Skip to navigation – Site map

HomeIssues5PapersA Corpus-Based Case Study in Prep...

Papers

A Corpus-Based Case Study in Prepositional AT/TO- Infinitive Alternation using the Lemma “AIM”

Kate James

Abstracts

The lemma ‘AIM’ in its verbal form presents a case of prepositional (at)/to-infinitive alternation. This study demonstrates that whilst users of monolingual and bilingual dictionaries would conclude that the pair is equivalent in use, a corpus study of the two forms enables certain clear distinctions to be made. The most significant of these may be found to be the predominant pattern of use of AIM+at in more formal styles with AIM+to being prefered in less formal discourse. Other uses are examined in context concerning clausal polarity, tense, and collocational elements of semantic significance. The relevance of such a corpus-based study is examined in relation to possible use in the field of stylistics. This study concludes by demonstrating the importance of corpus study to supplement information given in dictionaries, and suggesting the highlighting of distinctions in underlying principles through the use of such data, as well as noting possible relevance to the stylistician.

Top of page

Full text

Introduction

1Traditionally, the corpus-driven lexicographer is concerned with determining the meaning and pattern of words through large-scale data analysis, whereas the stylistician can be seen to concentrate on the variability of linguistic forms in a text and the resulting connotations and artistic effects thus created. Both, however, can be described as using empirical approaches, whether based on language usage determined through corpora analysis, or linguistic evidence in the text, and as pointed out by Mahlberg [2006]:

[B]oth are interested in HOW we say what we say.

2Thus, some overlap may be seen in corpus-based lexical and stylistic investigations. As Stubbs [2005] points out:

Individual texts can be explained only against a background of what is normal and expected in general language use, and this is precisely the comparative information that quantative corpus data can provide.

3As well as the more usual questions of a semantic nature, grammatical features may also interest the lexicographer, keen to distinguish patterns of usage. This may also be of concern to the stylistician, desirous to identify inherent stylistic qualities of a text. In this paper, a study of preposition choice following the lemma ‘AIM’ will be examined from a corpus-based viewpoint for lexicological purposes. Attempts will then be made to determine relevance to stylistics.

4The lemma ‘AIM’ in its verbal form presents a case of prepositional (at)/to-infinitive alternation. Faced with such a choice, the user or learner of English may confront certain difficulties as to the use of the “correct” form, and seek to discover patterns of differing use based on a variety of sources. According to Sinclair [1991: 37], the three main sources of lexicographic evidence listed in order of popularity are:

1. dictionaries
2. [native speaker] users’ ideas about their language
3. observation of language in use

5These categories, based on a lexicographer’s study of the language, may mirror the procedure adopted by a learner seeking to justify the choice of one word form rather than the other although the accessibility or value of each source will vary. As Stubbs [2000] points out, defending the value of observing language in use by means of corpora in such a search:

Native speakers have only very poor intuitions about many pragmatic aspects of language. If they had better intuitions then these pragmatic aspects would be recorded in dictionaries, but they often aren’t. Corpus studies show up patterns which are not visible in single examples...Corpus linguistics is based on methods of observation which make repetitions visible, and the fact of repetition makes quantitative methods essential.

6Echoing these concerns, Aarts [1991: 46] states that intuitive grammar, when combined with a corpus study, becomes an observation-based grammar. Structures such as ‘AIM’ at/to may necessitate such a study, described by Itkonen [1980: 344] as being suited to

unclear cases, i.e. less than well-established rules, as well as actual linguistic behaviour in general, [which] must be investigated by means of observation.

7If the observation of language allows certain rules of use to be established, this may also be of significance to the stylistician.

1. Corpus use

8Although modern dictionaries are corpus-based, Stubbs [2000] points out that

patterns are not explicitly recorded [...] The examples in the dictionary confirm [corpus] data, but the underlying principles are not given.

9This may imply that only a corpus study of a given word will enable correct patterns to be found by comparing data using quantitative techniques. This idea is supported by Aarts [1988] who claims that

only linguists who use corpus data themselves will know that a corpus yields a much greater variety of constructions than one can either find in the literature or think up oneself.

10Using corpora, the great quantity and variety of constructions may be accessed and visualised as a whole and, as Sinclair [1991: 100] states,

language looks rather different when you look at a lot of it at once.

11The corpus used in this study is the smaller collection of corpora from the Bank of English (BoE), the CobuildDirect (CobuildBoE) corpus comprising 56 million words. The individual corpora which make up this corpus may be found in Appendix 1. The corpus was accessed using Telnet and data collected using COBUILD corpus access software (CAS). After loading saved files for home use, certain data were further treated with VisualBasic programming using Microsoft Excel software for ease of access.

2. Corpus analysis

2.1. Corpus query

2.1.1. Query form

12The data analysed were obtained using the queries aim@+0,2at+VERB and aim@+0,2to+VERB. The results obtained were then further selected by retaining only the verbal forms of AIM, thus rejecting the nominal usages aim and aims. As corpus tagging is automatic, accuracy may not be total [Potter 1999: 35] and errors are thus likely be found; indeed, Kennedy estimates the margin of error as being 3% to 4% [Kennedy 1998: 212].

13In order to check the accuracy of the data obtained for this study, a manual verification was carried out on the data bearing verbal tagging as well as the non-retained data with nominal tagging. No obvious errors were noted concerning data obtained for verbal group tagging whereas in data selected for nominal tagging, 35 cases of the verbal form of AIM were found. The incorporation of these omissions to collected data allowed a slight error margin of 1.7% to be corrected. The errors are consistent with De Roses’s identification of the five most frequent tagging errors concerning, in this case, singular nouns assigned instead of a verb [De Rose 1991: 9-14]. Figure 1 shows the number of examples found in the corpus as well as in each sub-corpus, expressing in each case the estimated number of occurrences found per million words of text. 843 cases of AIM at and 1195 instances of AIM to can thus be found.

Figure 1: Frequency of occurrence and average frequency per million words of AIM at and AIM to

Corpus

aim@+0,2at+VERB

Corpus

aim@+0,2to+VERB

Number of occurrences

Average/

million words

Number of occurrences

Average/

million words

bbc

207

79.3

sunnow

218

37.4

oznews

167

31.3

ukephem

177

56.7

npr

115

36.8

times

171

29.7

times

100

17.3

today

167

31.8

today

59

11.2

ukmags

141

28.8

sunnow

47

8.1

oznews

113

21.2

ukbooks

39

7.3

ukbooks

74

13.8

ukmags

34

6.9

bbc

52

19.9

ukephem

33

10.6

usbooks

37

6.6

usbooks

27

4.8

ukspok

28

3.0

ukspok

9

1.0

npr

11

3.5

usephem

6

4.9

usephem

6

4.9

TOTAL

843

TOTAL

1195

14It may be noted that the average number of occurrences changes, sometimes considerably, depending on the sub-corpus examined and on the form requested. Possible reasons for this difference will be proposed in 3.2.

2.1.2. Distance span

15The corpus query was defined in order to allow the inclusion of sentences with two words occurring between the lemma AIM and to or at. This distance was determined by changing the distance values of the original query, starting with 0,0 and continuing up to 0,5. The optimal distance span for AIM at was found to be 0,2, the number of relevant cases steadying after this value. This was also found to be a reasonable distance span for AIM to, evoking a maximum of relevant concordances with a minimum of non-relevant ones, only two occurrences being excluded. As as and to complement the meaning of AIM, no relevant cases seem likely after a span of 0,5.

16Results show that both AIM to and AIM at are separated by intervening words in relatively few cases (seventeen times with AIM at and twelve times with AIM to), that most cases involve quantifiers or adverbs and that very few are used more than once in the data collected. The only co-occurrence pattern which is significantly repeated is AIM only/not only at which occurs six times. However, the small proportion of occurrences with intermittent words does not on its own allow conclusions to be made as to possible differences of use concerning aim to and AIM at.

2.2. Lexical profile

17The study of AIM at/to separated from its surrounding context would give little results or, as Sinclair [1996] states

the idea of a word carrying meaning of its own [can] be relegated to the margins of linguistic interest.

18Stubbs [2001: 63] develops this idea stating that

meaning is typically dispersed over several word-forms which habitually co-occur in text […] [and] these co-occurring word forms ‘share’ semantic features.

19The lemma AIM at/to, as well as having an individual semantic definition, may also be seen as being strongly attached to its surrounding context and as such could be considered part of an extended unit of meaning as defined by Sinclair [1996].

20The analysis of AIM at/to will be based on Sinclair’s model of extended lexical units [1996] & [1998] as developed by Stubbs [2001: 87], who specifies possible constituents of the unit and the relation between them. Both aspects will be analysed using corpus data according to the aforementioned model, given as follows:

  1. collocation

  2. colligation

  3. semantic preference

  4. discourse prosody

  5. strength of attraction

  6. position and positional mobility

  7. distribution in text types

21As Sinclair [1998] mentions, the first four categories are increasingly abstract and according to Stubbs [2001: 88] are all “probabilistic and non directional”, thus needing to be complemented by the last three relations which give more concrete statistical data and define the limits of application. Collocation and semantic preference seem here to be closely related and will be treated together. The strength of attraction, defined by Stubbs [2001: 88] as being “the probability of occurrence of a collocate, grammatical category, lexical set or discourse prosody” given the occurrence of a node will be discussed in each individual part of this study, statistical percentages being given according to actual occurrence.

2.2.1. Collocation and semantic preference

22Sinclair [1991: 170] describes collocation as ‘the occurrence of two or more words within a short space of each other in a text’ and has further noted that the probabilities of lexical items are affected mainly within a span of around four [Sinclair et al. 1970] & [Sinclair 1985].

23Looking at collocations occurring directly to the left of the node, it can be seen that the choice of lexical item tends to vary according to the subsequent use of AIM to or AIM at. Comparing the t score pictures in figure 2, it may be noted that the subsequent use of AIM at seems to result from the choice of more precise nominal subjects, such as talks, campaign, measure or strategy, thus reflecting the possibly more journalistic or formal tone of the surrounding text. This point will be examined in more detail when considering the distribution in text types. The t scores of nouns occurring in the first twenty collocates of the node at N-1 are also given in figure 3. It may be noted that AIM at is not only preceded by a significantly higher number of nominal occurrences but also that the t-score is always greater than 2, a value that Barnbrook [1996: 98] considers to be indicative of the most interesting cases of collocation using this method of calculation. The immediate left collocations (N-1) of AIM to contain more grammatical words, such as which or quantifiers; both aspects shall be studied in 3.2.2. when considering colligation, which may here be seen as the grammatical company a word keeps, or avoids keeping, either within its own group or at a higher rank.

Figure 2: t score pictures for the queries aim@+0,2at+VERB and aim@+0,2to+VERB

AIM at

a

a

is

NODE

at

ending

the

the

new

talks

NODE

only

reducing

a

of

which

was

NODE

particular

improving

out

announced

move

are

NODE

solely

getting

an

talks

spending

campaign

NODE

deliberate

helping

up

new

talks

were

NODE

specifical

bringing

its

series

visit

measures

NODE

mainly

resolving

difference

launched

peace

be

NODE

directly

forcing

people

an

bill

strategy

NODE

less

preventing

them

in

economic

plan

NODE

not

curbing

relations

tough

week

package

NODE

<f>

boosting

computer

operation

advertisin

policy

NODE

more

giving

iraq

conference

un

laws

NODE

in

increasing

young

meeting

latest

course

NODE

it

making

president

round

training

initiative

NODE

protecting

tensions

by

plan

project

NODE

securing

tension

week

special

programme

NODE

establishi

criticism

monitor

yesterday

been

NODE

providing

violence

welcome

intensive

today

NODE

creating

students

scheme

proposals

initiative

NODE

developing

parents

believed

policies

proposals

NODE

finding

down

visit

would

programmes

NODE

keeping

public

these

project

efforts

NODE

at

both

AIM to

p>

we

we

NODE

to

make

the

said

<p>

which

NODE

continuous

help

a

season

the

he

NODE

successful

provide

up

</h>

this

is

NODE

generally

raise

them

new

<h>

it

NODE

eventually

give

an

<h>

now

will

NODE

therefore

get

back

boss

london

are

NODE

instead

keep

out

national

programme

should

NODE

above

build

pound

the

will

also

NODE

low

cut

students

now

which

they

NODE

further

show

its

unit

i

i

NODE

at

reduce

his

trade

and

company

NODE

also

put

it

lives

group

course

NODE

them

bring

all

says

jason

project

NODE

</h>

promote

together

year

how

now

NODE

not

achieve

each

war

</b>

m

NODE

one

develop

awareness

south

scheme

would

NODE

for

improve

improve

this

tour

degree

NODE

he

offer

people

1998

campaign

programme

NODE

in

become

scotland

preparator

bill

book

NODE

<p>

win

parents

hindu

labour

re

NODE

end

your

maintenanc

start

always

NODE

prove

local

raising

service

labour

NODE

attract

within

Figure 3: t scores and number of actual occurrences of nouns appearing in the top 20 N-1 collocates of AIM at and AIM to

aim@+0,2at+VERB

aim@+0,2to+VERB

N-1

t-score

Number of occurrences

N-1

t-score

Number of occurrences

talks

7.13

51

company

3.19

11

campaign

3.97

16

course

3.16

11

measures

3.73

14

project

2.96

9

strategy

3.45

12

degree

2.41

6

plan

3.12

10

programme

2.38

6

package

2.99

9

book

2.34

6

policy

2.78

8

laws

2.63

7

course

2.51

7

initiative

2.44

6

project

2.42

6

programme

2.40

6

24A study of verbs falling directly to the right (N+1) of AIM at reveals that a total of 286 different verbal types are used with 116 being used more than once. With AIM to, these figures are 407 and 168 respectively. The collocations representing more than 1% of all usage at N+1 are given in figure 4 in order of frequency. This indicates that a wide semantic choice is available and the study of each individual case would be unlikely to show general patterns. However, when comparing the most popular collocations of both forms of the lemma, it can be seen that AIM to tends to favour the use of verbs possessing less semantic precision such as MAKE, GET, PUT. These are often used as phrasal verbs (make up, get back, put up) or as part of idiomatic use (make the most of it, get a crack at, put it right), implying text types of low perceived formality. The occurrence of these three verbs is lower following AIM at. Greater semantic precision may once again be detected in N+1 collocates of AIM at, reflecting observations made concerning nouns at N-1.

Figure 4: N+1 collocation of aim@+0,2at+VERB and aim@+0,2to+VERB

AIM at

AIM to

N+1 collocate

Number of occurrences

%

of usage as collocate

N+1 collocate

Number of occurrences

%

of usage as collocate

ending

50

5.93

make

47

3.93

improving

29

3.44

help

37

3.10

reducing

29

3.44

provide

37

3.10

helping

25

2.97

get

26

2.18

getting

24

2.85

raise

26

2.18

bringing

19

2.25

give

25

2.09

forcing

17

2.02

keep

20

1.67

resolving

17

2.02

build

18

1.51

preventing

15

1.78

improve

18

1.51

curbing

14

1.66

cut

17

1.42

boosting

13

1.54

put

17

1.42

giving

13

1.54

bring

16

1.34

increasing

12

1.42

show

16

1.34

making

12

1.42

develop

15

1.26

protecting

11

1.30

become

14

1.17

providing

11

1.30

end

14

1.17

establishing

10

1.19

promote

14

1.17

finding

10

1.19

win

14

1.17

securing

10

1.19

offer

13

1.09

creating

9

1.07

prove

13

1.09

developing

9

1.07

reduce

13

1.09

keeping

9

1.07

attract

12

1.00

25Although a study of individual collocates does not seem pertinent, semantic preference may be seen as being significant, notably through various synonyms conveying the idea of increase, reduction and halting. Corresponding verbs of these denotational synonyms placed at N+1 are shown in figure 5. These account for 22.8% of collocates of AIM at compared with 13.3% when AIM to is used.

Figure 5 : Verbs occurring at N+1 of AIM at and AIM to representing denotational synonyms of ‘increase’, ‘decrease’ and ‘halt’, no occurrences of negative polarity being found

AIM at

“increase”

“decrease”

“halting”

verb

number of occurrences

verb

number of occurrences

verb

number of occurrences

improving

29

reducing

30

ending

50

boosting

13

curbing

14

stopping

8

increasing

12

cutting

6

stabilising

3

developing

9

slashing

2

blocking

2

raising

6

showing

2

restricting

2

bolstering

2

limiting

1

halting

2

8.4%

6.5%

7.9%

AIM to

“increase”

“decrease”

“halting”

verb

number of occurrences

verb

number of occurrences

verb

number of occurrences

raise

26

cut

17

end

14

improve

18

reduce

14

stop

7

develop

15

halve

2

finish

5

promote

14

limit

2

halt

2

boost

7

minimise

2

stabilize

1

increase

6

shrink

2

add

2

slash

1

maximise

2

gain

1

7.6%

3.3%

2.4%

2.2.2. Colligation

26Stubbs [2001: 65] defines colligation as being

the relation between a pair of grammatical categories or, in a slightly wider sense, a pairing of lexis and grammar.

27Taking the larger sense of this definition into account and considering both grammatical words and concepts to be included, three areas of interest will be examined, namely tense frequency, relative clause expression and negation.

28Firstly, one main point demonstrating a marked preference concerning colligation is the grammatical form of each lemma as presented by the returned corpus query (the number of cases for each form of AIM at and AIM to may be seen in figure 6). Of the sixteen examples of the form aim used with at, 8 are caused by a preceding modal verb including one instance of will, 3 denote imperative clauses, 1 is an infinitive form and only 4 are used as a declined, present tense verb form. It can thus be seen that AIM at attracts the form V-ed in 95% of cases in contrast to AIM to where only 6.9% of V-ed can be seen. The notion of V-ed forms implies that a hypothesis may be confirmed as having been successful or not due to its qualities as past tense. This may account for its high frequency of use in the media where precise information is given and confirmation of success or failure is expected. The high rate of occurrence of present tense forms in the sub-corpus ukephem may highlight the potential, but not yet accomplished nature of desired outcomes. The passive voice, inducing the V-ed form, may be seen 223 times with AIM at compared with only 7 times with AIM to. This may in part be due to the use of AIM at in a majority of sub-corpora of relatively high formality which often favour the comparatively frequent use of the passive voice.

Figure 6: Frequency of occurrence of individual forms of the lemmas AIM to and AIM at according to the number of occurrences and the relative frequency expressed as a percentage

AIM at

AIM to

form

number of occurrences

% of occurrences

form

number of occurrences

% of occurrences

aim

16

1.9

aim

387

32.4

aimed

805

95.5

aimed

83

6.9

aiming

8

0.9

aiming

213

17.8

aims

14

1.7

aims

512

42.8

29Observations may also be made concerning negation as very few examples indicating negative polarity may be seen concerning AIM at/to. In order to compare the number of instances found, it may be useful to compare results of negation found in other studies. For example, taking a random sample of 18 million words of the BoE corpus, Halliday and James [1993] concluded that the relation of positive to negative clauses was 9:1. In a slightly different aspect, Tottie extracted a 50.000 word sample of the Survey of English Usage Corpus and, whilst 12.8 cases of negation were detected per 1000 words in written discourse, the occurrence in spoken discourse was discovered to be more than twice as frequent, reaching 27.6 cases per 1000 words [Tottie 1991]. If Tottie’s results are to be compared with those of Halliday and James, and also with those found in this study, it is first necessary to find the corresponding ratio of positive to negative clauses by estimating the number of sentences per thousand words. Using Francis and Kucera [1982]’s model of sentence length, given by Kennedy [1998: 158], based on the whole of the Brown Corpus, the average number of words in a phrase amounts to 18.4. Applied to Tottie’s statistics, it may be deduced that the relation of positive to negative clauses would amount to 8:1 in written discourse, which roughly corresponds to Halliday and James’ findings, and 4:1 in spoken discourse, which may however seem high. Tottie explained the differences in results as being due to the low degree of interaction characteristic of written communication compared with spoken discourse.

30A study of AIM at/to, however, shows a highly significant difference in statistics compared with the two studies mentioned above as, when considering AIM at/to, the relation of positive to negative clauses can be seen to be 100:1. There appears to be no grammatical or semantic reason preventing negation with the lemma studied yet when requesting examples of AIM expressing negative polarity, only 19 cases are found: 10 concerning AIM at and 9 affecting AIM to, with only one appearing in the British spoken sub-corpus. This may be the result of the high proportion of cases of AIM at/to found in media or ephemera-based sub-corpora where in both text types goals are asserted. This surprisingly low ratio of positive to negative clauses may also be influenced by the speaker’s projected involvement in the action, the semantic notion attached to AIM at/to expressing what is desirable from the Agent’s point of view; a positive clausal polarity seems here to be adapted to this purpose. The lack of negative clauses thus appears to be influenced by other factors in the unit of meaning which may support Sinclair’s notion of semantic prosody [Sinclair 1996].

2.2.3. Discourse prosody

31Discourse prosody is defined by Stubbs [2001: 88] as being

a descriptor of speaker attitudes and discourse function.

32He further explains this stating that discourse prosodies are evaluative and “often express the speaker’s reasons for making the utterance” [Stubbs 2001: 65], echoing Sinclair [1986]’s notion of semantic prosody where meaning is spread over a unit of meaning and not confined to any particular word.

33The lack of negative clauses has already been noted in relation to AIM at/to, and this is further reinforced by the association of the verb at N+1 as well as collocates at N+2. The desirable or undesirable sense of the main verbal collocates of AIM at/to cannot be determined in many cases without reference to the surrounding units. For example, verbs such as improve, develop or promote may carry desirable semantic associations, yet others such as end, force, cut, or reduce need to be examined in their surrounding context as they may be associated with undesirable notions. Figure 7 shows twenty randomly selected Key Word in Context (KWIC) concordance lines (with added line numeration) of AIM at/to + reducing/reduce. It may be seen that the use of a verb at N+1 having potentially undesirable connotations collocates with a noun at N+2 which also possesses such associations. This collocation of two items having, individually, undesirable connotations tends, when used together, to project a desirable goal, for example to reducing teenage smoking (l. 10) or reduce emissions of carbon monoxide (l. 16). This pattern can be seen as being repeated equally throughout the positive clauses of both AIM at/to and is also reflected in the negative clauses. For example, in the following line from the sub-corpus ukbooks the basic notion of “cultivating a sweet tooth” carries widespread undesirable connotations which can appear to be tentatively neutralized by the speaker’s prior negation:

this approach is not aimed at cultivating a sweet tooth

Figure 7: KWIC concordance lines for AIM at/to+reduce/reducing

1

bbc/06

has donated nine-million dollars to an international programme

aimed

at

reducing

the use of chemicals known as CFCs (chlorofluorocarbons),

2

bbc/06

next month for a second round of talks

aimed

at

reducing

tension over Kashmir. <h> India has made fresh

3

bbc/06

has agreed on large cuts in public spending

aimed

at

reducing

the country’s growing budget deficit, caused by unification.

4

oznews/01

2.3 million over four years to develop programmes

aimed

at

reducing

family violence including specific training for relationship counsellors.

5

oznews/01

<p> Goodman and Defiance had been in talks

aimed

at

reducing

operating costs by rationalising some of their rival

6

oznews/01

<p> The State Government had launched several initiatives

aimed

at

reducing

alcohol consumption among young people. <p> <b> AAPINTNEWS

7

sunnow/17

plan is part of a <KPD> 22million package

aimed

at

reducing

truancy and expulsions unveiled yesterday by schools standards

8

ukbooks/08

attention can then be directed at practical policies

aimed

at

reducing

dependency. However, although it might be useful to

9

ukephem/02

part at Glasgow Dental Hospital in a project

aimed

at

reducing

the level of tartar build-up. <p> To qualify,

10

usbooks/09

an `Adult Choice." Many tobacco company programs ostensibly

aimed

at

reducing

teenage smoking tell teenagers that they are too

11

sunnow/17

Black scoring avalanche. World champions women not to

aim

to

reduce

weight drastically. Perhaps the best guideline is to

12

sunnow/17

just couldn’t get it together. to save electricity

aimed

to

reduce

energy costs by a relatively-small 10 percent over

13

sunnow/17

appalling conditions. The squaddies were goods. Er they

aimed

to

reduce

the budget deficit of from twenty per cent

14

sunnow/17

when you leave Gazza at The bank is

aiming

to

reduce

its equity later this week, offering buyback on

15

sunnow/17

three-quarters full when Sam stepped SELA officials which

aims

to

reduce

foreign debt payments, mainly by encouraging countries to

16

sunnow/17

were extraordinary. We want to a plan that

aims

to

reduce

emissions of carbon monoxide by the year 2000.

17

sunnow/17

blamed for 68 deaths and by police nationally,

aims

to

reduce

the number of car accidents on roads over

18

sunnow/17

Josephs in today’s DJ Laing million. <p> It

aims

to

reduce

trade barriers in goods and services, promote investment

19

sunnow/17

club. They have already drafted J </b> Group

aims

to

reduce

sports injuries By JAMES SWANWICK AN Australian sports

20

sunnow/17

up to 500 jobs. The PBF, a charity,

aims

to

reduce

the financial impact of paralysis at the time

2.2.4. Distribution in text types

34Before looking at statistical data on the distribution of AIM at compared with AIM to in various text types, it is first necessary to define the field concerned. For Biber, text types have a strictly linguistic basis being “sets or groupings of texts such that the texts within each set are linguistically similar while sets are linguistically distinct” [Biber and Finegan 1991: 213]. Types are determined by their predominant linguistic features and given functional labels. Readily distinguished categories such as newspapers, novels or broadcasts are considered to be genres. Biber’s definition of genres conforms to the notion of text types as defined by Stubbs [2001: 20] who states that

different text types have different patterns of expectation.

It is this description of text types which shall be considered in this study.

35It has been recognised that grammatical categories, vocabulary, and semantic patterns vary according to text type. The distribution of the lexical units AIM to and AIM at may be seen to follow these predictions as can be noted in the representation of average occurrence in sub-corpora shown in figure 8. In the CobuildBoE corpus, AIM at is used with the greatest frequency in the sub-corpus bbc which accounts for 36.1% of its total occurrence in all sub-corpora. This strongly contrasts with the use of AIM to in the same sub-corpus, accounting for only 7.5% of its use throughout all the sub-corpora. This pattern is reflected less markedly in the highest use of AIM to seen in the sub-corpus ukephem where the lemma’s occurrence accounts for 17% of its total corpus use. Parallel to this, AIM at in this sub-corpus amounts to a mere 4.8% of its use throughout the CobuildBoE, thus emphasising preference for one specific form for a given text type.

Figure 8: The distribution of AIM at and AIM to in each sub-corpus expressing the average number of occurrences per million words as a percentage

AIM at

AIM to

sub-corpus

% average/million words

% average/million words

    bbc/06

36.1

7.5

npr/07

16.7

13.6

oznews/01

14.3

0.8

sunnow/17

3.7

10.8

times/10

7.9

14.2

today/11

5.1

12.2

ukbooks/08

3.3

5.2

ukephem/02

4.8

17.0

ukmags/03

3.2

13.6

ukspok/04

0.4

1.5

usbooks/09

2.2

1.9

usephem/05

2.2

1.8

36It may also be noticed that the lemma AIM at appears with significantly relative frequency in oznews and npr, showing the high representation of AIM at in radio and newspaper journalism. The sub-corpora sunnow and ukmags are high in the frequency order of AIM to in contrast to their number of occurrences of AIM at, possibly due to aspects of N+1 collocates, as shall be seen below. One sub-corpus where all forms of the lemmas appear rarely is usephem which only accounts for roughly 2% of occurrences in both cases; this may seem surprising if compared with ukephem especially in the case of AIM to.

37Collocations may also appear significant if examined according to text type. This is, not surprisingly, the case with sub-corpora with marked tendencies to favour one form in particular. If collocates appearing 10 times or more in any one sub-corpus are selected, certain lexical preferences may be seen. Searching all sub-corpora, 60% of the use of the verb ending as immediate N+1 collocate may be found in the sub-corpus bbc, 70% of the use of resolving and 30% of reducing. With AIM to, the sub-corpus ukephem, for example, appears to favour the use of provide (50% of use as a collocate) and help (28% of use). The sub-corpus sunnow appears to favour more semantically general collocates such as make, accounting for 40% of its use in N+1 collocation. This is coherent with evidence found of the high occurrence of AIM to in this sub-corpus as well as the marked preference shown by the lemma for more semantically general verbs at N+1.

2.3. General discussion of results

38It may be seen from corpus evidence that there are no notable semantic differences between AIM at and AIM to and thus other factors may lead to the preference of one form rather than the other. Drawing on all aspects treated, several hypotheses may be proposed.

39The relatively high frequency of AIM at has been noted in radio corpora, especially bbc, thus leading to possible conclusions as to the predictability of occurrence of at rather than to. The relatively high level of perceived formality of corpora such as bbc or npr may partly be responsible for this choice due to aforementioned collocational considerations, whether to the left or right of the node. Indeed, precise and formal N-1 collocates appearing simultaneously with verbs possessing precise semantic meaning at N+1 seem to indicate two linked tendencies: firstly, the high probability of such a combination appearing in sub-corpora such as bbc, npr, or oznews and secondly, the high number of occurrences of AIM at. This is a stark contrast to observations made when considering ephemera, magazine or tabloid corpora (ukephem, ukmags, sunnow) where the occurrence of AIM to largely predominates. In such corpora, less semantic precision may be noted whether at N-1 or N+1 and the level of perceived formality may be considered as being low.

40The tense of each form of the lemma may also be considered of importance. The high majority of the form V-ed used with AIM at has already been discussed yet reasons for this may be proposed by considering the general pattern of V-ing clauses as complements which express actualisation contrasting with the notion of to-infinitive clauses which indicate potentiality. This latter may result in the majority of present tense use seen with AIM to, as the Agent expresses present desires which may or may not be actualised, the unsure outcome not justifying the choice of a past tense; this may especially justify the frequent use of AIM to in ukephem where the potentiality of actions is emphasised.

41The absence of the lemma in all its forms in spoken discourse may also be noted, leading to conclude on the possibility of preference for more semantically general verbs or units such as want or would like to.

42The results given have been found using only the BoE. However, further scope may be given to this issue by examining in greater detail the structure “it is aimed at/ it is aimed to” by looking it up through a readily available web-based search engine. Using Yahoo, for example, one notices that the structure ‘AIM to’ is much more common than ‘AIM at’ with a ratio of over 90% to 10%, mainly because “it is aimed at” accepts both a noun (people, children…) or a “V-ing” structure while “it is aimed to” is almost always followed by a verb and hardly ever by a noun.

43From a stylistician’s point of view, the patterns revealed by corpus analysis may be revealing, as they often determine the impact of a phrase. The seemingly natural tendency, for example, of using more semantically precise lexical items as collocates of AIM at, is an indication of the register of language desired by an author. Also, the observation that both lemmas tend not to collocate with occurrences of negative polarity may draw attention to the exceptions.

44The main area of overlap between a corpus-driven study and stylistics may be seen in applications to discourse prosody. For example, the stylistician may note in an individual text that collocates at both N+1 and N+2 positions may individually have undesirable connotations, but when used together become a rather positive goal (see examples 2.2.3). What may be of generally more use to the stylistician is the possibility given by corpus tools to indicate that this is a norm, thus allowing deviations from this norm to be highlighted.

Conclusion

45Despite few distinctions being made in dictionaries, AIM at seems to follow different patterns of use compared with AIM to. The corpus evidence may be seen to support Stubbs’ idea that data confirms dictionary examples without giving underlying principles. This may be applied to the frequent use of the V-ed form with AIM at; thus, in a dictionary such as CCELD, it may be useful to mention that this is the most usual form met. Two equivalent examples, namely AIM at/to + VERB, should also be given to reduce confusion.

46Certain points or typical patterns which may not be obvious when comparing relatively few examples of such a word form may be observed more efficiently using corpus techniques, confirming the idea formulated by Stubbs [2000] that “the crucial shift is from studying what is possible to what is probable”. From a learner’s point of view, it may also seem important for words to be studied according to surrounding context and not as isolated units.

47The description of patterns and highlighting of repeated occurrences may therefore be of use not only to the lexicographer, but also to the stylistician. By determining typical usage with large amounts of data, deviation from this norm may be easier to spot. In a broader scope, corpus tools compiled by and for the lexicologist may prove to be a useful resource for the stylistician in all areas of stylistics, and not only where lexical matters are concerned. Continuing technical advances and the growth in available resources may hasten further the sharing of the lexicographer’s toolkit with mainstream stylistics.

Top of page

Bibliography

Aarts J., “Corpus Linguistics: an appraisal”, Paper read at the fifteenth International Conference on Literary and Linguistic Computing, Jerusalem, June 1988, in Ajimer K. & Altenberg B., English Corpus Linguistics, New York, Longman, 1991: 45-46.

Aarts J., “Intuition-based and observation-based grammars”, in Ajimer K. & Altenberg B., English Corpus Linguistics, New York, Longman, 1991: 46.

Barnbrook G., Language and Computers, Edinburgh, Edinburgh University Press, 1996.

Biber D. & Finnegan E., “On the exploitation of computerized corpora in variation studies”, in Ajimer K. & Altenberg B., English Corpus Linguistics, New York, Longman, 1991: 213.

De Rose S., “An analysis of probabilistic grammatical tagging methods”, in Johansson S. & Stenstrôm A-B., English Computer Corpora, Berlin, Mouton de Gruyter, 1991: 9-14.

Francis W. N. & Kucera H., Frequency Analysis of English Usage: Lexicon and Grammar, Boston, Houghton Mifflin, 1982.

Halliday M. A. K. & James A., “A quantitative study of polarity and primary tense in the English finite clause”, in Sinclair J., Hoey M. & Fox G. (eds), Techniques of Description: Spoken and Written Discourse, London, Routledge, 1993: 60.

Itkonen E., “Qualitative vs quantitative analysis in linguistics”, in Perry T., Evidence and Argumentation in Linguistics, Berlin, de Gruyter, 1980: 344.

Kennedy G., An Introduction to Corpus Linguistics, Longman, 1998.

Mahlberg M., “Corpus stylistics: methodology, theory, and patterns in literary texts”, 2006: http://ahds.ac.uk/litlangling/events/approaches/mahlberg.htm, accessed 10/02/10.

Potter L., A Guide to Using the CobuildDirect Corpus, Birmingham, University of Birmingham, 1999.

Sinclair J., “Lexicographic evidence”, in Ilson R. (ed.), Dictionaries, Lexicography and Language Learning, Pergamon and the British Council, 1985: 81-94.

Sinclair J., “The search for units of meaning”, Textus, 9, 1996: 79-106.

Sinclair J., “The lexical item”, in Weigland E. (ed.), Contrastive Lexical Semantics, Amsterdam, Benjamins, 1998: 1-24.

Sinclair J., Corpus, Concordance, Collocation, Oxford, Oxford University Press, 1991.

Stubbs M., “Society, education and language: the last 2,000 (and the next 20?) years of language teaching”, in Trappes-Lomax H. (ed.), Change and Continuity in Applied Linguistics, Clevedon, BAAL and Multilingual Matters, 2000: 15-34.

Stubbs M., Words and Phrases: Corpus Studies of Lexical Semantics, Oxford, Blackwell Publishers, 2001.

Stubbs M., “Conrad in the computer: examples of quantitative stylistic method”, Language and Literature, 14: 1, 2005: 5-24.

Tottie G., Negation in English Speech and Writing, London, Academic Press, 1991.

Top of page

References

Electronic reference

Kate James, A Corpus-Based Case Study in Prepositional AT/TO- Infinitive Alternation using the Lemma “AIM”Lexis [Online], 5 | 2010, Online since 15 October 2010, connection on 31 March 2023. URL: http://journals.openedition.org/lexis/505; DOI: https://doi.org/10.4000/lexis.505

Top of page

About the author

Kate James

Université d’Auvergne, IUT, 100 rue de l’Egalité, 15000 Aurillac
kate.james@iut.u-clermont1.fr

Top of page

Copyright

CC-BY-NC-ND-4.0

Creative Commons - Attribution-NonCommercial-NoDerivatives 4.0 International - CC BY-NC-ND 4.0

https://creativecommons.org/licenses/by-nc-nd/4.0/

Top of page
Search OpenEdition Search

You will be redirected to OpenEdition Search