TOWARDS A FRAMEWORK FOR BUILDING A TOOL

TO ASSIST L2 WRITING BASED ON SEARCH

ENGINES CAPABILITIES

The Case of English Phrases and Collocations

Grami M. A. Grami

Department of European Languages, King Abdulaziz University, Jeddah, Saudi Arabia

Basem Y. Alkazemi

Department of Computer Science, Umm Al-Qura University, Makkah, Saudi Arabia

Keywords: ESL Writing, Collocation, Lexical Phrases, Search Engines.

Abstract: Writing is one of the most difficult skills to learn and it gets more complicated when students learn to write

in another language. In fact, results of language proficiency tests such as IELTS shows a systematic

tendency of Arab students scoring less in writing than any other skill. Obviously there are various reasons

that complicate the task of ESL writing but we focus here on the incorrect combination of words, more

specifically collocations and lexical phrases, and its relation to L1 interference. We propose an alternative

approach in teaching ESL writing which utilizes common search engines in finding out not only correct

usage of words but systematic types of errors so they can be avoided. Moreover, students can use such an

approach to validate their writing style in their coursework.

1 INTRODUCTION

Writing is probably the most difficult language skill

for many ESL/EFL students which becomes evident

when we examine results of proficiency tests such as

IELTS. In 2009 for instance, the average score of

IELTS academic test takers in writing according to

Cambridge ESOL: Research Notes (2010) was 5.51

(the maximum score is 9) which was the lowest band

scored ever and well below the overall average of

5.88.

When we closely inspect the writing result of

Arab test takers (See table 1 below) we discover that

they scored the lowest mean (4.89) of any linguistic

background which begs the questions ‘why?’ and

‘how can their writing be improved?’

There are many reasons that make Arab ESL

writers struggle but we are trying to focus on the

area of combining words in this project. L2 writers

in general encounter difficulties when attempting to

produce accurate English sentences using the right

combination of words that also fit into the correct

Table 1: Mean IELTS score of some first languages.

Academic Listening Reading Writing Speaking Overall

Amharic 4.78 5.64 5.62 6.11 5.60

Arabic 5.14 4.96

4.89

5.65 5.23

Bengali 5.85 5.44 5.54 5.87 5.74

Chinese 5.72 5.85 5.19 5.28 5.57

Dutch 7.95 7.79 6.79 7.60 7.60

contexts of usage. A key element for such difficulty

usually corresponds to the interference of the mother

language when constructing sentences, followed by

instant interpretations into English which we believe

is a contributor to ESL writing difficulty among

Arab learners.

One resource learners can use to check their

sentences and the context in which they occur is the

Internet search engine. In fact, the literature shows

that search engines could function as free online

resources readily available to many ESL learners

and for various purposes. In our case, the use of

225

M. A. Grami G. and Y. Alkazemi B..

TOWARDS A FRAMEWORK FOR BUILDING A TOOL TO ASSIST L2 WRITING BASED ON SEARCH ENGINES CAPABILITIES - The Case of English

Phrases and Collocations.

DOI: 10.5220/0003331902250230

In Proceedings of the 3rd International Conference on Computer Supported Education (CSEDU-2011), pages 225-230

ISBN: 978-989-8425-49-2

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

search engines could yield further information about

how phrases and collocations are used, by who and

to what extent. Apparently mainstream engines are

general tools by default and as a result are very

likely to produce different results some of which are

confusing or even misleading. To filter the general

outcome, certain measures are required including

observing the number of results and where they

come from. In other words, the more pages that use

exactly the same phrase/expression, the more likely

the phrase in question is correct. Similarly, the more

results that originate from carefully edited sources

such as recognised institutions, media organisations,

government bodies and global corporations, the

more likely these results are to be trusted. The

hypothesis of this research therefore is that "search

engines can improve learning English writing at the

lexical phrases/collocations level."

This project would be divided into three stages,

the first which is the subject of this study involves

testing our theory against results generated by

Google and setting criteria for judging incorrect

phrases and collocations used by Arab ESL learners

and this stage is the main focus of this paper. The

second stage would be to conduct an empirical study

in a university-level ESL writing course where

students would be using Google to check phrases or

collocations they are unsure about. The final part of

the study would build on the findings and

recommendations of the two previous stages and

design a tool which can be incorporated into readily

available document processing programmes (e.g.

MS-Word

and Open Word) and which can

compare certain phrases in texts against the actual

results of recognised search engines like Bing

and

Google

. We intend to identify a new application

of an already existing technology. In theory, the tool

should assist L2 writers by significantly reducing the

time required to verify every phrase needs checking

and by suggesting various validating techniques

(filters) which can be selected individually.

2 LITERATURE REVIEW

2.1 Collocations

These are closely related concepts and are often

considered to be problematic areas among L2

learners even advanced ones especially in the lexical

level of ESL writing as reported by various experts

in the field (Halliday & Hassan, 1976; Sanders &

Pander-Matt, 2006; Xiu-lian, 2007; Yin, 2009; and

Stapleton & Radia, 2010)

Halliday & Hassan (1976) mention that

collocation is part of lexical cohesion, and it is

somehow associated with corpus linguistics as

mentioned by Lombard (1997). It is defined by

Halliday (1961) as “the syntagmatic association of

lexical items, quantifiable, textually, as the

probability that there will occur at n removes (a

distance of n lexical items) from an item x, the items

a, b, c ...” In layman terms, collocation refers to

certain words commonly used together which co-

occur more often than only by chance. Examples of

collocation in English include the use of verbs like

‘do’ and ‘make’ and adjectives ‘quick’ and ‘fast’

with certain nouns, for instance one can say ‘do your

homework’ and ‘make a sandwich’ but it is unusual

to swap the verbs in these commands even if

syntactically correct and the same applies to ‘fast

train’ and ‘quick shower’. (Guo & Zhang, 2007)

This is one reason why collocation is confusing

because ‘do’ and ‘make’ are almost synonymous to

many ESL students who would assume, probably

when applying L1 analogy, that they are

interchangeable. Verb + noun collocation is possibly

the most common type but there are also verb +

adverb (vividly remember), adverb + adjective (fully

aware), adjective + noun (excruciating pain), and

noun + noun (ceasefire agreement) collocations.

2.2 Lexical Phrases

On the other hand, English phrases - more

specifically lexical phrases - are defined by

Nattinger and DeCarrico (1992: 1) as “chunks of

language of varying length … that occur more

frequently and have more idiomatically determined

meaning than language that is put together each

time.” Other definitions like the one provided by

Lindstromberg (2000) also suggest that phrases

should be treated like units of language rather than a

collection of words which therefore makes phrases

largely inflexible. For example, ‘by and large’

means ‘generally’ and ‘as well’ means ‘too’. In this

sense, collocations and lexical phrases are similar as

both refer to fixed association of words.

Lindstromberg (ibid.) mentions that collocation is

the wider term and it refers to both fixed lexical

phrases and relatively loose association of words.

2.3 Teaching Collocation

and Phrases in ESL Classroom

Language teaching experts have recognised the need

to teach and learn more collocations and phrases in

L2 classroom. The common approach is to elicit lists

CSEDU 2011 - 3rd International Conference on Computer Supported Education

226

of words that commonly collocate with others (e.g.

verbs such as ‘do’, ‘make’, ‘take’, ‘have’ and

‘break’) and explicitly teach them to students. The

same technique applies to common English phrases.

One good example of such effort is Oxford

Collocations Dictionary for Students which is also

available online.

To us however, teaching word lists is not always

a practical solution. There are many reasons why we

take this line of thinking one of them is neatly

explained by Altenberg (1991) who mentions that

almost 70% of words are part of recurrent

combinations and English phrases count to the

thousands. Another reason would be the time and

effort required to go through each and every

word/phrase and the limited success when it comes

to actual production later on. We therefore argue that

teaching lists of words with no reference to the

context in which they are used could significantly

reduce students’ learning achievement, a point

confirmed by Lindstrobmerg (2000), and Nattinger

and DeCarrido (1992).

2.4 L1 Interference

Language transfer or L1 interference has been a

central point in second language acquisition (SLA)

and language teaching and therefore it has been

well-documented and researched (e.g. Odlin, 1989;

White et al., 1991; Lightbrown & Spada, 1997;

Brown, 2000; Picard, 2002; and Bordag &

Pechmann, 2007). In general terms, this

phenomenon happens when language learners apply

knowledge from their mother tongue to a second

language, which in our case would be applying

Arabic structures into English. (Ryan & Meara,

1991; and Fender, 2008) We more specifically argue

that a major contributor to the incorrect usage of

collocations and phrases among Arab ESL learners

is the interference from similar structures in L1.

In fact, we believe there is ample evidence from

the literature and our own investigation to support

this theory. For example, one unusual combination

of words, supposedly to form an awkward

expression, is what organisers of the 2010 Saudi

students’ conference in the UK used for a slogan

which reads ‘from different soils into one soil’. As

far as we are aware, no such expression exists in

English and to make sure we consulted Cambridge

Dictionary of Idioms in addition to more general

search engines to look for similar combination of

words but came up with nothing, a simple search for

the exact phrase using quotation marks in Google

returned no results. We believe the aforementioned

slogan can only be traced to a relatively common

expression used in Arabic journalism and simply has

just been literally translated into English. Another

example we encountered was learners writing "I

want to register my voice in MP3 format" indicating

that they actually want to record or tape their voice

into a digital recorder.

2.5 Search Engines and ESL Writing

Many experts recognise the important role played by

technology and online resources in modern ESL

learning. Stapleton & Radia (2009) for instance

believe technology contribution to the field of L2

writing has been known as early as when word

processing programmes became widely available.

Lincoln (2003) more to the point of this study

recommends ESL teachers to explicitly teach their

students how to use search engines as part of their

learning.

However, although the literature of educational

technology acknowledges the existence of such a

technique among ESL student writers as using

search engines to check phrases and collocations

(Stapleton & Radia, 2009; and Guo & Zhang, 2007),

it vaguely describes how these students actually use

these resources. The available literature in fact

hardly answers basic questions like how widespread

is this practice?, from where have students learned

this technique?, what measures do they use to filter

search results?, what renders a phrase/collocation

acceptable?, how often do students use this

technique?, and are students qualified to use general

search engines in demanding situations like assessed

ESL writing?

Another issue that may affect available text

processing software such as MS Word is that they

cannot identify certain incorrect collocations nor can

they show how popular a phrase/collocation is. For

instance, a phrase we considered in this paper was

‘from different soils into one soil’, which if was

searched in Google returns no results, i.e. it does not

exist, unlike MS Word which shows no style errors

at all.

3 METHODOLOGY

For the first stage of this study, we gathered samples

of possibly incorrect phrases and collocations from

original texts written by Arab ESL students (n = 37).

We then checked these combinations of words

against Google by using some preset criteria to filter

the returned search results which are the number of

TOWARDS A FRAMEWORK FOR BUILDING A TOOL TO ASSIST L2 WRITING BASED ON SEARCH ENGINES

CAPABILITIES - The Case of English Phrases and Collocations

227

returned results of the possibly incorrect

combination and the alternative combination, the

format (.doc(x) and .pdf against other formats of less

academic association) and their source (institutional,

academic, governmental against other sources of less

restricted nature).

As for the raw number of results, we were

looking for figures to indicate the popularity or

otherwise of an expression used by ESL learners.

Sources of documents were checked to determine

whether results come from trusted websites and/or

official documents (academic, organisational,

institutional and/or governmental).

Additionally, results of incorrect phrases and

collocations usage can be used to determine the

scale of the problem, and by inspecting the

geographical information of these results, it can also

be determined if these errors are more prone to be

committed by ESL learners of specific linguistic

backgrounds e.g. Arab learners. In other words, we

intend to establish whether certain errors are

originated from geographic domains more than it

would be possible only by chance.

We identified three possible categories of

incorrect usage and set measures to deal with these

different possibilities accordingly; if a

collocation/phrase yields no similar results then we

judge it ‘isolated and incorrect’ then try to guess

what the writer intends to say usually by referring to

corresponding ideas in his/her L1. If however its

return results are found commonly in Arab domains

but not other sources then we examine the

possibility of L1 interference and how widespread it

is. Finally, if a collocation/phrase does not exist in

English but is very common among ESL learners of

different backgrounds then we categorise it under

‘common errors’ regardless of the writer’s mother

tongue. In every case we proposed an alternative

option which we think is more accurate and we

check our alternatives against Google as well.

4 RESULTS AND DISCUSSION

The results show a definite answer when it comes to

choosing between two possible collocations/phrases;

in every case investigated, the alternative option

significantly outnumbered the original

phrase/collocation. We therefore recommend using

search engines results to indicate which string of

words is more likely to be correct if students are in

doubt choosing between more than one possible

combination. (See table 2 below)

As for the different filters used, the location filter

(Arabic domains) can tell us - to some extent - if an

error is common among learners from this particular

background. We for instance have identified two

incorrect usages of the preposition ‘from’ which

were found in texts written by Arabs and we can

relate this phenomenon to L1 interference. In other

words, a language teacher can now address this

Table 2: Google Results of Original Phrases.

Original Text

Raw Search

Results

Filtered Results

(by location)

Filtered Results

(by format)

Filtered Results

(by source)

Alternative

Choice

Search Results

“register my voice” 28,500 2 Pages

8 .pdf

1 .doc

none

“record my

voice”

146,000

“different soils into one soil” none none None none -- --

“Speak in English” 797,000 33,900 Pages

33,700 .pdf

3,820 .doc

3,060 .gov

7,860 .edu

1,120 .ac.uk

“Speak

English”

8,220,000

“Get my advantages” 3 none None none

“Get my

rewards”

34,100

“near from my family” 5 1 Page None none

“Near my

family”

985,000

“better from ” (comparison) * 1,330 Pages None none

“better than

”

91,900,000

“to talk English” 299,000 7,630 Pages

14,100 .pdf

1,020 .doc

8 .ppt

4,240 .edu

321 .gov

489 .ac.uk

“to speak

English”

9,040,000

“His days are finished” 2,310 none

3 .pdf

1 .doc

none

“His days

are over”

89,900

CSEDU 2011 - 3rd International Conference on Computer Supported Education

228

problem by asking students not to constantly

translate meanings from their mother tongue.

Another example of L1 interference is the

incapacity to differentiate between verbs such as

‘record’ and ‘register’ and how to use them in their

proper contexts. Again, this mistake has occurred in

texts written by Arabs which further supports the

hypothesis that L1 interference is widespread

indeed.

However, in extreme cases when a whole phrase

originated only in Arabic is wholly translated into

English with no regard to L2 conventions we found

that no similar results were found in any other

website. The only case we came across was using a

relatively common expression in Arabic by a group

of Arab students in the UK to promote their

conference but we also accept that it is not

uncommon to see more of the same. Google yields

no results and we could not come with an expression

that conveys the same meaning.

In few occasions, the choice of words might be a

personal preference or follows conventions of

formality and having a fewer number of results does

not always mean that our alternative choice is

correct and the original is not. For example ‘talk

English’ shows much less results than our preference

‘speak English’ but as the former collocation

abundantly appears in edited academic and official

websites as well as revised documents, one cannot

reject it and accept the latter simply because it shows

significantly more results.

The filters are interesting methods to determine

various characteristic of phrases. We already have

discussed that ‘location’ helps us understand if an

error is common among certain group of learners.

We also found that results filtered by source can be

stricter than results by format.

5 RECOMMENDATIONS FOR

ESL CLASSROOM

Our recommendation for language teachers therefore

is to avoid translating whole expressions from the

mother tongue to the target language and focus

instead on teaching expressions and phrases

commonly used in English. We also suggest that it is

recommended that when students write in a foreign

language that they follow its conventions without

constantly referring back to their mother tongue. In

fact, almost all the errors we identified can be traced

back to Arabic in some cases with no regard to L2

conventions at all as in the case with ‘from different

soils into one soil’.

As for using search engines, we suggest that if a

result appears in great numbers in educational and

official websites then it should be treated as a correct

combination of words. Finally, although the main

purpose of the study is to aid ESL writers the tool

we aim to develop can be used innovatively to serve

other purposes as well. For instance, search engines

can show results from specific regions and in

websites of certain languages which means one can

check how widespread an error is among learners

from a specific background and compare that to

others. The identification of these errors can further

help the research into L1 interference and the role of

context in ESL learning.

6 LIMITATIONS

AND FUTURE RESEARCH

Our investigation of the writing samples indicates a

widespread problem of incorrect usage of phrases

and collocations among Arab ESL writers chiefly

due to the interference of their L1. However, it is

difficult to find systematic patterns of errors within

writing of students of a certain background without

an empirical study that involves asking students to

write about topics very likely to generate such

patterns and - in our case - students can also check

combination of words they are not sure about against

search engines and from observing them doing so

we shall be more informed about the techniques

used and if these can be integrated in our tool model.

Our attention therefore should move to actual

ESL Arab writers who use Google and other search

engines to help them determine whether a

phrase/collocation they use is acceptable. This

proposed empirical study would be the second stage

of our project in which we aim to gather as much

information about search techniques and incorrect

usage of words as possible. The results should help

us better understand how search engines work and

how can we further develop various methods to have

as accurate results as possible. All the data gathered

would then be considered when we finally design an

open-source support tool which can be prompted to

search certain phrases and classify results according

to the filters we suggest.

7 CONCLUSIONS

Having reviewed the literature and assessed the scale

and widespread of the problem, we would argue that

there is a feasible chance ESL students’ writing can

TOWARDS A FRAMEWORK FOR BUILDING A TOOL TO ASSIST L2 WRITING BASED ON SEARCH ENGINES

CAPABILITIES - The Case of English Phrases and Collocations

229

be improved using Google and similar search

engines in tackling the problematic areas of using

collocation and lexical phrases. The incorrect usage

of these items happens in large part due to L1

interference as we attempted to establish in this

study. It is however not a very practical solution to

expect students to read and remember every lexical

phrase and collocation list available. We therefore

propose an alternative approach in using search

engines which is based on two concepts; simplicity

and learning by doing.

REFERENCES

Altenberg, B. (1991) ‘Amplifier collocations in spoken

English’, In S. Johansson & A. Stenstrom (eds.)

English computer corpora, pp 127 – 147, Berlin:

Mouton de Gruyter.

Brown, H. D. (2000) Principles of Language Learning and

Teaching, 4

ed., London: Longman.

Cambridge ESOL: Research Notes, Issue 40/ May 2010,

Cambridge University Press.

Fender, M. (2008) 'Spelling Knowledge and Reading

Development: Insights from Arab ESL Learners',

Reading in a Foreign Language, 22, pp. 19-42.

Guo, S. and Zhang, G. (2007) ‘Building a customised

Google-based collocation collector to enhance

language learning’, In British Journal of Educational

Technology, Vol. 38 No. 4, pp 747 – 750.

Halliday, M. A. K. (1961) ‘Categories of the Theory of

Grammar’, In Word, Vol. 17, pp 241 – 492.

Halliday, M. A. K. & Hassan, R. (1976) Cohesion in

English. London: Longman.

Lightbown, P. M. and Spada, N. (1997) ‘L1 constraints in

interlanguage judgments of grammaticality’, Paper

presented at the 17th Annual Second Language

Research Forum, Michigan State University, East

Lansing.

Lincoln, K. (2003) ‘Teaching Search Engines to ESL

Students: Avoiding the Avalanche’, In The Internet

TESL Journal, Vol. 9 No. 6, available on <iteslj.org>,

accessed on 11 August 2010.

Lindstromberg, S. (2003) ‘My good-bye to the lexical

approach’, In Humanising Language Teaching, Vol. 5

No.2.

Lombard, R. J. (1997) Non-native speaker collocations a

Corpus-driven characterization from the writing of

native speakers of Mandarin, PhD Thesis, MI: UMI.

Nattinger, J. R. & DeCarrico, J. S. (1992) Lexical Phrases

and Language Teaching, Oxford: Oxford University

Press.

Odlin, T. (1989) Language Transfer: Cross-Linguistic

Influence in Language Learning, Cambridge:

Cambridge University Press.

Ryan, A. and Meara, P. (1991) 'The Case of the Invisible

Vowels: Arabic Speakers Reading English Words’, In

Reading in a Foreign Language, Vol. 7 No.2, pp 531

– 540.

Sanders, T. & Pander Maat, H. (2006) ‘Cohesion and

Coherence: Linguistic approaches’, In Brown. K. et al.

(eds.) Encyclopedia of Language and Linguistics, 2nd

edition, volume 2. London: Elsevier.

Spada, N. and Lightbrown, P. M. (1993) ‘Instruction and

the development of questions in L2 classrooms’, In

Second Language Acquisition, Vol. 15, pp 205 – 224.

Stapleton, P. and Radia, P. (2009) ‘Tech-era L2 writing:

towards a new kind of process’, In ELT Journal, Vol.

64 No. 2, pp 175 – 183.

White, L., Spada, N., Lightbrown, P. M., and Ranta, L.

(1991) ‘Input enhancement and L2 question

formation’, In Applied Linguistics, Vol. 12, pp 416 –

432.

Xiu-lian, Y. (2007) ‘Lexical approach to college English

teaching’, In Sino-US English Teaching, Vol. 4 No.

10, pp 22 – 24.

CSEDU 2011 - 3rd International Conference on Computer Supported Education

230