Should Vocabulary Instrucion Be Integrated or Isolated?

In one of the treatments participants weretaught the words in isolation prior to reading the article, and in the otherthe vocabulary instruction was integrated with reading the article.P

Should Vocabulary Instruction Be

Integrated or Isolated?

KIERAN ANDREW FILE

Victoria University of Wellington

Wellington, New Zealand

REBECCA ADAMS

University of Auckland

Auckland, New Zealand

This study compares integrated and isolated form-focused instruction forvocabulary development in an English as a second language (ESL)reading lesson Two classes of ESL learners (N 5 20) from a universitypreparation academic English course were involved in the study Eachclass did two reading treatments in which they read an article and studiedvocabulary from that article In one of the treatments participants weretaught the words in isolation prior to reading the article, and in the otherthe vocabulary instruction was integrated with reading the article.Paribakht and Wesche’s (1997) vocabulary knowledge scale was used asthe instrument to measure learning and retention gains for words taught

in isolated instruction, in integrated instruction, and words acquiredincidentally Statistical analysis showed both types of instruction led tomore learning and retention of vocabulary knowledge than incidentalexposure alone Although retention rates were similar for isolated andintegrated instruction, there was a trend for isolated instruction to lead tohigher rates of learning

doi: 10.5054/tq.2010.219943

S everal empirical studies (e.g., Norris & Ortega, 2000) have indicatedthat form-focused instruction (FFI) is beneficial for languagelearning However, as Spada and Lightbown (2008) have recentlypointed out in a review of FFI research, there is little agreement as tohow FFI should be implemented in meaning-focused language-learningcourses In their review, Spada and Lightbown make the primarydistinction between two types of FFI, isolated and integrated In isolatedFFI, attention is given to form in instructional segments that are separatefrom communicative language practice but are still carried out incourses that are primarily meaning focused, for example, content-basedinstruction and communicative language teaching Integrated FFI differs

in that attention to forms is carried out during a communicative activity

in content-based instruction and communicative language teaching

programmes (such attention has alternately been called focus on form; cf.,Ellis, Basturkmen, & Loewen, 2001a) Spada and Lightbown consideredthe role of both kinds of FFI based on features of the learners, learningenvironments, and language items They suggest that these types of FFImay play complementary roles in language learning and call for furtherresearch exploring the roles and contributions of these two types of FFI.The current study provides a preliminary investigation of isolated andintegrated FFI in the context of vocabulary teaching in a meaning-focusedreading lesson, comparing isolated and integrated FFI with incidentallearning of new vocabulary According to Laufer (2005), vocabularylearning has been ‘‘neglected’’ (p 225) by focus-on-form researchers whohave tended to focus on the role of FFI on grammar learning or on theeffect of corrective feedback (see, e.g., Ellis et al., 2001a; Ellis, Basturkmen,

& Loewen, 2001b; Lightbown & Spada, 1990; Loewen, 2005) However,research has consistently found that vocabulary items make up a largeproportion of unplanned focus on form (e.g., Ellis et al., 2001a, p 290; Ellis

et al., 2001b, p 407), highlighting the importance of investigating the waythat different types of FFI might enhance vocabulary learning

VOCABULARY LEARNING IN READING LESSONS: FFI VERSUS NO FOCUS

Vocabulary learning from reading has long been a topic of interest forresearchers One of the primary questions has been whether to drawattention to forms in a reading lesson Nagy, Herman, and Anderson(1985) demonstrated that vocabulary could be incidentally acquired andretained from reading in a first language environment, and furtherresearch (e.g., Horst, Cobb, & Meara, 1998) has indicated that incidentalvocabulary learning is possible for second language learners as well.Although the possibility of incidental vocabulary acquisition is wellestablished, its efficiency has been called into question (Hill & Laufer,2003) with studies indicating that incidental learning accounts for onlysmall gains over quite large periods of exposure (Nagy, Anderson, &Herman, 1987; Nagy et al., 1985; Horst et al., 1998) Laufer (2005) alsoalludes to this trend and suggests that, ‘‘input, particularly reading input,alone is unlikely to be the best source of second language vocabularyacquisition’’ (p 228) She asserts that it may be necessary to draw learnerattention to lexical form in order to maximize vocabulary gains frommeaning-focused input Although meaningful input is essential tolearning, language learning can be facilitated through consciousattention to form, because attending to form in a meaning-focusedactivity may allow learners to notice the form, acquire it, and retain itover time (Long & Robinson, 1998; Schmidt, 1990)

INTEGRATED OR ISOLATED FFI

As mentioned earlier, FFI can happen in an integrated or isolatedapproach In an integrated approach, focus on linguistic form (in this casevocabulary items), whether planned or incidental, occurs in the context ofcommunicating meaning (e.g., Ellis et al., 2001b; Long & Robinson, 1998)

In a reading lesson, this kind of focus on form would occur while studentswere engaged in reading a second language text (in other words, whiletheir primary focus was directed at understanding the meaning in thetext) On the other hand, an isolated approach would entail shifting thelearners’ attention away from the meaning in order to focus on thefeatures of a particular vocabulary item in isolation An example of this ispreteaching vocabulary before a reading lesson Therefore, during theisolated FFI, the primary focus is on the language forms, rather than on themeaning of the text (although focus may shift back to meaning in anothersection of the lesson) The primary difference then, between integratedFFI and isolated FFI, is that isolated FFI requires learners to focus primarily

on language forms for an extended lesson segment, whereas integratedFFI requires them to shift their focus between the meaning of the text andlanguage forms for each item taught (Spada & Lightbown, 2008).Studies on the role of integrated FFI in promoting vocabularylearning from reading have overall uncovered positive effects Knight’s(1994) study of English first language Spanish learners found that focus

on form (in the form of dictionary use while reading a magazine article)led to substantial learning of previously unknown words, suggesting thateven short digressions from a purely meaning-focused activity toconcentrate on a form can promote notable gains in the learning ofthat vocabulary item Similarly, Fraser (1999) found evidence to suggestthat participants who inferred the meaning of a word and consulted adictionary (using a meaning-based and a form-based strategy) whilereading a text retained more word knowledge than those who justinferred the meanings Watanabe (1997) also found that attention tovocabulary during reading activities promoted learning In his study,subjects who read a text that included glosses of difficult wordsoutperformed subjects who read the text without glosses Such findingsalign with views on focus on form from researchers like Lightbown(1998) and DeKeyser (1998), who have claimed superior learning fromintegrated FFI, because it links both form and meaning in context.Although the above studies have shown that an integrated approach

to vocabulary FFI during a reading lesson enhances vocabulary learning,studies in which vocabulary instruction has been isolated have alsounearthed positive results For example, Paribakht and Wesche (1997)found strong evidence of the effectiveness of teaching vocabulary items inisolation They compared vocabulary learning scores of a reading-only

group and a reading-plus group In the reading-plus treatment, subjectswere required to read eight texts and do extra vocabulary-focusedactivities The reading-only group were required to read the same andother supplementary texts Tests of word knowledge indicated that,although the reading-only group did gain a significant amount of wordknowledge of the target words, knowledge was mainly at word recognitionlevel (Paribakht & Wesche, 1997, p 196), whereas the reading-plus groupgained more words and higher levels of word knowledge for many of thewords There is some evidence that isolated vocabulary focus may promotemore learning than an integrated approach Laufer (2005), in a review offocus on form (integrated FFI) and focus on forms (isolated FFI) researchthat compared gains from several studies, found that the studies whereforms were treated separately from meaning processing (focus on forms)showed larger gains than those where attention to form was integratedwith attention to meaning (focus on form; Hill & Laufer, 2003; Laufer,2003) Laufer considers this evidence that criticism of focus on forms istherefore unwarranted, arguing that, whereas both promote learningmore than an input-only approach, isolating forms as a teachingtechnique was especially effective Laufer’s reasoning is that, by isolatingforms, teachers can teach and cover the many different aspects of knowing

a word in a short amount of time—essentially offering rich instruction(Laufer, 2005, p 244) Hill and Laufer (2003) conducted a study thatcompared input only (incidental learning), meaning glosses in themargins (integrated focus on form), and decontextualized vocabularyexercises (isolated focus on form) Although they found evidence ofvocabulary learning in all three conditions, it was the isolated approachthat proved the most effective (Hill & Laufer, 2003, p 103).1In anotherstudy, Laufer (2003) found that low learning scores were obtained fromreading alone and that reading plus additional word-focused activities wasthe most effective approach

Although these studies show an advantage for isolated over integratedand incidental vocabulary learning, it should be noted that, in severalstudies, the isolated treatment was more time extensive than theintegrated one, and researchers may have conflated a number of factorswhen investigating the different approaches to FFI For example, in theHill and Laufer (2003) study and the Laufer (2003) study, the groupthat received isolated vocabulary focus was given a richer form ofinstruction (explanations and translations) than the group receivingintegrated vocabulary focus They were allowed to write sentencescontaining the target words, pushing them to process the target itemsboth receptively and productively The integrated group was only given

1 Scores were taken from tests that elicited passive knowledge of the target words, given both immediately after the treatment and as a delayed test given 2 weeks later.

receptive exposure Additional time and opportunities for productiveuse, therefore, may have resulted in a greater depth of processing of thetarget words, which could have led to more sustained word knowledgegains This possibility is supported by the results of Webb’s (2005) studycomparing the effects of meeting a word in a reading task andproducing a word in a sentence-production task, which suggested thatlearners who did the sentence-production tasks acquired greater wordknowledge than those who did the reading task This supports thepossibility that the differences between the groups in prior vocabularystudies may be due to different types of processing tasks, rather than todifferences between these integrated and isolated approaches to FFI.Theoretical support for this position comes from the output hypothesis(e.g., Swain, 2000), which asserts that producing output requireslearners to process language differently than when simply comprehend-ing it and that the process of producing output may promote the type ofprocessing that can lead to greater linguistic accuracy and learning (cf.,Loewen, 2005) The present study, therefore, attempts to create acomparison of isolated and integrated approaches to vocabularyteaching in a reading lesson, by providing equal instruction time andpromoting similar processing for the target words in both conditions.

THE CURRENT STUDY

From the preceding literature, it seems clear that some sort of FFI (be

it isolated or integrated) is more effective for vocabulary learning thaninput only (Hill & Laufer, 2003; Knight, 1994; Laufer, 2003; Paribakht &Wesche, 1997; Watanabe, 1997) Although vocabulary learning doesoccur incidentally through reading (see e.g., Horst et al., 1998; Nagy etal., 1985; Paribakht & Wesche, 1997), incidental vocabulary learning iscontingent on truly extensive reading in the second language, as well asreading conditions that allow learners to meet the same word multipletimes in meaningful contexts Although reading may still be animportant means of increasing vocabulary knowledge, it may also bemore effective for teachers to include FFI of new vocabulary inmeaningful contexts This approach may represent a more efficientmeans of learning vocabulary, particularly for learning lower frequencyvocabulary or vocabulary for specific purposes, like the academic wordlist (Coxhead, 2000)

The question now is which type of FFI (isolated or integrated) is moreeffective for learning new vocabulary through reading The purpose ofthis study is to examine the effectiveness of FFI for increasing vocabularyknowledge and in particular to contrast integrated and isolated FFI AsSpada and Lightbown (2008) point out, ‘‘to our knowledge, no

empirical classroom-based research directly compares the effects ofisolated and integrated instruction’’ (p 193).

The aim of the current study, therefore, is to examine which type offocus on form (isolated, integrated, or no focus) best suits vocabularylearning and retention from classroom-based reading lessons Thefollowing research questions guided this study:

1 Does any form-focused vocabulary instruction during a reading lesson lead

to more vocabulary learning and retention than no focus on form at all?

2 Does an isolated or integrated approach to vocabulary instruction in areading lesson promote more learning and retention of previouslyunknown words?

METHODOLOGY

Participants

Two classes of adult intermediate proficiency students from auniversity preparation programme (Class A and Class B) were theparticipants for this study All 27 students in both classes agreed toparticipate, but only data from 20 participants could be used in the finalanalysis because of absenteeism during the project Of the 20participants who completed the study, there were 10 males and 10females The regular teachers allowed one of the researchers (who wasalso a teacher in this programme) to teach each of the sessions involved

in the study, to ensure that the instructional sessions were equivalent.Table 1 summarises some general information about the participants.Design

This study followed a pretest–posttest–delayed posttest design todetermine whether FFI influenced learning and retention of newvocabulary Pretesting of vocabulary items can draw attention to the

Chinese 2 Thai 2

target words; however a 1- to 2-week period between the pretest andtreatment (see Horst et al., 1998, p 213) has been used by other studiesand is generally considered to be a reliable amount of time to mitigatethis effect Additionally, the regular teachers of Class A and Class B weregiven a list containing all target words and were asked to avoid using ordealing with them in class for the duration of the research project.Eighteen days after the pretest (test procedures will be discussed later),reading lessons and experimental treatments were carried out Bothclasses participated in both treatments; the order of treatmentpresentation was counterbalanced Therefore, a total of four readinglessons were carried out: two lessons (one for each class) where thewords were pretaught (isolated) and two lessons where they were taughtduring (integrated) the reading of the passage Figure 1 shows how thetexts and treatments were counterbalanced.

Immediately following each reading treatment, participants completedmultiple-choice comprehension tests followed by vocabulary tests (seelater for more detail) One entire treatment session, which included areading lesson, vocabulary teaching before or during the reading, and alearning test, took roughly 55 minutes to complete Sixteen days after thefinal treatment, a further posttest was carried out, to measure retention ofvocabulary knowledge Figure 2 outlines the procedures of the study.Materials

The two articles used in this study were written by a published writer.Both articles needed to be written in an academic style and in

FIGURE 1 Order of texts and treatments.

accordance with the study theme for that week: work, recreation, andleisure This was done to contribute to face validity of the readings forthe participants and to avoid disrupting the classes The titles of thecompleted articles were

1 ‘‘Girls Can Do Anything, but Are They Doing Too Much?’’

2 ‘‘Achieving a Balanced Life—What New Zealanders Are Saying Aboutthe Work–Life Balance’’

Both articles also contained New Zealand content, making them relevant

to the context the learners were studying in Both articles were roughlythe same length, with text one containing 1,300 words and text two1,250 A copy of one of the articles used in this study can be seen inAppendix A

FIGURE 2 Research design.

Target Words

The two articles were written first, and the vocabulary that appeared

in the articles was selected and adapted to fit the needs of the project.This was done to avoid producing stilted prose written around specificvocabulary After the articles were written, they were put through Cobb’svocabulary profiler (Cobb, n.d.) to identify which frequency bands thewords in each of the two articles fell in To find target words at a suitablelevel for the participants, scores from Nation’s Vocabulary Levels Test(see Appendix 3 in Nation, 2001, pp 416–424), carried out by theuniversity to help place students in appropriate classes, were used Thescores from this test suggested that words from the 5,000-word level andabove were not likely to be known to the students and therefore weresuitable to target for these participants Thus, words from the articles at

or above the 5,000-word level were selected as the target words for thisproject No words from the academic word list (Coxhead, 2000) wereselected as targets, because participants had already been studying thisword list as part of their regular instruction

A total of 18 words per article (36 target words for the project) wereselected for use as the target words for this study Where possible,remaining low frequency words in the article were simplified with higherfrequency equivalents (from within the first four 1,000-frequencybands) This substitution process was done with the writer of thearticles, to ensure that the higher frequency substituted words or phrasesdid not alter the meaning of the text It was also done to help make thetwo texts comparable in difficulty, by controlling the vocabulary load.Each of the 36 target words appeared once in the texts, to ensure thatnone of the target words received extra attention in the text A list of the

18 target words in alphabetical order for each text can be seen inAppendix B

Incidental Exposure

One of the purposes of this project was to examine the differences inlearning and retention of words that were taught and words that wereacquired incidentally in an intensive reading lesson From each word set

of 18 target words, 12 words were explicitly taught, and 6 words were nottaught but were included in tests to examine incidental learning The 6incidental words for each reading were randomly selected from each setand can be seen in column 2 in Appendix B All the explicit instructiontarget words in both texts were put in bold font in the articles The 6words that were being tested for incidental learning, however, were notput in bold font, to maintain the conditions for truly incidental learning

It is possible that additional incidental learning beyond these target

words may have occurred; however, to avoid test fatigue, only words mostlikely to be unknown prior to the study were included in the testing.Instructional Treatments

The purpose of this study was to examine vocabulary instruction andlearning in an authentic lesson plan Care was therefore taken to ensurethat the reading lesson used in this study followed the readinginstruction to which the participants were accustomed Each lessonfollowed similar steps First, the participants were given 3 minutes todiscuss the title of the article in pairs and make predictions about thearticle These predictions were discussed by the class Following thiswarm-up, the reading treatments began

In each treatment, the researcher read the text aloud while theparticipants followed along, reading their own copy This procedure wasadopted to control the pace of the reading lesson and to ensure that allparticipants had the same processing time for the text (cf Horst et al.,1998) At the end of each paragraph, the researcher would stop andsummarise the main ideas of the paragraph in a short two- or three-sentence oral summary The summaries did not include the target words.Isolated FFI Treatment

In the isolated treatment, all 12 vocabulary items were taught before theparticipants read the text Each word was presented individually on anoverhead transparency and pronounced by the researcher The researcherthen orally defined the words (from a scripted definition that was also used

in the integrated reading treatment) After the meaning was provided,

an example sentence was shown to the participants on an overheadtransparency, and the meaning of the word was explained in relation to thecontext of this example sentence Two synonyms for each target word werealso given After all 12 target words were taught, the researcher started toread the text orally The 12 target words were bolded in the text, but theresearcher paid no further attention to them when they came up in thereading This process has also been included in diagram form for one of thetarget words in Figure 1A, Appendix C

Integrated FFI Treatment

When teaching a word in the integrated treatment, the researcherwould teach the word in the context of the reading Instead of preteachingthe words, the researcher started the oral reading of the text immediatelyafter the warm-up After reading a sentence that contained one of the 12

targeted words, the researcher would then return to the target word(bolded in the article) and draw participants’ attention to the form, alsoproviding the correct stress After that, the researcher would give themeaning of the word orally (the same as in the isolated teaching routine).Then the researcher would explain the meaning of the target word inrelation to the context it appeared in, in the article Just as in the isolatedtreatment, this part was not scripted but was practiced several times toensure the same amount of time and instruction was being offered in theintegrated treatment as in the isolated treatment After providing ameaning, the researcher then gave two synonyms for the target word Thisintegrated instruction process is also highlighted in Figure 2B, Appendix

C The researcher carried out all instructional treatments, and the classteacher was not present during any of these treatments

Throughout the study, care was taken to ensure that the same amount

of time was dedicated to explicit vocabulary instruction in each of thetreatments To achieve this, all the parts of the vocabulary instructionwere planned and scripted for each of the two sets of words Theexplanation of the word in its context was left unscripted to promoteauthenticity; however, these unscripted explanations were practicedseveral times and timed to ensure that explanations were similar in bothtreatments All of these scripted definitions and synonyms for the targetwords were submitted to the vocabulary profiler (Cobb, n.d.), to ensurethat only high-frequency words were used in the oral definitions and assynonyms for the target words in order to keep them comprehensible tothe participants To maintain consistency of exposure and preventadditional study outside of class, participants were instructed not to takenotes during any of the reading treatments

Piloting

All treatments were piloted with a student from a different class in thisprogramme In the study, the treatments were also timed, and the timesfor the isolated and integrated reading treatments for each article wereroughly the same.2

of the articles, the postparagraph summaries, and the word teaching were all inclusive in this treatment.

Wesche, 1997) was used to determine knowledge of the target words.Additionally, a short comprehension test was created for each reading tohelp give participants the impression of a reading practice lesson.Because the comprehension tests do not form part of the data for thisstudy, they will not be discussed further.

The Vocabulary Knowledge Scale (VKS)

Paribakht and Wesche’s (1997) VKS was used to measure vocabularylearning and retention The main strength of the VKS is that it allows theresearchers to distinguish different levels of knowledge about aparticular word The VKS requires learners to self-report their knowl-edge of a target word on a five-point scale, ranging from no knowledge

of the word to the ability to use the word productively, as outlined inFigure 3

Each test included the 36 target words (24 taught words and 12incidental words) as well as a further 12 nonsense words which wereselected as distractors The distractors were selected from other studies(e.g., Waring & Takaki, 2003) and Web sites (Nonsensicon, n.d.).Nonsense words were selected based on the same ratios of syllables as thetarget words and were included in the tests to ensure that participantshad not previously seen all tested words

Before each test, learners were given verbal explanations on the testand an opportunity to try the test with several practice words FollowingParibakht and Wesche (1997), self-report scores of 1 or 2 were accepted.However, for score categories 3, 4, and 5, the participant had todemonstrate his or her knowledge of the target word by providing asynonym, definition, translation, or sentence

In total, four vocabulary tests were created: a pretest, two vocabularylearning tests (one posttest following each treatment), and a delayedposttest All four followed the same format (VKS); however, word

FIGURE 3 Vocabulary knowledge scale test.

order was randomised for each test Although the pretest and thedelayed posttest included all 24 taught targets, all 12 implicit targets, and

12 nonsense words, the learning tests that were administered followingeach reading treatment contained only the 12 targets taught for eacharticle, the 6 incidental targets contained in the article, and one-half ofthe nonsense words

Scoring and Interrater Reliability

Scoring of the VKS tests was based on Paribakht and Wesche’s(1997) and Joe’s (2006) scoring operationalisations, whereby self-report scores from 1 to 5 translated into a word knowledge scorefor a particular target word Answers above Level 2 on the VKS werejudged based on the quality of the response To receive a word score

of 3, the participant needed to provide a suitable synonym or definitionfor the target word Failure to do this meant the participant’s wordscore for that particular word was reduced to 2 If the participant self-selected 4 and provided a suitable synonym or definition for thetarget word, then they were credited with a word score of 4 To receive

a score of 5, participants were required to provide a suitable synonym

or definition and a sentence that showed accurate semantic andgrammatical knowledge of the target word Grammatical errors wereoverlooked if they appeared in other parts of the sentence

One of the researchers and an independent researcher scoredall three VKS tests (pretest, learning, and posttest) for all 20 partici-pants Interrater reliability was high (95%), and all disagreementswere discussed, paying close attention to the operationalisations, untilagreement was reached Pretest, posttest, and delayed posttestscores for each participant for each learning condition (integrated,isolated, and incidental) were created by totalling scores for eachword

Posttreatment Survey

A posttreatment survey was also administered to gather insights intothe difficulty and interest levels of the articles, as well as the preferredteaching styles of the participants in the study (i.e., did they like theisolated approach or the integrated teaching approach?) There was also

a section on the back of the survey for participants to note anything theywanted to communicate to the researcher about the study More thanone-half of the participants took the opportunity to make observations

in this section Some of the insights that were gathered from thesesurveys are discussed later where relevant

ANALYSIS AND RESULTS

As explained earlier, learning and retention of vocabulary weremeasured through the use of the VKS For each of the participants,separate VKS scores for the pretest, posttest, and delayed posttest werecalculated for each of the treatment conditions (isolated, integrated,and incidental) For each of these scores, the maximum possible was

60 points (5 points per word for 12 words) Descriptive statistics for theseare displayed in Table 2

All of the pretest scores were less than 20 points, indicating thatlearners had little or no knowledge of the words prior to the readingtreatments This result is supported when the distribution of item answers

is examined Of the 720 target items in the pretest (20 students multiplied

by 36 target items), 673, or 93%, were scored as 1 or 2, indicating noproductive word knowledge prior to the treatments By the posttest, thisfell to 444, or 62%; learners received scores of 3 to 5, signalling knowledge

of the words beyond simple recognition for 276, or 38% of the targetwords The descriptive data in Table 2 shows that, for both learning andretention, the highest scores were found in the isolated instructiontreatment, followed by words taught in integrated instruction In eachcase, there is a large gap between the two instructed conditions andincidental learning

Item scores on the VKS at each time of testing for each learningcondition are displayed in Figure 4 To simplify the visual data, scores of 1

or 2 (little or no word knowledge) and scores of 3 and higher are grouped

At the pretest, the target words were almost uniformly unknown(scores of 1 or 2) By the posttest, more than one half of the itemslearned in isolation were scored 3 or higher (139 items, 58%), indicatingthe ability to at least define the target items A similar pattern was foundfor words learned in integrated instruction However, for words learnedincidentally, the distribution of VKS scores remained fairly constant Todetermine whether learner knowledge of the words differed at the

TABLE 2

Descriptive Statistics: Isolated, Integrated, and Incidental

Pretest (N 5 20) Isolated 17.95 2.33

Integrated 18.30 2.83 Incidental 19.60 2.66 Posttest (N 5 20) Isolated 37.10 7.16

Integrated 33.55 5.65 Incidental 21.90 3.02 Delayed posttest (N 5 20) Isolated 25.10 5.03

Integrated 24.60 3.68 Incidental 22.95 2.65