CHAPTER 3 CORPUS ANALYSIS
3.2 Results of the Corpus Study
3.2.5 Syntactic Roles of ‘Pretty’ and ‘Charming’
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
However, when looking deeper through the score of collocates, one can see that the scores of modified nouns of charming were much higher than pretty. This may suggest that charming had a higher tendency to modify nouns related to person compared to pretty in the spoken discourse.
3.2.5 Syntactic Roles of ‘Pretty’ and ‘Charming’
In this section, the syntactic roles of pretty and charming will be presented. Table 3-16 shows the distribution of the adjectives’ tendency toward syntactic positions based on the analysis of the retrieved data in the study.
In this analysis, we only considered the adjectives which occupied either attributive or predicative position. In other circumstances, the instances were excluded from this stage of analysis and they were labeled as ‘others’, for example, the idiomatic expressions (e.g., sitting pretty, pretty penny, etc.), the adjective form of nouns (i.e., the pretty and the charming), and the convention use of charming (used to describe one’s disapproval when someone has been rude). In addition, ‘mistagged’ instances with errors in part-of-speech annotation were deleted from the analysis as well (e.g., It’s pretty cool.).
Table 3- 16 Distribution of Syntactic Positions
pretty charming
Roles Freq. Percentage Freq. Percentage
Attributive 242 46.01% 245 47.85%
Predicative 204 38.78% 196 38.28%
Mistagged 62 11.79% 69 13.48%
Others 18 3.42% 2 0.39%
Total 526 100.00% 512 100.00%
Table 3-16 indicates that both pretty and charming can appear in both attributive and predicative positions. Furthermore, both the target words tended to pre-modify the head noun (i.e., in attributive position), accounting for over 45% for both the adjectives.
‧
However, what makes them different was that the syntactic position which they applied when describing different categories of modified nouns (i.e., person, object, or animals) as showed in Table 3-17.
Table 3- 17 The Syntactic Roles Applied in Different Categories of Modified Nouns pretty
attributive predicative others mistagged Total Freq. (%) Freq. (%) Freq. (%) Freq. (%) Freq. (%) attributive predicative others mistagged Total
Freq. (%) Freq. (%) Freq. (%) Freq. (%) Freq. (%) position when it was used to describe nouns related to person (accounting about 57%);
whereas, pretty didn’t show its tendency apparently. Both attributive and predicative positions were commonly occupied by pretty in terms of descriptions of person.
Comparatively, both pretty and charming were more likely to pre-modify nouns related
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
to objects (accounting about 56% for pretty, and 69% for charming). One of possible reasons may be related to the difference in meaning between attributive and predicative adjectives. As Bolinger (1967) suggested that attributive adjectives were more often used to describe something’s or someone’s permanent characteristic (as shown in example 3-21-a); while, predicative adjectives can reflect temperate states or particular events (demonstrated in example 3-21-b).
(3-21)
(a) The wasps seem to be more discerning than human taxonomists. There's more to
attractive scent than a pretty flower though. [attributive]
(b) This plant's small yellow-centered white flowers are unpretentious but pretty,
borne in abundance throughout the year, and extremely fragrant. [predicative]
In example (a), the wasps were attracted by the scents from other species. There was something more attractive than a pretty flower. Here, the pretty flower referred to any kinds of flowers. The attributive adjective, pretty, was used to express the author’s opinion. To the author, flowers were all pretty in principle. Whereas, in example (b), flowers referred to a particular kind of flower grown on the plant and it was yellow. In this case, pretty took the predicative position aimed to indicate a temporary state. The scenery of flowers may disappear due to weather or other event.
Overall, from the syntactic roles of the two adjectives, it seemed that the language users preferred to use both pretty and charming to describe permanent characteristics or states of something by taking the attributive position (e.g., a pretty flower, a charming book). Whereas, in terms of description of person, the two adjectives were commonly used in both attributive and predicative position.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
CHAPTER 4
PSYCHOLINGUISTIC JUDGEMENT TASK
In the second study of the thesis, we intended to investigate what factors might influence EFL learners’ use of pretty and charming. Two versions of a questionnaire featuring an acceptability judgement task were designed for the EFL learners. In each version of the questionnaire, sixteen questions within two types of modified nouns (person and object) were included. Through the acceptability judgement task, it was expected that some interpretations of the participants’ judgements could be obtained. In the following section, we will firstly introduce the design of questionnaire in 4.1.1.
Then, in 4.1.2, the grading criteria will be presented. As for the participants, procedure and data analysis will be displayed in section 4.1.3 to 4.1.5.
4.1 Methods of Psycholinguistic Judgement Task
In this thesis, both corpus analysis and psycholinguistic judgement task mainly focused on two types of modified nouns—person-related modified nouns and object-related modified nouns. For the psycholinguistic judgement task, the questionnaire was mainly designed to investigate what factors might affect EFL learners’ uses of the two adjectives and the methods of psycholinguistic judgement task will be presented in this section.
4.1.1 The Design of Questionnaire
Three main variables. For the questions which focused on person-related
modified nouns (e.g., pretty girl, charming boy), they were designed mainly based on three variables: (a) gender, (b) patterns of modified nouns (i.e., pretty-only, charming-only, and both (called ‘common’)), and (c) source of beauty (i.e., inner and outer beauty). Each of them will be introduced respectively below.
The first variable of the questionnaire is gender. The gender of the person who has the qualities being described was controlled in the example sentences. In example
(4-‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
1), a female person and a male person, respectively, were described (All examples were retrieved from the COCA).
(4-1)
(a) She has a charming manner and she always likes to make jokes about herself.
(b) He has a charming manner and he always likes to make jokes about himself.
Secondly, in order to select the stimuli of the questions, the patterns of collocates were considered. In total, there were three patterns of the stimuli: pretty-only, charming-only, and common. The ‘only pattern’ refers to the strongest collocates of the target word. For example, according to the corpus result, clothes can only collocate with pretty as in example (4-2-a), while manner can only collocate with charming, as demonstrated in (4-2-b). As for ‘common pattern’, collocates are shared between the two target words. In example (4-2-c) and (4-2-d), the word voice frequently collocates with both pretty and charming.
(4-2)
(a) Judy sat there imagining all the pretty clothes she would have when she went to
the party. pretty-only (b) He has a charming manner and he always likes to make jokes about himself.
charming-only
(c) It's her good fortune that she can use her pretty voice to earn a living. common (d) He always uses that charming voice to say ‘Good morning’ to everyone.
common
Thirdly, source of beauty can be further divided into two types: inner beauty and outer beauty. It is hypothesized that this variable may affect participants’ uses of pretty and charming. The attractiveness of the female in example (4-3-a) is her charming smile,
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
and we categorized it as an ‘outer beauty’. As for (4-3-b), the man has a charming manner which may make people like him, and it belongs to an inner beauty.
(4-3)
(a) She lowered her window and gave the man her most charming smile.
outer beauty (b) He has a charming manner and he always likes to make jokes about himself
inner beauty From the definitions of pretty and charming shown in the dictionaries (see Chapter
3, section 3.1.2), one could know that both these two adjectives can be used to describe someone or something being ‘pleasant’ or’ attractive’.
In order to understand the participant’s use of the two adjectives in describing someone’s attractiveness, sixteen sentences in total (i.e., ‘gender’ ⅹ ‘pattern’ ⅹ ‘source of beauty’ =2ൈ3ൈ2=16) were constructed. In addition to the three controlled variables in the questionnaire, the gender of the participant was seen as a monitor variable in the study. A monitor variable may affect the outcome, but it cannot be kept constant in its value. It needs to be monitored. From the result of corpus analysis indicated in the previous chapter, we hypothesized that there was a relation between the gender of the participant (‘P’) and the gender of the person in the sentence (‘S’) whose qualities are described in the sentences. To test the assumption, in total, there were four combinations of gender in the study— the first gender is the participant; the second is the person-stimuli, [FemalePⅹFemaleS], [FemalePⅹMaleS], [MalePⅹMaleS] and [MalePⅹ FemaleS]. For instance, [FemalePⅹFemaleS] refers to a female participant who made a judgement on the sentence containing a female person. The former gender represents the gender of the participant, and the latter one stands for the gender of the person in the sentences.
‧
with different genders of the person described in the sentences were evenly divided into different versions of questionnaires. For example, manner, in the charming-only pattern has both person’s genders in two versions of the questionnaires, as shown in Table 4-1.Table 4- 1 Stimuli Used in Person-Related Modified Nouns Part of Questionnaire Patterns Adjective Source of
beauty
charming Inner beauty manner Female A
Male B
For example, (4-4-a) depicts a female person, and it was distributed into questionnaire A, while example (4-4-b) describes a male person distributed into questionnaire B.
(4-4)
(a) She has a charming manner and she always likes to make jokes about herself.
(b) He has a charming manner and he always likes to make jokes about himself.
As for ‘common pattern’, stimuli were shared between both pretty and charming. For example, traits frequently collocates with the two target words and there were four
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
sentences describing traits in total (i.e., genderⅹtwo patterns=2 ⅹ 2=4). Then, these sentences were distributed evenly into the two versions of questionnaires. Each version of questionnaire included two sentences describing the same person-stimuli in common pattern (one male and one female). Overall, there were eight questions in total focusing on person-related modified nouns in each version of questionnaires.
In terms of the questions which focused on object-related modified nouns, the stimuli were selected based on similar criteria—categorization of ‘patterns’ and ‘source of beauty’ were used. For object-stimuli, there were two different types of source of beauty—‘nature’ and ‘artifact’ as demonstrated in example (4-5).
(4-5)
(a) I like the idea of you shooting the pretty scenery of beach. … nature (b) This charming story is drawn from the true adventures of the author. artifact
For example, (4-5-a) scenery was categorized as a ‘nature’ beauty. Example (4-5-b) belonged to ‘artifacts’, which were made by human beings.
Similar to the person-stimuli, eight object-related sentences were designed to test the understanding of the learners’ use of the two adjectives. However, all the stimuli of the object-related nouns in the task were gender-neutral, there was no need to distribute the sentences into different versions of questionnaires. All of these sentences appeared in both versions of questionnaires. That means that the object-stimuli received double the number of responses from the two versions.
‧
Table 4- 2 Stimuli Used in the Object-Related Modified Nouns Part of Questionnaire Patterns Adjective Source of beauty R1 collocate
charming-only charming Nature nature
charming Artifact inn
common
pretty-only pretty Nature flower
pretty Artifact harbor
Overall, each version of the questionnaires contained sixteen questions (i.e., eight for person-related modified nouns and eight for object-related modified nouns), but the questions for person-stimuli were divided into two genders; the questions for the object-stimuli were the same for both versions. After showing the design of questionnaires, we then demonstrate the grading criteria in the following section.
4.1.2 Grading Criteria
All the sentences in the questionnaires were adopted or modified from the corpus data. In the acceptability judgement task of the study, five-point Likert scale was used to measure participant’s acceptability. Point-one means a sentence was ‘totally unacceptable’; whereas, point-five means that a sentence was ‘perfectly acceptable’.
Thus, it was expected that the results of the judgement task could reflect the participants’
degree of acceptability toward the different uses of pretty and charming. They tested the suitability of collocation rather than grammatical acceptability. If a question was left blank, the answer would be labeled as missing. The missing data would be deleted in the stage of data analysis.
4.1.3 Samples of the Questionnaires A and B
In the first part of the questionnaire, the questions were designed to gather the participants’ background information—gender, nationality, native language, and
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
language proficiency. Then, in the following part of questionnaire was the psycholinguistic judgement task including sixteen questions in total. Instructions of the questionnaire were shown in (4-6) below. The participants were required to rate their degree of acceptability from ‘totally unacceptable’ to ‘perfectly acceptable’ (i.e., from point-one to point-five).
(4-6) 第二部分:
作答說明: 此問卷共 16 題。請仔細閱讀每一題句子，圈選您對每一個句子的接 受度。1 為非常不接受----5 為非常接受。
Translations:
Part II:
Instruction: There are sixteen questions in total. Please read each sentence carefully, and circle the acceptability rating of each sentence. 1 means a sentence is ‘totally unacceptable’ to you--- 5 means it is ‘perfectly acceptable’ to you.
In terms of the presentation of the sentences in the questionnaire, they exemplified in example (4-7). The whole questionnaires are attached in this thesis as Appendix A and B.
(4-7)
(a) He dares say she is much too young to bother her pretty head (=智慧, which means intelligent in English) about such dilemmas.
(b) Let yourself be seduced by the charming nature and extraordinary landscapes.
In example (4-7-a), the subject of the sentence is he and the modified noun, head, is a strongest collocate of pretty. It was a sentence under the ‘pretty-only’ pattern.
Furthermore, pretty head means ‘intelligence’ in the instance and it is used to describe
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
an ‘inner beauty’ of the person (male in 4-7-a). Overall, the sentence was composed of three variables; that is, ‘pretty-only’, ‘male’, and ‘inner beauty’. To avoid the confusion of meaning, if a word may cause confusion, such as pretty head, the Mandarin translation was provided in brackets. In (4-7-b), nature is an example of an object-stimuli and it is a strong collocation of charming. It is a sentence of ‘nature’ and
‘charming-only’ category.
The hypothesis of the questionnaire designed for person-related nouns is addressed in (4-8), and the hypothesis of object-related nouns is addressed in (4-9):
(4-8)
Hypothesis 1: Pretty seemed to be more commonly used to modify females’ outer beauty, whereas charming was more commonly used to describe males’ inner beauty.
Here, we supposed that ‘source of beauty’ and’ ‘patterns’ have an effect on the participant’s judgement on the sentences.
(4-9)
Hypothesis 2: For sentences focusing on object-related nouns, it was hypothesized that the participants tend to use both pretty and charming to modify the beauty of
‘artifact’; while, they are more likely to use pretty to depict the beauty of ‘nature’.
Aside from the three controlled variables in the study (i.e., ‘gender’, ‘source of beauty’, and ‘patterns’), the participant’s gender is an important monitor variable for investigating whether there are different preferences of using pretty and charming for the participants with different genders. The hypothesis of the relation between the genders of the participants and the genders of the person in the sentences was established in (4-10).
(4-10)
‧
Hypothesis 3: For the sentences focusing on person-related nouns, ‘genders’, including the gender of the participant and the gender of the person in the sentence whose qualities are described in the sentences, have an effect on the participant’ s acceptability of the sentences.
4.1.4 Participants and Procedures of Questionnaires
In the acceptability judgement task, there were two groups. Group A was constituted by forty-four undergraduate students majoring in Education in a university in northern Taiwan, twenty-two for each version. All of them are EFL learners with Mandarin as their native language. Their English proficiency was categorized into three levels: high, middle, and low based on their performances of the college entrance exam of English subject. The categorization was the same as the class grouping in the freshman’s compulsory English course. The participants were randomly given a paper-based questionnaire.
Table 4- 3 Background Information of Participants in Group A
Version Proficiency Participant’s Gender
Total
On the other hand, Group B was constituted by forty-five participants taking part in a Mandarin course in a university in Taiwan. These learners are foreign learners in Taiwan. Regarding the participants’ native languages, Japanese (N=7) and Spanish (N=7) were the top one native language spoken by them, followed by German (N=6),
‧
languages are members of the West Germanic language branch. Overall, in spite of the varied native languages spoken by the participants in Group B, English can be regarded as their foreign language. The two groups were recruited to investigate whether the participants with different native language backgrounds would perform differently in judging the use of the two adjectives in describing someone’s or something’s attractiveness.Table 4- 4 Background Information of Participants in Group B
Group B
The participants were informed that the task was not a language test, and it would not affect their grades and scores. Additionally, they were free to ask questions about the meaning of the sentences during the task. They were first asked to write down their background information in Part A of the questionnaire, such as gender, level of English class, nationality, and native language. They then were informed to read each sentence carefully and answer the questions as honestly as possible. The instruction were shown in Mandarin. Most participants in both groups completed the questionnaires within 15 minutes.
4.1.5 Data Analysis
In the study, two-way and three-way ANOVA were used for investigating the effects between the variables in SPSS. In addition, the cases with any types of missing data were deleted in the study. Via this procedure, we expected to understand which factors may affect learners’ uses of the two adjectives—pretty and charming.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
4.2 Results of the Person-Stimuli
In this section, we will first present the results of the two versions of questionnaires by using the person-stimuli. Since the questions were selected mainly based on ‘gender’,
‘patterns’, and ‘source of beauty’, there will be three different variables in the analysis.
4.2.1 Overall Analysis of Group A and Group B
Test of homogeneity of variances. From the Levene’s test, the result indicated that
the variances (based on median) in the two groups were homogeneous, F (23,685)
=1.337, p=.135. Thus, we can run an ANOVA test. Since all the participants were English learners, we will firstly include all the data of the two groups of participants to perform a three-way ANOVA. Then, Group A and Group B will be discussed separately.
The analysis of Group A and Group B. Among all the participants in Group A
and Group B, the ‘patterns’ of collocates had a significant effect on the learner’s uses of the adjectives to describe people’s attractiveness, F (2, 703) =4.839, p<.01. From the estimated marginal means, one could find that the ‘charming-only’ pattern was the most acceptable use among the three ‘patterns’, with the mean of 3.601, followed by the
‘common’ pattern (M=3.531), and ‘pretty-only’ pattern (M=3.254). The results may indicate that the participants had no idea about the use of pretty. Using pretty to describe
‘common’ pattern (M=3.531), and ‘pretty-only’ pattern (M=3.254). The results may indicate that the participants had no idea about the use of pretty. Using pretty to describe