A Penny For
Your
Thoughts?
Developmental differences in Word Association Behaviour
Masterscriptie opleiding Engelse Taal- en Cultuur Faculteit der Letteren
Rijksuniversiteit Groningen
Saskia Evers
1331809
Contents Page Number Summary
3
Chapter 1: Introduction
4
Chapter 2 Word Associations
6
2.1 What Are Word Associations 6
2.2 Concept Learning and Word Knowledge 6
2.3 Variables in Word Association 8 2.3.1. Mode of Experiment 9 2.3.2. Different Stimuli 10 2.3.3. Individual Differences 12 2.3.4. Types of Responses 15 2.3.5. Conclusion
16 2.4 Differences in Word Knowledge Development in L1 and L2 Learners 17
2.4.1 First Language Learners
17 2.4.2. Second Language Learners
19 2.4.3. Conclusion
23
2.5 Expectations 24
Chapter 3: The Test
25
7.1 Tables 53 7.2 Form
62
7.3 Table used for SPSS 63
7.3.1 Concrete 63
7.3.2 Syntagmatic 65
7.3.3 Same Part of Speech 67
Summary
After a short introduction describing the mental lexicon, concept learning and the possible depths of word knowledge, it is explained that word association is an approriate manner of testing one’s word knowledge. The many variables that play a role can be found on different levels, such as the mode of experiment, different influences, individual differences and types of responses. There is a difference between the manner in which a first language is learned and how a second language is learned. In this study, the focus will lie on the changes taking place between the first and the fifth grade at a Dutch secondary school and especially on the shift from concrete to abstract, from syntagmatic to
1. Introduction
Even though people are not continually conscious of the fact, associations are continuously made in everyday life and especially in language. Word association can be looked at from many different points of view. Studies have been
performed in – among others – linguistics, psychology, and medical science. In this study the focus will solely lie on linguistics.
Word associations provide information about the level of someone’s word knowledge, about the development of language over time, about the type of words that are used when associations are made and so on. Variables that could influence these associations or how these associations change over time may differ.
1. Individual differences may play an important role such as age, years of prior education, education level, personality variables, intelligence, self-reflected thoughts, and speech disorders. Also gender, social status, surroundings etc could play a role.
2. The manner in which an experiment is conducted – for instance, written versus oral testing – may influence associations due to various reasons.
3. A participant may also be influenced by different stimuli, such as word frequency, how often a word is paired with other words, frequency of usage, semantic connections, role-taking, etc.
4. The type of responses also can differ between participants, for example part of speech.
The question asked in this particular study is: which variables influence the
A few variables were tested in an experiment in which three different classes at a secondary school received the words house, drive, time and know, after which they had to give three words they associated with the stimuli. The
variables looked at are age in combination with prior years of English education, realised by looking at the concrete/abstract, paradigmatic/syntagmatic nature of the responses, the occurrences, the frequency of the answers, and the part of speech in which the answer is given.
2. Word Associations
2.1 What are word associations?
According to Pollio (1966), word associations are not as easy as some more “rule-based” aspects of language; however, research led to some knowledge and understanding of word association behaviour. There are many theories claiming that in the case of word association two events take place, initiated by a stimulus word: “hypothesized fractional conditioned meaning responses, and the specific words aroused as associates by a stimulus word” (Pollio, 1966, p. 1).Within this theory two different kinds of word associations are mentioned: free word
association and forced choice word association. In free word association the subject is presented with a stimulus word and has to give only the first word that comes to his or her mind. In forced choice word association the subject is presented a given list out of which the subject is supposed to choose a certain prescribed number of words that he or she associates with the stimulus word. In both cases the response should be the first word(s) that come(s) to mind, without an effort to concentrate, focus or think. Fisher’s definition is slightly more
elaborate: “Free word association represents one means of examining an individual’s semantic network […] Those words elicited are presumed to be the ones that are accessed first. The specific nature of the links, that is, the
perceived relationships between concepts, is not identified by the free-word association task” (Fisher, 1985, p. 390).
2.2. Concept Learning & Word Knowledge
To make associations one needs sufficient knowledge of a word. It takes a long time before people have fully acquired a language, for the knowledge is acquired piece by piece. A number of steps need to be taken before an idea can be
expressed. First of all, one has to think of a word, but is that specific word really the first word one thinks of or is it another one?
“Initially, many of the concepts have biographical information associated with them: the what, where, when, and how the concept was acquired may be as prominent as or more prominent than the features of the concept” (Fisher 1985, p. 389). However, these concepts are connected to other concepts, meaning that they are related because they either have something in common or occur
together often or perhaps another reason that causes people to associate certain concepts with one another. However, connected concepts do not necessarily indicate that all possible connections are made automatically, for one’s knowledge of a word might still be incomplete, which means that not all connections are present.
Word associations are a helpful tool for looking at someone’s depth knowledge of words. Schoonen & Verhallen (1998) describe that the focus in research lies on the number of words children know (breadth knowledge), and not on the depth knowledge of these words. How well do children know the words they ‘know’? According to Greidanus and Nienhuis one’s knowledge about a word can be divided into its meaning, its form, its function and its position.
However, one does not necessarily need to know all four aspects to have (partial) knowledge of a word. The lack of knowledge may influence the associations people make when hearing or seeing a word. This holds for both first and second language learners. To complicate the matter: the meaning of a word can be divided in several categories as well. However, “at the most basic level, meaning consists of the relationship between a word and its referent (the person, thing, action, condition, or case it refers to in the real or an imaginary world). This relationship is not inherent; rather, it is arbitrary until formalized by the people using the word” (Schmitt, 2000, p. 23). However, most often it is described as the relation between a word and its concept, as some referents may occur in
different variations and/or states, leading to a varied set of words used for one concept.
about one of the four aspects mentioned by Verhallen – for example its form – or also has acquired knowledge about the other aspects. It is important when learning a language and be fully able to understand it – no matter if it is a first or second language – to know the deeper meaning of words, as possessing only partly knowledge may cause misunderstanding. For instance, a child may have seen a word before and therefore recognize its form, but not know how it is used and thus not be able to use it. While education progresses deeper word
knowledge becomes required. Therefore, when acquiring a vocabulary it is important to not only become acquainted with the word, but also learn its deeper meaning. The Word Association Test’s (WAT) goal is to test the quality of how children assign meaning to words. In the case of word knowledge tests that test the depth knowledge, the outcome should show the quantitative differences made between words.
Tests are being developed to test both the breadth of the word knowledge as the depth. Schoonen and Verhallen have developed “deep-word knowledge tests based on word associations” (Greidanus, 2001, p. 567). These not only valuate the word knowledge, but also look at “the knowledge of the meaning of the word, the words with which it [is] associative, and the collocations in which it occurs” (Greidanus, 2001, p. 567). The tests developed by Schoonen and Verhallen (1998a) and a test developed by Reid have proved to be quite useful.
The difference between the tests lies in the distractors: in Reid’s test the distractors have no semantic affinity to the target word, while the distractors in Schoonen and Verhallen’s test are semantically linked. This, for example, is one of the variables that can occur in word association tests.
In sum, concepts are important to make word associations, as they are connected to each other and so are words connected to these concepts,
to more possible connections. In Verhallen’s view word association tests are a method to measure the word knowledge in both depth and breadth.
2.3. Variables in Word Associations
Many factors influence the results that come forward in word association tests, although one factor may be more logical than the other. Not much obvious information has been found about the different factors that seem to influence the results of word association tests. In most studies the variables are mentioned, but a clear explanation about how this works, is lacking. Different variables are mentioned in different texts and studies, such as: age (maturation), education, intelligence, proficiency level, region, social status, sex, written vs oral testing, free vs free WAT-tests, role-taking, semantically related vs semantically non-related, frequency of occurrence, group vs individual testing, priming, induced stress, context of other stimulus words, etc. Elaboration upon the mode of experiment, the different stimuli, the personal characteristics and the types of responses will be shown in the following sections.
2.3.1. Mode of Experiment
Not only the form and the group size possibly influence the results. The decision between a free word association test and a forced choice association test also may prove to be of significance. First of all, in case of a free word association test the participant is allowed to think of a word him/herself and can choose each word that comes to mind. In contrast, in a forced choice word association test scenario, the participant is presented with a few words, which may influence the direction in which the participant is thinking, meaning s/he may be induced to give a certain answer because of which the results would not be fully valid. However, the two different tests have a different aim. For instance, when one conducts the free word association test one could be looking for
differences between certain groups (age, education, gender, region, etc) or in the case of tests in the psychological or medicine area the tests are used to find out what areas in the brain are damaged. When using a forced choice word
association test on the other hand, one might be aiming for the difference
2.3.2. Different stimuli
Not only is the mode of experiment important, it also appears to matter how often the participant comes across a specific word used in the test in everyday life. When a word has a higher frequency in a language the participant probably will be better acquainted with the word than in case of a lower frequency word, meaning that specific word will lead to certain associations. “Cofer and Shevitz have shown that the number of word associations produced in a 10 minute period is significantly affected by the frequency of occurrence of the stimulus word” (Pollio, 1966, p. 13). Not only the frequency of occurrence is of importance,
but also how often it is paired with other specific words.
“The frequency with which a stimulus word has been paired with other specific words, as well as the evaluative connotation of the stimulus word probably also affect the size of the associative hierarchy” (Pollio, 1966, p. 14). Furthermore, the chance of pairing certain words - that is the associative word to the stimulus word - is much higher when the two have been paired together many times before. This may also influence the response time, for it is likely that a subject will give a quicker response when two words have been paired before. McGeoch and Irion, performed a test using stimulus words that had either a positive or a negative inclination. Apparently pleasant words led to more
associative words than unpleasant words, meaning the positive relationships had better results in their study (Pollio, 1966).
Closely related to the frequency of occurrence is frequency of usage, mentioned by Milgram and Goodglass (1960). When a word is used regularly in everyday language, it is easier to find another word to associate it with, because there are simply more words that remind of the frequent used word. Milgram and Goodglass explain this in the following manner: “in order for such verbal
the word with the highest transitional probability in everyday English” (Milgram, 1960, p. 86). Subsequently, in another study it was suggested that the frequency in which certain individual words from a list that had been studied, at least
partially, is connected to “the extent that these elements were evoked as associates by other words in the same list” (Pollio, 1966, p. 19). It can be concluded from these tests and others as well, that there is some kind of correlation between words that belong to a certain list or as Pollio (1996, p. 19) puts it: “a collection of meaningful words” and that this relation influences the recollection of certain items and words.
Except for the former mentioned stimuli, the issue whether a stimulus word is semantically related or not to the associative words can create diversions in the results as well. Read created a test in which the words that could be chosen were not semantically related to the stimulus word, neither morphologically nor orthographically. Schoonen and Verhallen (1998a) on the other hand, did choose
words that were semantically related to the stimulus word. Their reason for doing so lies in the idea that three aspects play an important role in word association, namely: categorization, generalization and abstraction. Read distinguished three different types of associations: paradigmatic, syntagmatic and analytic
answers, compared to 53% in the child role. This might prove that adults unconsciously use abstract responses to distinguish themselves from children. Because of these results, Milgram wondered what would be “the age at which such role taking becomes possible and whether, at any age, children might show the paradoxical capacity for role playing at a level of abstraction higher than that of their spontaneous behaviour” (Milgram, 1960, p. 87), which consequently led to a new study. Three hundred children in different grades (2-8) grouped by academic achievement were asked to take the test. The average and superior children participated in the test because they had less trouble reading and
understanding and, therefore, would function as a control group. The written tests were conducted in the classroom. The children had to answer “by drawing a line from the stimulus word to the response of their choice” (Milgram, 1960, p. 87). There were 48 items in the test, of which the stimulus words were all nouns. The association words were nouns coupled with either a noun, an adjective or a verb. To guarantee the validity of the test, there were two different versions in which the word frequency was equal but which had a different yet systematic bias that favoured the concrete response. In other words most participants would probably choose a concrete word in their response, because the test made it most likely to do so. Unfortunately, Milgram does not explain how this was exactly
accomplished. One group took either the child or adult role and the other group had to answer in the self-role or spontaneously after which these conditions were compared to one another. The outcome held that the second and third year subjects mostly gave the same answers, no matter what role they had in the test. This indicates that the children were not able to distinguish the different roles and to “discriminate between associative styles” (Milgram, 1960, p. 88). One grade higher, that is to say fourth grade and so on, the children were able to make this discrimination, meaning they chose fewer abstract associations for the child role and more for the adult role, while this difference was not so obvious when children answered in a self-role.
role-taking are all important as they tend to influence the results in one way or another. One should keep that in mind while designing a test.
2.3.3. Individual Differences
Except for differences in the tests, one should also take account of differences between individuals. Every person is different and will, therefore, respond differently to a test. The role-taking experiment showed that the concept of child versus adult plays a role; an idea that may have risen because age of the subjects itself is important. First of all, children first have to learn the language, which means that their word knowledge is of an equal level as that of adults. Moreover, when a child has some basic knowledge of a word, it is not
necessarily true that the child has deep knowledge of that specific word. As stated before, deep word knowledge can be tested through word association tests. These tests have indicated that children respond differently from adults. For example, there are three types of word-associations, namely: paradigmatic, syntagmatic and analytic. Responses of children tend to be mainly syntagmatic, and consequently also are contiguous, whole-part and rhyming. Responses coming from adults on the other hand tend to be paradigmatic, which means that they are based on contrast, coordination and similarity. The type of response changes when children grow older. This shift does not necessarily become evident in children’s everyday language use; however, through word association tests it becomes possible to look into this. These tests have proven that the first word category that shifts is the noun and, soon after, the adjective. The category of the verb however, shifts much later, and unlike in the previous mentioned categories, this process takes place gradually. It is also said that children’s responses are action-orientated. Their associations deal more with actions than the associations adults make.
use more concrete words in their associations, while adults use more abstract words. The reason is that concrete words are used in a functional or situational context, while abstract words are “associated with intellectual maturity” (Milgram, 1960, p. 81). Concrete words are words such as synonyms, superordinates and subordinates, while abstract words are words that are part-whole, adjectives, verbs, and concrete contexts.
A third difference that is found besides category-shifts and the use of concrete/abstract words, is the notion that one will or will not answer in the same category as the stimulus word. Milgram and Goodglass claim – just as Read – that adults tend to give an answer that is of “the same part of speech as the stimulus word” (Milgram, 1960, p. 83). In other words: when the stimulus word is a noun, the adult’s answer could very likely also be a noun. Children, on the other hand, did not show really different behaviour, although they do not
necessarily answer in the same category. They can also give a quality, function or similar elaboration of the stimulus word. When the children grow more mature this pattern changes into one showing responses that contrast the stimulus word. Milgram and Goodglass have named these answers coordinate responses.
The conclusion of their studies stated that children tend to think mainly in concrete concepts, while it is possible for adults to think both abstractly and concretely, meaning that adults still include the concrete thinking in their own. In other words they have not lost their childlike associative abilities, but these abilities have developed during maturation and can now coexist.
Age, however, is not the only individual difference that plays a role. Thomas Merten (1995) studied word-association tests and looked at various
variables that could influence the results. Merten mentioned among others age, personality variables, intelligence, self-reflected thoughts and speech disorders. His research mainly dealt with schizophrenics, therefore his research will not be dealt with in depth. A few of his general points will be elaborated upon however, as he did a control test on normal subjects.
was put together that included “25 common German nouns (plus 2 warm-up items) for which free association norms had been established on a normative sample of 100 persons” (Merten, 1995, p. 252). Words that could be ambiguous or at least arouse confusion were excluded from the test. The testing took place on an individual basis. To test the verbal intelligence, another multiple choice test followed afterwards.
As for the results, only the parts relevant for this study will be mentioned. According to Merten “it is clear that, in contrast with the free association test, the common and the individual response conditions are much better at discriminating the subjects on the background of personality characteristics” (Merten, 1995, p. 254). Moreover, when the multiple regression analysis is applied, the
concordance scores are predicted by age, verbal intelligence, psychotism and, most of all, neuroticism. These variables, therefore, might be of importance when conducting a word-association test.
Thus, in sum, when healthy people are involved, age and intelligence lead to difference results in word association tests, as may many other personal differences that have not been mentioned here. This may require a need for more in depth studies concerning individual differences.
2.3.4. Types of responses
Individual differences lead to different responses, but what different types of responses are possible? Children’s responses differ from those of adults and as a consequence, the type of responses is also different. These differences can be found on the field of paradigmatic versus syntagmatic, and also in semantics. Paradigmatic responses are based on responses of the same word class, while syntagmatic responses, on the other hand, have a “sequential relationship to the stimulus word” (Schmitt, 2001, p. 40) and mostly do not belong to the same word class as the stimulus word. Schmitt lists another type of word association,
“types of links between a stimulus word and a response” (Greidanus, 2001, p. 569), which are co-ordination, collocation, superordination and synonymy. These four types can all be paradigmatic as well as syntagmatic in nature.
Children’s and adults’ answers do not only differ in their paradigmatic or syntagmatic nature, but also in their part of speech. Brown (1960) studied the
phenomenon that adults answer in the same part of speech while children do not have this strong tendency. To test this, four groups of each 20 participants – in total 80 participants – were formed. The first three groups consisted of children from grades I, II and III respectively who all lived in a middle income area. The fourth group consisted of adults, students or staff at the school. The free
association test consisted of 36 items out of which always 6 words represented one part of speech. In the end 6 parts of speech were represented, namely transitive nouns, intransitive nouns, count nouns, mass nouns, adjectives, and adverbs. All words were high frequency words in the Speech of American Elementary School children. Furthermore, the researchers claim that in earlier studies, these words proved to create differences between adults and children. A second test was conducted in which 12 non-existing nonsense words were presented; each was put in two sentences making it possible for the participants to connect the non-existing word to one of the six parts of speech. The
participants could score on the association test by assigning the answers to a part of speech, and on the making-up test by looking at the part of speech the participant assigned the new word to. The results showed that age and part of speech proved to be important variables, and more importantly, these two
variables interact. When the participants grow older, they give less homogeneous responses in the case of count nouns than in the other parts of speech. Adult responses are mostly “transitive verbs, naming human actions which ordinarily have some small inanimate thing as object” (Brown, 1960, p. 11).
2.3.5 Conclusion
Children acquire concepts and word knowledge gradually. Word association tests are a useful measure to look at someone’s depth word knowledge (e.g. its
meaning, its form, its function and its position), because they show whether the full knowledge of a word has been acquired or if the present knowledge is only partial.
Besides concept learning and word knowledge, several variables concerning word associations were introduced. First of which is the mode of
experiment, which includes written versus oral testing, and free word association
test versus forced choice word association tests. The second one is different
stimuli. Frequency of occurrence and frequency of use are important factors, but
also the difference between semantically related or semantically non-related is of great influence. Another factor that is mentioned is role-taking. Personal
characteristics also proved to be of importance, for example the subject’s age,
differences between children and adults, education, intelligence and personality variables. The type of responses can also be important, such as the factors of semantics and the different word choices of children and adults that play an important role. Moreover, also the different parts of speech influence the test results.
It can be concluded, therefore, that the type of responses changes during an individual’s development. A child gives different answers, because its word knowledge is still developing rapidly, while an adult already has already acquired quite an amount of knowledge. Even though an adult’s word knowledge will also develop, this will not cause any major changes in response types. A child does not make the same connections between words as adults do, for they lack the basis an adult has and may, therefore, give different types of responses in comparison to an adult.
2.4 Differences in Word Knowledge Development in L1 and L2 Learners
process or whether these processes cannot be compared. When a child acquires his/her first language a basis establishes holding concepts and constructions. Even though the second language may be of a different nature, these concepts and some constructions may very likely be the same in both languages,
implicating that the L1 provides a basis for the L2. However, although the learners have acquired this basis, they lack a sufficient word knowledge in their L2, which may lead to different associations in a second language. They do not know the same words in this language as in the first and therefore, may
experience difficulties making connections between different words.
This chapter deals with the stages both learners of a first language and learners of a second language respectively go through and how this influences word associations made.
2.4.1 First Language Learners
As more research is being carried out, various results, ideas and test methods concerning language acquisition have come into existence, which led to the conclusion that children associate certain words differently in comparison to adults. Native language learners first need to acquire sufficient word knowledge – among others – before they can fluently speak their native tongue, which
requires the expansion of the number of concepts in one’s mind.
The acquisition of a first language during early childhood shows similar characteristics in different languages around the world. Babies only start crying when they are, for example, hungry and uncomfortable, but later on the children start to make other noises, implicating that they are able to
1960, p. 2). It is expected that the child will have knowledge of the basic structures that are important in the language.
Even though the process of learning words seems to develop quite quickly during childhood, first language learners also experience difficulty grasping the complete lexicon, just as second language learners do. “Studies have estimated that English native-speaking university graduates will have a vocabulary size of about 20,000 word families” (Schmitt, 2001, p. 3). As most university graduates will be just over twenty years old, they seem to learn about 1,000 word families a year. Roughly speaking a five-year-old would have acquired knowledge of 4,000 to 5,000 word families.
But knowing all these word families does not necessarily mean a child – or even an adult – knows all possible aspects of a word. The phase in which one can recognize a word, understand it, but not use it in his/her own speech is called receptive knowledge. Only when able to actually use the word in his/her own language one has productive knowledge about the word. It is also possible to be able to use the word in your own – spoken – language, but to not understand the word when seen in its written form.
When learning a new word, meaning is considered to be one of the most important aspects. However, recent studies showed that knowledge about the written form of a word is also a “key component to both vocabulary knowledge and language processing in general” (Schmitt, 2001, p. 45). Knowledge about written form also connects different words to one another for their similarities in for example spelling.
Words can be connected in different manners by children. Verhallen (1980b) named a few: schoolse relaties, deel-heel relaties and essentiële
betekeniskenmerken, or in other words: relations that are prescribed according to
certain rules, part-whole relations and essential meaning qualities. These connections belong to the deeper knowledge one can acquire of a word. The knowledge is far less deep, when the associations are made because of
However, the associations made by children and adults differ. For instance, children tend to name words that are concrete, while adults generally prefer abstract words. In a certain period in children’s development, children reach a stage that Fisher calls “formal operational”. In this stage children “are able to manipulate abstract ideas as well as concrete objects and, therefore, can recognize set-member, type-example, and other hierarchical organization of [a certain kind of] knowledge” (Fisher, 1985, p. 392). This probably is the stage in which the child changes its thinking and starts thinking of words in an adult-like manner.
Even though it is already general knowledge that this period of
adolescence is a time of change, Fisher claims that during adolescence the word associations change too. The group of adolescents differs from both children and adults, for it is the phase in between child- and adulthood, meaning this transition period will show characteristics of both children and adults. It is the time that cognitive abilities of the young people become clear. The knowledge framework of the group of adolescence is the victim of many processes like reorganization and reorientation. Not only the age of the children changes, causing a difference in word associations, but boys and girls also tend to give different associations (Fisher, 1985).
When the end of adolescence is near and the children reach the stage of adulthood, it becomes clear that these changes have taken place. This change is mainly based upon semantic differences. The responses given by children are more contiguous, and whole-part. The responses given by adults are more based on contrast, coordination, and similarity. Children tend to make a sound which has more of a rhyming effect, meaning that it sounds like the sound of the stimulus word. Another difference between the two age groups appears to be that adults more often give a paradigmatic response, while children give a syntagmatic response.
People have been learning second languages – at least – since the time of the Romans and perhaps even earlier. At one moment in time the focus in second language teaching lay on memorizing vocabulary, while the next moment grammar seemed to be of more importance. One of the newest methods is that of learning through communication and interaction. Other parts of learning a language that have been emphasized over time were reading and writing,
listening, and interacting in the second language. It is also possible that grammar and vocabulary interact and are, therefore, both of major importance. Sometimes it was thought that word choice was created by a sufficient knowledge of
grammar, while in reality, the choice was made out of lexical considerations. Even present day, linguists do not agree to which method is best to acquire a second language.
A person already knows a language and because of this the person already has some basic knowledge when learning a new one. This basis is most useful when the second language is closely related to the first one. It also
matters whether the learner is cognitively mature, whether the general
knowledge of the world is relatively large or small, whether the person is afraid to make mistakes, whether there is room for silence or whether s/he should speak from the start, whether the person has time to learn, whether s/he comes in contact with proficient speakers, and whether s/he receives feedback, etc. All these factors influence the language proficiency (Lightbrown, 1999, p. 31-32).
However, even though it gives a basis, already having knowledge of a language when learning a second language is not only beneficial. First language knowledge can also cause errors to arise in the second language, as these two different languages will influence each other. Learners will combine different aspects of the two languages, because of which wrong expressions might arise.
Second language learners share a common characteristic with first language learners, namely access to modified input, which is the same as child-directed speech in the case of a first language. Just as may be the case in a first language situation, errors are often ignored as long as the meaning of the
message is still clear, and consequently the learner might not learn the language correctly. Only when the error does intervene with the message, the speaker is most likely to be corrected. One of the rare places where people are often corrected is in the language classroom. Not correcting errors is not the only obstacle that occurs when learning a language, as there is also the issue of noticing. Teachers tend to think that the extent of children’s word knowledge and therefore also the depth word knowledge is far better than it really is. They do not notice when a child only has partial or even no knowledge of a word until they give a completely wrong definition or appear to not know the word at all.
Dutch and Turkish children and claims that this difference not only exists between different languages but also between different age groups, no matter what the native language is. Because of this it can be concluded that second language learners know fewer words than native speakers do.
One of the reasons that children have not acquired extensive knowledge of a word might be that adults tend to speak at the level of a child. Even though adults already have sufficient knowledge of their first language, they use simple structures that resemble the grammar of a child. “Many of the sentences
produced by second language learners in the early stages of development would be quite ungrammatical in their first language. What is more, some
characteristics of these simple structures are very similar across learners from a variety of backgrounds, even if the structures of their respective first languages are different from each other and different from the target language” (Lightbrown, 1999, p. 36). This might indicate that languages are universal, as correlates to Chomsky’s theory about first languages. Chomsky does not have a lot to say for second languages, but other linguists claim that second language learners know more about the language than they could have known if their knowledge
depended only on the input they had been exposed to (Pinker, 1995). Even though Second Language Learners already have some basic knowledge of how languages work, the associations made in the second
language still need not be similar to those made in the first language in regard to the proficiency level. The maturation process has taken place, which means that the chosen words might already show the adult characteristics that have been mentioned before. However, when the second language word knowledge has not sufficiently developed yet, responses may differ from responses given in a first language. As people keep on learning their whole lives, their knowledge changes. It is a dynamic process, meaning situations change due to various influences and therefore, word knowledge and word associations change.
Each environment has a different set of words typically used in this particular environment. Some people will have this input and others will not, depending on which environment is ‘visited’ often.
Not only has the environment influenced word association behaviour, but in the first language, native speakers tend to produce associations that are stable. Generally they seem to be giving similar semantic related responses. Second language learners on the other hand produce associations that are much more diverse and unstable; often their responses are based on purely phonological, rather than semantic, links with the stimulus words (Greidanus, 2001, p. 569-570). The produced associations are of an irregular nature and mostly deviate from the type a native speaker would produce. “This is partly because L2 responses often include clang associations. It is also presumably because the organization of L2 learners’ mental lexicons is usually less advanced. [Not only is their vocabulary insufficient, but also the quantity and the kind of knowledge that accompanies these words are inadequate. There is more to learning words than only filing them into one’s memory. When one learns a new word, it not only broadens one’s word knowledge, but it also deepens the already acquired knowledge of other words, as also is shown by Greidanus and Nienhuis. New connections concerning meaning between words will arise. When this process lacks behind, L2 learners often have insufficient knowledge of the second language]. Second, L2 subjects frequently misunderstand the stimulus words [because of their insufficient knowledge], leading to totally unrelated associations. Third, non-native speakers, like L1 children, tend to produce more syntagmatic responses, whereas native-speaking adults tend toward paradigmatic responses. Fourth, L2 responses are relatively unstable. However, with increasing proficiency in the language, L2 responses seem to become more like those of native speakers. This suggests that the associations of L2 learners, like other elements of word knowledge, evolve in an incremental fashion” (Schmitt, 2001, p. 41).
to fossilized errors in the second language use that are difficult to overcome. Also a lack of knowledge leads to ‘weird’ unexpected associations, for a person has not made the expected connections yet but also to a different nature of
responses.
2.4.3 Conclusion
Concluding from the previous sections, in general learners of a second language have not acquired the language just as thoroughly and in depth as the first language. Not only will the first language serve as a basis for the newly acquired language, but the learner will also have acquired an insufficient number of words and an insufficient amount of knowledge of these particular words. For example, the learner will not see all possible connections between different words; s/he will not know the meaning of each word and will certainly not know all four aspects of all these words. Especially younger people, who have just started to learn the second language, will not have a lot of knowledge of this language – especially because their first language may not have adequately developed yet – while adults who have studied the language in more detail and for over a long time will have deeper word knowledge. In comparison with the first language learners, the second language learners have to start from scratch and start by learning
vocabulary, with as only guideline, their meaning and their form. Then its function and its position will not be clear to the learner yet. Later on, when one has
only know many words, but also the deeper meaning of these words. Besides that, grammar, pronunciation and communication will be almost flawless.
2.5 Expectations
As for word associations, the idea of beginning versus advanced word
knowledge will have a few consequences. The question raised here is: which variables influence associations and when does this becomes clear? To look into this matter a simple free word association experiment was designed in which three different age groups of Dutch origin – all learners of English – were asked to give the first three words that came to their mind when reading a few simple high frequency English words. The following expectations ought to be kept in mind.
- A beginner language learner will not see many connections between different words and therefore, have the language knowledge level of a child and will make more concrete associations;
- A beginner might tend to think primarily of words that are used quite often and are therefore higher in frequency, due to a lack of knowledge of less frequent words;
- The answers of a (young) beginning learner will be more syntagmatic; - An advanced learner will show more of a variety in his/her associations
than a beginner would;
- An advanced learner will also give more abstract and paradigmatic responses as s/he will probably be older, and moreover, will have more word knowledge and be able to make various connections;
3. The Test
3.1 Method
During adolescence people develop into different human beings. Earlier tests appear to indicate that also word associations undergo a transformation. This test aims to see whether this is the case and if so, what exactly changes and when this change takes place. A form containing some questions aiming to individual differences and four words to which a response needed to be given was handed out to three different age groups at a secondary school.
3.1.1 The Participants
To study the different variables, the test was conducted at Christelijke
Scholengemeenschap Liudger; a secondary school offering all different levels of
education. CSG Liudger is located in diverse areas in Drachten and its
surroundings. The location where the test was held is de Raai in Drachten. Three different groups were asked to fill in a form. The pupils were all following the VWO program, the highest level of education in the Netherlands, meaning they have the possibility to enter university after having passed their exams. All groups followed an English language course.
The first group were first graders of the ages 12 and 13. Mostly this was their first year of profound English education, but they should have enjoyed two prior years of English education at their primary school, with the exception of one girl who claimed to have had English lessons for 4 to 5 years. The class
consisted of twenty children, of whom nine were girls and eleven were boys. Because these pupils were part of a new education system that began this year, their lessons involved more spoken English interaction than the other groups.
The third group consisted of fifth graders, which is the year preceding the exams. Their level of education – and thus knowledge – ought to be (nearly) sufficient enough to pass the exams in the following year. There were eighteen girls and six boys all between the ages of 15 to 19. At this stage, the pupils’ ages are so diverse because many students did not succeed to pass every year immediately, while other pupils achieved so well that they were possibly allowed to skip a year at this school. Because of this, it is more difficult to say whether the pupils have had only five years of English at the secondary school, and two years at the primary school, especially because numerous pupils in all classes tended to forget that they had enjoyed English lessons at their primary school. Moreover, who did note down that they had had 2 years of English at the primary school, they all added that these two years were negligible as the level of English was not ‘that high’. Therefore, it is safe to conclude that all of the pupils from the third group at least had five years of English, but the medium will be around seven with a maximum number of nine years.
As for the individual differences: age was an important factor, as well as gender and the number of years every pupil had experienced English education. There probably were more, but these were the only ones accounted for in this test.
3.1.2 The Materials
As for the materials, the test itself was a free word association test, meaning the participants received stimulus words, but they had to come up with associations themselves. In a form the participants were asked to fill in their: group, age, gender and how many years they enjoyed previous English education. In the following section four words were presented to which the participants had to come up with three responses that came to mind first. The chosen words needed to be understandable for all levels of English proficiency and thus needed to be high in frequency of occurrence. Because all pupils would – probably – be
house, drive, time, and know. As one can see in table 1 they all have quite a high
frequency of occurrence.
Table 1. Stimulus Words
Word Frequency House 49,295 Drive 8,800
Time 150,233
Know 120,632
Prior to the stimulus words a short introduction said that the pupil had to read the word and fill in the first three English words that came to his or her mind. Below this, the four words were listed, with room for the pupils’ responses.
3.1.3 The Procedure
The procedure was fairly simple. After a brief introduction, the class would get an explanation about the form that needed to be filled in. In the first group this introduction had to be given in English because this was part of their own bilingual program. The other classes received a Dutch introduction, explaining they had to fill in their own personal information on top of the page, after which they had to read the words below; the first three words that came to their minds had to be written down in English and they were not allowed to think too long about either the words or the answers. After this introduction, the forms were handed out, the pupils filled them in and after five minutes, when everyone was finished, the forms were handed in. A few pupils had difficulties with the word
know - even though they were third graders – but overall no difficulties came up
3.2 The Results
After the pupils had filled in the forms and wrote down the first three words that came to their minds when reading the stimulus word, the forms were collected again. The three words were looked at in different manners. Furthermore, even though the children probably had more time to think of the last word than of the first, no differentiation was made between the first and the last word that was written down. The material has been analyzed according to the discussed literature. In order to look closely at the different words, they were taken
separately and put in tables. In the following section, attention will be paid to the concreteness versus the abstractness and the syntagmatic versus the
3.2.1 Concrete Vs Abstract
Milgram and Goodglass claimed – among others – that children tend to give responses that are of a concrete nature, while adults tend to give more abstract responses. To study whether this is indeed the case the results were put in a table. First when one looks at the number of concrete versus abstract responses in percentages it becomes clear that the number of concrete responses dwindles, while the abstract responses become more standard, as seen in table 2.1 to 2.3.
Table 2.1 Concrete Vs Abstract Group 1
Table 2.2: Concrete vs Abstract group 2
The last group gave a better view of the shift from concrete to abstract. As in this group 59% of the pupils gave a concrete response, which is in contrast to the first group where 68% did. This is not a major drop, but the difference is there.
However, 41% of the pupils gave an abstract response. This is a bigger number when compared to the first graders of course, but the real shift from concrete to abstract did not occur in these groups for the majority of the answers were still concrete.
Table 2.3: Concrete vs Abstract Group 3
Striking, however, is that in all cases the verb know leads to more abstract responses than to concrete responses. Although the pattern is lacking, for the
second group has the least abstract responses of all three groups. Still, the number of abstract responses is considerably higher than the number of concrete responses.
Table 2.4 Concrete Vs Abstract all groups
Group 1 Group 2 Group 3 Concrete 68,8% 65,8% 59%
Abstract 31,2% 32,5% 41%
However, to see whether this shift is a significant one a one way anova has been carried out. The shift previous showed in percentages apparently is not of a significant kind, for when all words are taken together, the significance level appears to be 0,051 while p<0,05. For the result to be of a significant nature, the significance level ought to have been below 0,05.
3.2.2 Syntagmatic Vs Paradigmatic
In the case of syntagmatic versus paradigmatic associations, an odd occurrence came forward. In the preceding chapters the conclusion was drawn that when children grow older, their responses become more paradigmatic. In the case of this test, this shift is not as obvious, because most responses already are
paradigmatic. However, the number of paradigmatic responses is not equal in the different groups.
Table 3.1: Syntagmatic Vs Paradigmatic Group 1
As can be seen in table 3.1 the majority of the responses are paradigmatic. Only in the case of the verb drive the responses were mostly syntagmatic. Strikingly though, this shift apparently takes place slower for verbs, for know also has notably more syntagmatic responses than the two nouns, but still the
paradigmatic answers are in the advantage.
The same conclusion can be drawn for the second group, the
third-graders. Their responses were for 70,11% paradigmatic and 28,45% syntagmatic. So it is safe to say that the majority of their answers are already paradigmatic and probably the shift has already taken place. Also in this group the only exception can be found in the case of the word drive which has more than 75% of syntagmatic responses in the test, whereas the two nouns house and drive only have 8,05 and 4,6 percent syntagmatic responses respectively and the other verb know only 25,29 percent which is still strikingly more than the nouns.
However, the paradigmatic responses are in the advantage again, even though the total numbers are smaller than that of group 1, as 74,17 percent of the responses were paradigmatic in group 1, while only 70,11 percent of the
responses in group 2 were paradigmatic, but this difference might be negligible.
Table 3.2: Syntagmatic Vs Paradigmatic Group 2
The answers are different in the case of the last group, the fifth-graders. In the third group the number of paradigmatic responses rose again, because 73,96 percent (table 3.3) of the answers was paradigmatic which is very similar to the number of group 1. The difference between this group and the first two can be found in the fact that time gets more syntagmatic responses, while know gets less. It seems that these words have changed roles. Because of this the end percentage remains the same. This means that the paradigmatic responses do not rise in number; instead they come from different sources. To be precise, 22,22 percent of the responses for time was paradigmatic, while this was 4,6 percent in group 2 and 5 percent in group 1, which is quite a difference. For know this difference was 9.72 percent in group 3, versus 21,67 percent and 25,29 percent in groups 1 and 2 respectively.
Table 3.3: Syntagmatic Vs Paradigmatic Group 3
Table 3.4 shows the averages for the syntagmatic and paradigmatic answers of all groups. Here the impression is given that – on average – there is no shift from syntagmatic to paradigmatic answers as both group 1 and group 3 have the same percentage of answers for both categories. Only the second group deviates from the other two groups.
Table 3.4 Syntagmatic vs Paradigmatic all groups
Group 1 Group 2 Group 3 Syntagmatic 25,83% 28,45% 26,04%
Paradigmatic 74,17% 70,11% 73,96%
The one way anova that has been carried out in order to prove the significance of these results, indicates that both the results for time and know appear to be significant as the significance level is 0,014, while p<0,05. This also holds for the overall score. For house and drive however, these results do not appear to be significant at all, as the significance levels appears to be 0,942 and 0,394 respectively.
When this pattern is compared to that of the abstractness of the responses, an obvious similarity lacks. It is clear that the number of abstract responses grow when the pupils become older and have enjoyed more years of English language education, when only looking at the percentages, but this does
Average 18,75 26,04 53,25 73,96 - -
not seem to be the case for the paradigmatic responses. There is a small drop, but overall it is almost negligible.
Table 3.5 Abstract vs Paradigmatic all groups
Group 1 Group 2 Group 3
Abstract 31,2% 32,5% 41%
Paradigmatic 74,17% 70,11% 73,96%
3.2.3 Occurrence
In order to look at the number of occurrences, these were translated into percentages. Then these percentages were put – for each word in each group separately – in categories. As a result it becomes clear whether the older children have a more varied pattern of associations or that their associations become more similar. For example, in the category house, garden was used by 9 of the 20 pupils. Having 60 possible answers, this means that 15% of the
possible answers would be garden and therefore, a 1 is found noted down in the second column.
When looking at tables 4.1.1, 4.2.1 and 4.3.1 one should keep in mind that the number of pupils in the classes differed. The first group only had 20 pupils, the second group 29 and the third group 24, which means that the number of possible different words is 60, 87 and 72 respectively for each category. The total of possibilities is then 240, 348 and 288.
Table 4.1.1 : Group 1 Occurrences – 240 Possible
Group 1 35%-25% 25%-15% 15%-5% 5% or less Total of different Words House - 1 8 19 28 Drive 1 - 5 22 28 Time - 1 3 26 30 Know - - 6 29 35 Total 1 2 22 96 121
Table 4.1.1 makes clear that different the pupils came up with various
words are mentioned by quite a number of pupils. These words can be found in table 4.1.2 with the belonging percentages of occurrence and the frequency of the association according to the National British Corpus. Only the stimulus word
know did not lead to one response word that was mentioned by many students in
group 1. It also may be surprising that both garden and car occur quite often in the National British Corpus, while clock only does so 2,800 times.
Table 4.1.2 Top Associated Words (More than 15%)
Word Association Percentage of Occurrence Frequency of Association (out of 1,000,000,000) House Garden 15% 10,851 Drive Car 31,67% 26,731 Time Clock 23,33% 2,800 Know - - -
Group 2 on the other hand shows slightly different results. There are quite a number of words that only occur a few times (5% or less), indicating that the answers of the participants in group 2 show more variation. However, one should remember that there are more participants in group 2 than in group 1, meaning that the difference in variation may have a different reason than age and years of education, namely the number of pupils that can mention different words. On the other hand, in group 1 quite a lot of words occurred between the 5-15%, which declines in group 3, as only 9 words occur that often. This suggests that there was a shift from words that occurred reasonably often to words that barely anyone mentions.
Table 4.2.1. Group 2 occurrences
Group 2 35%-25% 25%-15% 15%-5% 5% or less Total of different words
House - 1 2 36 39
Drive 1 - 1 32 34
Know - 1 4 35 40
Total 2 2 9 128 141
Table 3.2.2 shows that the subjects keep on using the same words often, which may be caused by the frequency of occurrence in the NBC, with the exception of the word clock. The percentage of occurrence is almost the same as was the case in group 1, except for the stimulus word know. In the results of group 1, there was not a response word that occurred strikingly often, but in group 2 suddenly the word school occurs in 16.67% of the responses. In table 3.4 shows this is a rise of 6.67%. However, school is one of the words that occur often in the National British Corpus.
Table 4.2.2 Most Associated Words (More than 15%)
Word Association Percentage of Occurrence Frequency of Association House Garden 15,12% 10,851 Drive Car 31,03% 26,731 Time Clock 25,58% 2,800 Know school 16,67% 36,862
After having seen that the answers of group 2 became more varied, one will notice that the opposite takes place in group 3. Even though the difference between the two groups is not huge, these results seem to be in between group 1 and 2, while one would expect there would be even more variation than in group 2. Group 1 shows even more variation than group 3, for the first group has 96 words that occur 5% or even less, while this is 82 words in the third group. This cannot be explained by the difference in number of pupils, for group 1 had fewer pupils than group 3, which means that group 1 theoretically should have shown less variation instead of more. On the other hand, group 1 does have more words that occur slightly more often as 22 words occur between 5 to 15 per cent of the occurrences, while in group 3 this is only the case with 16 words.
Table 4.3.1. Group 3 Occurrences
different words House - 1 2 27 30 Drive 1 - 4 19 24 Time 1 - 4 19 24 Know - - 6 27 33 Total 2 1 16 82 101
In table 4.3.2 there is a striking difference with both the first and the second group, for the most often occurring word in the house category is family and not
garden as was the case in the first two groups. Family, however, is used even
more than garden was. Another striking point of observation is that again, the stimulus word know does not lead to one specific word that is used extremely often, as was the case in group 2.
Table 4.3.2 Most Associated Words (More than 15%)
Word Association Percentage of Occurrence Frequency of Association House Family 16,67% 33713 Drive Car 33,33% 26.731 Time Clock 27,78% 2800 Know
Tables 4.1.2, 4.2.2, and 4.3.2 only showed a few of the most mentioned words. To make clear what kind of change took place in the three groups, the five most frequent occurring words of each stimulus word were collected and subsequently put in table 3.4.
Table 4.4 Most occurring words
House Garden 15,00 15,12 12,50 House Home 3,33 4,65 12,50 House Kitchen 8,33 2,33 2,78 House Window 8,33 3,49 1,39 Know Knowledge 8,33 4,76 11,11 Know Learn 3,33 5,95 13,89 Know School 10,00 16,67 12,50 Know Science 0,00 8,33 0,00 Know Smart 5,00 2,38 6,94 Time Clock 23,33 25,58 27,78 Time Hour 8,33 8,14 6,94 Time Minute 8,33 2,33 5,55 Time School 0,00 4,65 12,50 Time Watch 10,00 13,95 11,11
When looking at the stimulus word drive a weak positive progression of the occurrence of both bike and drive can be seen, and a strong progression of occurrence of the word fast and a strong regression of the word motor, which was used 5% of the times in group 1 while it did not occur at all in the last group.
Road on the other hand, was used around the same amount of times in all
groups, although there appears to be a weak progression at the end.
As for the stimulus word house most striking is that family did not occur in group 1, while there was a huge rise in group 3. This also – even though less strong – is the case for the word home, while the opposite seems to be happening for both kitchen and window.
When looking at the results of the stimulus word time it appears that clock is mentioned most often in all three groups and a larger percentage of the group responds with this particular word. School is also a response occurring often in case of time even though it is not mentioned by the first group at all, but in contrast it is mentioned strikingly often in the third group.
However, these data do not give a clear picture of whether the answers become more varied over the years of whether every group just associates different words often. In order to see whether there really is more variation, the type token ratios are calculated for each of the stimulus words. In table 4.5 it is clear that the type token ratio slinks down for each of the stimulus words,
Table 4.5 Type Token Ratio
Stimulus Word Group 1 Group 2 Group3
House 0,467 0,453 0,417 Drive 0,467 0,390 0,333 Time 0,500 0,442 0,333 Know 0,583 0,476 0,458 Total 0,504 0,411 0,350 3.2.4 Part of Speech
In chapter 2 it has been mentioned that adults tend to give responses in the same part of speech as the stimulus word, while children do not have this tendency as strong. In table 5.4.1 the most striking occurrence can be found. Immediately, one notices that the responses are of the same part of speech in case the stimulus word is a noun, while this is not the case when the stimulus word is a verb. This conclusion can be drawn for all three groups, when looking at tables 5.4.2 and 5.4.3.
Table 5.4.1: part of speech gr. 1
In group 1 there seems to be especially a preference for nouns in the case of
house and drive while this preference is slightly less in cases of the two other
stimulus words, even though there is still clearly a preference for nouns. In case
of time and know 68,33% and 80% respectively preferred a noun, while this was above 90% with the first two words.
Table 5.4.2 shows something similar, although fewer people responded with a noun in the case of house – but still near 90% of the class did – while almost 99% responded with a noun in case of drive, so there is a rise in this case. Also when looking at time it is striking that also here more nouns – and thus the same part of speech – are used then was the case in the first group.
Table 5.4.2: part of speech gr. 2
Table 5.4.3. however, strikingly resembles table 5.4.1. except for the stimulus word know, which led to more answers in the same part of speech – and thus a verb – than was the case in the first two group, for both in group 1 as in group 2, 20% or even less of the answers were in the same part of speech as know, while this is 33,33% in the third group.
Table 5.4.3. part of speech gr. 3
Group 2 Same part of speech Yes No Percentages Yes No No answer Nr % House 78 8 89,66 9,20 1 1,1 Drive 1 86 1,11 98,89 - - Time 66 20 75,86 22,99 1 1,1 Know 14 70 16,09 80,46 3 3,4 Average 39,75 46 45,68 52,89 1,3 1,28 Total 159 out of 348 184 out of 348 182,72 Out of 400 211,54 Out of 400 5 out of 348 5,6 out of 400
Table 5.4.4. shows that barely anything has changed over the years. On average the three groups responded the same and gave half of the answers in the same part of speech and half of the answers in another category, even though they used another part of speech more often. What none of the tables actually show, but what has been mentioned, is that irrespective of the nature of the stimulus word the subjects appeared to have a strong preference for nouns.
Table 5.4.4: averages all groups
Group Speech Yes Nr No Nr Yes % No % Group 1 17,75 32,25 46,25 53,75 Group 2 39,75 46 45,68 52,89 Group 3 34,75 37,25 48,72 51,49
A one way anova that has been carried out over the results reflecting the number of responses that were in the same part of speech as the stimulus words; that is nouns as an answer to house and time and verbs as an answer to drive and know. Unfortunately, the significance level proved to be above p<0,05 for all words, meaning these results cannot be regarded as significant.
In the section “occurrence” it was discussed how often the subjects associated certain words, but clarity whether this has anything to do with the frequency of occurrence in everyday language use lacked. In this section the focus will lie exactly on the frequency in normal British English. With help of the National British Corpus (NBC) the words are divided into categories. In table 6.1 it is shown how these categories are created.
Table 6.1. Categories
Table 6.2 shows the results for frequency of the responses to the stimulus word
house for every group. It can be seen that the frequency of the response words
moved from less frequent to more frequent when the subjects become older. More words were mentioned that occur more than 25,000 times. Also the words that occur less frequent in the National British Corpus, experience a drop from 8 to 5 words. Strangely enough though, the responses of group 2 seem to be rising in every aspect except for the 4K category. However, this table only represents the different words that were mentioned, but not the number of times these words were mentioned; meaning that if all classes had named a same number of
different words the results might look slightly different. When looking at the percentages though, there seriously seems to be a drop from less frequently used words to more frequent words.
Table 6.2. House
Frequency Occurrences in the
National British Corpus
1K >25,000
2K 25,000-10,000
3K 10,000-5,000
4K 5,000-2,500
5K 2,500-0
Group Stimulus Word 1K % 2K % 3K % 4K % 5K % Total
1 House 2 6,9 6 20,7 6 20,7 7 24,1 8 27,6 29
2 House 4 10,3 10 25,7 9 23,1 5 12,9 11 28,2 39
The responses to drive, however, seem to follow a different pattern. There is a drop in more frequent words and a rise in less frequent words although this rise is the highest for the second group and then drops again. Group 3 did not mention any words at all that belong to the 4k group. Moreover, after a fall in group 2, the subjects name quite a lot of responses that belong to the 2k category in group 3.
Table 6.3 Drive
The responses to time are of a curious nature as well. For the number of frequent words (1K) drops between the first and the second group, after which in the third group it rises again to almost the same percentage as was the case in the first group. This also holds for the 2K and the 3K categories, except for the fact that group 3 has a higher percentage of words than the first group. The 4K category contains less frequent words and seems to experience a rise in number of words. The 5K category experiences the same rise, were it not for the fact that the fifth-graders use less infrequent words than the other two groups did.
Table 6.4 Time
Group Stimulus Word 1K % 2K % 3K % 4K % 5K % Total
1 Drive 5 17,9 2 7,1 7 25,0 2 7,1 12 42,9 28
2 Drive 2 6,1 0 0,0 8 24,2 2 6,1 21 63,6 33
