Teaching of formulaic sequences
4.1 Introduction
The current chapter aimstointroducethe backgroundtothe study conducted forthis dissertation first with a consideration ofthree fundamental questionsthat were raisedinthe review and synthesis offeredinthe previous chapters,then with a review of studies ontext memorizationthat helpto clarifythe rationale forthe present study.
Thethree central questionsthat have emergedin discussion so far are (1) Should teaching practitionersfocus onformulaic sequencesin aforeignlanguage context withthe particulartarget population being adultlearners?, (2)Ifthey should, whichformulaic sequences shouldtheyteach?, and (3)How shouldtheyteachthetargeteditems? To answer the first question, a number of reasons fortheteaching of formulaiclanguagetothis
particulartarget population are put forward. Addressingthe second andthird questions, principles and proposals drawn from theideasinthe previous chapters areoffered.
4.2 Rationale for focusing on FSsin teaching
The set of reasons for choosingto focus on formulaic sequencesinteaching are outlined here. The primary motivation should be evident by now, however:the more readily available formulaiclanguage isto alanguage user,the better prepared he or sheisto stayin control withinthe chaos of authentic communication. Put another way,the more analytic processing a user hasto exert,theless he or she can handle comprehension or production. Formulaic language makesit morelikely for a personto accomplishthethreetypes of
socio-interactional goals (or functions)introducedin Section 2.3. Equally—orinthe case of the adult FLlearner moreimportantly—reliable formulaiclanguage knowledge enables the language userto free up his or her attentional resources at a given moment, andthe set-aside
41
attentional capacity can be directedtoward a number of cognitive processes,including noticing (a necessary condition for anylearning). Although formulaic sequences are, bytheir own nature, restrictive with regardto flexibility and novelty of expression, and can be
construed as a ‘straitjacket’ forthelanguage user (Wray & Fitzpatrick, 2010), flexibilityin processing can ensue from opportunities for variation within formulaic word strings, and entirely new combinations can be generated at any point by switchingtothe analytic processing forthe selection of smallerlexical units (Wray, 2002). Allthese claims strongly pointtothe recommendationthat adult FLlearners be supportedin obtaining formulaic knowledge. Eventhoughitis simplyimpossible for FLlearnersto be given encountersinthe classroomthat are sufficientto enabletheideal kind of underlying pattern extraction
produced by native speakers, even a small amount of formulaic knowledge should be helpful intheir overall L2learning, givingthem morethanjust holisticlanguage knowledge,
especially whenthey processthelanguage outside of realtime communication. For example, itis easytoimagine a situation where alearneris explicitly studying a grammar rule andthe metalinguistic knowledgeis strengthenedif exemplarsin whichthe ruleis embedded are already part ofthelearner’s formulaic knowledge.14
4.3 What formulaic sequences should be taught?
The choice of which formulaic sequencestoteachis rather difficult. Since classtimeis limited, practitioners needto make careful choices asto which onesto focus onintheir classes. Although criteria are hardto agree on, ‘serviceable’ sequences are probablythe best candidates. In determining which are serviceable, however,itisimportantthat practitioners do not blindly rely upon corpus frequency data, but rather continually assesstheir students’
needs and wantsin orderto be ableto make on-demand selections.
14 The converse ofthis would bethe case when FLlearners are first equipped with metalinguistic rules,thenlater encounter exemplars containingthose rules and are ableto recognizethe exemplars as manifestations of some underlying rulesthatthey have previously studied (see Section 4.4.4).
42
Whileit may be difficultto determine what messages a given group of students will find beneficial fortheir current and futurelanguage use,there seemto be atleast a few
non-controversial recommendationsthat caninformthe selection ofitems. Oneisto focus on flexible formulaic sequences (see Section 2.3.1). The versatility or flexibility of formulaic sequencesthat are made up of frames with slots makesthem good candidates. Nekrasova (2009) foundthat L2learners rely more on fixed multiword unitsthan on flexible ones, and pointsto the possibilitythatthe composition (and potential versatility) of a flexible sequence is difficult forlearnersto graspinthe first place. Therefore, althoughlearningjust one ortwo example possibilities out of a flexible sequenceis unlikelyto be enoughto gain procedural knowledge ofthe versatility ofthat sequence,thereis good reasonto expect, as discussedin the previous section,thatthe knowledge can facilitate deeperlearning ofthe sequencein future encounters. Another recommendation would betoteach figurative multiword strings (see Section 2.3.2). Ifthelearners’ objectiveisto approximateto a nativelikelevel, figurative sequences can bethe biggest challenge. Evenifthe goalis notthat high, a carefully selected set of suchitems should beintroducedtolearners of alllevels because,likeit or not, native speakers will usethoseitems astheir natural choice. Atleast forthe sake of comprehension, inthis case, themost frequently used figurative FSs should betaught. Thelast
recommendation here has to do withtheteaching of formulaic sequences consisting entirely of wordsthat are familiartolearners. Inthe study by Peters (2012), an additional review of which will be giveninthe next section, her participants were askedto copy words and
phrases from atextthatthey felt merited attention, and eventhoughthe students were briefed abouttheimportance of formulaiclanguage,theytendedto write down unfamiliar single words ratherthanthe complete formulaic sequencein whichthese words appeared. This finding,then, pointstothe possibilitythat formulaic sequences consisting entirely of known words forlearners may not attracttheir attention.
43 4.4 How should formulaic sequences be taught?
Forteachingin any situationto be successful,the question of ‘how’toteach islikelyto be asimportant asthat of ‘what’toteach, and probably moreimportantin most cases. Accordingly,this sectionis goingto considerthe potential effectiveness of a number of different waystoteach formulaic sequences.
4.4.1 Input enhancement
Unobtrusiveinstructional means, collectively referredto as ‘input enhancement’
techniques (Sharwood Smith, 1991, 1993), might be a candidateto effectivelyteach formulaic sequences. In a recent study, Peters (2012)investigatedthe effect oftypographic enhancement and of glossing, a form ofinput enhancement, on retention. Peters foundthat the participantsinthetreatment group, who worked ontexts withtypographic enhancement (underlining and bold font),were better ableto recollect glossed formulaic sequences from reading withthe enhancement. One methodologicalissue with her study, however,isthatthe students had beeninformedthat a vocabulary posttest would follow, andtherefore,they may have made more of an effortto rememberthe highlighteditemsinthetextthanthosethat were not highlighted. For another more recent example, Webb, Newton, and Chang (2013) exploredthe effects ofinput flooding, another well-knowntechnique ofinput manipulation, on facilitatinglearning of formulaic sequences. Webb, Newton, and Changincorporated 18 verb-noun sets (e.g., buytime;cut corners;losetouch)in a graded reader and prepared four versions, differinginthe number oftimes each ofthose collocations appeared: only once, fivetimes, 10times, and 15times. Theintermediate-level EFLlearner participants, who were randomly assignedto four groups working on one version different fromthe otherthree, read the story whilelisteningto a recording ofit,and their retention ofthetarget collocations was measured by unannouncedimmediate posttests. Not surprisingly,the more often a collocation was repeated,the higherthe recall ofthe given collocationsinthe posttests, withtests on
44
receptive knowledge showing better scoresthantests on productive knowledge. However, as many as 15 encountersin such a short amount oftime was still far from a guarantee for full scores on any ofthe posttests. Forinstance, even after 15 encounters, collocations were correctly recalled only half ofthetimeinthe L1-cued productive knowledgetest. Those who encounteredthe collocations only onceinthetext did not gain better posttest scoresthanthe control participants who had not even readthetext. The results ofthis study,therefore, suggestthatifthe goal ofteaching formulaic sequencesisto bring about productive knowledge,input enhancement may not be an optimal way.
4.4.2 Chunkingin text comprehension
Alittleless unobtrusiveintervention oninput processingthaninput enhancement may betext chunking. As an example, Boers, Eyckmans, Kappel, Stengers, and Demecheleer (2006)investigatedthe effect of awareness raisingthroughtext chunking. Thetreatment groupintheir study, a group of advanced EFLlearners, worked ontext chunking as a regular activityinthe course of a school year, whilethe contrast group engagedin other activities usingthe sametexts. Duringthe course-endinterview,the studentsin both groups were asked to orally retellthe content of a new Englishtext. Boerset al. reportthat significantly more formulaic sequences were foundinthe narratives produced bythe students inthetreatment group. However, Boers and Lindstromberg (2012) point outthatthis was becausethose students recycled more word strings verbatim fromthe newtext, whilethe comparison group tendedtoincorporatejust single words fromthetextintotheir retelling. Stengers, Boers, Housen, and Eyckmans (2010) replicatedthe study by Boers et al. with a slight
methodological change. In Stengers et al., with a viewto avoidingthe possibility of recycling language verbatim fromtheinputtext forthe L2 retelltask,they used aninputtextinthe L1 oftheir participants (new cohorts oflanguage majors). Pretest-posttest comparisons showed no significant differenceinthe uptake of formulaic sequences betweenthe participantsthat
45
had regularly engagedintext chunking andthosethat had not. The results ofthesetwo studies suggestthat simply directinglearner attentionto multiword segmentationsis
insufficienttoleave durable memorytraces, althoughit maypotentially pushlearners’ mode of processing toward being more holistic, a possibility speculatedon by Boers and
Lindstromberg (2012).
4.4.3 Exercise of schematic and contextual knowledge
Ifthe assumptionthatthreeinter-related sources of schematic, contextual, andlanguage knowledge (see Section 3.2) are storedin our memoryis accurate,thenthereis an attractive proposal fortheteaching of formulaic sequencesthat followslogically, and which also resonates,in a way, withtransfer-appropriate processing (see Section 3.5). This proposalisto invoke and engagelearners’ schematic and contextual knowledgeinthe encoding of
formulaic sequences. The fundamental principle ofthisidea seemsto be closely relatedtothe dual coding hypothesis proposed by Paivio (1986) and Sadoski (2005). This dual coding hypothesis assumesthat cognition occursintwoindependent but connected codes,thatis, a verbal code forlanguage and a nonverbal code for mentalimagery, andit holdsthat concrete vocabularyis easierto rememberthan abstract vocabulary. Concretenessis strongly
associated withimageability (Hamilton & Rajaram, 2001)15, andtheimageability of
something presumably dependslargely onthe extentto which one’s schematic and contextual knowledgeis applicableto the creation ofitsimage. For an example oftheimageability effects of certaintypes of formulaic sequences (most notably figurativeidioms), Steinel, Hulstijn, and Steinel (2007) showedthatidiomsthat evoke a mental picture relatively often (e.g., stickto your guns) were better retainedin an L1—L2 paired associateslearning experimentthanidioms which less readilycall up animage (e.g., hangfire). Thisline of
15The use of mentalimagery is atthe core ofteaching approachestoidiomsinspired byideas from cognitive semantics (e.g., Boers & Lindstromberg, 2005;Lakoff, 1987).
46
thinkingis also applicabletothelearning of formulaic sequences comprised of familiar and unfamiliar words. Hsu (2010) and Kasahara (2010, 2011), forinstance, showthat collocation learning fosters recall of a new word containedinthe collocation atleast as well aslearning the new word as a singleitem. In fact, oncethe collocationislearned,the word which was already familiar (e.g., business) can serve as a cue forthe recall ofits newlylearned
syntagmatic partner (e.g., acumen). Their studiesindicatethe power of syntagmaticlearning of familiar plus unfamiliar word strings.
4.4.4 Teaching grammar to facilitate formulaiclearning
The claim forteaching grammarin orderto fosterthelearning of formulaic sequences might at first sound contradictory. Giventhe argumentslaid outthus far, however,the claim should not come as a surprise. The generative potential of a multiword string consisting of fixed frames with open (thoughin many cases semantically constrained) slots can be harnessed at a maximumlevel onlyifthelanguage useris familiar withthe underlying blueprint. However,the adult L2learner hastremendous difficulty extractingthe patternjust through normal communicative engagementinthelanguage. Therefore,instruction for familiarizinglearners with explicit grammar rules,thatis, gettingthemto gain explicit knowledge aboutthelanguage (e.g., N. Ellis, 2005; Krashen, 1981; Takashima, 2011),is doubtlessly beneficial forthem. Explicit knowledgeis not as serviceable asimplicit
knowledge ofthelanguagein realtimelanguage processing. Nevertheless, giventhat adults learningin a foreignlanguage context seriouslylackinput from whichtoimplicitly extract underlying patterns (see Section 2.6 for pattern-basedlanguage acquisition) onthe one hand, and do not enjoythe socio-interactional protection ofthe bubble while having reliable resources of schematic and contextual knowledge onthe other, reliance on explicit grammar knowledge, albeit not derived fromimplicit knowledge andthus not as convenient, seemsthe only viable optionleft forthemifthey are strivingto approximateto nativelike proficiency.
47
The more explicit knowledgethey have,the more successfully theycan engagein analysis, both within and outside of communicative engagement.
4.4.5 Use ofinteractionally non-demanding tasks
Whenteachinglearnersto developtheir communication skills,the success of which,it should be clear by now, dependslargely uponthe use of formulaiclanguage,thereis no question aboutthe fundamentalimportance of engagingthemin actual communication (becausethatis ultimatelythe only place wheretransfer-appropriate processing for authentic communication cantake place). The case also has been established, however,thatin normal communication,they predominantly rely ontheir existing formulaiclanguage knowledge withlittle chance oflearning sequencesthat are deemed ‘difficult’ forthem because of a combination of factors contributingtotheirinherentlearning difficulty coveredin Section 3.6. Thisis so becausetheirlimited attention capacity is exhaustedin managingthe
socio-interactional demandsthat are placed onthem orthatthey perceiveto be so, and when thelanguage knowledge readily availabletothemis not helpful enough,their schematic and contextual knowledgeis called up, allowingthemto bypasstime-consuming and
attentionally-draining “online” analytical computation. Evenifthereis a residue of attentional resources available at a giventime,itis hypothesized thatthisisnotdirected, unlessingeniously directed otherwise,towardlinguistic analysis (VanPatten, 1990, 1993). Grammatical analysis of formulaic sequences, even known ones,is farlesslikely because morpho-syntactic features carry,in most cases,less communicativeload (or significance for comprehension or message conveyance). Guidingthemto processlanguage without recourse totheir schematic and contextual knowledgeis atall orderinthe first place, and also
inadvisableinthelight ofthe discussion above (Section 4.4.3). The crucial stumbling block, then,isthe non-existence ofthe social ‘bubble’ (see Section 2.5). Wray (2000) commented that “[a]rguably,the classroom offers such a bubble” (p. 481). She seemsto suggestthisin
48
regardto classroom communicative activities—I drawthis conclusion based onthe context of the discussioninthat paper. I am compelledto addthatit really depends, especially when teaching mono-demographic adultlearnersin a foreignlanguage context. To start with,the teacher needsto be proficient enoughto create and keep such a bubbleinthetargetlanguage. Second,the extentto which he or sheis successfully ableto do so depends heavily onthe complex combinations ofthe class size,level (or differinglevels) ofthe students,their motivation, and other factorsthat affectlanguagelearning within a classroom. Last, but equally non-negligible,the effects ofthe socio-interactional normsintheir L1, which are, at leastinthe case of Japanese studentslearning Englishin Japan, quite different fromthose normally expectedinthe L2, arein most casesinevitablein communicating with other students havingthe same cultural background,thereby putting extra pressure on
communication, evenif itisinthe L2. As Wray addedinthe same paper, “the best analytic learning will occurin a contextin whichthere are nointeractional challenges or surprises” (p. 482). In summary, whilethe requirement for communicativetasks forthe development of communicationis never negated,it appearsto methatthere also hasto be someteaching place, not authentically communicativein nature, where students feel non-threatenedto be communicatively operational, andtherefore ableto devotetheir attentiontoward analysis of language whetherlexicalin nature or grammatical.
4.4.6 Text memorization
Text memorization, a form ofinteractionally unchallenginglearning, can be one useful wayto help students tolearn formulaic sequences. Clearly,there has been general reservation inthelanguageteaching profession about employinga repeat-and-memorize approachin class although, astouched onin Chapter 1,there areteachers who have a favorable view towardthe effectiveness of memorization. This reservationislargely “duetolong-rehearsed criticisms of audiolingual-type approaches andjustifiable skepticism aboutthe value of any