My paper for the Lexis module of MA TESOL. Do contact me if you are interested in the actual responses from learners or my references.
1.0 Introduction
“Alright, alright,” said my advanced student after
long debate, “If you say it’s ‘to fulfil an ambition’ then I’ll write ‘to
fulfil an ambition’, but I just don’t get it. An ambition is something that, by
definition, just cannot be fulfilled.”
Teaching lexis is much more than simply teaching words.
We are instructed to present new items to our students in terms of meaning,
pronunciation and form, and specifically in that order, but ‘knowing’ a word by
no means ends there. It would be virtually impossible to present all the
possible meanings of a word at the same time, and even if we could, it would
certainly only confuse our students with much unnecessary information, but even
then we would be omitting a huge aspect of what it means to ‘know’ a word: the
associations connected to it.
In the case of my advanced student quoted above, the
difficulty in learning this collocation is that the word ‘ambition’, as he is
translating it from his native Russian, is always something in the future and
therefore ceases to be an ambition as soon as it has been fulfilled. This is an
interesting example of how associations in the mother tongue may carry over to
items in the target language, and thus provide a justification for teaching
connotation alongside meaning, pronunciation and form.
I actively encourage my students to make their own
associations with lexis to help them memorise new items, so this project was
very interesting to me as a means to discover which links students actually use
to retrieve their knowledge.
2.0 The Mental Lexicon
At what point can it be said that a student ‘knows’ a
word? Definitions may differ, but what many agree on is the necessity for the
knowledge of associations with the target word (Richards in Meara 2009; Deese
in Coleman 1964; Nation in Schmitt and Meara 1997).
1. The spoken form of a word.
2. The written form of a word.
3. The grammatical behavior of the word.
4. The collocational behavior of the word.
5. How frequent the word is.
6. The stylistic register constraints of a word.
7. The conceptual meaning of a word.
8. The associations a word has with other related
words.
(Nation 1990:31
quoted in Schmitt and Meara 1997)
Three of the eight points on Nation’s list (3, 4 and
8) are related to associations: grammatical, collocational and connotational.
If a teacher is to ensure that a word is truly ‘known’, then he has to make
sure target vocabulary is both presented and learnt in relation to these
factors.
Additionally, once a word has been ‘learnt’, what
evidence is there that it will be actively used? Previously, theorists have
spoken of a receptive-productive cline and of moving words from the active to
the passive; however Paul Meara (2009) proposes an alternative to the storage
of vocabulary with his graph theory, which suggests that words are
stored as part of a complex network in which one word provides (or likewise,
fails to provide) ‘access’ to another, rather than being at a particular point
on a continuum.
Of course, we cannot expect a learner to be able to
make all the associations that a native speaker may. As McCarthy states, ‘The
language learner has a formidable task in emulating the complexities of L1
storage in an L2 and, naturally, the L2 mental lexicon will have to develop
from a few initial strands to a the goal of labyrinthine connections between
words’ (1990:42).
Nonetheless, as teachers, we have to ensure that we
provide and utilize those ‘initial strands’ because ‘seeing patterns in
associations can help teachers present new vocabulary and evaluate student
comprehension’ (Sökmen 1993:135).
Besides the role that associations play in knowing a word, they also contribute to
rapid retrieval. McCarthy makes this clear when he states that, from the point
of view of the learner, there can be great frustration in failing to recall the
exact word required (it’s on the tip of my tongue!), and also from the point of
view of the teacher:
‘If a language learner cannot actively use a
particular word when it is needed, without too much mental searching, then we
might feel that we are dealing with an incomplete knowledge of the word, or at
the very least we will want to distinguish between receptive knowledge and productive
knowledge’ McCarthy (1990:43).
But how can this
ability be assessed?
3.0 The Word Association Test
If the mental lexicon is seen as a network, then ‘one
of the most accessible and most easily understood methods of studying the structure
of semantic relationships in bilingual lexicons is the use of word
associations’ (Meara 1980:13). Meara points out that such studies are
particularly useful to teachers of English, due to the high levels of associational
stereotypy, ie. native speakers exhibit a tendency to produce the same response
to a given stimulus.
Word association tests (WATs) may have different
forms, but typically involve the student responding to an aural stimulus word
by writing down the first word which comes to mind. Alternatives to this have
been Coleman (1964), who provided 5 context words before the association was to
be made, tests requiring multiple responses and even interlingual associations.
Sökmen (1993:137) neatly summarises the conclusions
drawn from previous studies.
‘Past word association research with second language learners leads us
to expect that nouns are most likely to solicit nouns (Ludwig, 1984), and verbs
will get more varied responses (Ruke-Dravina, 1971). Previous research also
indicates that beginners have fewer primary responses because their lexicons
are small and less organized (Meara, 1978). Advanced students have more
synonyms and contrast words (Soudek, 1981). Regarding age and education,
Riegel's (1968) study shows that older and more educated students have fewer
primary responses.’
Both Sökmen and Meara criticize WATs for focusing on
words as individual items and thus dealing with vocabulary breadth, rather than
depth, or a combination of the two. As Meara states, tests on the word-level
are a case of ‘not being able to see the wood for looking at the trees’ (2009:74).
Nevertheless, they remain the most popular method of testing, perhaps due to
their ease of administration.
Besides variation
in the administration of the test, responses have also been classified in
different ways. Typically, and for the purposes of this study, responses are
classified as paradigmatic, syntagmatic, phonological (or clang)
and encyclopaedic.
3.1 Paradigmatic
Paradigmatic, or choice, associations involve the
replacement of one word by another of the same word class. They can further be
classified as co-ordination (words of the same level of generality, eg. husband
– wife), hyponymy (superordinates and subordinates eg. vegetable – cabbage) and
synonymy (near synonyms, eg. beautiful - pretty).
3.2 Syntagmatic
Syntagmatic associations refer to both collocations
and colligations, and seem to be the association of choice for advanced
learners. ‘Since they have more words in their lexicons and more detailed word
clusters, it appears they are less likely to reIy on a contrast association’
(Sökmen 1993:146). Schmitt and
Meara (1997:18) suggest that learners master grammatical and semantic
associations but may never master the connected collocational aspects.
3.3 Phonological
Associations
These involve a response which has neither semantic
nor grammatical link to the cue-word, but which bears some phonological
similarities to it. Meara describes this category as ‘phonetically motivated
rather than semantically motivated’ (1980:17) and claims that they occur much
more often at low levels. High occurrences of phonological, or clang associations, could be due to
instruction using the keyword technique, which encourages learners to associate
target vocabulary with similar-sounding words in the L1, albeit with an
unrelated meaning (Meara 1980). Fitzpatrick (2007) reports assumptions in
studies that as proficiency rises, there is a move away from phonological
associations.
3.4 Encyclopaedic
Associations
These refer to ‘personal knowledge about the world’ (Khazaeenezhad
and Alibabaee 2013:110), including affective associations. Previous studies
show that the response is dependent not only on the cue-word, but also, naturally,
on the respondent himself. In Sokmen’s (1993:146) study, she found that
beginners were unlikely to give affectual responses.
An interesting aspect of this category are interlingual
associations, which, according to Meara are ‘associations which are made in a
language which is not the same as the one in which the stimulus word was
presented’ (1980:17). Meara was referring to participants recording their
results in another language to the cue-words, but I believe this also applies
to an L2 cue which evokes an L1 association, which is then translated into the
L2 and recorded. This could be because the student does not know that the L2
cue does not have the same connotation, or because the L1 association with this
word is simply so strong that it overrides any other associations.
4.0 The Study
4.1 Choice of Cue-Words
According to McCarthy’s stipulations, 8 items of
vocabulary were chosen to include a variety of parts of speech, as shown in the
table below (Figure 1). I had intended to include one adverb, but was
restricted by the size of vocabulary of my A1 level participants. All 8 cue-words
necessarily appear in the A1 textbook so I could be sure students were familiar
with these words, but in selection I tried to choose items which had multiple
contexts and collocations in order to give higher level students as much
freedom as possible to their associations.
Figure 1 also shows the frequency of the items based
on the number of results on the Internet using the Google search engine. For
the purposes of this study, pineapple
was chosen as the required low-frequency word. According to the Collins Cobuild
online dictionary, 18 out of the 26 languages given translate pineapple as ananas, and four further languages have a very different-sounding
translation, leaving very little chance of L1 influence in understanding this
word in comparison with other common fruit such as orange or banana.
Therefore, in this case just as with the other seven test words, students need
to understand the word in order to retrieve an association, and if the
association is phonological, then it will not simply be the translation of
pineapple in the L1. Studies mentioned in Fitzpatrick (2007) suggest that
high-frequency cue-words result in a narrower range of responses than
low-frequency items. In this study, we can expect that pen and blue will yield
less variety than pineapple.
Traditionally cue words have been chosen for tests
based on frequency. The most famous example of a frequency-based word
association list is that used in the Kent-Rosanoff test of 1910, but as Schmitt
and Meara note, ‘word frequency is not a reliable index’ (1997:25). McCarthy (1980)
states that there has been little agreement on which cues to use.
As already stated, all eight test items are common
words which even students at A1 are familiar with. These are presented without
a context and thus without an ‘anchor word’ to help direct meaning. McCarthy
quotes Moore and Carling’s (1982:196) definition of anchor words as ‘words of low semantic variability which {…} narrow
down the meaning options’ (1990:44).
The absence of such anchors allows complete freedom to the retrieval process,
thereby ensuring as far as possible unbiased results.
Sökmen (1993:137) names previous studies which
indicate that nouns evoke nouns, while verbs result in a wider variety of
responses. Both run and stand were deliberately chosen because
they belong to both parts of speech, ie. they are both verbs and nouns. At
lower levels, however, learners are likely to perceive them as verbs.
Table 1: 8 Cue-Words
|
|
Classification
|
Cue-word
|
Justification
|
Frequency
|
|
1.
|
Function word
|
in
|
Usually the first preposition learnt, preposition of
place, also features highly in collocations, idioms, phrasal verbs. Can be
mistaken orally for ‘inn’.
|
25,270,000,000
|
|
2.
|
Common object (noun)
|
table
|
A word used in virtually every English lesson, yet
also representing other objects besides being a synonym for desk.
|
1,990,000,000
|
|
3.
|
Common object (noun)
|
pen
|
Taught in the first lessons of English, yet may take
many different forms.
|
469,000,000
|
|
4.
|
Adjective
|
blue
|
Common colour deliberately chosen because of the
many associations connected to the Russian word ‘голубой’.
|
14,970,000,000
|
|
5.
|
Verb / noun
|
run
|
Common verb, especially familiar to young learners,
also part of phrasal verbs, literally and metaphorically. Also a noun.
|
2,080,000,000
|
|
6.
|
Verb / noun
|
stand
|
Every student is familiar with the phrasal verb
‘stand up’, yet it also features as a noun at higher levels and is part of
numerous other phrasal verbs.
|
1,510,000,000
|
|
7.
|
Adjective
|
strong
|
Initially learnt as an adjective describing physical
strength, but can be used in many contexts where the Russian 'сильный’ cannot.
|
1,550,000,000
|
|
8.
|
Low frequency word
|
pineapple
|
Tropical fruit often encountered in early stages of
textbooks yet less common in itself in Europe.
|
83,500,000
|
4.2 Choice of Participants
I was primarily interested in the mental lexicon of my
own learners, and so 25 students were chosen at each of the CEFR levels I
currently teach: A1, B1 and C1. A further 25 non-native teachers volunteered to
take part in the test, totalling 100 participants. All the A1 students are
children between 6 and 14; 9 of the B1 level participants and 17 of the C1
participants are teenagers.
4.3 Conducting the
Test
The WAT was
carried out among the students as part of a regular lesson, while teachers were
interviewed individually. Stimuli were given orally only, even when students
asked for clarification. ‘Blue’ and ‘in’ could be interpreted as ‘blew’ and
‘inn’ due to the lack of context, and I was interested in seeing which word
came to the participants’ minds first. Respondents recorded their answers in a
designated form (Appendix A).
4.4 Limitations of
the study
The inclusion of an adverb as a stimulus word would
have been interesting as I encourage the use of adverbs wherever possible and
it would be seen whether students have taken this on board. Adverbs collocate
with both adjectives and verbs, so this would have been a further area of
investigation. However, my A1 students do not know any adverbs except ‘very’,
which I felt was so general a word as to not reveal any patterns at all, and so
an adverb was omitted.
During testing it became clear that my numbering
system (from left to right) went against the natural logic of most of my students,
who filled in their boxes from top to bottom, then realized their mistake and
had to re-number their responses. In future, I would have a single row of 8
boxes to avoid confusion.
Classification of responses is problematic, as there
is a good deal of overlap. Respondents
were not asked why they had chosen their association; indeed, it is doubtful
whether they would be able to say. Where there seemed to be no obvious reason
for the response, the encyclopaedic category was chosen, as it was assumed that
the learners had some personal association which the researchers could not know
about.
It must be remembered that the choice of stimuli was
subjective and based on my knowledge both of my students and of their L1. Fitzpatrick
warns us that ‘the findings of any word association study are {…} dependent to
a great extent on the selection of cue words’ (2007:323). She also states that
we must ‘be wary of concluding that word association behaviour is independent
of task language’ (2007:320). Therefore, any results can only be considered
relevant to these groups of Russian-speaking participants learning English.
5.0 Results
5.1 Trends by
Cue-Word
Predictably, the greatest number of emotive
associations resulted from the word pineapple,
which has little semantic variability, few collocations, no colligations, and
for the majority of people has a strong positive association.
The most frequent response to pen was pencil, and here
it is difficult to say whether the association was that of a lexical set (paradigmatic)
or phonological, so results have been included twice to show both
possibilities. In general, it is assumed that the link is paradigmatic, but
scores in brackets show the number of phonological links if pencil is categorised as such.
Results do not confirm the hypothesis that
higher-frequency words result in more homogenous responses, as pineapple, the lowest-frequency word in
this study, produced fewer responses than commoner words such as table.
It can be seen that nouns produced more paradigmatic
responses than other parts of speech, with the greatest number coming from the
word pen. Blue gave the highest number of syntagmatic responses.
Finally, there is no evidence to suggest that nouns
result in a narrower range of responses than verbs.
Table 3 shows the two commonest responses by cue-word
and level. Only frequencies higher than 2 were included in the table; those
instances where only one response is present indicate that all other responses
were present only once or twice.
The most significant result is the homogeneity between
levels. It seems to suggest that, whatever the student’s proficiency, the main
associations are nonetheless very similar.
Table 2: Responses by Cue-word
|
|
Range of responses
|
Paradigmatic Responses
|
Syntagmatic Responses
|
Phonological Responses
|
Encyclopaedic Responses
|
|
In
|
40
|
51
|
40
|
11
|
3
|
|
Table
|
43
|
49
|
37
|
1
|
12
|
|
Pen
|
27
|
65
|
28
|
44 (41 = pencil)
|
4
|
|
Blue
|
25
|
22
|
71
|
1
|
6
|
|
Run
|
45
|
34
|
56
|
1
|
7
|
|
Stand
|
43
|
28
|
57
|
2
|
6
|
|
Strong
|
44
|
21
|
65
|
1
|
3
|
|
Pineapple
|
39
|
39
|
38
|
14
|
19
|
Table 3: Top Responses by Cue-word
|
|
A1
|
B1
|
C1
|
NNS
|
|
In
|
On (5), box (3)
|
Out (8), on (4)
|
Out / home (4)
|
Out (8), room (4)
|
|
Table
|
Chair (5), desk (3)
|
Chair (6), school (3)
|
Chair (7)
|
Chair / desk (6)
|
|
Pen
|
Pencil (9), school (4)
|
Pencil (15)
|
Pencil (8)
|
Pencil (9), write / blue (3)
|
|
Blue
|
Sky (8), sea (4)
|
Sky (9), pen (3)
|
Sky (11)
|
Sky (11), red (3)
|
|
Run
|
Jump (5), fast (3)
|
Fast (4), walk (3)
|
Away (6), park (3)
|
Fast (7), jump (5)
|
|
Stand
|
Up (11), sit (4)
|
Sit (9)
|
By (3)
|
Up (10), sit (4)
|
|
Strong
|
Man (9), enough (3)
|
Weak (8), hard (3)
|
Muscles (4), man (3)
|
Weak (13), man (3)
|
|
Pineapple
|
Fruit (4), eat / yellow (3)
|
Apple (8), fruit (4)
|
Yellow / apple (3)
|
Yellow (6), juice (4)
|
5.2 Trends by
Language Proficiency
Fitzpatrick (327)
suggests that the more proficient a learner, the greater the range of
responses. This study does not support this statement, as the widest range of
responses were from C1.
Syntagmatic
responses account for just over half of all given, with the greatest number
among C1 level students, and, in stark contrast, the fewest among B1, who gave
well over half their responses as paradigmatic associations.
The high
frequencies of syntagmatic and encyclopaedic responses at C1 suggest a high
level of creativity and semantic associations among these learners.
Table 4: Responses by Proficiency
|
|
Range of responses
|
Paradigmatic Responses
|
Syntagmatic Responses
|
Phonological Responses
|
Encyclopaedic Responses
|
|
A1
|
105
|
74
|
103
|
5 (+ 9 = pencil)
|
14
|
|
B1
|
99
|
118
|
67
|
10 (+ 15 = pencil)
|
11
|
|
C1
|
137
|
53
|
119
|
14 (+ 8 = pencil)
|
26
|
|
NNS
|
98
|
75
|
113
|
4 (+ 9 = pencil)
|
9
|
|
Total
|
306
|
320
|
402
|
33 (+41 = pencil)
|
60
|
5.3 Noteworthy
Responses
Some responses worthy of discussion are, firstly,
those associated with the word blue.
The Russian translation of this word (голубой) is slang for homosexual, and we can see that the L1 association is so strong that it carries over into
the L2 and gives us gay, Viktor (a homosexual student in the
group) and guy, which I believe to be
a misspelling or misunderstanding of the word gay. There are no such occurrences among the teachers but there are
five at all levels of students, suggesting a powerful L1 association and
confirming Carter (1998), who states that lexical choice marks the evaluation
of the speaker.
One teacher who participated in my test had been
lamenting the lack of blu-tac in Russia just before we administered the test,
so her response to blue was tac, showing that fresh associations may
occur first. Participating on another day, or without my presence, her response
would almost certainly be different. This reinforces Fitzpatrick (2007), who
notes that the respondent brings just as much to the task as the stimulus
offers in itself, and also Schmitt and Meara (1997), who claim that there is no
direct connection between the frequency of the item and the chances of it being
‘retrieved’.
6.0 The Task
6.1 Mental Links
Some of the most frequent responses may be categorised
as what Schmitt and Meara call ‘classroom English’ (1997:32):
Pen
– pencil, school
Pen
– red (from teachers only!)
Table – desk
Stand - up
This illustrates
both the beneficial result of repeated, frequent exposure to vocabulary and the
strength of links made when an item is presented visually. No doubt pen and table were initially introduced by pointing at them in front of a
student’s eyes, rather than in the context of a text, or by definition, or by
any other common presentation method.
We also see that
lexical sets help students to retrieve words:
Pen
(school object) – pencil, paper
In (preposition) – out,
on
Pineapple (fruit) – apple, mango
Run (verb of motion) – jump, walk
Where possible, the superordinate was named, eg. pineapple
– fruit. Perhaps a failing of the
choice of cue-words is that there was little opportunity for students to give a
subordinate and so there are only isolated examples such as pen – fountain (pen). This shows that words
are arranged in the mental lexicon in their lexical sets, as opposed to a
random list or alphabetical order. It also suggests that cue-words give a
response at the same level of generality or less general responses (the
superordinate) and are less likely to result in the subordinate. Meara’s
evidence supports this in his example:
Sleet
– rain
but Rain ≠ sleet Meara
(2009:61)
Fitzpatrick claims that ‘high-frequency cues will lead
to more homogenous responses’ (2007:323). This, however, is not borne out in my
research. As seen in Table 2, the highest frequency word in resulted in 40 different responses, while the low-frequency pineapple resulted in 39. The scattered
nature of the responses indicates that each learner makes his own association,
be it grammar-based, meaning-based or affective.
Finally, Fitzpatrick names research showing that
native speakers vary wildly in their response type and asks whether, with
increasing proficiency, learners are ‘moving towards a response profile which
represents a “native-speaker norm”’ (2007:327). The results of this study
indicate the widest variety of responses at C1, yet with A1 also demonstrating
quite a range of responses (Table 4). Regardless of range, the top two
responses for most cue-words nonetheless show many similarities (Table 3).
6.2
Phonological Similarities
Results do not confirm the hypothesis that low-level
learners make more phonological associations than more proficient students. In
this study, A1 students made almost the same number of phonological
associations as the non-native teachers, while the greatest number were by B2
students. If we assume that the cue - response pen – pencil is a
phonological rather than paradigmatic link, there is little change to the overall
trend.
In two cases, the NNS misunderstood the cue-word ‘in’ as ‘inn’. I suggest that the greater the size of the listener’s
vocabulary, the greater the possibility of phonological associations. Low-level
learners may simply not know enough similar-sounding words to be able to make
this link.
6.3 Characteristic
Responses
The overwhelming result for pen was pencil, but it is
difficult to decide whether the classification of this response should be
paradigmatic (co-ordination) or phonological. In Aitchison’s Reith lecture, she
mentions the bathtub effect, whereby the beginning of a word is easier to
recall than the middle segment, which could be playing a role here..
Meara’s (1980) statement
that English has high associational stereotypy and therefore a limited range of
responses is not clearly proven. The Kent-Rosanoff test also used table as a stimulus, and found that 78%
of respondents gave chair. In this
study, we see only 24% giving this response, although it is nonetheless the
most frequent.
We certainly see
that, among paradigmatic responses, co-ordination is the most frequent response
type, with answers such as blue - black.
7.0 Conclusion
There are very few useful conclusions which can be drawn
from a study of this small scale. When only 8 cue-words can produce over a
hundred different responses, any inference can, at best, apply only to these
test subjects. As Sinclair explains, ‘lexical patterns are very difficult to
observe because they are realized by a large vocabulary of infrequent words,
and so it is not easy to work out the recurrent patterns’ (2004:165).
The wide variety of responses in many cases supports
Fitzpatrick’s assertion that individuals have a preferred word association
response profile (2007:328), and we may also agree with her tentative
suggestion that prediction is virtually impossible since each learner has thus
to be analysed separately.
Nonetheless, one significant pattern is the tendency
among low-level learners to give paradigmatic rather than syntagmatic
responses. Widdowson (1989) mentions that a speaker is deemed ‘competent’ by
his knowledge of lexical chunks. The implication for teachers of low-level
students is therefore that words should be taught in phrases as soon as
possible in order to enable fluency. ‘Overconcentration on learning single
words may hinder the development of the L2 phrasal lexicon and deny the
opportunities this gives for rapid retrieval and fluent, connected speech’
(McCarthy 1990:45).
Overall, literature suggests that ‘seeing patterns can
help teachers present vocabulary’ (Sökmen 1993:135), therefore there is clearly
a call for large-scale research in the future.
