against other megacorpora, as well as more specialized corpora, to establish their validity. 2. A reanalysis of the lists across major registers (e.g., spoken versus written English) and within subregisters of those major groupings (e.g., ction versus news article reports versus academic prose) could provide additional insights relative to English for specic purposes,

English for academic purposes, content-based instruction, literaturebased curricula, and other such pedagogical orientations. 3. Corpus-based research must continue to nd better ways of locating, tagging, and counting multiword items. Such measures will surely move us closer to the psychological reality of linguistic forms and the preservation of their meanings. We feel that too many frequencybased vocabulary studies have ignored this concept. 4. Corpus-based research should continue to explore ways of identifying, tagging, and preserving the meaning senses of multiword items, as well as single-word items (cf. Landes, Leacock, & Tengi, 1998). Only then will we be able to more accurately describe natural language in terms of the intricate relationships between linguistic forms and their context-sensitive meanings. 5. Future research regarding frequency of PVs and other multiword items might also benet from probabilistic analyses such those made possible through Bayesian and frequentist statistical applications (Bod, Hay, & Jannedy, 2003; Manning & Schtze, 1999). Finally, we recognize that the narrow scope of this study leaves many questions unanswered regarding PVs: What about the large body of less frequent PVs? What about literal versus gurative meanings? What about separable versus nonseparable PVs? What about register variation among PVs? In addition to these and other linguistic questions are the lingering concerns of how to use our growing corpus-based understanding of PVs and other multiword items to affect successful pedagogical outcomes. Recently, several works by noted experts in the eld have begun to bridge the gap between corpus-based ndings and fruitful instructional practices (e.g., McCarthy & ODell, 2004; Schmitt, 2004; Sinclair, 2004). We are certain that many more efforts like these will be needed as we obtain new insights regarding multiword items.

Dee Gardner is an associate professor of applied linguistics and TESOL at Brigham Young University in Provo, Utah, United States. He specializes in vocabulary acquisition, literacy development, and applied corpus linguistics. His current research interests include corpus semantics, including semantic frequency analysis, and strengthening the ties between corpus-based research and English language teaching. Mark Davies is a professor of corpus linguistics at Brigham Young University, Provo, Utah, United States. He has published widely on corpus design, contruction, and use, as well as variation in syntax. He is the creator of the interface for the 100-millionword British National Corpus and is currently nishing a 360+-million-word, webaccessible corpus of American English, 1990present.

Biber, D., Johansson, S., Leech, G., Conrad, S., & Finegan, E. (1999). Longman grammar of spoken and written English. London: Longman. Bod, R., Hay, J., & Jannedy, S. (Eds.). (2003). Probabilistic linguistics. Cambridge, MA: The MIT Press. Bolinger, D. (1971). The phrasal verb in English. Cambridge, MA: Harvard University Press. British National Corpus Consortium. (2000). British national corpus: World edition. Oxford: Humanities Computing Unit of Oxford University. Celce-Murcia, M., & Larsen-Freeman, D. (1999). The grammar book: An ESL/EFL teachers course (2nd ed.). Boston: Heinle & Heinle. Coady, J. (1997). L2 vocabulary acquisition: A synthesis of the research. In J. Coady & T. Huckin (Eds.), Second language vocabulary acquisition (pp. 272290). Cambridge: Cambridge University Press. Cobb, T. (1997). Is there any measurable learning from hands-on concordancing? System, 25, 301315. Cobb, T. (1999). Breadth and depth of lexical acquisition with hands-on concordancing. Computer Assisted Language Learning, 12, 345360. Condon, N., & Kelly, P. (2002). Does cognitive linguistics have anything to offer English language learners in their efforts to master phrasal verbs? ITL Review of Applied Linguistics, 133-134, 205231. Cornell, A. (1985). Realistic goals in teaching and learning phrasal verbs. International Review of Applied Linguistics in Language Teaching, 23, 269280. Courtney, R. (1983). Longman dictionary of phrasal verbs. Harlow, England: Longman. Darwin, C. M., & Gray, L. S. (1999). Going after the phrasal verb: An alternative approach to classication. TESOL Quarterly, 33, 6583. Darwin, C. M., & Gray, L. S. (2000). Comments on Clayton M. Darwin and Loretta S. Grays Going after the phrasal verb: An alternative approach to classication: The authors respond. TESOL Quarterly, 34, 165173. Davies, M. (2005). VIEW: Variations in English words and phrases. (An online interface with the British National Corpus: World Edition). Provo, UT: Brigham Young University. Available from Devoto, B. (Ed.). (1922). Mark Twain in eruption. New York: Harper & Bros. Fellbaum, C. (Ed.). (1998). WordNet: An electronic lexical database. Cambridge, MA: The MIT Press. Folse, K. S. (2004). Vocabulary myths: Applying second language research to classroom teaching. Ann Arbor, MI: University of Michigan Press. Gass, S. M., & Selinker, L. (2001). Second language acquisition: An introductory course (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum. Gavioli, L., & Aston, G. (2001). Enriching reality: Language corpora in language pedagogy. ELT Journal, 55, 238246. Hulstijn, J., & Marchena, E. (1989). Avoidance: Grammatical or semantic causes? Studies in Second Language Acquisition, 11, 241255. Hunston, S. (2002). Corpora in applied linguistics. Cambridge: Cambridge University Press. Landes, S., Leacock, C., & Tengi, R. I. (1998). Building semantic concordances. In C. Fellbaum (Ed.), WordNet: An electronic lexical database (pp. 199216). Cambridge, MA: The MIT Press. Laufer, B. (1997). The lexical plight in second language reading: Words you dont know, words you think you know, and words you cant guess. In J. Coady & T.

Huckin (Eds.), Second language vocabulary acquisition: A rationale for pedagogy (pp. 2034). Cambridge: Cambridge University Press. Leech, G., & Smith, N. (2000). Manual to accompany the British National Corpus (Version 2) with improved word-class tagging. Lancaster, England: Lancaster University. Liao, Y., & Fukuya, Y. J. (2004). Avoidance of phrasal verbs: The case of Chinese learners of English. Language Learning, 54, 193226. Manning, C. D., & Schtze, H. (1999). Foundations of statistical natural language processing. Cambridge, MA: The MIT Press. McCarthy, M., & ODell, F. (2004). English phrasal verbs in use. Cambridge: Cambridge University Press. Miller, G. (2003). WordNet (Version 2.0) [software]. Available from http:// Moon, R. (1997). Vocabulary connections: Multi-word items in English. In N. Schmitt & M. McCarthy (Eds.), Vocabulary: Description, acquisition and pedagogy (pp. 4063). Cambridge: Cambridge University Press. Nesselhauf, N. (2003). The use of collocations by advanced learners of English and some implications for teaching. Applied Linguistics, 24, 223242. Read, J. (2000). Assessing vocabulary. Cambridge: Cambridge University Press. Read, J. (2004). Research in teaching vocabulary. Annual Review of Applied Linguistics, 24, 146161. Sawyer, J. H. (2000). Comments on Clayton M. Darwin and Loretta S. Grays Going after the phrasal verb: An alternative approach to classication: A reader reacts. TESOL Quarterly, 34, 151159. Schmitt, N. (Ed.). (2004). Formulaic sequences: Acquisition, processing and use. Amsterdam: John Benjamins. Sheen, R. (2000). Comments on Clayton M. Darwin and Loretta S. Grays Going after the phrasal verb: An alternative approach to classication: Another reader reacts. TESOL Quarterly, 34, 160164. Side, R. (1990). Phrasal verbs: Sorting them out. ELT Journal, 44, 144152. Sinclair, J. M. (Ed.). (2004). How to use corpora in language teaching. Amsterdam: John Benjamins. Sinclair, J. M., & Moon, R. (Eds.). (1989). Collins cobuild dictionary of phrasal verbs. London: Collins. Spears, R. A. (1993). NTCs dictionary of phrasal verbs and other idiomatic verbal phrases. Lincolnwood, IL: National Textbook. Stubbs, M. (2001). Texts, corpora, and problems of interpretation: A response to Widdowson. Applied Linguistics, 22, 149172. Stubbs, M. (2002). Words and phrases: Corpus studies of lexical semantics. Oxford: Blackwell. Sun, Y., & Wang, L. (2003). Concordances in the EFL classroom: Cognitive approaches and collocation difficulty. Computer-Assisted Language Learning, 16, 8394. Walter, E., & Pye, G. (Eds.). (1997). Cambridge international dictionary of phrasal verbs. Cambridge: Cambridge University Press. Widdowson, H. G. (2000). On the limitations of linguistics applied. Applied Linguistics, 21, 325. Wray, A. (2000). Formulaic sequences in second language teaching: Principle and practice. Applied Linguistics, 21, 463489. Wray, A. (2002). Formulaic language and the lexicon. Cambridge: Cambridge University Press. Wyss, R. (2003). Putting phrasal verbs into perspective. TESOL Journal, 12, 3738.

Frequency and Coverage of Top 100 Phrasal Verb Lemmas in BNC
Rank 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 Verb GO CARRY SET PICK GO COME GO POINT FIND COME MAKE TAKE COME COME COME GO WORK SET TAKE GET SIT TURN TAKE GIVE GET LOOK CARRY GO GET TAKE COME PUT PUT TURN GET BRING BRING LOOK LOOK BRING BREAK TAKE GO BRING GO SET PUT LOOK TAKE HOLD GET AVP on out up up back back out out out up up over out on in down out out up back down out on up up up on up out out down down up up on up in back down back down off off about in off out out back up down # 14,903 10,798 10,360 9,037 8,065 8,029 7,688 6,984 6,619 5,523 5,469 5,420 5,022 4,830 4,814 4,781 4,703 4,633 4,608 4,552 4,478 4,284 4,199 4,186 3,936 3,871 3,869 3,678 3,545 3,426 3,305 2,873 2,835 2,710 2,696 2,507 2,505 2,251 2,221 2,200 2,199 2,163 2,104 2,083 1,974 1,869 1,660 1,641 1,628 1,624 1,538 % of PV 2.9 2.1 2.0 1.7 1.6 1.5 1.5 1.3 1.3 1.1 1.1 1.0 1.0 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.8 0.8 0.8 0.8 0.7 0.7 0.7 0.7 0.7 0.6 0.6 0.5 0.5 0.5 0.5 0.5 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.3 0.3 0.3 0.3 0.3 Cum % of PV 2.9 5.0 6.9 8.7 10.2 11.8 13.3 14.6 15.9 17.0 18.0 19.1 20.0 21.0 21.9 22.8 23.7 24.6 25.5 26.4 27.2 28.1 28.9 29.7 30.4 31.2 31.9 32.6 33.3 34.0 34.6 35.2 35.7 36.2 36.8 37.2 37.7 38.2 38.6 39.0 39.4 39.8 40.3 40.7 41.0 41.4 41.7 42.0 42.3 42.7 43.0 % of LV 0.14 0.10 0.10 0.09 0.08 0.08 0.07 0.07 0.06 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.03 0.03 0.03 0.03 0.03 0.03 0.03 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.01 Cum % of LV 0.14 0.25 0.35 0.43 0.51 0.59 0.66 0.73 0.79 0.85 0.90 0.95 1.00 1.05 1.09 1.14 1.18 1.23 1.27 1.32 1.36 1.40 1.44 1.48 1.52 1.56 1.59 1.63 1.66 1.69 1.73 1.75 1.78 1.81 1.83 1.86 1.88 1.90 1.92 1.95 1.97 1.99 2.01 2.03 2.05 2.06 2.08 2.10 2.11 2.13 2.14



Frequency and Coverage of Top 100 Phrasal Verb Lemmas in BNC (Continued)
Rank 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 Verb HOLD PUT BRING MOVE TURN PUT GO BREAK COME SIT TURN GET COME MAKE GET TURN BRING COME BREAK GO TURN GO HOLD PICK SIT HOLD PUT MOVE LOOK TAKE PUT COME GO LOOK SET TURN GIVE MOVE COME MOVE BREAK GET GIVE COME TAKE GIVE SET MOVE TURN AVP out on out on back back round up along up round in round out off down down over out over over through on out back back in in around down off about along round about off in out through back off through out off in back down up around # 1,507 1,428 1,425 1,419 1,373 1,369 1,366 1,286 1,270 1,158 1,146 1,127 1,107 1,105 1,086 1,051 1,022 1,004 996 991 975 972 908 856 834 823 810 790 779 775 742 741 717 694 645 594 579 573 567 566 549 533 532 518 509 507 504 477 423 % of PV 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 Cum % of PV 43.2 43.5 43.8 44.1 44.3 44.6 44.9 45.1 45.4 45.6 45.8 46.0 46.2 46.4 46.6 46.9 47.0 47.2 47.4 47.6 47.8 48.0 48.2 48.3 48.5 48.7 48.8 49.0 49.1 49.3 49.4 49.6 49.7 49.8 49.9 50.1 50.2 50.3 50.4 50.5 50.6 50.7 50.8 50.9 51.0 51.1 51.2 51.3 51.4 % of LV 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.00 0.00 0.00 0.00 0.00 0.00 Cum % of LV 2.16 2.17 2.18 2.20 2.21 2.22 2.24 2.25 2.26 2.27 2.28 2.29 2.31 2.32 2.33 2.34 2.35 2.36 2.37 2.38 2.38 2.39 2.40 2.41 2.42 2.43 2.43 2.44 2.45 2.46 2.46 2.47 2.48 2.49 2.49 2.50 2.50 2.51 2.51 2.52 2.52 2.53 2.53 2.54 2.54 2.55 2.55 2.56 2.56

Note. # = token frequency. Cum % = cumulative frequency percentage. AVP = adverbial particle. PV = phrasal verb. LV = lexical verb.



Spoken Grammar and ELT Course Materials: A Missing Link?

Canterbury Christ Church University Canterbury, England


National Cheng Kung University Tainan, Taiwan

Drawing on the evidence of a growing body of corpus research over the past two decades, this article investigates the phenomenon of spoken grammar in conversational English and the extent to which our current knowledge of the area is reected in contemporary textbooks for English as a foreign language (EFL) learners. The article reports on a survey conducted by the authors of 24 general EFL textbooks published in the United Kingdom since the year 2000 and concludes, on the basis of the survey, that coverage of features of spoken grammar is at best patchy. Where it is dealt with at all, there tends to be an emphasis on lexicogrammatical features, and common syntactic structures peculiar to conversation are either ignored or conned to advanced levels as interesting extras. We argue that this is inadequate for many learners, particularly those for whom the development of oral uency in informal interactions with native speakers is an important goal.

ver the past decade applied linguists have had a considerable and growing interest in the grammar of spoken English. This interest has been stimulated by the development of electronic corpora of natural spoken English discourse (for a useful survey, see Leech, 2000, pp. 681 682) and the striking ndings that these corpora have revealed. These ndings are reported by, for example, Brazil (1995), Carter and McCarthy (1995, 1997, 2006), and Biber, Johansson, Leech, Conrad, and Finegan (1999), from whose Longman Grammar of Spoken and Written English (LGSWE) we take many of the examples used in this article. In view of our increased understanding of the ways in which grammar is used in speech, as distinct from writing, it is hardly surprising that some prominent researchers in the eld (e.g., McCarthy & Carter, 1995, 2002) have argued that much greater attention should be given to spoken grammar in materials for EFL teaching and learning.

TESOL QUARTERLY Vol. 41, No. 2, June 2007

Behind this demand lies the assumption that not enough is being done at present. In other words, there is a missing link between corpus research ndings and current pedagogical practice. In this article we seek to test this assumption by surveying a selection of EFL textbooks published in the United Kingdom with respect to their coverage of spoken grammar, both quantitatively (how much attention is given) and qualitatively (what kind of attention is given). If there is indeed a missing link, what is the precise nature of the omission, and what, if anything, should be done to forge the link? In order to investigate these questions, we begin by dening what we mean by spoken grammar, exploring its role in communication and examining commonly occurring features that typify the way it is used. These features then form the basis of our textbook survey. In the nal part of the article, we present a case for giving spoken grammar a higher prole in contemporary English language textbooks and thus for connecting the ndings of corpus-based research into language use more closely to language learning and teaching.


The subject of this article is the grammar of informal, conversational English, rather than of spoken discourse characteristic of more formal settings, such as debates or speeches. We are therefore using the term spoken grammar synonymously with conversational grammar, in keeping with the way the term is normally used in the literature (e.g., Biber et al., 1999, pp. 10371125; Brazil, 1995; Carter & McCarthy, 1995, 2006; Leech, 2000). Because our focus is on pedagogical applications of spoken grammar for second language teaching, we are not concerned with vernacular or nonstandard forms of grammar (Biber et al., p. 1121), that is, forms which are restricted to regional dialects or widely felt to be signs of ill-educated usage.1 We have also excluded forms associated with particular age groups.2 All the grammatical features exemplied in this article are taken from descriptions of standard, nondialectal conversational English, although the issue of whether a particular grammatical

Examples of nonstandard features of spoken grammar taken from Biber et al. (1999) are the use of aint and double negation, as in She aint never given me no problems (p. 1121). 2 An example of a feature associated with the speech of younger speakers would be what Yule, Mathis, and Hopkins (1992) have labeled quotative structures, which make use of the verbal elements be like or go to introduce direct speech clauses. For example, He goes, someday I might have a kid and <laugh>. Im like, No! (American English) (Biber et al., 1999, p.1120).



form is standard or nonstandard may sometimes be open to question. Nor are we considering planning errors in speech, such as false starts or mixed structures (i.e., where speakers begin an utterance planning one grammatical structure and change in midsentence to another closely associated one, e.g., Theres not necessary to introduce to a new law). Whereas these phenomena clearly affect syntax and grammar, they do not result in regular, systematic use of speciable grammatical features. Spoken grammar is the manifestation of systematic grammatical phenomena in spoken discourse that arise from the circumstances in which speech (i.e., conversation) is characteristically produced. Following Biber et al. (1999) and Leech (2000), we take the view that speech and writing draw on the same underlying grammatical system (rather than on two separate systems) but that the system is adapted in various dynamic and often ingenious ways to meet the particular circumstances in which each medium is used. Speech, rst of all, is usually spontaneous and unplanned (unlike most written discourse) and thus has to be adapted to the needs of real-time processing (Leech, 2000). This process results in the step-by-step assembly of a spoken utterance (Brazil, 1995, p. 17), as speakers compose their utterances impromptu as they speak, with no opportunity for editing (Miller & Weinert, 1998). Second, speech normally occurs face to face, ensuring not only that there is a shared context between the participants (Leech, 2000, p. 694), but also that speakers tend to be more aware of interpersonal factors, factors which will affect the way they use language, including grammatical choices. Indeed, in their analysis of casual conversation, Eggins and Slade (1997) suggest that speakers engage in such conversation primarily to clarify and extend the interpersonal ties that have brought them together (p. 67). Third, speech is highly interactive, requiring cooperation and contextual sensitivity from all participants, who take turns to speak and listen, to negotiate meanings, and to respond immediately to one anothers contributions (Brown, 1989). Fourth, conversation frequently takes place in informal settings in which the participants have a close symmetry of relationship, a factor which will also impinge on the choice of language used. The combination of these characteristics of speech is reected in certain general linguistic properties that permeate conversational discourse, as well as a range of more specic grammatical constructions, which speakers do not use to the same extent in writing. Following are examples of some of the general linguistic properties noted in the literature. The staging of information across syntactic boundaries, with a small quantity of information being assigned to each phrase (Miller & Weinert, 1998, p. 22), is manifested, for example, in the use of syntactic

devices such as noun phrase prefaces, discussed in the following section, and in the lower lexical density of spoken language compared with written language, noted by Halliday (1985). A tendency to avoid syntactic elaboration and complexity (Leech, 2000; Miller & Weinert, 1998) results in a very low mean phrase length (Leech, p. 695), particularly of noun phrases, and in the use of chains of clauses linked by coordinating conjunctions like and or simple subordinators like cos or so (Carter, 2004), in preference to more elaborate sequences of main and subordinate clauses. A more exible approach to the positioning of constituents in spoken utterances than in written texts affects, for example, the placement of adverbials and question tags, as these examples from Carter (2004, p. 31) illustrate (italics added):
I was worried I was going to lose it and I did almost. Spanish is more widely used isnt it outside of Europe.

The frequent use of vague language to hedge or modify what the speakers are saying (e.g., sort of and kind of used before noun phrases, verb phrases, adjectives, and adverbs) has a softening effect, deliberately making the speakers statements sound less precise and assertive, thus leaving space for negotiation and interaction (Carter, 2004; McCarthy, 1998). These general properties of spoken language are likely to occur to a greater or lesser extent across languages and will form part of the general communicative and pragmatic competence that learners bring with them from their rst language to the task of learning a new one. However, the way these properties are manifested in the deployment of specic grammatical features will inevitably vary from one language to another. These featuresthe way that the grammar of speech is encoded in speciable linguistic constructions in the target languagewould thus become a focus of instruction in a teaching programme that included a component on spoken grammar. These features have also formed the focus of much of the recent corpus-based research into the grammar of spoken English.


We have grouped the individual constructions commonly found in spoken grammar into three categories, which we have found useful in describing and assessing the way these features are coveredor not

coveredin EFL textbooks. The particular grammatical features we are focusing on in this categorisation are those especially characteristic of conversation. Category A features are productive grammatical constructions, that is, constructions which involve a degree of grammatical encoding in their production or grammatical decoding in their interpretation. An example of such a construction, found more commonly in speech than in writing, would be the question tag isnt it, placed in midutterance, as cited earlier from Carter (2004). It would be classied as a Category A feature because it would need to change its form to arent they if the subject were a plural nounthereby requiring the speaker to draw on the grammatical resources of language (in this case subjectverb agreement) to use it accurately. Category B features, on the other hand, are xed lexicogrammatical units which do not undergo morphological change and are inserted into the utterance at an appropriate place, typically to modify a constituent in the utterance. Their forms are not affected by the surrounding grammatical context, but their position within an utterance will be subject to certain syntactical restrictions. Again, we are referring here to lexicogrammatical units more commonly used in spoken than in written English. An example would be the hedging devices sort of and kind of, also noted earlier in Carter (2004). Category C consists of a small set of grammatical features associated with prescriptive and proscriptive attitudes to grammatical acceptability, in that they appear to violate a surface-level rule of grammar. Examples are the use of less rather than fewer with countable nouns (e.g., less people), or the use of the indicative was rather than the subjunctive were in second conditional structures (e.g., If I was a rich man . . .). They are common in informal contexts of use, typically found in conversation (although not conned to it) and, unlike Category A and B features, would be regarded by linguistic purists as ungrammatical. For this reason, they are less common in formal written communication.3 What distinguishes Category C from Category A is that each Category C feature has a clearly identiable, more formal alternative and selection of one or the other is purely a matter of stylistic preference and conveys no difference in emphasis or meaning. In the following sections we exemplify and discuss these features in more detail.

Some famously proscribed items like split innitives or dangling participles may in fact be as frequent, if not more frequent, in written rather than spoken language and hence would not form part of Category C, which is reserved for proscribed items more clearly associated with informal, conversational contexts.



Category A
We are concerned in Category A with syntactical constructions of the type dened earlier that corpus data show are used extensively in speech but found much less frequently in corpora based on written English outside genres such as email messages or computer chat room communication, which in many respects imitate informal spoken language. If listed in standard reference grammars of English (e.g., Quirk, Greenbaum, Leech, & Svartvik, 1985), these constructions are marked as features of conversational or colloquial style. We focus on four examples of such features that have attracted the attention of many scholars in the eld because they frequently occur and which we have consequently targeted in our survey of textbooks. In our discussion we show their role and function in conversational discourse, relating this to the circumstances discussed earlier in which such discourse takes place. The specic examples of each feature are taken from Biber et al. (1999) unless otherwise stated.

Noun Phrase Prefaces

The rst example we discuss is the use of noun phrase prefaces, coreferential to a pronoun in the following clause, as in this example (the noun phrase preface is in bold type):
This little shopits lovely (p. 1074)

The grammatical encoding involved here is in the agreement between the preface itself and the following pronoun. The terms used to describe this feature vary in the literature. Thus noun phrase prefaces (the term used by Biber et al., 1999) are also known as left dislocations (Stubbs, 1983; Quirk et al., 1985; Leech, 2000) or heads (Carter & McCarthy, 1997). For ease of reference, we follow Carter and McCarthy and use the term head to describe this feature. The head is a fronting device used to orient the listener to the topic the speaker is introducing. It makes explicit, in the form of a noun phrase, the referent of a pronoun used in the main proposition. The noun phrase may be the subject of the proposition, as in the example, but does not have to be, as these further examples from Biber et al. (1999, p. 1074) illustrate:
Those Marks and Sparks bags, can you see them all? (British English) you know, the vase, did you see it? (American English)

Noun Phrase Tags

In using noun phrase tags, the speaker makes explicit the identity of a pronoun used earlier in the utterance, as in this example (noun phrase tag in bold type) from Biber et al. (1999):
I reckon theyre lovely. I really do, whippets. (p. 1080)

Noun phrase tags, the term used by Biber et al., are also known as right dislocations (Quirk et al., 1985) or tails (Aijmer, 1989; Carter & McCarthy, 1995). To maintain symmetry with the term heads, we shall henceforth use the term tails for this phenomenon. The tail acts as an immediate reminder of what has been said, or what is important (Carter, Hughes, & McCarthy, 1998) and is typically used when making a comment relating to the topic being discussed in the conversation. Although it frequently takes the form of an appended noun phrase (Biber et al., 1999, p. 1080), it may also be expanded to include the verbal element contained in the preceding clause, as the following two examples from Carter et al. (1998), taken from the Cambridge and Nottingham Corpus of Discourse in English, show:
Hes a real problem is Jeff. Shes got a nice personality Jenny has. (p. 70)

Both heads and tails are grammatical features that break up the normal subjectverbobject sequence of sentence constituents and are highly unusual in written prose (Biber et al., 1999). They serve important purposes in conversation. First, they help to reduce the burden for the listener involved in processing language in real timein the case of heads, reminding the listener what the topic was. Carter et al. (1998, p. 71) refer to them as listener-sensitive devices. Second, it seems reasonable to assume that head structures involving relatively long or complex noun phrase prefaces allow the speaker, not just the listener, more processing (i.e., planning) time. Compare, for example, the following two utterances, one with and one without a head preface:
Those Marks and Sparks bags, can you see them all? (preface + proposition; Biber et al., p. 1074) Can you see all those Marks and Sparks bags? (single proposition)

Past Progressive Tense

Speakers use the past progressive tense, in contrast to the more standard past tense, to introduce reported speech structures, typically with the

verbs say and tell. The following examples are taken from Biber et al. (1999):
but Yvonne was saying on my wages I wouldnt get a mortgage. He was telling me that theyd died of the frost or something. (p. 1120)

The past progressive tense is, of course, widely used across spoken and written registers. However, its use in speech with reporting verbs, where it can be used to introduce both direct and indirect speech clauses, is a feature listed in Biber et al. as a peculiarity of the grammar of conversation. Carter and McCarthy (1995) similarly note its striking frequency in their own minicorpus. As with heads and tails, this use of the past progressive in reported speech structures reects some of the characteristics of casual conversation. Biber et al. (1999) refer to its evidential functionto provide supporting evidence in passing of what the speaker is saying, whereas McCarthy and Carter (2002) note its interpersonal function, because its effect is to make the reported statement sound a little less denite, as if the speaker is adopting an indirect or nonassertive stance (p. 58). This use of the past progressive in reported speech structures is entirely overlooked in most standard descriptive and pedagogical grammars (Carter & McCarthy, 1995).

Although ellipsisthe omission of elements which are precisely recoverable from the linguistic or situational context (Biber et al., 1999, p. 156)is a feature of both speech and writing, its use and distribution in each medium is not identical. In particular, situational ellipsis, that is, the omission of items which are retrievable from the immediate situation (as opposed to the co-occurring text), has been shown to be a feature identied with conversation, rather than with written texts (Biber et al., 1999, pp. 156158; Carter & McCarthy, 1995), arising as it does from a combination of informality and shared context (Leech, 2000). Situational ellipsis particularly affects elements at the beginning of a turn or a clause, for example, subject pronouns and operators in questions, as the following examples from Biber et al. illustrate:
Saw Susan and her boyfriend in Alder weeks ago. (ellipsis of I; p. 158) Too old to change, arent we? (ellipsis of Were; p. 158) Why arent you working? Got a day off? (ellipsis of Have you?; p. 1105)

The terms initial ellipsis (Biber et al., 1999) and front ellipsis (Leech, 2000) are variously used for this phenomenon, which Biber et al. (1999,

p. 1105) note is more frequent in British English than in American English. At a surface level, ellipsis appears to reduce the degree of grammatical encoding required. However, the speaker still needs to be capable of retrieving the omitted elements when decoding elliptical utterances. Also, when using ellipsis productively, as the second example from Biber et al. (1999, p. 158) shows, the correct use of the question tag arent we depends on the speakers awareness of what has been omitted from the rst part of the utterance. For these reasons, situational ellipsis is classied as a Category A feature.

Category B
Category B consists of lexicogrammatical units (Nattinger & DeCarrico, 1992) which are xed in the sense that they cannot themselves be grammatically modied (e.g., through inection or change in person or number) but which can combine with other structures to form larger syntactic units. They may consist of single lexemes, for example, stance adverbials such as really or actually, both shown by the Biber et al., (1999) corpus to be signicantly more frequent in conversation than in written registers (p. 869) or short phrases, such as sort of and you know. Because they are xed phrases, they are stored and accessed by the speaker as ready-made lexical units of language requiring little encoding work. The encoding is in the task of slotting them into an utterance in a syntactically acceptable place but does not involve changing the form of the phrase itself. For this reason, they are likely to be easier to learn than Category A features and more readily accessible to the learner for immediate use. The four examples of such units, which follow, have all been shown to be used frequently in spoken communication (Biber et al.) and form the focus of our textbook survey of items in this category. The rst three are associated with vagueness and imprecision, allowing the speakers to appear less categorical about what they are saying and giving the listener room to ll the gaps and form a personal interpretation of the speakers meaning. Although not all Category B features have this function, we have selected these examples because vague expressions of this kind are interesting reections of the interactive, interpersonal dimension of spoken language and make an important contribution to the informal, convergent tenor of everyday talk (McCarthy, 1998, p. 118).

Example 1 of a Category B unit is use of the particles sort of and kind of as hedging devices to make the reference of an item deliberately vague

(Aijmer, 1984), for example, a funny sort of place, he sort of looked at me and gasped. Aijmer also notes the function of these phrases as conversational llers, indicating to the hearer that the speaker is pausing to nd the correct formulation of what the speaker wishes to say. They are thus further examples of how speakers adapt their use of language to the requirements of real-time planning.

Vagueness Tags
Example 2 of a Category B unit is use of vagueness tags (De Cock, Granger, Leech, & McEnery, 1998) or vague category identiers (Channell, 1994), such as and things like that, or something, and and stuff like that, used frequently at the end of utterances in conversation to allow the listener to identify a general set of items based on the characteristics of the items given before the tag (p. 122). Like sort of and kind of, they are associated with imprecision in speech and, for this reason, are described by Biber et al. (1999,) as retrospective vagueness hedges (p. 1080).

Modifying Expression
Example 3 of a Category B unit is use of the modifying expression a bit or a little bit with adjectival and noun phrases (e.g., a bit better, a bit suspicious, a bit of a bore). Channell (1994) notes that it appears particularly in spoken language as a hedge or modalizer of the speakers attitude (p. 111), adding that it thus has the pragmatic force of politeness. In this, it reects the interactive nature of conversation and its consequent association, noted in the literature, with expressions of feelings, attitudes, and politeness (Biber et al., 1999; Leech, 2000).

Discourse Markers
Example 4 of a Category B unit is use of the discourse markers you know and I mean, as in these examples from Biber et al. (1999):
American English: And they spend hundreds of dollars on those dogs, you know. British English: Theres this pandaand hes really bored with, I mean, hes getting no sex so he breaks out of ermLondon Zoo to go off and nd a partner. (p. 1078)

Biber et al. (1999) classify these features as inserts, that is, standalone words which do not enter into syntactic relations with other structures but which tend to attach themselves prosodically to a larger structure (p. 1082). Leech (2000) describes them as belonging to a class of rou370 TESOL QUARTERLY

tinized particles loosely integrated with clause grammar which strongly reect the interactiveness of conversation (p. 696). Biber et al. note that they are among the most frequent inserts in both British and American conversational English.

Category C
In Category C we have placed items identied as being associated with informal, conversational contexts which would be considered grammatically incorrect by prescriptive guides to correct usage. For each item there is a correct alternative. Examples from MacAndrew (1991) include the use of less instead of fewer with countable nouns:
You would have less cars on the road and less accidents. (p. 56)

and the use of more as a comparative marker with adjectives of one syllable, in preference, or as an alternative, to the -er morpheme:
Its denitely cheaper and more fresh. (p. 20)

Carter and McCarthy (2006) include both of these features as examples of informal usage. With reference to the rst item, they state that although, traditionally, fewer is the comparative form used with plural count nouns and less with singular noncount nouns, increasingly, in informal spoken English, less is used with plural count nouns (p. 103). This suggests that some of these Category C features may also be associated with language change. We are interested in nding out in our textbook survey whether learners are made aware of such phenomena or whether they are only presented with the forms traditionally felt to be correct.


Twenty-four mainstream textbooks at ve levelsbeginner/ elementary, pre-intermediate, intermediate, upper intermediate, and advancedwere chosen for the survey. The books are listed in Table 1, and are henceforth referred to as B1, B2, and so forth, according to this list. All the books were published in the United Kingdom between 2000 and 2006 and designed for students in EFL courses both in the United Kingdom and internationally. The start date of 2000 was chosen to allow time

TABLE 1 List of Books Included in the Survey Book reference B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11 B12 B13 B14 B15 B16 B17 B18 B19 B20 B21 B22 B23 B24

Book details Beginner/elementary Soars and Soars (2000a) New Headway Beginner Soars and Soars (2000b) New Headway Elementary Potten and Potten (2001) Clockwise Elementary Cunningham and Moor (2005a) New Cutting Edge Elementary Pre-intermediate Dellar and Walkley (2005) Innovations Pre-Intermediate Acklam and Crace (2005) Total English Pre-Intermediate Cunningham and Moor (2005b) New Cutting Edge Pre-Intermediate Intermediate Forsyth (2000) Clockwise Intermediate Gairns and Redman (2002) Natural English Intermediate Soars and Soars (2003a) New Headway Intermediate Cunningham and Moor (2005c) New Cutting Edge Intermediate Clare and Wilson (2006) Total English Intermediate Intermediate/upper intermediate Dellar and Hocking (2000) Innovations Intermediate/Upper Intermediate Upper intermediate Naunton (2000) Clockwise Upper Intermediate Haines and Stewart (2000) Landmark Upper Intermediate Kay and Jones (2001) Inside Out Upper Intermediate Dellar and Hocking (2004) Innovations Upper Intermediate (2nd ed.) Cunningham and Moor (2005dd) New Cutting Edge Upper Intermediate Acklam and Crace (2006) Total English Upper Intermediate Advanced Walton and Bartram (2000) Initiative Advanced Jones and Bastow (2001) Inside Out Advanced Haines (2002) Landmark Advanced Students Book Cunningham and Moor (2003) Cutting Edge Advanced Soars and Soars (2003b) New Headway Advanced

for the published ndings from research (e.g., Brazil, 1995; McCarthy & Carter, 1995) to inuence textbook content. To compare like items as closely as possible, we selected books which are published for use in general EFL courses for adult learners and avoided books published for specic contexts (e.g., for a particular country), for specic groups (e.g., young learners), or for specic purposes (e.g., exam preparation or ESP/EAP classes). The books are broadly communicative in nature, with a range of authentic listening and reading texts, communicative tasks to develop receptive and productive language skills, and exercises to develop competence in the language systems of grammar, vocabulary, and pronunciation. Students are often required to work in pairs or small groups on a wide variety of tasks related to topics of general interest, such as family and friends, likes and

dislikes, eating and drinking, leisure and lifestyle, jobs and work, and fame and fortune. Textbooks at all levels (from beginner to advanced) were chosen to see whether attention to specic features of spoken grammar was more evident at one level than another. We did not, however, expect to nd much overt attention to these features at lower levels, that is, beginner and elementary, and hence we weighted the selection of books slightly in favour of those at intermediate, upper intermediate, and advanced levels.

The aim of the survey was twofold. First, we wished to nd out the extent to which common features of spoken grammar, revealed in corpus studies of conversational English and highlighted in the literature, have found their way into a representative sample of mainstream EFL textbooks. Second, we wished to nd out what kind of attention was given to these items in cases where they were included in a particular book. Thus, the survey had both a quantitative and a qualitative dimension. The items of spoken grammar which formed the basis of the survey are those that, as a result of their frequency in conversational English, have attracted the attention of researchers in the eld. We have used these items to exemplify our description and categorisation of spoken grammar described in the rst half of this article.

Each textbook was searched page by page for reference material, exercises, or activities that overtly focused on the features of spoken grammar identied in our Categories AC. Because our interest was in the extent to which overt attention was given to spoken grammar, examples of grammatical features that only appeared in text material (e.g., textbook dialogues or transcript material for listening texts) were not included in the search unless these features were highlighted in some way for the learners attention. This highlighting would usually be a brief explanation accompanying an example of the feature, followed by a short practice activity requiring the learner to use the feature productively in speech. Whatever the level of overt attention, each feature of spoken grammar highlighted in one way or another in each book was recorded by a check mark on a grid (see Tables 2 and 3). The items of spoken grammar listed on the grid include the eight examples discussed earlier, four in Category A and four in Category B. We were interested to

374 B1 B2 B3 B4 B5 B6 Pre-intermediate B10 B8 B9 B7 Beginner/elementary Category A B11 B12 Intermediate Category B Category C

TABLE 2 Features of Spoken Grammar in a Survey of British EFL Textooks: Beginner/Elementary, Pre-Intermediate, Intermediate Levels


Heads Tails Past progressive reporting verbs Initial situational ellipsis Other

Sort of/kind of Vagueness tags A bit/a little bit I mean/ you know Other: OK, now, right, well, you see, anyway, actually

Theres + plural noun: Theres lots of things left to do. Me + NP in subject position: Me and my brother. . . If + was in 2nd conditional structures: If I was younger, the job would be easier.


TABLE 3 Features of Spoken Grammar in a Survey of British EFL Textbooks: Intermediate/Upper Intermediate, Upper Intermediate, Advanced Levels B13 B14 B15 B16 B17 B18 B19 Upper intermediate B20 Intermediate/ upper intermediate Category A B21 B22 B23 Advanced Category B B24


Heads Tails Past progressive reporting verbs Initial situational ellipsis Other: Unnished conditional for polite instruction/ request: If youd like to hold on a minute Other: Affirmative question tag after affirmative verb: So hes rich, is he?


Sort of/kind of Vagueness tags A bit/a little bit I mean/you know Other: Well, you see, anyway, a lot, really, actually, basically, just

Theres + plural noun: Theres lots of things left to do. Less + plural noun: Less people read the classics now.


see which, if any, of these items, all noted as being of high frequency in corpus studies, were selected for attention in the textbooks. We did not, however, restrict our search to these preselected items but also noted any other grammatical structures specically associated with speech which were included in the books. These structures were organised as either Category A or B according to the operational denitions discussed eariler, and noted under Other in Tables 2 and 3. For Category C, we did not have predetermined items but left the selection open to record whatever features happened to appear in any of the textbooks.

Tables 2 and 3 show which features of spoken grammar identied in Categories AC are included or not included for special attention in the 24 books surveyed. The table is organised by level of textbook, distinguishing between beginner/elementary, pre-intermediate, intermediate, upper intermediate, and advanced. Points worth noting are elaborated in the following sections.

Variation in Overt Attention

Although all 24 books surveyed include work on speaking and many on informal conversational English, usually in special sections with titles like English in Use (B8, B14), Features of Natural Conversation (B15), Natural English (B9), and even Spoken Grammar (B13), the textbooks vary considerably in the extent to which they provide overt attention to particular features of spoken grammar of the kind identied in Categories AC and listed in Tables 2 and 3. A number of books provide fairly extensive coverage, with the Landmark and Innovations series being perhaps the most systematic of the selection, whereas nearly half of the books surveyed (11 out of 24) do not include any of the features listed in Tables 2 and 3, or any other features under Other in Categories A and B. As expected, the lack of explicit treatment is most apparent in the four books at the beginner/elementary level (B1B4).

Category B Attention
Most of the space allocated to spoken grammar is for Category B features, together with a range of conversational llers, idioms, and other xed expressions presented as conversational gambits, for ex376 TESOL QUARTERLY

ample, What are you up to this weekend? (B13, p. 57) and How interesting! (B9, p. 16). Category B features seem to be given more attention in the upper intermediate and intermediate textbooks as opposed to the advanced books, which in view of their easier learnability and availability for immediate use, is perhaps not surprising. Examples of these features included in the books, and highlighted for the learners attention, are
Hes a bit strange, a bit weird. (B13, p. 15) She was sort of annoyed with me. It was kind of expensive. (B13, p. 99) My dad thought I was a pick-pocket or a drug addict or something. (B13, p. 99) Im not really interested in art and all that stuff / and stuff like that. (B8, p. 57) Yeah, but I mean youve got to live with each other . . . (B15, p. 62) Well you know, I mean I havent really got much time for that, you know. (B15, p. 72)

Category A Attention
By contrast, there is little overt focus on Category A features. Heads and tails only feature in the advanced-level textbooks, tail structures appearing in just one of the books (B24) in a section on conversational tags, and head structures appearing only in B22 under Exploring Natural Speech. Examples of each structure included in the two books are provided in bold type:
Shes one of the all-time greats, Liza Minnelli is. (tail structure, B24, p. 46) ConictIm getting better at itmuch better than I used to be. (head structure, B22, p. 97)

Situational ellipsis features in just 2 out of the 24 books (B5 and B15). Examples drawn to the learners attention include
Sounds a bit risky to me. Got the time? Must dash. (B15, p. 102)

One of the books (B5) includes the use of the past progressive tense to report what people say, with this example taken from a conversation the learners had previously listened to:
Paul was telling me about your new car. It sounds really good. (B5, p. 114)

Perhaps understandably at this pre-intermediate level, the learners are not shown how this tense can be used to introduce a reported speech

clause (with backshift rules, etc.), nor are they given any information about its particular use in conversational as opposed to written English. This books also does not differentiate between the use of the past simple or past progressive in the reporting verb, again not surprisingly at this level. There are three entries in Table 3 under Other in Category A. Two concern the use of an affirmative question tag after an affirmative verb in the preceding clause, for example: So hes rich, is he? (B24, p. 46) and So he helped you, did he? (B13, p. 42). Biber et al. (1999, pp. 208209) noted that a speaker uses this feature of conversational grammar as a device to echo or comment on something the previous speaker has just said. B13 points our that it is often used to express surprise (p. 42), a point perhaps reinforced by the fact that all their examples begin with So . . .. The third entry is the use of conditional clauses in unnished sentences to give polite instructions: If youd like to hold on a minute (B15, p. 122). This construction would be very unusual in writing in its elliptical form but would not be so unusual if the second half of the sentence were to be completed, for example,
If you would like to hold on a few days, we will have the articles sent to you.

Category C Attention
Four of the books draw the learners attention to Category C features and to the ways in which observation of certain rules of grammatical agreement may be relaxed in informal usage. Examples from B22 and B13 include (relevant features are once again in bold type)
Theres lots of things left to do. (B22, p. 73) Theres a lot less/more people there. (B13, p. 122) Less people read the classics now. (B22, p. 73) Theres more/not as many cars on the road there. (B13, p. 122)

Whereas this indicates an attempt to make learners aware of how grammar may be affected by informal contexts of use, it is interesting to note that only one of the intermediate-level textbooks surveyed points out that If I was . . . is an acceptable spoken alternative to the more formal If I were . . . in its unit dealing with second conditional clauses. The example given is
If I was younger, the job would be easier. (B9, p. 108)

One of the books (the pre-intermediate-level B5) draws the learners attention to the use of the object pronoun me, in preference to I, in the

subject position of an utterance like Me and my brother both support Hull City football club. This use is noted by Carter and McCarthy (2006) as one which is usually found only in speech (p. 380). In the textbook, it is presented for receptive rather than productive use, with this explanation:
We dont usually start with me like this when we write. Some older people even think it is wrong. Are there things like this in your language? (B5, p. 16)

As this comment shows, this structure would be regarded by many as nonstandard, and for this reason presumably, it is highlighted in the textbook for recognition purposes only. It serves to illustrate the fact that, when investigating features of spoken grammar, the borderline between standard and nonstandard usage is not always clear.

Mode of Presentation
Where features of spoken grammar were highlighted, the textbooks in general followed a similar mode of presentation and practice, whereby the feature in question is initially encountered in a stretch of authentic, or plausible, semiscripted conversational discourse, presented as a listening text. After a task to check global comprehension of the text, the learners attention is drawn to the target feature of spoken grammar (e.g., by listening again or reading a transcript of the text and underlining examples of the feature), and its communicative purpose and use in the listening text is either explained or explored through someusually rather briefquestions for discussion. There is then typically a short practice activity, where the learners are required to use the feature in a fairly controlled setting. The procedure is thus similar to the three Is (illustration, induction, interaction) approach to teaching spoken grammar advocated by McCarthy and Carter (1995). It is also consistent with the task sequence of global understanding, noticing, and language discussion suggested by Timmis (2005), although the discussion questions are notably less critical and analytical than Timmis (p. 121) advocates. The following extract from B22, in which head structures are introduced, is fairly typical. The advanced-level students have just listened to a tape in which seven people talk about their attitudes toward conict. They have answered questions relating to the comprehension of the tape and have proceeded to talk briey with each other about real-life situations where they had to deal with conict. The language focus is illustrated in this extract.

Exploring natural speech

1. Why do the speakers start their answers with the word conict in these extracts rather than use the normal word order? a. ConictIm getting better at itmuch better than I used to be, but its not enjoyable, its never enjoyable having to deal with conict. b. Well conictId rather not talk about it, thats basically how I deal with it. 2. Reply to these questions in a similar way: a. How do you feel about family rows? b. Whats your attitude to war? c. Do you nd it easy to deal with disappointments? d. Do you ever have problems with your neighbours? (Haines, 2002, p. 97)

The textbook analysis shows that the British ELT market has attempted to include various phenomena of spoken grammar shown to be frequent in corpora of spoken English over the past 20 years. Approaches to teaching these features, as illustrated by the textbook material, are also consistent with contemporary approaches to teaching grammar in general, with an emphasis on noticing and exploring language in contexts of authentic use. It should, however, be noted that there was a marked preference for Category B features and that structures in Category A were, by and large ignored. Only 4 of the 24 books surveyed included any reference to Category A features. Those that were included were only introduced as additional pieces of language, mainly at the upper intermediate or advanced levels. Given that the majority of learners never continue classes to the point at which advanced-level textbooks are used, they would not come across these features except informally in their exposure to English outside class. In considering the question in the title of this article, thenIs there a missing link between spoken grammar and ELT materials?the answer in the case of the British market must be a qualied yes, certainly with respect to Category A structures. The question whether there are good pedagogical reasons for devoting more attention to Category A features of spoken grammar is discussed in the following section. The preference for dealing with Category B features may well reect the textbook writers perceptions regarding teachability and learnability alluded to in the previous section. As xed phrases, these bits of language might be regarded as easier to learn (and teach) on the grounds

that they can simply be added to the learners lexicon without involving a restructuring of her grammatical knowledge. Second, given the fact that grammatical items we have classied under Category A are by and large ignored, the concentration on Category B features suggests a rather limited view of what spoken grammar actually is. There seems to be a tendency among some of the textbooks to associate distinctive features of conversational English solely with language chunks, rather than with generative grammatical structures. This in itself may be due to other inuences from applied linguistics in the past 15 years emphasising the importance of formulaic language and routines in language learning and language use (Aijmer, 1996; Nattinger & DeCarrico, 1992; Pawley & Syder, 1983; Widdowson, 1989).


The case for teaching spoken grammar rests on an acknowledgment of the fact that written and spoken language are not replicas of each other and that the teaching of speaking skills should reect this difference. In conclusion, we present four arguments for including a more serious treatment of spoken grammar in ELT textbooks, particularly those designed for use in contexts where learners look to native speaker models, from which most of the available corpora of spoken English are taken, as a point of reference for acceptable usage.

Frequency of Use
The rst argument relates to frequency of use. One of the main functions of electronic corpora is to provide hard evidence about which items of language occur most and least frequently in natural situations, so that curricular decisions about syllabus content can be based on actual attested use, rather than on intuitions which are often unreliable or simply wrong (Biber & Conrad, 2001; Biber & Reppen, 2002). The evidence from corpus studies of spoken English shows that all the features discussed in the rst part of this article occur frequently in conversational discourse. Heads and tails, for example, occur in the Biber et al. (1999) corpus over 200 times per million words in conversation (p. 957), which makes them twice as frequent in the same corpus as ought to modals, or get passives (We got married, They got arrested, etc.), both fairly standard items in intermediate-level language syllabuses. Similarly, Carter and McCarthy (1995) point out the striking frequency in their

Cambridge and Nottingham Corpus of Discourse in English of past progressive reporting verbs in indirect speech, in the specic genre of casual conversation, as distinct from other genres of informal spoken language, such as service encounters and narratives.

Justication of Inclusion
Frequency of occurrence does not in itself justify inclusion of an item on a teaching syllabus. As Cook (1998) has argued, frequency and desirability are two different things and something is not a good model simply because it occurs frequently (p. 61). However, with respect to the items of spoken grammar discussed in this article, there are no reasons to suppose that they are bad or in any other way inappropriate models of usage. Rather, as has been argued, they serve a range of important communicative functions. These functions, relating to the unplanned, interactive, and interpersonal nature of conversation, mean that features like heads or tails cannot simply be covered by more conventional structures and, in contexts where learners need English to interact with native speakers, deserve to be taught for more than just recognition purposes.

Consequences to Learners
Neglecting to teach these features may lead to learners either avoiding them or transferring equivalent features from their rst language, as an illuminating study by De Cock et al. (1998) has shown. Comparing speech data of nonnative and native speakers, the authors found some striking differences between a group of advanced French EFL learners and a group of native English speakers in the United Kingdom in their use of vagueness tags and other common formulae of the kind grouped in our Category B. For example, the French learners made heavy use of the vagueness tag and so on, which rarely occurred in the native speaker data, and little or no use of tags like and everything, and things, and or something, which were used frequently by the native English speaker group. Similar discrepancies were found in the use of sort of, I mean, and you know, which were frequently used by the native English speaker group but considerably less so by the French learners, and in the use of the insert in fact, where the reverse pattern was found. The study suggests that these advanced learners of English experienced the need to use vague language and inserts, but either due to lack of contact with the target language or lack of explicit teaching, they were left alone to formulate their own expressions, drawing on their own resources (e.g., transfer from L1 or from a written model of L2). The resulting expres382 TESOL QUARTERLY

sions they used inevitably differed in kind and in application from those used by native English speakers.

Evidence Based on Research

Whereas mismatches between English language learner and native English speaker usage may not be cause for great concern in the global arena in which English is used, and in which the great majority of its users are not native speakers, learners still need models of some kind as a point of reference. And some evidence, from research conducted by Timmis (2002) into the kind of English that students actually want to learn, suggests that native speaker norms of English exert a very strong appeal to learners from a diverse range of countries and contexts of language use, and that these norms include the kind of informal, spoken grammar highlighted in the work of Carter and McCarthy (1997) (p. 246). Just over half of Timmiss 400 respondents agreed that it was important for them to be able to use the kind of English illustrated by short authentic examples of informal spoken English provided, whereas 22% disagreed. Perhaps it is more interesting that of those who agreed, two thirds were living and studying in countries where English was not the dominant spoken language.

Further Study
Clearly, further studies of this kind are needed to ensure that the learners voice is not disregarded in decisions about whether to devote more attention in ELT textbooks to spoken grammar than presently exists. In the absence of such studies, it is surely patronizing, as Carter and McCarthy (1996) have argued, for materials writers to decide on the learners behalf that they do not need to concern themselves too much with features of spoken grammar. The real challenge, they suggest, is to provide descriptions and to develop materials which serve the needs of teachers in all situations, whether they be native or non-native, so that they can decide how best to make such hitherto unrecorded aspects of English more widely accessible (p. 370).

Richard Cullen is Head of the Department of Language Studies at Canterbury Christ Church University, Canterbury, England. He has worked as a teacher and teacher trainer in Nepal, Greece, Egypt, Bangladesh, and Tanzania. His research interests

include teacher talk and classroom discourse, teacher education, and the teaching of grammar. I-Chun (Vicky) Kuo is currently a lecturer at National Cheng Kung University, Taiwan, where she is helping to develop an Internet-based programme of English learning and testing. She is in the process of completing her doctorate from Canterbury Christ Church University on student interaction in a British EFL setting.

Acklam, R., & Crace, A. (2005). Total English pre-intermediate. Harlow, England: Longman. Acklam, R., & Crace, A. (2006). Total English upper intermediate. Harlow, England: Longman. Aijmer, K. (1984). Sort of and kind of in English conversation. Studia Linguistica, 38, 118128. Aijmer, K. (1989). Themes and tails: The discourse function of dislocated elements. Nordic Journal of Linguistics, 12(2), 137154. Aijmer, K. (1996). Conversational routines in English: Convention and creativity. Harlow, England: Longman. Biber, D., & Conrad, S. (2001). Quantitative corpus-based research: Much more than bean counting. TESOL Quarterly, 35, 331340. Biber, D., Johansson, S., Leech, G., Conrad, S., & Finegan, E. (1999). Longman grammar of spoken and written English. Harlow, England: Longman. Biber, D., & Reppen, R. (2002). What does frequency have to do with grammar teaching? Studies in Second Language Acquisition, 24, 199208. Brazil, D. (1995). A grammar of speech. Oxford: Oxford University Press. Brown, G. (1989). Making sense: The interaction of linguistic expression and contextual information. Applied Linguistics, 10, 97108. Carter, R. (2004). Grammar and spoken English. In C. Coffin, A. Hewings, & K. OHalloran (Eds.), Applying English grammar: Functional and corpus approaches (pp. 2539). London: Arnold. Carter, R., Hughes, R., & McCarthy, M. (1998). Telling tails: Grammar, the spoken language and materials development. In B. Tomlinson (Ed.), Materials development in language teaching (pp. 2568). Cambridge: Cambridge University Press. Carter, R., & McCarthy, M. (1995). Grammar and the spoken language. Applied Linguistics, 16, 141158. Carter, R., & McCarthy, M. (1996). Correspondence. ELT Journal, 50, 369371. Carter, R., & McCarthy, M. (1997). Exploring spoken English. Cambridge: Cambridge University Press. Carter, R., & McCarthy, M. (2006). Cambridge grammar of English: Spoken and written English grammar and usage. Cambridge: Cambridge University Press. Channell, J. (1994). Vague language. Oxford: Oxford University Press. Clare, A., & Wilson, J. (2006). Total English intermediate. Harlow, England: Longman. Cook, G. (1998). The uses of reality: A reply to Ronald Carter. ELT Journal, 52(1), 5763. Cunningham, S., & Moor, P. (2003). Cutting edge advanced. Harlow, England: Longman. Cunningham, S., & Moor, P. (2005a). New cutting edge elementary. Harlow, England: Longman.

Cunningham, S., & Moor, P. (2005b). New cutting edge pre-intermediate. Harlow, England: Longman. Cunningham, S., & Moor, P. (2005c). New cutting edge intermediate. Harlow, England: Longman. Cunningham, S., & Moor, P. (2005d). New cutting edge upper intermediate. Harlow, England: Longman. De Cock, S., Granger, S., Leech, G., & McEnery, T. (1998). An automated approach to the phrasicon of EFL learners. In S. Granger (Ed.), Learner English on computer (pp. 6779). Harlow, England: Longman. Dellar, H., & Hocking, D. (2000). Innovations intermediate/upper intermediate. Hove, England: Language Teaching. Dellar, H., & Hocking, D. (2004). Innovations upper intermediate (2nd ed.). London: Thomson ELT. Dellar, H., & Walkley, A. (2005). Innovations pre-intermediate. London: Thomson ELT. Eggins, S., & Slade, D. (1997). Analyzing casual conversation. London: Continuum. Forsyth, W. (2000). Clockwise intermediate. Oxford: Oxford University Press. Gairns, R., & Redman, S. (2002). Natural English intermediate. Oxford: Oxford University Press. Haines, S. (2002). Landmark advanced students book. Oxford: Oxford University Press. Haines, S., & Stewart, B. (2000). Landmark upper intermediate. Oxford: Oxford University Press. Halliday, M. A. K. (1985). Spoken and written language. Victoria, New South Wales, Australia: Deakin University Press. Jones, C., & Bastow, T., (2001). Inside out advanced. London: Macmillan. Kay, S., & Jones, V. (2001). Inside out upper intermediate. London: Macmillan. Leech, G. (2000). Grammars of spoken English: New outcomes of corpus-oriented research. Language Learning, 50, 675724. MacAndrew, R. (1991). English observed: The shock of real English. Hove, England: Language Teaching. McCarthy, M. (1998). Spoken language and applied linguistics. Cambridge: Cambridge University Press. McCarthy, M., & Carter, R. (1995). Spoken grammar: What is it and how can we teach it? ELT Journal, 49, 207218. McCarthy, M., & Carter, R. (2002). Ten criteria for a spoken grammar. In E. Hinkel & S. Fotos (Eds.), New perspectives on grammar teaching in second language classrooms (pp. 5175). Mahwah, NJ: Lawrence Erlbaum. Miller, J., & Weinert, R. (1998). Spontaneous spoken language: Syntax and discourse. Oxford: Clarendon Press. Nattinger, J., & DeCarrico, J. (1992). Lexical phrases and language teaching. Oxford: Oxford University Press. Naunton, J. (2000). Clockwise upper intermediate. Oxford: Oxford University Press. Pawley, A., & Syder, F. (1983).Two puzzles for linguistic theory: Nativelike selection and nativelike uency. In J. Richards & R. Schmidt (Eds.), Language and communication (pp. 192226). London: Longman. Potten, H., & Potten, J. (2001). Clockwise elementary. Oxford: Oxford University Press. Quirk, R., Greenbaum, S., Leech, G., & Svartvik, J. (1985). A comprehensive grammar of the English language. London: Longman. Soars, L., & Soars, J. (2000a). New headway beginner. Oxford: Oxford University Press. Soars, L., & Soars, J. (2000b). New headway elementary. Oxford: Oxford University Press. Soars, L., & Soars, J. (2003a). New headway intermediate. Oxford: Oxford University Press.

Soars, L., & Soars, J. (2003b). New headway advanced. Oxford: Oxford University Press. Stubbs, M. (1983). Discourse analysis: The sociolinguistic analysis of natural language. Oxford: Basil Blackwell. Timmis, I. (2002). Native-speaker norms and international English: A classroom view. ELT Journal, 56, 240249. Timmis, I. (2005). Towards a framework for teaching spoken grammar. ELT Journal, 59, 117125. Walton, R., & Bartram, M. (2000). Initiative advanced. Cambridge: Cambridge University Press. Widdowson, H. G. (1989). Knowledge of language and ability for use. Applied Linguistics, 10(2), 128137. Yule, G., Mathis, T., & Hopkins, M. (1992). On reporting what was said. ELT Journal, 46, 245251.




TESOL Quarterly will occasionally feature debates and discussions on emerging topics of interest in the profession. Leading scholars who represent diverse perspectives will comment briey on the topic. In this issue, we focus on the place of pedagogical implications in the research articles we publish and, by extension, the connection between research and pedagogy in general. This symposium was occasioned by a reader response from ZhaoHong Han. I invited a former TQ editor, editors of two leading research journals, and the editor of an international practitioner-based journal to comment briey on the topic.

Pedagogical Implications: Genuine or Pretentious?

Teachers College, Columbia University New York, New York, United States

A central mission of TESOL Quarterly (TQ), as stated in its submission guidelines is to publish manuscripts that contribute to bridging theory and practice in the profession, and then, as its main submission category, TQ invites full-length articles [that] present empirical research and analyze original data that the author has obtained using sound research methods (TESOL, 2007). Indeed, many such articles have appeared in the journal over the years. However, what has also become increasingly eye-catching is a tendency in authors, in particular, those who report on experimental research, to have an add-on section, variously labeled Pedagogical Implications, Classroom Implications, Applications to Practice, and the like, to ostentatiously link the research to practice. This gesture is often more pretentious than genuine, for it does not seem warranted by the research reported. It is therefore not clear whether the authors indeed felt that they had sufficient empirical basis for drawing pedagogical implications or they did so because the reviewers had mandated them to draw such implications, guided by the journals criteria for evaluating the suitability of a manuscript for publication, such as the following:
The manuscript strengthens the relationship between theory and practice: Practical articles must be anchored in theory, and theoretical articles
TESOL QUARTERLY Vol. 41, No. 2, June 2007


and reports of research must contain a discussion of implications or applications for practice. (TESOL, 2007; emphasis added)

Either being the case, this practice can have such far-reaching ramications given the journals wide circulation and diverse readership, which includes individuals who may not have familiarity with the subject matter addressed (TESOL, 2007), that it merits the elds attention.

By way of illustration, Kim (2006) reported on an empirical study that sought to measure the effects of input modications (via lexical elaboration and typographical enhancement) on incidental acquisition of lowfrequency vocabulary. The study deployed an experimental design of six 20-minute-exposure conditions and three test tasks to measure postexposure performance. The studys participants were 297 Korean learners of English as a foreign language (EFL). Vocabulary acquisition was operationalized as form versus meaning recognition of isolated words. The results, as Kim summarized, were (a) LE [lexical elaboration] alone did not aid form recognition of vocabulary. (b) Explicit LE alone aided meaning recognition of vocabulary. (c) TE [typographical enhancement] did not aid form and meaning recognition of vocabulary. (d) LE and TE combined did not aid form recognition of vocabulary. (e) Both explicit and implicit LE aided meaning recognition of vocabulary. (f) Explicit and implicit LE did not differ in their effect on form and meaning recognition of vocabulary. (g) Whether a text was further enhanced in addition to either explicit or implicit LE did not seem to affect the acquisition of the previously unknown words forms or meanings. (p. 341) In a nutshell, the study yielded the nding that explicit lexical elaboration helped meaning recognition of the target words but not form recognition, and in contrast, typographical enhancement of words helped form recognition but not meaning recognition. In addition, there was no signicant interaction between lexical elaboration and typographical enhancement.

Based on these ndings, Kim concluded the research report with six pedagogical implications: Explicit lexical elaboration emerges as an effective approach to help L2 learners recognize the meanings of low-frequency L2 vocabulary from reading. Different types of explicit lexical elaboration devices can be used for this purpose.
To maximize the effect of explicit lexical elaboration, the text can be enhanced typographically. Although not systematically manipulated in this study, combining different types of typographical input enhancement that are proven effective with college-level Korean learners of EFL can increase their effectiveness. This study enhanced only the L2 target words, not their lexical elaborations. However, as long as the least distracting type of typographical input enhancement can be used optimally, the target words can be enhanced along with their lexical elaborations. This double treatment should better draw L2 learners attention to form (i.e., target words) and meaning (i.e., lexical elaborations) at the same time. As with Laufer and Hill (2000) and Lomicka (1998), information technology can be harnessed to provide an electronic or online version of an elaborated and enhanced reading text. Extensive use of hyperlinked text and multimedia can maximize the effects discussed in this study for the teaching of reading comprehension and vocabulary acquisition. High speed Internet connections are available virtually anywhere in Korea at fairly reasonable cost, which affords Korean EFL a strategic advantage. As previous research into incidental L2 vocabulary acquisition from reading has found, a single exposure to L2 vocabulary in this study resulted in limited but selectively signicant acquisition. This result provides at least indirect evidence that multiple exposures to the same words in different contexts are necessary to consolidate knowledge of those words. Multiple exposures to the same words entail that L2 learners either meet those words in the same text multiple times in one sitting or encounter them repeatedly in different texts over an extended period of time. For example, freshman English programs in Korea can design a reading program that extends over a semester or so, where English language learners choose reading materials of interest that are relevant to their academic areas. The reading materials will present target words in explicitly elaborated and typographically enhanced form. (pp. 366367)

As expressed, these implications are a mixed bag of recommendations (i.e., explicit lexical elaboration and double treatment) and suggestions on how to fulll them, some of which are general (e.g., Different types of explicit lexical devices can be used [for lexical elaboration])

and some of which are specic (e.g., extensive use of hyperlinked text and multimedia). It is important to note that when these proposed implications are pitted against the ndings from the study, it is immediately clear that a preponderance of them are far-fetched; there is little or insufficient empirical support for them. Consider the two recommendations. First, with regard to explicit lexical elaboration, although the study generated some favorable results, it is necessary to keep in mind that the study was the rst published research of its kind,1 and as such, there is, as a whole, insufficient empirical evidence, and hence a very limited or inconclusive understanding, of its efficacy on incidental vocabulary acquisition. With regard to double treatment involving typographical enhancement, the study, as noted earlier, reported a signicant effect for typographical enhancement in form recognition of words, but existing research at large has depicted a conicting picture regarding its efficacy for second language morphosyntactic and vocabulary development (see, e.g., Barcroft, 2003; Izumi, 2002; Jourdenais, Ota, Stauffer, Boyson, & Doughty, 1995; Leow, 1997). Although the jury is still out on the ultimate efficacy of typographical enhancement (cf. Wong, 2003), two preliminary ndings from this body of research are especially relevant to the present discussion. One nding is that too much articial salience is not necessarily benecial to learning, for it can encourage learners to cultivate an overuse of linguistic features that becomes hard to undo (see, e.g., Alanan, 1995; Jourdenais et al., 1995; Leeman, Arteagoitia, Fridman, & Doughty, 1995; Overstreet, 1998). A second nding is that simultaneously focusing learners attention on meaning and form during input processing may lead to a trade-off effect, such that one is processed at the expense of the other (see, e.g., Overstreet, 1998; VanPatten, 1992). Kims own study has corroborated this particular nding insofar as it shows (a) that typographical enhancement aided recognition of the form but not the meaning of the target words, (b) that explicit or implicit lexical elaboration did just the opposite, and (c) that when explicit or implicit lexical elaboration and typographical enhancement were combined, there was no superior effect for either form or meaning recognition over that produced when either approach was used alone (see Kim, 2006, Table 8). If these ndings are conrmed by further research, double treatment should not be recommended for implementation in a pedagogical setting. Finally, as a further compromise to the overall empirical basis of the alleged implications, Kims study suffers two critical weaknesses in its

The limited research Kim cited was mostly unpublished masters theses.



design. First, the experimental duration was too short, lasting only 20 minutes and involving only a single exposure to input. Second, the study did not effectively control for prior knowledge. As such, the ndings from the study must perforce be interpreted with caution. Relating them to practice at this stage would be too far a stretch, if not misleading.

The practice of devoting a section in any research article to pedagogical implications might have stemmed from a fallacy that any research can be related to pedagogy. In the domain of SLA, not every topic (or study, for that matter) is relevant to second language teaching, and the ones that are relevant may bear a direct or indirect, actual or potential, and above all, complex relationship to teaching (Ellis, 2003; Gregg, 2001). Moreover, like it or not, it is as yet a reality that on many fronts, our knowledge of L2 development is too thin and eclectic to serve as a valid basis for application to practice (Pienemann, 1985, p. 54), or even for proposing provisional specications (Stenhouse, 1975), though it is equally the case that on as many others, where there has been sufficient accumulation of empirical ndings, generalizations are safely drawn that deserve serious attention by teachers, materials developers, and/or curriculum designers (see, e.g., Ellis, 2005; Larsen-Freeman, 1995; Lightbown, 1985, 2000; Norris & Ortega, 2000; Towell & Hawkins, 1994; VanPatten, 2002).

SLA is still a very young eld of study (Ellis, 2005, p. 209), and for this reason alone, any excessive concern to allege or show a relationship between an empirical study and classroom practice may be counterproductive to research and practice. For researchers, such practice can create a false validity that stops them from undertaking further investigation. For practitioners who are unfamiliar with the research, on the other hand, the alleged implications may be indiscriminately embraced as a blueprint for daily classroom practice. This is not to suggest that TQ should drop its mandate to prospective authors on linking research and practice, but it is to suggest that its editorial policy should be revised to reect a more cautious stance (Hatch, 1978). The change that results should then translate into evaluation criteria that legitimize a variety of ways to articulate the link. For example, a research study may reect this link by having a pedagogical question as its motivation; another research study may relate to an exSYMPOSIUM 391

isting database that may eventually shed light on a pedagogical concern; another study may conclude, based on a careful analysis of its methodological limitations, that it is too early to draw any pedagogical implications; yet another study may offer ideas for teachers to experiment with in their own classrooms, and so forth. In all likelihood, this broadened scope will lead to a greater variety of research articles than are currently published in the journal, covering, for example, interventional, noninterventional, interpretive, or experimental research. Even so, researchers need be ever mindful that as much as their studies are generalizable, pedagogy is largely local, as pointed out by one of the reviewers. Thus, any suggestions on pedagogical implications or applications must be predicated on a sound grasp of classroom realities. As the agship journal of TESOL, TQ is in a unique position to provide leadership in TESOL research and practice. Inasmuch as the eld is dynamic, it is imperative that its editorial practice be frequently subjected to review and modication to reect changing conceptions and practices as well as to provide directions. Such a course of action will, inter alia, serve to preempt any inadvertent disservice to the profession by, for example, licensing researchers to inate their ndings through extending immature results to pedagogy, and, as a consequence, misguiding educators into a practice that may inhibit rather than promote learning.
Dr. ZhaoHong Han is an associate professor of linguistics and education at Teachers College, Columbia University, New York, New York, United States. Her research interests include second language learnability and teachability. She is the author of Fossilization in Adult Second Language Acquisition and co-editor (with Terence Odlin) of Studies of Fossilization in Second Language Acquisition.

Alanan, R. (1995). Input enhancement and rule presentation in second language acquisition. In R. Schmidt (Ed.), Attention and awareness in second language acquisition (pp. 259299). Honolulu: University of Hawaii Press. Barcroft, J. (2003). Effects of questions about word meaning during L2 Spanish lexical learning. The Modern Language Journal, 87(4), 546561. Ellis, R. (2003). Task-based language learning and teaching. Oxford: Oxford University Press. Ellis, R. (2005). Principles of instructed language learning. System, 33, 209224. Gregg, K. (2001). Learnability and second language acquisition theory. In P. Robinson (Ed.), Cognition and second language instruction (pp. 152180). Cambridge: Cambridge University Press. Hatch, E. (1978). Apply with caution. Studies in Second Language Acquisition, 2, 123 143.

Izumi, S. (2002). Output, input enhancement, and the noticing hypothesis: An experimental study on ESL relativization. Studies in Second Language Acquisition, 24, 541577. Jourdenais, R., Ota, M., Stauffer, S., Boyson, B., & Doughty, C. (1995). Does textual enhancement promote noticing? A think-aloud protocol analysis. In R. Schmidt (Ed.), Attention and awareness in foreign language learning (pp. 182209). Honolulu: Second Language Teaching & Curriculum Center, University of Hawaii at Manoa. Kim, Y. (2006). Effects of input elaboration on vocabulary acquisition through reading by Korean learners of English as a foreign language. TESOL Quarterly, 40, 341373. Larsen-Freeman, D. (1995). On the teaching and learning of grammar. In F. Eckman, D. Highland, P. Lee, J. Mileham, & R. Weber (Eds.), Second language acquisition theory and pedagogy (pp. 131150). Mahwah, NJ: Lawrence Erlbaum. Laufer, B., & Hill, K. (2000). What lexical information do L2 learners select in a CALL dictionary and how does it affect word retention? Language Learning & Technology, 3, 5876. Leeman, J., Arteagoitia, I., Fridman, B., & Doughty, C. (1995). Integrating attention to form with meaning: Focus on form in content-based Spanish instruction. In R. Schmidt (Ed.), Attention and awareness in foreign language learning (pp. 217258). Honolulu: Second Language Teaching & Curriculum Center, University of Hawaii at Manoa. Leow, R. (1997). The effects of input enhancement and text length on adult L2 readers comprehension and intake in second language acquisition. Applied Language Learning, 8, 151182. Lightbown, P. (1985). Great expectations: Second language acquisition research and classroom teaching. Applied Linguistics, 6, 173189. Lightbown, P. (2000). Classroom SLA research and second language teaching. Applied Linguistics, 21, 431462. Lomicka, L. (1998). To gloss or not to gloss: An investigation of reading comprehension online. Language Teaching & Technology, 1, 4150. Norris, J. M., & Ortega, L. (2000). Effectiveness of L2 instruction: A research synthesis and quantitative meta-analysis. Language Learning, 50, 417528. Overstreet, M. (1998). Text enhancement and content familiarity: The focus of learner attention. Spanish Applied Linguistics, 2, 229258. Pienemann, M. (1985). Learnability and syllabus construction. In K. Hyltenstam & M. Pienemann (Eds.), Modelling and assessing second language acquisition (pp. 23 76). Clevedon, England: Multilingual Matters. Stenhouse, L. (1975). An introduction to curriculum research and development. London: Heinemann. TESOL. (2007). TESOL Quarterly submission guidelines. Alexandria, VA: Author. Available from 1031 Towell, R., & Hawkins, R. (1994). Approaches to second language acquisition. Clevedon, England: Multilingual Matters. VanPatten, B. (2002). Communicative classrooms, processing instruction, and pedagogical norms. In M. S. Gass, K. Bardovi-Harlig, S. Magnan, & J. Walz (Eds.), Pedagogical norms for second and foreign language learning and teaching (pp. 105118). Amsterdam: John Benjamins. Wong, W. (2003). The effects of textual enhancement and simplied input on L2 comprehension and acquisition of non-meaningful grammatical form. Applied Language Learning, 13, 1745.

The Research/Pedagogy Interface in a 21st-Century Publication Context

University of Adelaide Adelaide, South Australia

The issues raised by ZhaoHong Han represent a timely call to reconsider how we can best think and write about the relationships between research and practice in the eld of TESOL. The pressure to get research published has been growing steadily in the recent past and is now felt strongly at many levels of the TESOL profession. For many of us, it is not only the fact of publication but also the journal in which our work appears that is of signicance in the various accounting schemes used to evaluate and reward our work and that of our schools or universities. Hans introduction to her nal paragraph is important here: As the agship journal of TESOL, TQ is in a unique position to provide leadership in TESOL research and practice. TQ does have a very high prole, as well as, and perhaps because of, its very broad reading audience with a wide range of interests. This high prole means that many in the eld see TQ as a highly desirable place to publishwhich meets the goals of the editors and advisory board. However, it also means that the editors and referees have considerable power to ensure that articles meet stringent conditions before being deemed acceptable. There are two lines of thought I would like to follow in responding to Hans article, the rst of which is related to Hans suggestion to change TQ editorial policy, and the second to how additional advice to authors could be framed to help avoid the kinds of issues Han highlights. It seems to me that the change Han seeks relates not so much to policy as to how the policy may be applied by individual referees in particular cases. She endorses the requirement to link research and practice, but wishes to see evaluation criteria that legitimize a variety of ways to articulate the link. The current criteria ask only for research reports to contain a discussion of implications or applications for practice (TESOL, 2007); this is a broad form of words, and would seem already to allow for a variety of ways in which such a discussion could be couched. Nevertheless, it seems to be a common experience of many of us writing in TESOL-related elds (and not just for TQ) that referee comments ask for applications to teaching practice to be more explicit than we had originally thought appropriate. Authors responses to these demands represent one way in which published articles are in fact co-constructions of the authors and a range of others, including but not restricted to journal editors and referees (see e.g., Burrough-Boenisch, 2003).

Experienced authors may feel less intimidated by such requests than relative novices and be more prepared to take agency in explaining to the editor why the requested change is inappropriatethe expectation that authors will act in this way seems to be a very occluded aspect of the publication process, in my experience. Novice authors may therefore be particularly vulnerable to co-constructions with which they are ultimately unhappy, but to which they accede in order to have the article accepted. However, it is always the author who will be judged wanting if readers of the journal nd fault with the article after publication. The implications for practice in an article should ultimately be judged in the same way as the articles other claims, in terms of their t with the strength of the evidence presented in the article to substantiate themremembering that the evidence may be of widely different types depending on the type of article. At root, a perceived lack of such t is the basis of Hans concerns with the set of pedagogical implications she discusses in her article. My second point then relates to how additional advice to authors could be written to encourage the writing of pedagogical implications that avoid the kinds of problems Han notes. For me, avoiding the problems would require a three-way focus: on the range of contexts in which TQ readers work and research; on the skills and understandings that can be expected of TQ readers as professionals in the eld; and on intercultural variations in the ways in which implications are appropriately expressed. For many authors in my experience, the most obvious implications from a study relate to the specic context that gave rise to the study in the rst placeand this would be particularly the case for classroombased research. When addressing the much wider set of contexts represented by TQ readers (e.g., teachers in many sectors and countries, teacher-educators, academic researchers), authors may be helped by having readily available a snapshot list that highlights the breadth of the audience for whom they are writing. It may also be useful to remind authors that the readers bring considerable expertise to their reading of the implications and are unlikely to be looking for, or to accept, recipes that appear to be designed for direct application without considerable reection and adaptation. The TQ editors might also consider providing on the Web site a set of examples of appropriately worded implication sections from previous TQ articles, perhaps annotated to show how the evidence presented in the rest of the article in each example relates to the implications. In this way readers could be exposed to various ways of wording pedagogical implications that have succeeded in the TQ context, including question forms and various modal constructions and hedges. It therefore seems to me that guidelines for both referees and authors may be a useful way to move on this issue. To facilitate their development in a truly usable form, I would like to advocate some research on articles published in TQ in the past 5 years or so, looking perhaps at the various

types of research articles published, the level of prior publishing experience of the authors, and the ways in which the requirement for a discussion of pedagogical implications has been met, in terms of both language forms and content. Such an analysis could add substantially to our understanding of how the communication of research results in the TESOL eld is affected by a requirement for including implications for practice and provide the basis for an informed decision about the advisability of maintaining the requirement in its present form or advocating change.
I thank my colleagues Kate Cadman, Christina Eira, and Michelle Picard for their contributions to the ideas presented in this article, which emerge both from our teaching of research English as an additional language and from our efforts to publish our own research and help others publish theirs.

Margaret Cargill is a senior lecturer in researcher education and development at the University of Adelaide Graduate Centre, Adelaide, South Australia. She is also executive editor of TESOL in Context, the journal of the Australian Council of TESOL Associations.

Burrough-Boenisch, J. (2003). Shapers of published NNS research articles. Journal of Second Language Writing, 12, 223243. TESOL. (2007). TESOL Quarterly submission guidelines. Alexandria, VA: Author. Available from =1031



A Bridge Too Far?

Georgia State University Atlanta, Georgia, United States

I applaud ZhaoHong Hans call for more measured and varied presentations of pedagogical implications reective of the actual likelihood of a research articles pedagogical relevance. It is probably easy for all of us who read research to think of research articles that seem to reach for

pedagogical implications not clearly motivated by the research results or that read like afterthoughts the authors would rather not have thought of at all. In articulating what many of us have probably privately felt about pedagogical implications but not publicly expressed, Han has provided stimulus for a conversation the eld needs to have, a conversation that can be entered from a number of vantage points, including pedagogy, research, and the profession at large. From a pedagogical perspective, I suspect that few would disagree that cautiousness is needed in offering pedagogical implications. Lets hope, however, that authors do not become so cautious that they are wary of making any claims of possible connections between research and pedagogy. There is already a signicant, perhaps growing, divide between research and pedagogy in our eld (witness, e.g., the temporal and physical distancing between the more research-oriented AAAL and more pedagogically oriented TESOL conferences, and the attendant decreased interaction between participants in the two). Many of us involved in graduate TESOL education work hard to persuade our students of the benets of research-informed classroom practice, yet we all, faculty and especially former students, know the realitythat once in the eld, teachers, with more students than discretionary hours in the day, may have little time and energy for keeping up with research that could inform their decision making. If the research they do nd time to read makes little effort to speak to them, then it should come as no surprise if they have little incentive to read more of it. There have, of course, been notable efforts over the past decade or more to close the research(er) practice(ioner) gap by encouraging more classroom-based and -oriented research, with more obvious relevance to teachers, and by urging teachers to view themselves as researchers (e.g., Crookes, 1993, 2003). The divide is not lessened, however, if teacher or action research is perceived as not real research but something that teachers do for the sole purpose of reecting on their own practice in their own classrooms. As Casanave (2001) has observed, It is not enough to know thyself (p. 15), especially if one does not also look beyond the realm of personal or close colleagues classroom experience. It is interesting that, although there has been much discussion of teachers as researchers and critical consumers of research, there has been less public discussion of the need for language researchers to be, or be informed by, language teachers.1 If researchers are themselves far removed from second language (L2) classroom experience, it may indeed appear pretentious, to borrow Hans term, or presumptuous, for them to advise classroom practitioners on what to do or not do in their classrooms. This apparent presumptuous1

Teacher cognition is, though, a rapidly growing area of interest among researchers (see Borg, 2003).



ness is not, however, an argument for absolving researchers from responsibility for considering the possible value of their work for the language learners and teachers who might benet from it, but instead a reason for more collaboration between language teachers and researchers as professionals who can inform each other in research and teaching. Many researchers are, no doubt, well aware of not just how partial their knowledge of language pedagogy is but also how partial the state of knowledge in their research area is as well, and thus with good reason they may only reluctantly play the pedagogical advisory roles that journals often want them to take on. As Han notes, SLA is a young eld; much about second language learning (or acquisition) remains un- or underinvestigated. At the same time, from an epistemological standpoint, we might well ask when knowledge in any human science will ever be anything but partial (Lincoln & Guba, 1985, 2000). Teachers are faced with making countless pedagogical decisions, with or without complete condence in their efficacy, every time they enter a classroom, and, its worth noting, second languages are learned, and have been for millennia. This is not to suggest that the contributions of ongoing L2 research are not needed, that we need not aim for more efficient and effective teaching and learning (though it may suggest that teacher cognition deserves more credit than it often receives). Lets not forget that when teachers nd the time to read what research has to offer, they bring to their reading a fund of knowledge based on prior experience, prior reading, and, not infrequently, prior classroom-based research, but, perhaps most important, also current awareness of their local pedagogical situations, all of which place them in a strong position to judge the relevance and transferability of researchers pedagogical suggestions. Whats missing, however, and what research writers can proffer, as Hans comments suggest, is close knowledge of the researchers own particular projects and of how their ndings help advance (if they do) a broader research agenda that may be meaningful for pedagogyknowledge that can be shared in the research articles pedagogical implications section and thus contribute to the joint pedagogical knowledge construction of the eld. Of course, L2 research projects actually exist on a continuum of objectives, from the more theoretical to the more pedagogical (Ellis, 2006). Researchers primarily concerned with linguistic theory building may not anticipate an audience of teachers when they carry out and write up their research, well aware that their projects may have only the most tenuous connection to pedagogy. Given the wide variety of publishing forums now available, with more than 50 applied linguistics journals now in print (Braine, 2003), researchers can choose from a range of possible audiences, from the more theoretically inclined and highly specialized to a broader readership that includes both researchers and teachers. If au398 TESOL QUARTERLY

thors aim for publication in TESOL Quarterly, with its commitment to bridging theory and practice as the only refereed journal published by the largest professional organization for teachers of English to speakers of other languages, then the choice of a diverse audience has been made. Being responsive as an author to the spirit, not just the letter, of the TESOL Quarterly commitment to readers and the profession may well entail considering pedagogy early in ones research plans (long before the implications are written up), conceiving of research problems as nested in a number of research and real world contexts, and contemplating the needs of an audience that includes those eager to make the most of our elds partial knowledge on Monday morning.

I am grateful to Jim Belcher, Alan Hirvela, Lauren Lukkarila, and Kate Moran for very helpful discussion of the topic of this brief essay (i.e., pedagogy and published research).

Diane Belcher co-edits the journal English for Specic Purposes as well as the Michigan Series on Teaching Multilingual Writers. She has served as guest co-editor of the Journal of Second Language Writing several times and has also co-edited three books on academic literacy.

Borg, S. (2003). Teacher cognition in language teaching: A review of research on what language teachers think, know, believe, and do. Language Teaching, 36, 81 109. Braine, G. (2003). Negotiating the gatekeepers: The journey of an academic article. In C. P. Casanave & S. Vandrick (Eds.), Writing for scholarly publication: Behind the scenes in language education (pp. 7390). Mahwah, NJ: Lawrence Erlbaum. Casanave, C. P. (2001). Controversies in second language writing: Dilemmas and decisions in research and instruction. Ann Arbor: University of Michigan Press. Crookes, G. (1993). Action research for SL teachersgoing beyond teacher research. Applied Linguistics, 14, 130144. Crookes, G. (2003). A practicum in TESOL: Professional development through teaching practice. Cambridge, England: Cambridge University Press. Ellis, R. (2006). Researching the effects of form-focused instruction on L2 acquisition. AILA Review, 19, 1841. Lincoln, Y., & Guba, E. (1985). Naturalistic inquiry. Newbury Park, CA: Sage. Lincoln, Y., & Guba, E. (2000). Paradigmatic controversies, contradictions, and emerging conuences. In N. Denzin & Y. Lincoln (Eds.), Handbook of qualitative research (pp. 163188). Thousand Oaks, CA: Sage.

Gauging the Scholarly Value of Connecting Research to Teaching

University of Wisconsin, Madison Madison, Wisconsin, United States

ZhaoHong Han questions whether TESOL Quarterly should mandate pedagogical implications for research studies. Underlying this issue there are other questions. 1. Does the discipline place a high scholarly value on connecting research to teaching? 2. How is TQ positioned in these disciplines? How should TQ dene its focus in relation to other journals in second language studies and foreign language pedagogy? The eld of second language studies crosses many disciplines, each with its own history, focus, and community of researchers. Considering connections between research and language pedagogy, I will focus here on the elds of applied linguistics, second language studies or second language acquisition, and foreign language education. Grabe (2002) traces the origins of the broadest of these three disciplines, applied linguistics, to the rst issue of Language Learning: A Journal of Applied Linguistics in 1948. Offering a denition that conceives the eld as much larger than language learning, VanPatten (1999) says that applied linguistics is in essence any use of linguistic theory . . . for any domain other than the development of linguistic theory (p. 51). This denition ts the broad scope of the journal Applied Linguistics, which, according to its editorial policy, publishes research into language with relevance to realworld problems with its primary interest in making connections between elds, theories, research methods, and scholarly discourses, and less interest in the ad hoc solution of particular problems (see Aims in the front matter of any issue). Given the local nature of many teaching problems, as Han points out, pedagogical applications might not receive high scholarly value for this journal, a reection of their nonnecessity for the broad nature of the eld. Second language studies is typically seen as emerging on the scholarly scene in the late 1970s with a proposal by British and American scholars that language learners linguistic production should be studied as a language in its own right rather than a defective version of a target language. The origin of the eld is further marked by Hatchs (1978) col400 TESOL QUARTERLY

lection of interlanguage studies and by the appearance of the journal Studies in Second Language Acquisition that same year. Articles in Studies in Second Language Acquisition typically do not include sections on pedagogical implications, although their editorial policy allows empirical investigations of the interface between SLA and language pedagogy, such as classroom interaction or the effects of instruction (see front matter of any issue). Another potential beginning point for second language studies is Werner Leopolds (1939) study of the acquisition of English and German by his daughter Hildegard growing up in Milwaukee in the 1930s. Given the relatively young age of these disciplines as well as the theory- and empirical-base of their key publications, it is understandable that they would exercise considerable restraint against pedagogical applications of research, especially when these disciplines historically tended to reect learning that occurs naturally rather than to inuence how learning happens in a classroom. In contrast, the discipline of foreign language education is both much older and more narrowly dened than either applied linguistics or second language studies. For example, The Modern Language Journal (MLJ) originated in 1916 as the rst foreign language pedagogical journal in the United States to serve a variety of languages and teachers (Magnan, 2001, p. 92). As such, it was focused primarily on issues related to teaching and on the instructed environment in geographic areas where the language was not the vernacular. For the MLJ then, in its beginning, as well as for the foreign language education eld that looks rst to the classroom as a research site, pedagogical implications were welcomed, even expected. When I assumed the editorship of the MLJ in 1994, however, the debate about the premature and limiting nature of pedagogical implications for all language-learning research was gaining force (cf. Ellis, 1997), encouraging me to revise the mission of the journal to welcome language learning research without immediate pedagogical applications, a step that would constitute a change in editorial policy (Magnan, 2001, p. 115). In answer to the rst question I posed in this comment on Hans article, it would appear that the scholarly value of pedagogical implications for second language research is related to the discipline, with the highest value being for those scholars working historically in foreign language education. Katz and Watzinger-Tharp (2005) explain that language scholars working in foreign language departments often affiliate their work with all three related disciplines: applied linguistics, second language studies, and foreign language education. For them, then, the scholarly value of pedagogical implications of research would likely be higher than for scholars working in English departments who tend to affiliate themselves primarily, or perhaps exclusively, with only applied linguistics or second language studies. For TQ, the second question I

posed raises the following issue: With the increase of teaching English worldwide, does the organization wish to situate its agship journal clearly in the area of foreign language education or more broadly within applied linguistics and second language studies? Or does TQ prefer to have an editorial stance that would allow, but not mandate, pedagogical implications in order to extend its reach across disciplines? This latter route is the one I chose for the MLJ with my change in editorial policy. The gate-keeping role of scholarly journals denes and shapes a discipline. Han is suggesting that, by its insistence on pedagogical applications for research studies, TQ risks shaping research in a troublesome direction, where implications have a false validity when generalized beyond the local setting from which they came, making them premature, far-fetched, excessive, and even pretentious. I agree with Han that the danger of over-generalizing implications is very real, and that suggested implications from research should not be indiscriminately embraced as a blueprint for daily classroom practice. To guard against this misappropriation, I suggest the following, to authors and my fellow editors: 1. Pedagogical implications should be strictly limited to clear, substantial ndings. In quantitative articles, statistics should be reported carefully and pedagogical implications strictly limited to signicant ndings. Unfortunately, it is not uncommon for authors to state that a nding is marginally signicant, a term without statistical meaning. Authors need to establish an alpha level prior to conducting the study; results are then determined as signicant or not signicant in accordance with this alpha level. It is also not uncommon for authors to have nonsignicant ndings but say that they are still suggestive of trends on which pedagogical implications are drawn. In such cases, pedagogical implications should not be made. In qualitative studies, it may appear more difficult to draw a rm line for where ndings merit consideration for pedagogical applications, but rigorous analysis and reporting should make those boundaries clear. 2. Given that applications to teaching would always be local in nature, the setting of the study must be clearly detailed in the research report so that any pedagogical implications can be considered in terms of similarity of context. 3. In making pedagogical suggestions, authors should be encouraged to review other studies that point toward similar instructional applications, making clear and frank mentions of similarities and differences in the research contexts and their ndings. In cases where a group of studies indicate similar directions for teaching, recommendations would have increased value. 4. Pedagogical recommendations might be made in rhetoric that sug402 TESOL QUARTERLY

gests informed questioning (e.g., Might this observation suggest that . . . ?) rather than declarations (Teachers should . . . ). Authors need to remember that such statements are often cited out of the context of the study, where they are conveyed as instructional dictums. 5. We are rapidly approaching a time when the scholarly journal will have an electronic accompaniment, such as a discussion board. This forum, rather than journal pages, might be used for readers to speculate on pedagogical implications and share instructional material that concretizes certain recommendations. By dividing pedagogy and research in this way, it could be easier to maintain an appropriate distance between research ndings and pedagogical implications. The difference between allowing, encouraging, and mandating authors to include pedagogical implications with their research studies must exist on the continuum that it is. With increasing numbers of journals in elds encompassing second language learning and teaching, there is room for different stances on this important issue. It does not seem prudent for all journals to take the same stance for or against attributing pedagogical implications to research studies. Much as medical science has not waited to suggest any intervention until a cure is found, language instruction has immediate needs that might benet from best guesses anchored in research. In deciding whether to include pedagogical implications in research reports, editors of journals need to respond to the demands of the eld in which the journal situates itself, keeping in mind that, by their publications, journals shape the eld whose work they disseminate.

Sally Sieloff Magnan is the current editor of the Modern Language Journal. She began the annual volume, Issues in Language Program Direction, of the American Association of University Supervisors, Coordinators, and Directors of Foreign Language Programs in 1990 and served as series editor until 2006. She has also edited or co-edited three books.

Ellis, R. (1997). SLA and language pedagogy: An educational perspective. Studies in Second Language Acquisition, 19, 6992. Grabe, W. (2002). Applied linguistics: An emerging discipline for the twenty-rst century. In R. B. Kaplan (Ed.), The Oxford handbook of applied linguistics (pp. 312). Oxford: Oxford University Press.

Hatch, E. (Ed.). (1978). Second language acquisition: A book of readings. Rowley, MA: Newbury House. Katz, S., & Watzinger-Tharp, J. (2005). Toward an understanding of the role of applied linguists in foreign language departments. The Modern Language Journal, 89, 490502. Leopold, W. (1939). Speech development of a bilingual child: A linguists record. Vol. 1: Vocabulary growth in the rst two years. Evanston, IL: Northwestern University Press. Magnan, S. (2001). MLJ editorial policy: Reections on the profession, denition of its disciplines. The Modern Language Journal, 85, 92125. VanPatten, B. (1999). What is second language acquisition and what is it doing in this department? ADFL Bulletin, 30, 4953.



Pedagogical Implications in TESOL Quarterly? Yes, Please!

Iowa State University Ames, Iowa, United States

The desire for linking theory and research to practice is what drove the inauguration of TESOL Quarterly as the journal of the TESOL organization more than 40 years ago. Betty Wallace Robinetts (1967) editorial introducing the journal of the newly founded organization spoke to the issue of practical implications: Although the major emphasis will be on practical matters, our interpretation of what constitutes practicality is broad. . . . the classroom exercises and specic techniques will be here, but we would like to think that our readers are being led to search a little more deeply into the why of certain drills and the wherefore of certain techniques. (p. 2) It seems evident through these words and through the contents of the rst many years of the journal that practice was central to the mission of TQ. Indeed, it is our commitment to professional practice in TESOL that motivates most, if not all, of TESOL members today as it did 40 years ago. However, much has changed in our profession, and arguably our agship journal needs to evolve with these changes. Have those who read and write for TQ shifted their focus to the point that TQ articles should no longer be required to link to practice? Has the TQ crowd grown out of discussion of practice? In my view, TQ should continue to be dened by its relevance to TESOL practice. TESOL is not second language acquisition (SLA) even if some overlap exists between the issues in these two areas. Research in SLA encompasses work with and without pedagogical implications, and

accordingly journals of SLA do not require articles to link to pedagogy. What helps to dene TQ is its adherence to Robinetts vision, even as the profession has matured and become more complex. If an author can state no implications for teaching and learning, TESOL Quarterly is the wrong journal. The very issues ZhaoHong Han raises about the pedagogical implications of Youngkyu Kims study should be the substance for fruitful discussion in TQ. First, are practicioners not to take any guidance from research results unless they have been replicated? If so, where are teachers and teacher educators to get their knowledge? Are teachers better off using their best guesses about what works than drawing from results of unreplicated research? Second, if no implications are suggested, at what point can critics raise questions about the interpretation of results for practice, the consistency of results from other studies, and the authors suggestions for practice? The author is in the best position to make the rst attempt at interpreting results for practice, and without such an attempt, the discussion does not begin. Third, I have never seen a perfect research design implemented in language education. Nevertheless, many TESOL professionals and others see educational research as providing some useful insights into learning and teaching. In keeping with professional practice, when reviewers feel that a study warrants publication, readers deserve to read the study and the authors interpretations. These three issuesthe role of replication, the need for synthesis, and the infelicity of educational research designare critical for those of us who attempt to draw implications from research in hopes of improving teaching and learning. If pedagogical implications are not included in TESOL Quarterly articles, what is to prompt professionals to raise these questions? What is to prompt us to reect on how we interpret research in making pedagogical suggestions? Implications for practice are worthy of debate. The fact that Han disagrees with the implications suggested by Kim is cause for useful discussion of issues critical to TESOL rather that cause for changing the guidelines to preclude such discussion. Much has changed in our profession over the past 40 years. Robinetts 1967 editorial borders on apologetic for asserting the need to include anything but practice. Today we are asking whether articles included in TQ should state implications for practice. The irony of the question is striking in view of the dramatic increases in numbers of English language learners throughout the world and the variety of circumstances under which they learn. Professional knowledge has never been in greater demand. Never has the need been greater for TQ to strengthen its identity as the journal where readers nd pedagogical implications of the research as well as debate about the strength of support for authors claims.

Carol A. Chapelle is a former editor of TESOL Quarterly (19992004) and president of the American Association for Applied Linguistics.

Robinett, B. W. (1967). Editorial. TESOL Quarterly, 1, 2.

Editors Note
The editorial board discussed the desirability of revising the statement to our authors in its annual meeting on March 20, 2007. The board affirmed the long-standing commitment of the journal to bridge research and practice and decided to continue encouraging authors to provide a discussion of implications or applications for practice. Members felt that the current statement is appropriately broad and open. They emphasized the responsibility of the authors, referees, and the editorial team to see that pedagogical implications are fairly presented. However, TQ explicitly states in its information for contributors that positions taken by individual authors are not endorsed by the TESOL organization, the Quarterly, or the editorial board members.



TESOL Quarterly invites commentary on current trends or practices in the TESOL profession. It also welcomes responses or rebuttals to any articles or remarks published here in the Forum or elsewhere in the Quarterly.

The Author Replies to Z. H. Hans Pegagogical Implications: Genuine or Pretentious?

Ewha Womans University Seoul, Korea

One of the easiest criticisms that can be leveled against almost any data-based study is that it is of too short a duration. My study is not immune to this criticism, but it is not alone. For example, as Lee and Huangs (in press) meta-analysis shows, 4 (31%) out of 13 studies where information on treatment duration was reported had a single treatment session lasting less than 30 minutes. The logistical constraints of conducting empirical research in real-life educational settings limited the time available for the study. Thus, any effects found in studies of short duration, and claims made based on them, as in Kim (2006), with regard to the teaching and research interests of the TESOL Quarterly readership, should be interpreted with caution. I make very clear in my article the context-specic nature of my pedagogical recommendations and the directions for further research. As the name of the research design used in my study, a posttest-only control group design, indicates, no pretest of participants prior vocabulary knowledge was administered in advance of the main data collection. Several factors can motivate avoidance of a pretest, one being a desire not to sensitize participants to a studys focus. Such was the case in my study, which involved an incidental learning condition. A careful reading of Kim (2006) would clearly show that the allegation of ineffective control for prior vocabulary knowledge is unjustied. Careful attention was paid to minimize this threat to internal validity (see the denition of retrospective vocabulary pretest, and footnote 16, p. 353, and footnote 17, p. 356). As in Hulstijn, Hollander, and Greidanus (1996), a fairly conservative measure of participants prior knowledge of the target words was used, in the form of a retrospective test.
TESOL QUARTERLY Vol. 41, No. 2, June 2007


As regards the effect of double treatment (i.e., enhancement of both a target form and its adjacent meaning in the form of lexical elaboration), the literature has shown evidence of a competition between form and meaning (see, e.g., Lee, 2007; Overstreet, 1998). A debilitating effect of typographical enhancement, however, has been reported not on acquisition of L2 target forms themselves, which involves acquisition of both form and meaning of the L2 targets in question, but on overall content comprehension of reading texts containing typographically enhanced L2 forms (S.-K. Lee, personal communications, February 27, 2007; March 1, 2007). In this respect, the double treatment proposed in Kim (2006) differs distinctively from previous studies in the same research domain and needs further investigation.

Youngkyu Kim is an assistant professor in the Department of Korean Studies, Graduate School of International Studies, Ewha Womans University, Seoul, Korea. His research interests include acquisition, instruction, and assessment of Korean and English as a second language, particularly task-based language teaching, language for specic purposes, reading comprehension and vocabulary acquisition, and applied corpus linguistics.

Hulstijn, J. H., Hollander, M., & Greidanus, T. (1996). Incidental vocabulary learning by advanced foreign-language students: The inuence of marginal glosses, dictionary use, and reoccurrence of unknown words. Modern Language Journal, 80, 327339. Kim, Y. (2006). Effects of input elaboration on vocabulary acquisition through reading by Korean learners of English as a foreign language. TESOL Quarterly, 40, 341373. Lee, S.-K. (2007). Effects of textual enhancement and topic familiarity on Korean EFL students reading comprehension and learning of passive form. Language Learning, 57, 87118. Lee, S.-K., & Huang, H. (in press). Visual input enhancement and grammar learning: A meta-analytic review. Studies in Second Language Acquisition, 30(3). Overstreet, M. (1998). Text enhancement and content familiarity: The focus of learner attention. Spanish Applied Linguistics, 2, 229258.



A Reader Responds to J. Jenkinss Current Perspectives on Teaching World Englishes and English as a Lingua Franca
Leeds Metropolitan University Leeds, England

There is much that one can agree with in Jennifer Jenkinss article in TESOL Quarterly (Jenkins, 2006). Some of her points are, however, quite controversial.


Jenkins (2006) refers to English as a lingua franca (ELF) as if it were a well-established variety of English with its own norms and regularities, similar in kind if not degree to so-called nativized varieties. In her opening paragraphs, she refers to the way World Englishes (WEs) have gained increasing acceptance, setting the scene for the parallel acceptance of ELF, which is merged with WEs in collocations such as WEs and ELF. Although Jenkins claims to be using WEs in its narrow sense of new Englishes in Africa, Asia and the Caribbean (p. 159), she goes on to write the whole article as if ELF were a part of WEs, a cohyponym to the superordinate term. She refers to the consensus on WEs and ELF that is emerging (p. 157) and repeatedly conates ELF and indigenized varieties, sliding from one to the other as if the phenomena described were the same or comparable. This weaving in and out of the two concepts appropriates whole swathes of argumentation to the rhetorical position of ELF (as perceived by Jenkins) even in cases where the scholars concerned have explicitly dissociated their work from her position (e.g., Kachru, 2005, pp. 211 220; Holliday, 2005, p. 164).

Jenkins (2006) suggests that teachers begin exposing less procient learners to a range of ELF varieties (p.174); it is important to ponder

the implications of this statement for classroom practice, bearing in mind the only indication she gives of what ELF varieties are is the list of unproblematic forms taken from Seidlhofer: e.g., She look very sad; a book who I like (Seidlhofer, 2004, p. 220). To describe such forms in a nonjudgemental way is a kind of descriptive linguistics and is a legitimate enterprise, but Jenkins occasionally conates descriptive grammar and pedagogic grammar. The fallacy of much native-driven corpus linguistics was that description and prescription were assumed to be the same thing (Widdowson, 2003, p. 88). Jenkins may be committing the same fallacy.


Jenkins (2006) argues that the pure form of ELF excludes users of English as a rst language (L1) from the description (p. 161). Indeed, the spirit of the native speaker haunts ELF by its absence; it is always there, hovering in the background, as a slightly malignant presence, exerting a norming effect (Leung, 2005, p. 128) and obliging users of English as a second language (L2) to defer to its dictates (Jenkins, 2006, p. 174): Natives are always lurking (Seidlhofer, 2002, p. 204). Jenkins (2006) does at one point accept that native speakers play a minor role in ELF; however, this small minority of L1 users will have to follow the agenda set by ELF speakers and use any language items which may have been codied in ELF (p. 161). The modal verb of obligation suggests that ELF as far as its L1 user interlocutors are concerned is a one-way street; Jenkins would apparently seek to promote Seidlhofers list of core items for L1 and L2 users of ELF alike. The very act of listing so-called common core items in a supposedly emerging variety of international English suggests the potential for codication of these forms, which will collectively constitute the agenda of the majority of ELF users, an agenda which all and sundry will have to follow. Although one can agree with Jenkins that it is unacceptable for the native English speakers to impose their agenda on ELF, it is equally unacceptable for the putative L2 users of ELF as an L2 to impose their agenda on users of ELF as an L1.


Jenkins (2006) confusingly refers to ELF both as an emergent phenomenon (p. 166) and as comparable to the emerging Englishes of the expanding circle (p. 167). This telescoping of the two terms blurs an important distinction. The distinction between emerging and emergent has

a crucial bearing on whether one sees ELF as a product which one can capture and codify or whether one sees ELF as a process, made up of a number of lingua francas depending on the purposes to which the language is put and thus elusive in terms of the possibility of codication.
emerging means in the course of development toward completion; emergent by contrast suggests a perpetual process in which movement toward a complete structure is constant but completion is always deferred. (Lantolf & Thorne, 2006, p. 14)

In emergent views of the utterance, a word absorbs the sense of preceding and subsequent words, thereby extending almost without limit the boundaries of its meaning (Vygotsky, quoted in Wertsch, 1991, p. 43). Lexis and grammar are, thus, emergent concepts, not xed abstractions; their structure is always deferred, always in a process but never arriving (Hopper, 1998, p. 156). An approach to language varieties as emerging, on the other hand, sees the development of language forms moving toward a xed or stable point, after which the language can be codied; this is the approach taken by Jenkins (2006), who sees the ELF train coming closer in the tunnel of native speaker norms: She clearly assumes that at some point (p. 161) ELF will stabilize and become institutionalized and codied; she sees this as happening possibly in another 15 years (p. 162). The point about ELF, then, is that it can seen by different scholars as both emerging and emergent. ELF is emerging because it may be moving toward its own norms, but we do not know yet what these norms look like, and it would be premature to try to capture it once and for all, to pluck the heart of its mystery; it is emergent because it is elusive, everevolving, and dialogic (Bakhtin, 1981).


Jenkins (2006) refers to ELF as a distinct variety and distinguishes it from English as a foreign language (EFL). She asserts that EFL users are those who are learning English for use in communication with native speakers (p. 161). This denition of EFL is unusual if we take into account both traditional denitions of EFL and more recent ones, all of which dene EFL as English learnt in a community where it is not an official language of communication (McArthur, 1992; Richards, Platt, & Weber, 1985; Strevens, 1977; Thornbury, 2006). Saville-Troike (2006) gives the standard denition of foreign language in contrast to second language as a language which is not widely used in the learners immeTHE FORUM 411

diate social context, but rather one that might be used for future travel or other cross-cultural communication situations (p. 188). SavilleTroike points out, in contrast to Jenkins, that learners of EFL have little opportunity to interact with native English speakers and are not likely to need to participate in a native speaker community (p. 101). Kramsch (2002), in her discussion of ESL and EFL, goes further in distancing EFL from an exclusive focus on native speaker models: There is little concern here with an approximation to any NS norm of language use (p. 8). This approach to EFL (diametrically opposed to that suggested by Jenkins) assumes that there are no borders as far as ELF is concerned: EFL, ESL, and ELF users will, at some time or another, all need to negotiate encounters with diverse interlocutors, including users of ELF as an L1.


Jenkins (2006) argues that exposing learners to their own English varieties will reduce their dependence on native-like varieties and at the same time reduce the linguistic capital that learners believe nativelike English to possess (p. 174). The term linguistic capital is associated with the work of Bourdieu (1991). Bourdieu sees language as part of a network of power relations which includes economic, social, and cultural forms of power. Economic capital is only one form of power; symbolic capital, of which language and culture are the main embodiments, is an additional form of power, and it affects our capacity to get access to economic and social power. The English language can be seen as symbolic capital in the hands of the colonial powers (Phillipson, 1992) or a weapon in the hands of the oppressed. Phillipsons (1992) rejection of the role of English internationally, though morally understandable, in practical terms leaves the status quo as it is; it leaves the power of English in the hands of the dominant elites. Jenkins (2006) does not reject English outrightshe offers a halfway house where, on the one hand, power structures remain infused with the common core grammar of standard English but, on the other hand, the resistance from the periphery has in its hands a broken weapon; Jenkins does not condemn her ELF users to voicelessness, but, in my view, she risks bringing them stuttering onto the world stage: English is a linguistic capital and we ignore it at our peril (Canagarajah, 2006, p. 205).

Luke Prodromou obtained his doctorate from the University of Nottingham, Nottingham, England. He is a visiting fellow of Leeds Metropolitan University, Leeds, England, and also teaches young learners in a private language institute in Thessaloniki, Greece.

Bakhtin, M. (1981). The dialogic imagination. Austin: University of Texas Press. Bourdieu, P. (1991). Language and symbolic power. Cambridge, England: Polity Press. Canagarajah, S. (2006). An interview with Suresh Canagarajah. In R. Rubdy & M. Saraceni (Eds.), English in the world: Global rules, global roles (pp. 200212). St. Leonards, New South Wales, Australia: Allen & Unwin. Holliday, A. (2005). The struggle to teach English as an international language. Oxford: Oxford University Press. Hopper, P. (1998). Emergent grammar. In M. Tomasello (Ed.), The new pyschology of language (pp. 155175). Hillsdale, NJ: Lawrence Erlbaum. Jenkins, J. (2006). Current perspectives on teaching World Englishes and English as a lingua franca. Tesol Quarterly, 40, 157181. Kachru, B. (2005). Asian Englishes: Beyond the canon. Hong Kong SAR, China: University of Hong Kong Press. Kramsch, C. (2002). Beyond the second vs. foreign language dichotomy. In K. Spelman Miller & P. Thompson (Eds.), Unity and diversity in language use (pp. 121). London: Continuum. Lantolf, J., & Thorne, S. (2006). Sociocultural theory and the genesis of second language development. Oxford: Oxford University Press. Leung, C. (2005). Convivial communication: recontextualizing communicative competence. International Journal of Applied Linguistics, 15, 119144. McArthur, T. (1992). The Oxford companion to the English language. Oxford: Oxford University Press. Phillipson, R. (1992). Linguistic imperialism. Oxford: Oxford University Press. Richards, J., Platt, J., & Weber, H. (1985). Longman dictionary of applied linguistics. Harlow: Essex. Saville-Troike, M. (2006). Introducing second language acquisition. Cambridge: Cambridge University Press. Seidlhofer, B. (2002). Habeas corpus and divide et impera: Global English and applied linguistics. In K. Spelman Miller & P. Thompson, (Eds.), Unity and diversity in language use (pp. 198220). London: Continuum. Seidlhofer, B. (2004). Research perspectives on teaching English as a lingua franca. Annual Review of Applied Linguistics, 24, 200239. Strevens, P. (1977). New orientations in the teaching of English. Oxford: Oxford University Press. Thornbury, S. (2006). An AZ of ELT. Oxford: Macmillan. Wertsch, J. (1991). Voices of the mind. Cambridge, MA: Harvard University Press. Widdowson, H. (2003). Dening issues in English language teaching. Oxford: Oxford University Press.

The Author Replies

University of Southampton Southampton, England

Luke Prodromous comments on my 2006 article in TESOL Quarterly are welcome in that they provide me with a good opportunity to draw attention to the various misconceptions of English as a lingua franca (ELF) that he and others who favour a traditional native-speaker normative view of English language teaching continue to hold. This I can entirely understand, given that ELF is, for many ELT professionals, such a new and complex concept. On the other hand, I was more than a little surprised by the extent of Prodromous misreadings of my 2006 article and, indeed, by what appears to be a certain amount of selective quoting of the views of others. In some cases, for example, those who hold a perspective on ELF similar if not identical to my own have been quoted in ways that make the opposite appear to be the case (e.g., Leung 2005; Seidlhofer 2002; Widdowson 2003). In the short space that I have been allocated to respond to Prodromous comments, I would not be able to do justice to either ELF or myself in respect of the number and seriousness of his misreadings of my (and other ELF researchers) words. Rather than attempting to deal with all of these in summary form, I therefore provide just one example as an illustration, and refer readers to Jenkins (2007), where they will nd such issues discussed at length and will be able to draw their own conclusions. Prodromou claims that in referring to ELF as both emergent and emerging, I have blurred an important distinction. But this is not so; rather, the problem is his own failure to grasp the fact that ELF, a phenomenon without precedent, does not t neatly into pre-existing categories predicated on the tired old dichotomy of native/nonnative Englishes. By its very nature and use, ELF is both emerging as more and more L1 groups learn and use English and emergent as speakers of English from this growing number of L1 groups shuttle between communities (Canagarajah, 2005, p. xxvi) and increasingly shape their lingua franca English in ways that accommodate the needs of the particular interlocutor of the moment. This is entirely different from Prodromous assumption that lingua franca uses of English, whatever their context, are broken and stuttering where they differ from English as a native language. Accommodation of this kind is, of course, in many ways similar to what

happens in communication among native speakers of English (or, for that matter, of any language). The difference is one both of scale (there are vastly more lingua franca than rst language speakers of English) and of resulting forms (the forms used by ELF speakers exist in their own right, not through deference to the forms of an absent native English speaker). Prodromou is thus also mistaken about the nature of what he considers to be real language varieties (from which he excludes ELF), which he apparently sees as moving toward a xed or stable point. The fact that a language can be codied at any point in its development does not mean that it is xed once and for all or that its speakers do not accommodate to each other in various ways.

Jennifer Jenkins is Professor of English Language in the Modern Languages Department at the University of Southampton, in Southampton, England, where she conducts research into ELF and its implications for language learning and use, and teaches courses and supervises doctoral research in World Englishes in general and ELF in particular.

Canagarajah, A. S. (Ed.). (2005). Reclaiming the local in language policy and practice. Mahwah, NJ: Lawrence Erlbaum. Jenkins, J. (2007). English as a lingua franca: Attitude and identity. Oxford: Oxford University Press. Leung, C. (2005). Convivial communication: Reconceptualizing communicative competence. International Journal of Applied Linguistics, 15, 119144. Seidlhofer, B. (2002). Habeas corpus and divide et impera: Global English and applied linguistics. In K. Spelman Miller & P. Thompson (Eds.). Unity and diversity in language use (pp. 198220). London: Continuum. Widdowson, H. G. (2003). Dening issues in English language teaching. Oxford: Oxford University Press.



TESOL Quarterly publishes brief commentaries on aspects of English language teaching. For this issue, the editor focused on the theme of theorizing language teacher education. She asked the authors to consider the following question: How do you theorize language teacher education in your professional community?


University of British Columbia, Vancouver, Canada

Constructing Pedagogical Awareness With Brazilian Language Educators

Universidade Federal do Rio Grande do Sul Porto Alegre, Brazil

Universidade do Vale do Rio dos Sinos So Leopoldo, Brazil

This brief article focuses on research issues connected with our activities in the research group Foreign Language Acquisition in the Classroom1 at Universidade Federal do Rio Grande do Sul, in Porto Alegre, Brazil. For the past 6 years, the research group has focused on corrective feedback, collaborative tasks, and identity negotiations in the classroom and their impact on interaction and learning. The research group maintains strong links with foreign language teacher education by participating in courses that promote training and further education to English teachers in the wider community in a critical perspective (Pennycook, 2001). In this article we report on reactions by teachers to two research studies recently developed by members of our research group (Longaray, 2005; Fontana, 2005). The studies are located in two different public

Projeto ALESA (Aquisio de Lngua Estrangeira em Sala de Aula), linked to the CNPq Research Group Teoria e Prtica de Ensino e Aquisio de Lngua Estrangeira, is coordinated by Dr. Lima at the Post-Graduate Program in Letters at Universidade Federal do Rio Grande do Sul, Brazil. Dr Fontana is also a member of the group.

TESOL QUARTERLY Vol. 41, No. 2, June 2007


schools in the metropolitan area of Porto Alegre, Rio Grande do Sul, in the south of Brazil. Both studies include careful observation of the contexts in which the foreign language is taught, relying on videotaped and audiotaped data, interviews, questionnaires, and researchers eld notes.

Interaction is seen as the key component of the language learning process. Learners should be made aware of their own skills and processes leading to the development of the target language, and the teacher should assist them in participating fully in meaningful activities to stimulate their investment in the language they are learning. In this context, Nortons (1997) denition of investment is of central importance. She uses the term to signal the socially and historically constructed relationship of learners to the target language and the ambivalent desire to learn and practice it (p. 411). Thus, during the training courses, the main questions asked of the teachers were, What is your actual investment in the teaching of the foreign language? What do you think is the actual investment of your students in the learning of the foreign language? Where is the foreign language located? What is this foreign language for? These training courses are opportunities for the teachers to experience what language education should be like and to display the different social identities they themselves bring to the classroom. Such identities are connected to contexts both outside and inside the school where they are working. We follow Nortons (1997) denition of identity also. She uses the term to refer to how people understand their relationship to the world, how that relationship is constructed across time and space, and how people understand their possibilities for the future (p. 410). We also refer to Canagarajahs (2004) notion of safe houses to understand learners complex behavior. Safe houses are spaces in the classroom where learners can negotiate their identities usually without being noticed by the teacher, such as through the written messages they send to each other.


Findings from both studies, especially those related to the concepts of identities and investment as dened by Norton (1997), were discussed with English teachers on two occasions in 2006 when we had the opportunity to lecture in two training courses. Each group had approximately 30 participants, whose ages ranged from 25 to 42. They had chosen to

attend the courses to improve their English knowledge and their teaching skills. The teachers work at public and/or private elementary and secondary schools, and some work in private language programs. Teachers at public schools usually complain of low salaries and their hard routine, sometimes working 40 hours a week or more, in different schools. They are also concerned about learners lack of motivation and complain of having too many students in the classroom. In addition, most schoolteachers have to deal with the low status of English teaching at elementary and secondary levels. In spite of all that, most of the participants attending our courses showed great interest in the studies brought to them for discussion. The concepts of identity, investment, and safe houses proved to be extremely relevant to their experience. These concepts and the related results brought by Longaray (2005) and Fontana (2005) include issues not usually considered on a systematic basis in the professional lives of schoolteachers. The participants had the opportunity to ask questions and express their opinions about the topics and ndings brought to their attention. For example, Regina had this to say:
Im not convinced of the importance of teaching them a foreign language. Besides, they do not want to learn English at all. (Regina, March 27, 2006)

This is a very common remark among public school English teachers. The linguistic imperialism (Phillipson, 2003) we are exposed to in countries like Brazil makes people believe that everybody should have the same investment in learning English. Even though the governmental education authorities aim at promoting citizenship by means of offering access to the wider world culture through an international language, the unprivileged social classes do not acknowledge this aim as their legitimate interest. However, few teachers are sensitive to these students lack of investment and do not understand that they lack investment because they also lack a general perspective for their future lives. The following excerpt also demonstrates this idea:
Investment is interesting. How can we convince the students they really need to invest in their English learning? First, we really need to feel like investing in good teaching. The problem is that we do not have enough support from our schools. They should invest in us. (Elena, March 27, 2006)

The symbolic capital (Bourdieu, 2004) represented by the mastering of English as a foreign language is highly valued in more privileged social classes, as Rosana claimed. At the same time, she condently displays her sense of identity, deconstructing the nativelike accent as the only desirable outcome for foreigners.

Now English is our language. I am proud of my Brazilian accent. It took me years and lots of money to learn English. Now I want my students to learn it as well. Its good for them. (Rosana, May 5, 2006)

However, students nonparticipation seems to be a very common issue raised among public school teachers, as expressed by Cludia:
My students are really good at these safe houses. Sometimes they think I do not see they are making faces at me. I tell them off right there. Anyway, they are teenagers. Its not easy. (Cludia, May 5, 2006)

In contrast with privileged learners who can afford private language courses and who can see the crucial role of English in their lives, most students in the public school setting do not see this connection. There is a gap between what learners want and what they get in the classroom. Those who foresee this connectionsuch as getting jobs for which English is necessary or traveling to English-dominant countriesdo not believe it is an attainable goal. As analyzed by Longaray (2005) and suggested by Norton (2001), a new theoretical lens (p. 160) is needed to fully understand the reasons for this lack of investment, considering the relevance of a foreign language, especially English, for public school students. As this article suggests, supported by other studies (e.g., Lima, 2004), we have found that sharing research ndings as well as theoretical constructs with teachers is highly productive for foreign language teacher education in Brazil.

The authors thank Bonny Norton for her suggestions concerning the nal version of this article.

Marlia dos Santos Lima is a professor of applied linguistics and English language at Universidade Federal do Rio Grande do Sul, Porto Alegre, Brazil. Her research interests include grammar teaching, focus-on-form instruction, and teacher training in English as a foreign language. Beatriz Fontana is a professor of English as a foreign language at Universidade do Vale do Rio dos Sinos, So Leopoldo, Brazil. Her research interests include critical pedagogies for foreign language teaching, and issues related to language, gender, and power.

Bourdieu, P. (2004). A economia das trocas simblicas. So Paulo, Brazil: Perspectiva. Canagarajah, A. S. (2004). Subversive identities, pedagogical safe houses, and critical learning. In B. Norton & K. Toohey (Eds.), Critical pedagogies and language learning (pp. 116137). Cambridge: Cambridge University Press. Fontana, B. (2005). Aquisio de ingls como lngua estrangeira em uma escola pblica: Jogos de poder, produo e reproduo de identidades. Unpublished doctoral thesis, Universidade Federal do Rio Grande do Sul, Porto Alegre, Brazil. Lima, M. S. (2004) A autonomia do aluno de lngua estrangeira e a correo de seus erros. In L. Rottava & M. S. Lima (Eds.) Lingstica aplicada: Relacionando teoria e prtica no ensino de lnguas (pp. 205224). Iju, Brazil: Editora da Uniju. Longaray, E. A. (2005). Identidades em construo na sala de aula de lngua estrangeira. Unpublished masters thesis, Universidade Federal do Rio Grande do Sul, Porto Alegre, Brazil. Norton, B. (1997). Language, identity, and the ownership of English. TESOL Quarterly, 31, 409430. Norton, B. (2001) Non-participation, imagined communities and the language classroom. In M. Breen (Ed.), Learner contributions to language learning: New directions in research (pp. 159171). Harlow, England: Pearson Education. Pennycook, A. (2001). Critical applied linguistics: A critical introduction. London: Lawrence Erlbaum. Phillipson, R. (2003). Linguistic imperialism. Oxford: Oxford University Press.



English as Cultural Capital in the Oaxacan Community of Mexico

Universidad Autnoma Benito Jurez de Oaxaca Oaxaca, Mexico

Currently in Mexican public schools, the formal instruction of English starts in secondary school. To prepare elementary school children for this transition, our TESOL student-teachers at the Universidad Autnoma Benito Jurez de Oaxaca carry out their practicum in public elementary schools. This is the most direct way in which we purposely promote English as cultural capital in our community of Oaxaca, Mexico1 (Clemente & Higgins 2005), which I exemplify here with two brief vignettes. I will add, however, a third vignette to illustrate the connections that preservice teachers establish outside the English classroom. This last vignette corresponds to an episode involving a major political event currently taking place in Oaxaca: the teachers strike.

Oaxaca, one of the poorest states of Mexico, has the greatest ethnic diversity, and, on average, Oaxacan children reach only fth grade of elementary school.



The rst scenario involves a small and crowded classroom of middle class students: 49 ten-year-old kids attending their English lesson. The chairs are put together in ve rows along the room, which means that the knees of the taller students touch the backs of the students in front of them. This arrangement, along with the uniforms they wear and the age of the students, makes the scene look very homogeneous. They have been in school since eight oclock in the morning. Their English class is at one oclock, when the room is quite warm. The English lesson develops according to what one would expect of a regular classroom routine. When the student-teacher arrives everybody stands up and greets her. She greets them back, asks them to sit down, and takes attendance. Afterward, she explains the simple-present verb tense and then hands out a sheet with some exercises. This task is followed by an activity in which a student is nominated to ask another student a question. The activity involves asking or answering questions about everyday events (Do you watch TV everyday? Yes, I do/No, I dont). Although the teacher moves from here to there calling on different students, many are left with their hands raised. Too many students and too little time, she says. The lesson ends with the students asking her to revise their individual assignments. In this middle-class context, the students participate, pay attention, and carry out the activities. They have an overall sense of accomplishment. It is also evident that the students already know some English, and they seem to be aware of the possibilities of demonstrating it. For them, English represents a form of linguistic capital that they are adding to their cultural capital. They know that this is what schooling is about. The second scenario depicts an urban working-class population in an afternoon school with only 12 students. They range from 1013 years old and their clothing styles are not very homogeneous. Although a school uniform is compulsory, these students wear permanently soiled, faded, and worn out clothing of different colors and designs. These students have either failed some grades, started school later than others, or were rejected from other schools. Some already work, so the afternoon is their only option for nishing elementary school. When the student-teacher arrives, the class is not very quiet. Further, the students seem to have no seating arrangement. The students are not sitting up straight to face the teacher. Some are kneeling on the seats while others are slouching over them. All of them, however, show interest once the teacher starts explaining the activity, while she takes a tape recorder out of a bag. A student interrupts her, speaking in Spanish: Is it yours? Where did you buy it? Its nice! The student-teacher politely tries to ignore his comments: Yeah, its mine but lets start the activity! The second activity requires some materials she has brought: magazine cuttings, watercolor paints, brushes, glitter, and other things. (Neither

the students nor the school can provide materials.) They end up with colorful half-nished projects showing images such as three doors, four babies, and six owers. The lesson ends with the teacher trying to get the remaining materials back and begging the students to clean their desks and hands while two kids throw glitter at each other. Although these students had a great time, I am not sure whether any English learning took place, or if that was important. These working-class students did not seem to share the same feeling about the necessity of learning English as did the class in the rst scenario. Their forms of seeking cultural capital were not per se about attaining the use of English. However, their student-teacher believes that if they learn some English at this stage, they will be competent in some basic English or simply capable of nishing secondary school. Lets consider a third scenario in which our student-teachers role as linguistic providers crosses the boundaries of the classroom: For various months the main square of the city has been occupied by striking teachers from the public schools, who demand higher salaries and better learning conditions for their students, and other groups seeking political and social change in Oaxaca. In this context, our TESOL student-teachers have discovered that, as well as expressing their political views, they can also help the teachers to communicate with non-Spanish speaking tourists who, apart from being a major economic resource of the state, could tell the world about this social struggle. For example, they help the striking teachers to translate their messages into English:



With messages such as this, the tourists seem to be more understanding and sympathetic to the teachers cause; and above all, hopefully, the tourists will go back to their worlds to tell the story of the Oaxacan peoples struggle.2 Thus, by helping the teachers communicate their cause to the world, the student-teachers use their English knowledge not only within the school settings, but also in the political events that are at the core of this current social struggle. To me, these language activities suggest the kind of social diversity that is framing these pursuits of cultural and symbolic capital (Bourdieu, 1991). Bourdieus discourse on class distinctions can be read as claiming that funds of cultural and symbolic capital are xed quantities associated with particular class positions (Bourdieu, 1984; Shusterman 1999). In this view, those located in the upper spheres of the social system have pre-existing forms of cultural and symbolic capital, but those in the lower spheres lack them. A different reading is possible from these small vignettes. In the middle-class context, student-teachers are helping to reproduce already existing forms of cultural capital; whereas, in the working-class context, they are exposing students to new forms of cultural capital. Moreover, the student-teachers are using their particular cultural and symbolic capital to aid striking teachers in a profoundly political struggle. These examples show that the processes of cultural and symbolic capital are not static locations but expressions of the diversity found in different social class perceptions and actions (Garcia-Canclini, 2004). In this way we make a humble connection between English language use and the Oaxacan community.

Angeles Clemente works as a full-time lecturer and researcher at the Universidad Autnoma Benito Jurez de Oaxaca, Oaxaca, Mexico. She has published research on theoretical and practical issues about autonomy and self-direction, teacher education, and beliefs. Her current interest is ethnographic research on the construction of learning cultures, identities, social agencies, and critical stances in the context of English language learning and teaching.

Bourdieu, P. (1984). Distinctions: A social critique of the judgment of taste. Cambridge: Harvard University Press. Bourdieu, P. (1991). Language and symbolic power. Cambridge: Polity Press.

Ulises gets out of here refers to the strikers demand for the removal of the governor of Oaxaca.



Clemente, A., & Higgins, M. (2005). Whose English is it anyway?: Language, culture and identity. Papeles de trabajo sobre cultura, educacin y desarrollo humano. Madrid, Spain: Departamento Interfacultativo de Psicologa Evolutiva y de la Educacin, Universidad Autnoma de Madrid. Retrieved September 12, 2006, from http:// Garcia-Canclini, N. (2004). Diferentes, desiguales y desconectados: Mapas de la interculturalidad. Barcelona, Spain: Gedisa Editorial. Shusterman, R. (Ed.). (1999). Bourdieu: A critical reader. London: Blackwell.




This section presents brief synopses of empirical research and theoretical discussions in peerreviewed journals. The aim is to disseminate ndings and perspectives in elds related to TESOL and to provide access to the diverse conversations among scholars in the eld.


Rutgers, The State University of New Jersey

Its Not My Job: K12 Teacher Attitudes Toward Students Heritage Language Maintenance. Bilingual Research Journal, 30, 453477. Jin Sook Lee and Eva Oxelson, 2006.
Is linguistic diversity constructed as a resource among K12 teachers? Lee and Oxelson investigate how teachers trained in ESL and bilingual cross-cultural language and academic development (BCLAD) and teachers not trained in ESL and BCLAD understand the role of heritage language maintenance in K12 schooling. Their study found, through a survey of 69 teachers and an in-depth interview with 10 teachers, that teachers with BCLAD or ESL training had distinctly different views on the role schools should play in heritage language maintenance from teachers without BCLAD or ESL training. BCLAD/ESL teachers reported making heritage language maintenance an important part of their teaching practice and believed that by supporting heritage language maintenance their students would have a have a strong ethnic identity as well as strong family values (p. 461). However, Non-BCLAD/ESL teachers believed that the primary job of school was to teach English and believed that heritage language maintenance was not their job. Many of the non-BCLAD/ESL teachers advocated that multilingual parents speak English at home with their children. Another striking difference was that non-BCLAD/ESL teachers believed that their students could either learn English or maintain their heritage language, but not both. Sook and Oxelson suggest four issues that future research and teacher education programs should address: (a) examining the extent to which teacher training can shape teacher attitudes and practice, (b) supporting second language acquisition or multilingualism in all K12 teacher education programs, (c) repositioning current assessment policies, specically No Child Left Behind, so that they do not hold teachers accountable for students English acquisition
TESOL QUARTERLY Vol. 41, No. 2, June 2007


at the same time as content learning, and (d) creating synergy among schools, teachers, parents, and communities to support heritage language maintenance.

Recasts in Adult English L2 Classroom: Characteristics, Explicitness, and Effectiveness. Modern Language Journal, 90, 536556. Shawn Loewen and Jenefer Philp, 2006.
How are recasts used and taken up in ESL classrooms? Recasts are a fairly gentle and inconspicuous way for teachers to correct errors while maintaining the focus on meaning and the ow of the interaction. Loewen and Philp bring a nuanced examination to recasting in a traditional (face-to-face) adult ESL classroom. They found that 50% of all feedback takes the form of recasts. The recasts described were short and delivered with declarative intonation (p. 548). Teachers stressed the problematic form produced by the student and gave equal time to recasts of phonologic, morphologic, syntactic, and lexical forms. The authors also examined how recasts were taken up by second language (L2) learners. The adult ESL students successfully took up elicitations (83%) and recasts (60%). Informs (46%) had a lower rate of uptake. During the posttest, students were able to reproduce the recast 50% of the time. However, the accuracy in recalling different types of feedback was not signicantly different. The authors found that recasts had a benet in connecting learner response and test performance. The authors also reported that recasts with interrogative intonation and stress, recasts within extended focus-on-form episode, and recasts with only one change were predictive of successful uptake. However, learner uptake of recasts is not an indication of learning. Rather, interrogative recasts, shorter recasts, and recasts with only one change were predictive of accuracy in learners posttest performance. The authors suggest that the closer the recast is to the L2 learners initial utterance, the more likely the learner will be able to remember and reproduce the target language. In this case, explicitness does not appear to carry the most weight.

Corrective Feedback in the Chatroom: An Experimental Study. Computer Assisted Language Learning, 19, 114. Shawn Loewen and Rosemary Erlam, 2006.
How are recasts used and taken up in a chatroom? Loewen and Erlam also explore recasts, but in their study, the social context of the interac428 TESOL QUARTERLY

tions becomes key in understanding the role of recasts (implicit feedback) or metalinguistic information (explicit feedback). Loewen and Erlam examine the way that recasts are used and taken up in a synchronous computer-mediated communication (CMC), which pushes their empirical ndings into another, evolving language-learning context. Loewen and Erlam compared how previous research on recasts in traditional (face-to-face) classrooms transfers to L2 learning online by replicating previous research based on face-to-face classroom interaction to a synchronous CMC chatroom. The authors had elementary English learners complete two communicative tasks using past tense and found that the corrective feedback used during online meaning-focused interaction did not increase the learners performance on timed and untimed grammatical judgment tests. Additionally, the authors found that there was no difference in effectiveness between implicit and explicit feedback during online chatroom interaction. Loewen and Erlam examine why there was no signicant learning as a result of error feedback and suggest that the learners may not have been ready to learn past tense. However, the students in the control group (in a face-to-face study) made more errors per feedback session than those in the CMC session and received slightly less feedback than their classroom counterparts. The CMC learners also had several lines of text between the error and the feedback; thus, they did not receive immediate feedback and may not have perceived it as feedback in relation to the given error. In the face-to-face context, feedback was immediate in relation to the error and the group focused on the teacher student interaction. The CMC students had a reduced incidence of uptake in response to feedback. Most of the CMC context groups had instances of students chatting off-topic and the instructor struggling to get the whole group back on task. Loewen and Erlams research shows that even learners at a beginning level of L2 prociency are able to perform communicative tasks in the chatroom and suggests that CMC will continue to be a fruitful venue of teaching and research for learners at all levels.

From Content to Context: Videogames as Designed Experience. Educational Researcher, 35(8), 1929. Kurt Squire, 2006.
How can studying video games help us understand the evolution of learning? Squire suggests that digital video games are both a new medium for learning and a designed experience that educators should consider. With a focus on experience Squire examines the ways that

learners (a) develop problem-solving identities, (b) learn through failure (p. 26) with a low affective lter, and (c) construct situated meanings through interaction. Although Squire does not directly address TESOL or English language learners in his theoretical discussion, his designed experience framework can be applied to language learners. Squire considers two types of video games: exogenous games and endogenous games. Exogenous games, such as Math Blasters, present prescribed sets of skills and facts that the learner must transmit and memorize. These types of games are very common in language labs. The context of learning focuses on individual knowledge production and places predetermined skills and facts in a digital frame. Endogenous games, such as the Sims or Civilizations, create space where learners use a set of tools to solve problems and learn by experimenting, discovering, and constructing meaning through social interaction. Endogenous games draw on collaborative interaction. At the same time, Squire warns that learners involved in role-playing games, such as Civilizations and Grand Theft Auto, are enacting and interacting with the worldviews (cultural models) that the game espouses. If the player does not enact the games preferred cultural models, the player will not be successful. Squire suggests that teachers should closely examine the intersection of the design constraints and the players intentions (p. 26). The learning culture of video games in many ways is opposed to the dominant discourse of schooling. As groups such as the U.S. military invest in video gaming technology for both recruiting and training, Squire implores K12 and adult education to consider using this powerful new medium as a way to maintain inuence.



TESOL Quarterly welcomes evaluative reviews of publications relevant to TESOL professionals.


Simon Fraser University

Teaching English to the World: History, Curriculum, and Practice.

George Braine (Ed.). Mahwah, NJ: Lawrence Erlbaum, 2005. Pp. xx + 191. George Braine effectively captures the rich diversity of English language teaching (ELT) around the world. This text, a compilation of 15 brief chapters written by nonnative-English-speaking (NNES) teachers from different countries, provides the reader with a historical and curricular overview of English language education in each of the authors homelands. The result is an informative, insightful, and pleasantly readable look at the past and present state of English language education through the eyes of true insiders, a view that is often underappreciated and virtually absent in the literature today. Each chapter is comprised of three components: a history of ELT in the featured country, a concise summary of the English language curriculum at all levels, and a biographical sketch of one educatorusually one of the authors of the chapterwho is a native of that country (and consequently a nonnative speaker of English). Taken together, these three sections paint a vibrant picture of the development and current structure of ELT as well as the many obstacles which teachers in each country face. The book would be a valuable addition to the library of any ELT professional if only for the historical sketches of ELT throughout the centuries. To provide a few examples, the rst recorded English classes were taught in Germany as early as the Middle Ages for reasons of trade (the beginnings of English for specic purposes), and in other nations (e.g., Brazil and Hungary), English language education is a more recent development. These historical summaries also consider the original purposes of ELT in each setting: trade (Saudi Arabia), missionary activity (China, Singapore), colonization (Hong Kong, India), restructuring the national education system (Hungary, Poland, Turkey), and replacing other second languages (French in Lebanon and Brazil, Russian in PoTESOL QUARTERLY Vol. 41, No. 2, June 2007


land and Hungary, and Dutch in Indonesia). In each chapter, the authors are careful to consider the political, economic, cultural, and pragmatic factors involved in foreign language education. The reader comes to appreciate the complexity of such decisions and the difficulties of implementing these changes at the national level. Equally worthwhile are the sections on the ELT curricula of each nation. The contributors do an admirable job of tackling the daunting job of distilling the whole of their nations foreign language curricula (at the preschool, elementary, secondary, and university levels) into a few well-crafted, informative pages. Once again, the complexity and avor of differing national approaches to ELT come to the forefront. Educators in the 15 countries recount the shifts in teaching approaches (away from grammar translation and audiolingual to communicative and competency based), national policy and political agendas (from English as a desirable language to study to one that was banned or severely limited and back again to favorable status), and utility (from a language studied as a novelty to one learned out of necessity). These sections also devote considerable space to related issues such as teacher training and professional development; teacher qualications, the quality of language instruction, and the view of native speakers as de facto qualied educators; the status of English in the nation as an official, second, foreign, or auxiliary language or lingua franca; and the scarcity of resources that plague many national education systems. These curricular summaries will be of greatest interest to researchers at all levels, but Englishlanguage teachers in any setting should nd these concise insiders views informative and may even empathize with their colleagues from other parts of the world. Each chapter ends with a biographical vignette of one of the authors. Although these vignettes may strike some as a superuous add-on, the result is quite the opposite. The subject of each sketch is an NNES who grew up in the corresponding country and who became a teacher of English, sometimes in spite of their circumstances rather than because of an early dream of entering the profession. Some authors relate inspirational tales of the impact of motivating teachers they had early in their educational careers (Poland, Saudi Arabia), and others describe great personal obstacles and discouragement that they overcame to attain their goal of becoming an English teacher (China, Indonesia). What is perhaps most impressive about these brief personal encounters is the common thread of ongoing professional development. As we learn of each persons struggles and successes with English as an object of study and later as a subject of instruction, we discover the different routes each person took to acquire academic credentials (very often a doctorate in applied linguistics or related eld) as well as a high level of communicative and cultural competence in the language.

In a book of just under 200 pages, it is easy to question its breadth and depth on such a complex topic. This is a valid criticism of this particular book, but such an observation also requires the following qualication: The chapters are short and concise by design because a text of even twice the length would be subject to the same criticism. What is more, each chapter ends with a short bibliography for the interested researcher different points of departure for examining a countrys ELT curricula or educational history in greater depth. A second shortcoming of the book is anticipated by the editor, George Braine, in the introduction: No African country is represented. It seems that, despite his best efforts, he was unable to identify authors from African countries who were willing or able to write a chapter on the state of ELT in their home nations. Braine hopes that a future edition will have even broader geographical representation. Teaching English to the World is an engaging text on many fronts. First, it offers a concise overview of the history of ELT and current curricular features from 15 vastly different countries. Second, it introduces the reader to several NNES colleagues, each with an admirable story and unique window on the world of English language teaching. Finally, the book gives a much-needed forum for NNESs who conduct refreshingly original research on ELT issues to communicate their ndings to their counterparts around the world. The texta brief, informative, and enjoyable glimpse into ELT in various countrieswill be a nice addition to libraries of ELT researchers, educators, program administrators, and graduate students.
DENNIS BRICAULT North Park University Chicago, Illinois, United States



Language Planning in Higher Education: A Case Study of Pakistan.

Sabiha Mansoor. Karachi, Pakistan: Oxford University Press, 2005. Pp. xvii + 445. Sabiha Mansoor is the pioneer of sociolinguistic research in Pakistan. In 19901991, she surveyed the attitudes of the students in Lahores educational institutions toward English, Urdu, and Punjabi. This data were published as a book titled Punjabi, Urdu, English in Pakistan: A Sociolinguistic Study (Mansoor, 1993). It is still a landmark in this type of research in the country. The book under review is the authors doctoral thesis submitted to a

British university in 2003. The major aims of the study were to (a) investigate the language prociency of learners in different medium(s) in school and college and (b) to determine the level of success of students from different backgrounds in higher education. The author tests 18 hypotheses relating to the medium of instruction, the language actually used in classrooms and the availability of English, and other aspects. Besides the introduction, which expatiates on this objective and places it in the context of relevant academic work, the book has six chapters, the last of which summarizes the research ndings and gives a conclusion. The other ve chapters are devoted to the development of the main arguments. The book also includes lists of tables, appendices of data, and an extensive bibliography. These items increase the value of the work because so much data are generally not available in one source. Mansoor carried out her survey through questionnaires administered to students, teachers, and parents. In addition, she interviewed administrators and prominent persons connected with education. The subjects were chosen through two-stage cluster sampling. In the rst stage, educational institutions were chosen, and in the second, students were chosen from each institution. The sample size was 2,450 students from 35 educational institutions distributed in the capital cities of the four provinces of PakistanKarachi (Sindh), Lahore (Punjab), Peshawar (Northwest Frontier Province), and Quetta (Baluchistan). Teachers were also selected for the sample from the clusters (education institutions) located in these cities. The sampling was carefully designed to represent an array of colleges and universities in Pakistan. On the basis of this survey Mansoor presents her conclusions and recommendations. The major conclusion is that the desire to learn English by all groups of students is very high. This result is not surprising considering that power and prestige within the country as well as abroad is available only if one knows English. It is also not surprising that the regional languagesby which she means the indigenous languages of Pakistan such as Punjabi, Sindhi, Pashto, Balochi, and so forthare seen as decient for educational purposes (p. 241). After all, no government of Pakistan has developed them or given them any respect. But what is surprising is that Sindhi speakers use Urdu more in the informal domain (77%) than Sindhi (25%) (p. 242). Because the sample had only 4% Sindhi speakers and Sindhi speakers constitute 14.10% of the population, the sample may not have been representative of Sindhi speakers, though the author obviously worked hard to represent them by including Hyderabad, a largely Sindhi-speaking city, in her sample. (A nonrandom survey of 153 Sindhi students of 10th class in 2000 indicates much higher affinity for Sindhi; see Rahman, 2002, 592593). Mansoors recommendations about the language policy to be adopted in higher education need to be seen from the historical, political, and

economic perspectives. She cites what appear to be democratic reasons such as the demand by stakeholders for strengthening English at the higher educational level. In this context she says that the attitude of students toward English may have changed radically from ambivalence and hostility to an enthusiastic acceptance (p. 342). In fact, apart from some protest against elitist English-medium schools during the sixties, Pakistan never had any signicant level of hostility to English. English was never part of the anticolonial movement in South Asia even though Gandhi did try to recruit it in his anticolonial rhetoric. This general acceptance of English is not democratic consensus. It is merely the pragmatic acceptance of reality by the common people who are resigned to their fate and feel that because the market conditions which privilege English will not change, they should acquire it somehow. This acquiescence of the people should not blind us to the fact that English does privilege a narrow elite; it does make it difficult for the underprivileged to seek elitist employment and social prestige and it, along with Urdu, puts pressure on the languages of the people that are becoming weak and some of which may disappear soon. These problematic issues are not given the attention they deserve, though Mansoor does suggest that regional languages should be taught. But students see learning languages as a burden if learning them leads neither to a job nor to prestige. So, if the present policy of privileging English is to continueas the authors recommendations suggestits price would be paid by the common people of Pakistan and their languages. Essentially, Mansoor does not suggest any radical change in the present language policy. She presents four policy options and chooses one which appears to give a choice between Urdu and English as mediums of instruction at the school level (p. 353) and the strengthening of English at the higher education level (p. 257). This so-called choice is meaningless because English-medium schools are expensive and poor people simply cannot afford to send their children to study in them. Contrary to Mansoors claim, this policy does not aim at democratization and empowerment of the masses (p. 357). It may, of course, mean that English will be taught more prociently at the higher education level, but that does not change the fact that the children of the rich and the powerful will arrive at this level, just as they do now, with far greater prociency in English than their vernacular-medium counterparts. I have claried and emphasized my own disagreement with Mansoors recommended language-teaching policies because of the bearing it has on the language-teaching policy in Pakistan. I must also emphasize the fact that Sabiha was a pioneer in research on attitudes of Pakistanis toward languages, and this book establishes her as the leading authority in this eld. This book is, therefore, useful for many reasons. First, it is the rst major survey of language attitudes, with focus on English, in the

country. Second, it is the only major survey of the educational needs of students, teachers, parents, and administrators. And, third, it provides a major source of analysis and discussion of documents on language and education policies. For these reasons it will long remain a major landmark in educational linguistics in Pakistan.

Mansoor, S. (1993). Punjabi, Urdu, English in Pakistan: A sociolinguistic study. Lahore, Pakistan: Vanguard. Rahman, T. (2002). Language, ideology and power: Language-learning among the Muslims of Pakistan and North India. Karachi, Pakistan: Oxford University Press. TARIQ RAHMAN Quaid-i-Azam University Islamabad, Pakistan



Theory Construction in Second Language Acquisition.

Geoff Jordan. Amsterdam: John Benjamins, 2004. Pp. xviii + 294. In this book, Jordan provides a comprehensible exploration of the epistemologies in which second language acquisition (SLA) is grounded and from which it has evolved. Jordan opens his work with a set of guidelines. These guidelines are the principles that conduct our progress from Descartes and the paradigm of the Enlightenment to Poppers notion of falsication. Taking a rm stand on the side of rationalism, Jordan takes us on a journey among squabbling factions of scholars offering his readers the possibility to examine all theories, articulate the problems found in each, and then y any kite we like (p. 265). As we make the journey through general learning theories as well as theories of SLA, Jordan serves as an able tour guide who skillfully displays points of interest, but who periodically focuses our attention on the guidelines lest we lose our way in the labyrinth of epistemologies. A brief summary of the six guidelines appears immediately following the abstract. Jordan proposes six assumptions and offers ve criteria for the evaluation of SLA . I have taken the liberty of listing only the kernel of these assumptions and theories. The assumptions: 1. An external world exists independently of our perceptions. 2. Research is inseparable from theory.

3. 4. 5. 6.

Theories attempt to explain phenomena. Research is fundamentally concerned with problem solving. We cannot formalize the scientic method. There is no need for paradigmatic theories.

The criteria: 7. Research, hypotheses, and theories should be coherent, cohesive, expressed in the clearest possible terms, and consistent. 8. Theories should have empirical content. 9. Theories should be fruitful. 10. Theories should be broad in scope. 11. Theories should be simple. Starting with an account of the basic problem in the proliferation of SLA theories and expressing a need to examine the underpinnings of SLA research, Jordan uses these guidelines to examine and evaluate the contribution to SLA of an astonishing variety of principles, techniques, and epistemologies, none of which completely measures up to every one of the proposed guidelines. As a fascinated tourist in Jordans epistemological world, this reader headed for familiar landmarks and was rewarded by Jordans ability to both expand and contract information. Writing about Chomsky, for example, Jordan begins with a six-page summary of generative grammar (UG), follows this review with a succinct discussion of minimalism and internalism, gives a clear explanation of why UG continues to be relevant, poses the arguments of Chomskys critics, and moves on to summarize Chomskys central claims. This reader missed a discussion of Chomskys controversial notion of the idealized native speaker, a paradigm that she has observed to be an issue of concern in multicultural circles, where it seems to have been replaced with the model of a successful language user, and she assumes, perhaps unjustly, that Jordan would place such a switch among his much maligned relativist views. Jordan agrees with Larsen-Freeman and Long (1991) and Cook (1996), who note that although UG offers a clear and acceptable vision of how language operates, it nevertheless does not really offer much support for SLA because it is not clear how much of UG is actually accessible to second language learners. Speaking on a gut level, however, this reader, who herself is a learner of several languages, instinctively feels that there is quite a bit of UG there during the entire human lifespan. When it comes to Krashen, Jordan promotes the frequently voiced criticisms against the power of Krashens intuitively appealing theory by

noting the fuzziness of the dichotomy between learning and acquisition, the inadequacy of the input theory, and the lack of clarity in the notion of the affective lter. Drawing on Nicola (1991), Jordan does note the considerable support for Krashens explanatory powers. To this reader, however, the primary appeal of Krashen lies not in his faculty to produce a coherent theory but in his ability to label phenomena that speak so effectively to language learners and teachers. As a teacher educator, this reader has constantly seen the faces of her multilingual teacher-learners light up as they were presented with I + 1, the monitor model, and the affective lter. Be that as it may, Jordan sees Krashen as failing to satisfy Guidelines 9 and 10, and we proceed on our journey of professional tensions to Ellis and Tarone, whose work on interlanguage does not meet Jordans expectations of clarity in Guideline 7. This reader must admit that she eagerly awaited the one theory, that pearl that would satisfy all the guidelines, and Jordan did seem to approach satisfaction with the work of Schmidt (p. 216). Schmidt in the analysis of his own struggles with acquisition of Portuguese poses an interesting theory of learning awareness through noticing. But here again, Jordan pulls the rug from under his readers textual feet. Schmidt does not satisfy the demands for empirical content in Guideline 8 because he offers a porous denition of noticing. Although Jordan does not lead us to the Holy Grail in his journey through general theories of learning and theories of SLA, and although some readers might nd his journey circuitous, this reader found the book interesting in its encompassing approach and would recommend it as review text in a course on SLA.
Cook, V. J. (1996). Second language learning and language teaching. London: Arnold. Larsen-Freeman, D., & Long, M. H. 1991. An introduction to second language research. Harlow, England: Longman. Nicola, M. (1991). Theories of second language acquisition and of physics: Pedagogical implication. Dialogue on Language Instruction, 7, 1727. NATALIE HESS Northern Arizona University in Yuma Yuma, Arizona, United States



Bilingualism and Language Pedagogy.

Janina Brutt-Griffler and Manka M. Varghese (Eds.). Clevedon Hall, England: Multilingual Matters, 2004. Pp. v + 145. In an increasingly globalised world, bilingualism is more the norm than the exception. However, for too long, bilingualism has been viewed

as a subdiscipline of areas such as linguistics and applied linguists or psychology and education, where the bilingual speaker is interpreted as performing a balancing act between two binary opposites, two disconnected language systems. The editors, Brutt-Giffler and Varghese, argue that bilingualism should be accepted as a discipline in its own right, and they have put together a very welcome and insightful collection of articles that deal with some of the practical pedagogical issues surrounding bilingualism. In their view, culture and worldview are essential elements in planning for bilingual education because the bilingual speaker is involved in ongoing negotiation and mediation between his/her languages, cultures, and worldviews. Therefore, bilingualism should be viewed as a continuum rather than as negotiation between binary opposites. The eld of bilingualism, as the editors see it, has become more and more politicised, pitted as it is against essentialist notions of nationality, culture, and ethnicity, and involving issues far wider than that of communication (p. 2). The book draws together arguments from studies of curriculum, pedagogy, teacher training, and linguistic imperialism in an invaluable contribution to a vital issue. The articles in the collection illustrate the necessity of widening the debate beyond such essentialist interpretations by looking at vital components of the bilingual speakers world: denition of the language needed for survival and success academically in the dominant culture, positive aspects of being a bilingual speaker as well as teacher and student issues in a number of bilingual contexts. Valdess discussion, though centred on the U.S. environment, has relevance for all other educational situations drawing attention as it does to the multiple denitions of academic language. Problems stem, she argues, from a lack of clear consensus on what constitutes academic language, little or no communication between those involved in English language teaching, and a very different approach to the academic language requirements for speakers of English as a rst (L1) and as a second language (L2). For L2 speakers of English, the focus may never move beyond basic interpersonal communication skills and stylistic conventions, but academic language courses for L1 English speakers focus on the necessary cognitive language prociency required for success in the academic forum. Lack of a clear denition of academic language has repercussions throughout the school system. In suggested solutions, Valdes identies a number of areas that involve a clear denition of academic language, along with an efficient way of evaluating student and teacher prociency. Interlanguage studies have long viewed features of the developing language of the L2 English learners in a positive manner, an approach which Toribio argues should now be applied to the code-switching patterns used by bilingual speakers of Spanish and English in the United

States. Her research indicates that far from illustrating deciency, illicit language acts such as code-switching signal the strategic and efficient use of linguistic and cognitive resources in the appropriation and management of two language system (p. 42). This approach has both political and pedagogical implications, preferring as it does to view such attempts at communication as a form of empowerment and indicating rule-governed and systematic linguistic behaviour. Four contributions discuss teacher training as a vital component of widening the debate on bilingualism. Hornberger, along with advocating a continuum model of biliteracy and bilingualism, also outlines how this model can be achieved through teacher education. The University of Pennsylvanias Graduate School of Education programme, she argues, stresses the inclusive and contextualized nature of communicative competence. She discusses issues such as context, content, and media of biliteracy and deals with pertinent points involving topics such as global and local dilemmas, language imperialism, and language content. Creese describes a British situation where teachers of English as an additional language (EAL), often subject trained, work alongside a classroom subject teacher to make the curriculum accessible to all students in statutory education. Subject teachers may not be able to facilitate language development in the classroom, and this role also falls to the EAL teacher, though problems can arise if the EAL teacher is viewed only as a helper. The question of teacher identity may become a problem. The bilingual EAL teacher may use the students L1 to explain the lesson and thus create identity issues for the subject teacher. Morgans article explores teacher identity as a pedagogical tool in its own right. In a poststructural educational world, teacher and student understanding can be negotiated through practical activities in the classroom, and his description of his own experience with a group of Chinese students is both entertaining and enlightening. His image-text was a learning source for himself as well as his students but he cautions that such an approach should be unthreatening and respectful and should be always open to critical analysis and reinterpretation (p. 92). Benson discusses the context of developing countries, where the language of education often differs from the language spoken at home. Such situations present a distinct challenge to the bilingual teacher. Mozambique and Bolivia are cited as examples. Both countries have had educational systems that use the colonial, exogenous language for populations the majority of whom do not speak this language at home. Mozambique has at least 24 indigenous languages, and only about 24% of Mozambicans speak Portuguese as a rst, second, or foreign language. Knowledge of the colonial language has always given the elite group a position of superiority over those less versed in that language. It is therefore essential to build an inclusive educational system that provides op440 TESOL QUARTERLY

portunities to the less well off and at the same time accommodates their own indigenous languages. Similarly, Bolivia, with 33 indigenous language groups, faces problems even though about 50% of the population speaks Spanish as a rst or second language. Demands on bilingual teachers are tremendous in both contexts. Teachers with little formal training and no role models are expected to provide instruction in two languages. Challenges to be faced include looking outside the colonial frame of reference and developing a more organic, local approach that includes coordination, cooperation between teachers, team teaching, more classroom aides, and activities that encourage language acquisition. Overall, this book provides a vital contribution to the more practical issues of bilingualism and widens the scope of the discussion to include important political and pedagogical questions. The articles should be of interest to those already working in ESL situations in the monolingual world of the United States, the United Kingdom, Australia, and Ireland and should also be essential reading for a teacher preparation course.
JOSEPHINE OBRIEN Sandford Language Institute Dublin, Ireland