Avoiding the Target Language with the Help of Google: Managing Language Choices in Gathering Information for EFL Project Work

the interface with the Swedish version of the Internet search engineGoogle www.google.se,1and more specifically an examination of howsome pupils orient towards the emergent features of S

Avoiding the Target Language with the Help of Google: Managing Language Choices in Gathering Information for EFL Project Work

an avoidance strategy—that of avoiding the target language Thisqualitative study highlights several ways in which pupils can and doexercise their language choices in their computer-assisted projectwork using (Swedish) Google More specifically, after mapping outthe trajectories of pupils’ Internet searches and their gathering ofresources, conversation analysis is used to zoom in on key moves thatare jointly negotiated at the interface with Google, where differentlanguage options are made available and even flagged By makingincremental choices that avoid English (or favour Swedish), the visi-bility of English can diminish radically, if not altogether The nega-tive implications of avoidance are discussed as well as thepracticalities of solutions using Google itself

doi: 10.1002/tesq.102

This article has arisen as part of a larger study investigating theinformation literacy of pupils when gathering information for pro-ject work in the English as a foreign language (EFL) classroom Itemerged early on in different data sets from a variety of school settings

in Sweden that there was great variation in pupils’ language choices atvarious stages of the project work, as regards not only their talk, butalso their demonstrated language preferences in their searches andreading practices Among other things, this observation has precipi-tated a closer examination of language choices in search activities at

the interface with the Swedish version of the Internet search engineGoogle (www.google.se),1and more specifically an examination of howsome pupils orient towards the emergent features of Swedish Googlewhich directly affect the language of their information sources so thatthey (can) avoid English This is not to say that this is a path that all

or even most pupils have opted for in the data, but making use of thelanguage options and translation tools afforded by Google warrantsfurther attention because of the potential implications for the foreignlanguage classroom At this point it should be mentioned that becausethe equivalent language options are available for other languages too(with the computer’s default option normally being the country’s offi-cial language/languages), the implications of this study are of rele-vance in almost all settings apart from English-dominant countries(where Google’s national default search engine does not offer many

of the language features) One implication is that teachers may ously take it for granted that in the foreign language classroom, pupilswill be using the target language for speaking, reading, and writing,rather than developing avoidance strategies

errone-In summary, this article sets out to examine the language choicespupils can and do make when using Google, and particularly howavoidance of the target language is aided by (Swedish) Google andhow translation options are jointly negotiated over time The implica-tions of actual processes in the real world as opposed to those idea-lised in project-based learning are then discussed in the final sectiontogether with suggestions for dealing with avoidance

OVERVIEW OF PREVIOUS RESEARCH

Avoidance

In the second language acquisition (SLA) literature, avoidance istypically recast as a (negative) communication strategy Traditionally,communication strategies have been taken to mean “L2 [second lan-guage] problem-management efforts, dealing with language produc-tion problems that occur at the planning stage” (D€ornyei & Scott,

1997, p 177) However, there are a number of reasons why this sizablebody of research is marginal to the interests of this study First, as isevident from the above quotation, the focus has been on languageproduction, primarily oral production Second, the numerous taxono-mies do not always include the term avoidance (cf D€ornyei & Scott,

1 With few exceptions Google.se proved to be the sole search engine used in all our ish data sets.

Swed-1997, pp 196–197), and even Tarone’s (1978) early but frequentlycited study does not group language switching as an avoidance strat-egy, but rather as a subcategory of conscious language transfer.2 Eventhough language switching is included in most taxonomies (e.g., Bialy-stok, 1983; D€ornyei & Scott, 1995; Færch & Kasper, 1983; Poulisse,Bongaerts, & Kellerman, 1990; Willems, 1987),3 cited examples of lan-guage switching invariably involve one-word switches using the speak-er’s first language Third, few empirical studies on language strategiestake seriously the fact that communication strategies in general andavoidance in particular arise as situated activities in social interaction(but see Markee, 2011) Instead, this study focuses on reading (not alanguage production skill), which through a series of jointly negoti-ated language switches can be transformed into an activity performedsolely in the learners’ first language.

Apart from the term avoidance and the involvement of a languageswitch, the problem-orientedness of communication strategies (D€ornyei &Scott, 1997, p 182; Færch & Kasper, 1983, p 31; or “problematicity,”Bialystok, 1990, p 3) is relevant to this study That is, communicationproblems, especially those arising from deficient language resources,trigger learners to make use of strategies in order to solve them Thus,problems to do with reading proficiency may also result in avoidancestrategies Furthermore, drawing on a conversation analytic framework,

I concur with Markee’s (2011) respecification of avoidance as “a locallycontingent practice that is collaboratively co-constructed by partici-pants in real time as a topic of interaction during the course of natu-rally occurring institutional talk” (p 603)

Project Work and Project-Based Learning

Project-based learning (PBL) is generally attributed in some way oranother to Dewey’s problem-based approach to learning (Beckett,

2006, pp 3, 5) It was adopted in the field of English language ing from the mid-1970s, not least because it “espoused principles oflearner-centred teaching, learner autonomy, the negotiated syllabus,collaborative learning, and learning through tasks” (Hedge, 1993,

teach-p 276) Although project-based approaches may differ even within aforeign language learning context, many accounts of project-basedlearning enumerate the following features as central: a sequence ofactivities over time, including “the gathering, processing, and report-

2 Furthermore, Tarone (1978) classes transfer as a positive strategy.

3 Note, however, that language switching is often given other names, such as code-switching and transfer.

ing of information from target language resources” (Stoller, 2006,

p 24); the use of authentic materials; the integration of languageskills; learner collaboration; learner autonomy; and a process andproduct orientation (e.g., Debski, 2000a; Eyring, 1997; Hedge, 1993;Stoller, 2006) Although much of the literature seems to examine PBL

as an alternative approach to language learning, it is most likely used

as a complementary approach in the guise of occasional project work(van Lier, 2006, p xii), as in the case of this study

Some of the studies of project-based learning focus on what hasbeen given the acronym PrOCALL (project-oriented computer-assistedlanguage learning); that is, PBL which makes use of computer technol-ogy, including the Internet (e.g., Debski, 2000b; Jeon-Ellis, Debski, &Wigglesworth, 2005) One major difference in PrOCALL is what vanLier (2002) terms triadic interaction: When two students are workingtogether at the computer they are mainly working side by side (ratherthan face to face) with a joint focus on the computer screen, whichserves as “a third interlocutor of sorts” (p 147)

The few PrOCALL studies that focus on specific language skillsfocus mainly on spoken skills (Ewing, 2000; Jeon-Ellis et al., 2005) orwritten skills (e.g., Kessler, Bikowski, & Boggs, 2012) However, theskill most under scrutiny here is reading, and although all aspects ofthe above list of features central to PBL are of relevance to thisstudy, the focus is mainly on what occurs in the searching and gather-ing of the materials stage of project work Mostly it is assumed that thereading will be done through the target language, partly throughspecific reference to authenticity, but rarely is it made as explicit as inthe aforementioned quotation by Stoller (2006) At the same time,these features can be seen as idealised ones, either by framing them asrecommendations (e.g., Stoller, 2006) or occasionally even referring tothem as an idealisation (Debski, 2000a, p 311) This study adopts ashift in focus from the intended aims and underpinnings of PBL towhat actually happens in the classroom; in other words, focusing

on what Breen (1989, p 188) terms the task-in-process rather than thetask-as-workplan (see also Seedhouse, 2004, p 93) Indeed, Seedhouse(2004, p 251) posits that the major contribution which a conversationanalytic methodology can make in the field of second language acqui-sition research is to demonstrate the benefits of this shift in focus

Conversation Analysis

Over the past decade or so, there has been an upsurge in the ber of interactional studies of the second language classroom, examin-ing both teacher–pupil interactions as well as interactions between

num-peers Much of this work has been linked to the processes of secondlanguage learning, sometimes termed conversation analysis for second lan-guage acquisition (CA for SLA; e.g., Hellermann, 2008; Markee, 2008;Markee & Kasper, 2004; Mori & Hasegawa, 2009; Pekarek Doehler,2010; Seedhouse, 2004) However, there is still relatively little “strongsocio-interactionist” research (Mondada & Pekarek Doehler, 2004,

p 502) which investigates empirically the use of computers in the eign language classroom, and even less involving computer-assistedproject work (but see Cekait_e, 2009; Musk, 2011) One cornerstone of

for-a strong socio-interfor-actionist for-approfor-ach is thfor-at socifor-al for-action “involves thelearner as a co-constructor of joint activities, where linguistic and othercompetencies are put to work within a constant process of adjustmentvis-a-vis other social agents and in the emerging context” (Mondada &Pekarek Doehler, 2004, p 502) To understand the co-construction ofactivities, it behooves the analyst to adopt a participant (emic) perspec-tive, by drawing on how participants themselves signal their under-standing of each other’s actions This perspective, in turn, requiresattention to the unfolding details of interactional trajectories aided bytranscriptions (see the Data and Method section) As to the interac-tional data, “naturally occurring” data is preferred in CA studies(Hutchby & Wooffitt, 2008, p 12); that is, recordings of social interac-tions which would have taken place anyway without the researcher’sintervention

Data and Method

The data for this study comprise project work collected from theEnglish class of one Swedish secondary school Four pairs of pupils, sixgirls and two boys, in year 9 (about 16 years old) were video-recordedwhile they were carrying out Internet searches for information sources

To facilitate the recording process and to enable us to record one pair

at a time, the pairs took it in turn to carry out their searches in a sional computer room, with a table, chairs, and a laptop There was,however, no printer available in this room Therefore, rather thanbeing able to print out relevant sources directly, students were told tocopy and paste their sources into a Microsoft Word document and savethis for later use These provisional circumstances in themselves have

provi-no doubt impinged on students’ practices of gathering sources, butthis aspect has probably not significantly affected what constitutes themain focus of this study, namely the language choices they made asregards what sources to save and the language in which to save them.The video recordings were carried out by two student teachers atLink€oping University, Linn Olsen and Sara G€othman, in the autumn

of 2010 Two video cameras were used for this: one from behind ing on the screen, and one from the side capturing what the pupilswere doing These recordings were then synchronised in the computersoftware ELAN (Max Planck Institute, n.d.) to facilitate a fine-grainedmoment-by-moment analysis of what was going on simultaneously.The project work was about the infamous serial killer Jack the Rip-per The instructions were in Swedish, and they can be found in Fig-ure 1 with a translation in English.4 Although this study focuses only

focus-on the search and informatifocus-on-gathering stage of the project, pupilsthen went on to write a manuscript as a basis for an oral presentation

in front of the class approximately 3 weeks after the start of the ject Only two of the pairs agreed to have their oral presentationsrecorded, but otherwise all eight pupils signed forms of consent fulfill-ing the main ethical principles laid down by the Swedish ResearchCouncil (Vetenskapsradet, 2002).5

pro-Although conversation analysis is the main method of analysis when

it comes to examining the details of the pupils’ search and tion-gathering trajectories, it also became necessary to create more of

informa-a bird’s-eye view of the entire process to uncover globinforma-al pinforma-atterns oflanguage choices and possible language avoidance In order to visual-ise these patterns more clearly, I have adapted (and simplified) a flowdiagram originally constructed by Sara G€othman (2012) A brief exam-ple and a key to this flow diagram is found in Figure 2 The fourcolumns correspond to different steps in the search and information-gathering process The first column, “Google search word,” indicatesthe search word(s) typed into the Google search box The second col-umn, “Google suggestions,” relates to the interactive features of the

Fakta om Jack the Ripper—Jack Uppskäraren

en seriemördare som skakade England i slutet av

• teorier om vem han var

• något annat ni själva finner intressant

Facts about Jack the Ripper—Jack Uppskäraren

a serial killer who shook England at the end of the 1880’s

Search for facts about:

• the murders

• the police interrogations

• news articles

• theories about who he was

• anything else you find interesting

FIGURE 1 Task instructions in the Swedish original and in translation.

4 After the first two pairs were recorded (in focus here), the student teachers decided to translate the instructions into English for the second two pairs This probably had an effect on some of the search strings, particularly for the second two pairs The first two pairs searched almost exclusively for “jack the ripper” (with the addition of tidningsartik- lar [newpaper articles] in one single case).

5 These include information to participants, their consent to take part, confidentiality in respect of their identities, and restricting the use of the data to research purposes only.

search engine, whereby Google offers options such as translation ties or alternative search words (based on the ones already used) Thethird column, “Selected website,” shows which websites are activelyselected from Google’s list of search hits The last column, “Saved toWord,” shows when text (often including images) is copied and pastedinto a Word document and saved The arrows indicate the sequentialflow of the process over time, as do the numbered lines in the figure.The colour coding indicates whether Swedish or English was used andwhether translation options were adopted.

facili-After identifying significant language choices from the flow grams, transcriptions were made of these interactional sequencesaccording to general CA practice (cf Jefferson, 2004) However, apartfrom transcribing talk (indicated by the talking head icon ) and add-ing important features of gesture and gaze in double parentheses, thetranscriptions used here are augmented by additional lines to showthe use of the computer (using the computer icon ) The icons come

dia-at the beginning of each line At significant moments, screen shots arealso integrated into the transcription The benefit of these additionalfeatures is to portray the multimodal nature of the interaction moreaccurately On the downside, these multiple lines frequently representactions that are taking place simultaneously, which is difficult to showaccurately in a traditional CA transcription To compensate, the lines

Google search word Google suggesƟons Selected website Saved to Word

‘translate?’

Metropolitan Police Service

Metropolitan Police Service

English website translated into Swedish

are numbered according to speech and timed pauses, so that talk (orpauses) and simultaneous actions are given the same line numbers Afull key to transcription conventions can be found in the Appendix.

ANALYSIS

The analysis comprises three main sections: (1) the language-relatedoptions available in Google, (2) the search trajectories of the two pairswho activated the translation options, and (3) a detailed examination

of their use of translation tools

Google’s Language-Related Options

The data reveal six main ways in which users can make languagechoices, only two of which use machine translation.6 The following listcompares and explains the choices made by all four respective pairs.Language of typed-in search word(s) The very first action of typing

a search string into Google may involve a language choice The guage of the search string then largely dictates the language of thehits Proper nouns, such as “Jack the Ripper,” on the other hand, usu-ally generate a mixture of English and Swedish hits Table 1 comparesthe language choices of the four pairs

lan-Even at this stage it is evident that the first two pairs (in focus here)avoided using English, which had a direct effect on the language ofthe hits Proper nouns yielded hits in either English or Swedish, butSwedish words automatically limited hits to Swedish-language sources.Limit the search to Swedish webpages There is an option inGoogle designed to limit the searches to either “Pages written in Swed-ish” or “Pages from Sweden.” These are located in the left-hand col-umn as illustrated in Figure 3 (numbered 1 and 2, respectively) TheTABLE 1

Language of Typed-In Search Words

Maria and Johanna Nadja and Ronja Elin and Deborah Charlie and Max

effect of selecting this option is to limit the language to Swedish only(even in the case of “Pages from Sweden,” the vast majority of theseare in Swedish) Neither of these options were used, but a search tipthat appeared prominently at the top of the list of hits, saying S€ok end-ast efter svenska resultat [Search only for Swedish results] was noticed ontwo occasions: once by Nadja and Ronja (cf line 6 in Figure 7) andonce by Charlie and Max This is illustrated by Figure 4.7 Nevertheless,the appearance of this tip did not lead to any uptake.

Selecting English and/or Swedish hits The third option, selectingwhich Google hits to follow up on, was relevant to all four pairs, insofar

as each selection also involved choosing a language even if language

FIGURE 3 Swedish Google’s list of search hits for “jack the ripper.”

7 This is evident from their talk and the movement of the cursor, respectively.

was not always the main selection criterion The language selections arecollated in Table 2, where I have also added which selected hits weretranslated into Swedish (to be dealt with in greater detail later).

Besides the first two pairs selecting and entering far fewer webpagesfrom Google’s list of hits, what is notable here is that the only webpag-

es they selected in English were then translated into Swedish, which iswhy their selection of English webpages is in parentheses in Table 2.Google’s suggestions for alternative search words This is whenGoogle suggests alternative search words/strings, which can be found

at the bottom of each page of hits under the heading S€okningar ade till [Searches related to] [+ original search string] (e.g., Searchesrelated to jack the ripper).8These suggestions do not necessarily involve anew language choice However, some of the alternatives may be inSwedish, such as S€okningar relaterade till seriem€ordare [serial killers], asillustrated by number 5 in Figure 3 Consequently, clicking on asearch string including Swedish automatically limits search hits toSwedish-language websites

relater-As Table 3 shows, all of Google’s suggested search strings adopted

by the four pairs were in fact Swedish ones, which immediately

FIGURE 4 Swedish Google’s option of limiting search hits to Swedish ones.

TABLE 2

Selecting English and/or Swedish Hits

Maria and Johanna Nadja and Ronja Elin and Deborah Charlie and Max

eliminated all the English hits until new search words were typed in.(This is exemplified by Maria and Johanna in line 4 of Figure 6.)Translate this page This is the first of the main translation options

in the Swedish Google After the web address at the end of the secondline of each hit in English, there is a clickable option,9 which says

€

Overs€att den h€ar sidan [Translate this page] (see Figure 3, number 4)

By clicking on this option, the original webpage is automaticallymachine translated from English to Swedish whilst maintaining theformatting of the original English version This option was onlyadopted by the first two pairs, as can be seen in Table 4

In addition to Nadja and Ronja’s uptake of the Google option, theyalso used the translation function in Word (see Figure 7, line 2).Translate list of hits Another prominent translation option may beoffered by Google at the bottom of a page which includes English hits,where it says Se €oversatta engelska resultat f€or [See translated Englishresults for] [+ original search string], as is shown in Figure 5

This option was used only once, by Maria and Johanna (see ure 6, line 7) However, after they made this choice, the three subse-quent pages they entered were then machine translated intoSwedish

Fig-TABLE 4

Uptake of Google’s Translate This Page Option

Maria and Johanna

Nadja and Ronja

Elin and Deborah

Charlie and Max

TABLE 3

Uptake of Google’s Alternative English and Swedish Search Words/Strings

Maria and Johanna

Nadja and Ronja

Elin and Deborah

Charlie and Max

*Because Elin and Deborah did not log out after they finished, when Charlie and Max start

up their search, Elin and Deborah’s last search string, artiklar om jack the ripper [articles about jack the ripper], is already in Google’s search box Charlie and Max therefore simply reuse the girls’ search term.

9 When the data were collected, this option appeared instead at the end of the first line.

Search Trajectories

Now that we have examined the main language options afforded byGoogle, let us focus on the search trajectories of the two pairs whoactively made use of Google’s translation options Even though theirtrajectories were not identical, they shared many features These trajec-tories are illustrated by the flow diagrams in Figures 6 and 7 Even acursory examination of the colour coding in both figures reveals thatEnglish (indicated by the black/red boxes) is abandoned as a readingmedium as soon as they select their first website in English (the

“Metropolitan Police Service” in both cases).10

Apart from these two pairs making extensive use of the translationoptions, what also becomes immediately apparent from the flow dia-grams is the comparatively large proportion of machine-translated web-sites (indicated by black/red-rimmed grey/yellow boxes) that theydeemed to be useful sources, insofar as they were copied and pasted intoWord and then saved Indeed, in Maria and Johanna’s case (Figure 6)only one website originally in Swedish was saved as opposed to the twotranslated into Swedish For Nadja and Ronja, two were saved in the

FIGURE 5 Swedish Google’s Translate English Hits option.

TABLE 5

Uptake of Google’s Translate English Hits Option

Maria and Johanna Nadja and Ronja Elin and Deborah Charlie and Max Translate

hits

10 The only exception (Figure 7, line 8) is commented on later.

original Swedish version and two were translated into Swedish In ther case was anything saved in English This contrasts starkly with theother two video-recorded pairs, who selected English and Swedish web-sites, did not machine translate anything, and saved text in both Englishand Swedish (according to the language of the original website).11

nei-Zooming In on Significant Choices

Let us now turn our attention to how these pairs’ language choiceswere negotiated and managed at strategic points in the overall trajec-tories This focus involves going beyond the broad brush strokes of the

Google search word Google suggesƟons Selected website Saved to Word

‘translate?’

Metropolitan Police Service

Metropolitan Police Service

FIGURE 6 An overview of Maria and Johanna’s search trajectory.

11 Elin and Deborah saved text from three English sources and one Swedish source (which they used twice) Charlie and Max saved text from four English sources but only one Swedish source, and part of the Swedish text was even erased when they found the equivalent information in English.