2.2 Empirical Studies of First Language Acquisition of Metaphorical
2.2.2 Hsieh (2004)
During the course of language acquisition, metaphorical ability is considered to be acquired late. There are considerable studies concerning English metaphors;
however, little attention is paid to the development of Chinese metaphors. Hsieh (2004) explored the development of Chinese figurative language, especially metaphors and metonymies by investigating both children‟s comprehension and production. Metaphors and metonymies are both cognitive processes but work in different ways. Metaphors include two conceptual domains in which an abstract concept is understood based on the concrete object; that is, in the mechanism of metaphor, the metaphorical meaning is mapped from the source domain to the target domain. On the other hand, metonymies involve no domain change. Hsieh (2004)
24
conducted two tasks and recruited 80 subjects. They were divided into four age groups and each group consisted of 20 subjects. Their mean age were 6 (kindergartener), 10 (fourth grade of primary school), 15 (junior high student), and 19 (university student).
In Experiment 1, the metaphorical production of all age groups was investigated.
The production of metaphorical expression was elicited with two types of instruction:
the Nominal instruction (A is B) and the Comparative instruction (A is like B).
Sixteen test items were used and classified into two categories: concrete nouns and emotion adjectives. Thus, four factors were discussed between the instruction types and target types. Experiment 1 was conducted to see which type of target item (concrete nouns or emotion adjectives) was better for metaphor elicitation. Test items were selected from Gentner (1988) and Levorato and Cacciari (2002). Since some metaphors were culture-specific, some cultural unrelated items were modified to be close to Taiwanese culture. Besides, the nominal instruction was tested prior to the comparative instruction to make sure that the linguistic pattern „is like‟ does not prompt the mappings. The answers elicited from Experiment 1 were classified into three general categories: literal (including paraphrase, examples, description), metaphorical (including attributive, relational, double, conventional), and metonymical (synecdoche, producer-product, user-object, cause-effect).
Experiment 2 was a comprehension test which consisted of two parts and it was conducted based on the results of Experiment 1. First of all, Hsieh (2004) investigated the subjects‟ preference for comprehending metaphors of concrete nouns. That is, whether the subjects preferred attributive metaphors or relational metaphors was discussed. Two types of metaphors which differed in that attributive metaphors are mapped based on the shape or appearance of an object and relational metaphors are
25
mapped based on their function. The second part was to discover the subjects‟
preferences for metaphors or metonymies when they comprehended emotion adjectives.
In Experiment 2, 16 questions were given altogether with three choices: in the comprehension of concrete nouns, choices including (a) attributive metaphor, (b) relational metaphor, and (c) nonsense mapping, as in (8); as for the comprehension of emotion adjectives, choices comprised differently (a) metaphoric expression, (b) metonymic expression, and (c) nonsense item, as shown in (9). In addition to the eliciting questions, the subjects were asked to justify their answers to confirm the meaningful mappings and exhibit the mapping processes.
(8) The moon is like A. a ball B. a light bulb C. a rice dumpling, because … (Hsieh 2004:41)
(9) Being happy is like A. a smile face B. singing birds C. head down, because … (Hsieh 2004:42)
The results of Experiment 1 indicated that under concrete nouns, targets with comparable instructions (X2 value = 110.35, df = 9, p < .05) triggered more metaphors than nominal instructions (X2 value = 109.98, df = 9, p < .05). The results were collaborated by Levorato and Cacciari (2002) that the comparative instruction which served as a prompt indeed enhanced the performance in mapping domains. Moreover, the elicitation under concrete nouns with the nominal instruction, the younger children (Mean ages = 6;7, 10;11) used attributive metaphors more frequently (6-year-olds:
N=73, 45.625% and 10-year-olds: N=61, 38.125%) than relational-nominal metaphors (6-year-olds:N= 6, 3.75% and 10-year-olds: N=18, 11.25%); on the contrary, the rest of the age groups (mean ages = 15;10, 19;05) preferred to use relational-nominal metaphors (15-year-olds: N=51, 31.875% and 19-year-olds: N=30, 18.75%) than
26
attributive metaphors (15-year-olds: N=13, 8.125% and 19-year-olds: N=20, 12.5%).
Under the elicitation of emotion adjectives with comparative instruction, the elder groups (Mean ages = 15;10, 19;05) produced more metaphors (15-year-olds: N=94, 58.75% and 19-year-olds: N=82, 51.025%) than the younger groups (6-year-olds:
N=2, 1.25% and 10-year-olds N=16, 10.00%). Generally speaking, the emotion adjectives with the nominal instruction triggered metonymies frequently (N=16.25%, 21.25%, 21.25%, 26.25%, in four groups respecitvely) than metaphors (N= 0%, 2.5%, 7.5%, 15.625%). The triggered metonymy was only limited to the subtype- cause and effect. Overall, the junior high students (Mean age = 15;10) produced the most metaphors of the age groups.
The results of Experiment 2 showed that the younger children tended to comprehend concrete nouns based on the shape or appearance of the object (i.e., attributive metaphors) (6-year-olds: P = 81.875% and 10-year-olds: P = 76.875%). On the other hand, when comprehending emotion adjectives, the younger children preferred metaphoric mapping (6-year-olds: P = 57.5% and 10-year-olds: P = 55.625%) than metonymical mapping (6-year-olds: P = 39.375% and 10-year-olds: P
= 43.75%); with the growth of age, metonymical expressions were preferred (15-year-olds: P=59.375% and 19-year-olds: P = 53.125%). Overall, Hsieh (2004) indicated that the children inclined to elicit metaphors based on the appearance of the objects. Besides, concrete nouns were prone to elicit metaphors whereas emotion adjectives were apt to elicit metonymies. However, Hsieh‟s research design was inconsistent. For example, in the production task, the instruction was differently made for different age groups, resulting in a bias in the findings.
27 2.2.3 Hsieh (2008)
Metaphors are non-literal uses of linguistic expressions. In order to comprehend the metaphorical expressions, meta-linguistic ability needs to be developed. Since the simple processing ability may not be able to comprehend metaphorical expressions, it is considered that metaphors are acquired late in children‟s language development.
Few studies have been conducted on the first language acquisition of metaphorical expressions in Chinese. Instead of looking into metaphor phrases and idiomatic phrases, Hsieh (2008) focused on a single word si „die‟ and reclassified its syntactic and semantic properties. She investigated the acquisition of Chinese word si with metaphorical meanings.
One hundred forty-four subjects were recruited from Grades 1 to 6 in an elementary school, which were divided into six age groups (Mean ages = 6;4, 7;7, 8;7, 9;7, 10;7, 11;6). Five issues were discussed in the study: Markedness Theory, context effect, animacy effect, age effect, and categorical effect. Two tasks were employed;
Words in Isolation (WII) and Words in Context (WIC).
(10) Si-diao „die off‟
(11) Jinxian hen shangxin, yinwei tade xiaomao zuotien si-diao le.
„Jinxian was sad, because her kitten died last night.‟
The first task consisted of si items in isolation as in (10) and the second task presented si items in context with animate or inanimate NP subjects as in (11). Both tasks included multiple-choice questions. In order to test the subjects‟ comprehension of the literal and metaphorical meanings of si, three options were given in each question: (a) literal meaning, (b) metaphorical meaning, and (c) irrelevant meaning.
Hsieh (2008) categorized si into three types based on its syntactic categories: Type 1 as a verb (which further classified into three subtypes: Type 1-1 a single verb, Type
28
1-2 in an RVC construction (V+si), and Type 1-3 followed by a verb (si+V)), Typ2 as an adjective, and Type 3 as an adverb. Each type was divided into literal and non-literal meanings based on the semantic properties of si.
The results showed that the literal expressions were easier to comprehend than metaphorical expressions which supported the Markedness Theory that as long as the marked form was acquired (i.e., the non-literal meaning), the unmarked form would be acquired (Hsieh 2008: 69). Moreover, it was found that the children‟s performance was remarkably enhanced by the contextual clues (i.e., WIC task).
The overall results for the comprehension of metaphorical meanings indicated that the children performed better on the WIC task than on the WII task, supporting previous literature that metaphors were found much easier to comprehend in context than in isolation (Gardner and Winner 1978).
As far as the animacy effect is concerned, the younger children (i.e., Grades 1 to 3) encountered more difficulties in comprehending sentences with inanimate NPs than animate NPs. Hsieh claimed that since the animacy effect was violated, the children failed to interpret the non-literal meanings of si. On the other hand, the elder children (i.e., Grades 4 to 6) had no difficulties in comprehending sentences with inanimate NPs, showing that the NP with an animate feature can determine the meanings of si.
In addition, regarding the categorical effect, the answers to metaphorical meanings of si followed a difficulty hierarchy in which Type 1 (si as a verb) > Type 2 (adjective) > Type 3 (adverb). This tendency corresponded to the acquisition of word class that verbs and nouns were frequently used; thus acquired earlier than other syntactic categories. The fact that adverbs that were the most difficult to comprehend might be due to the peripheral and optional contribution to the meanings of si. Among the subtypes of si as a verb, Type 1-3 (si+V) was the easiest to interpret. Overall
29
speaking, it was found that the subjects‟ metaphorical ability was constrained by the categorical type of si.
Furthermore, age was found to play an influencing role in the development of metaphorical ability. For the children at higher grades, they were performed better at interpreting both literal and metaphorical meanings; also, they comprehended the syntactic types of si better than the younger children. The age of 10 was found to be the breakthrough of interpreting children‟s metaphorical ability. This finding was consistent with the previous results (Gardner and Winner 1978, Gibbs 1987, Levorato and Cacciari 2002, Prinz 1983). On the whole, Hsieh (2008) investigated the acquisition of si in Chinese and discussed five issues. She discussed si expressions in three types of parts of speech: as a verb, as an adjective, and as an adverb; thus, different syntactic constructions of si were compared. Although similar linguistic expressions (i.e., single word with different meanings) were discussed, whether the findings of si apply to the results of lao remains an issue.
2.2.4 Hsieh and Hsu (2010)
Much attention is paid to the studies of the development of idiomatic expressions in production and comprehension. Most of the studies of idiomatic development focused on the native speakers of English, French, and Italian. There was little research of the idiomatic expression comprehended by Mandarin-speaking children.
Thus, Hsieh and Hsu (2010) conducted two experiments to investigate the effect of familiarity, context, and linguistic convention on L1 idiom comprehension of Mandarin-speaking children.
A total of 32 subjects were recruited from an elementary school and divided into two age groups: the mean age of the first group was 6;1, and the second was 9;6.
30
Besides, 16 adults were recruited as a control group. In Experiment 1, the subjects were asked to answer the meanings of chosen idiomatic expressions without any linguistic context. In Experiment 1, a list of 30 plant name idioms was selected from the Academia Sinica Balanced Corpus of Mandarin Chinese. These idioms were rated on a four-point scale by 45 elementary school teachers to confirm the familiarity degree to children. A total of 14 idioms were chosen from the assessment: seven familiar and seven unfamiliar. The subjects‟ answers were coded and classified into two categories: „Correct‟ (i.e., correct idiomatic meaning) and „other‟ (i.e., literal meaning or unrelated meaning).
In Experiment 2, a comprehension task along with a meta-pragmatic task was conducted. Hsieh and Hsu (2010) aimed to investigate how context and familiarity influenced children‟s comprehension and children‟s meta-pragmatic knowledge. The children were asked to choose a picture describing either the literal meaning or the idiomatic meaning. The selected 14 idioms were embedded in two different stories:
one depicting the literal meaning of the idiom and one depicting the idiomatic meaning. Each idiom with a story included four pictures in which the fourth picture contained two sub-pictures: one describing the literal meaning of the idiom and the other the idiomatic meaning.
The subjects were asked to choose one of the sub-pictures based on their comprehension of the idiom. After the picture selection, they needed to give explanations for their choice which was a meta-pragmatic task. In the meta-pragmatic task, the subjects‟ explanations were classified into four categories: (1) relating to linguistic convention, (2) relating to literal meaning, (3) relating to the context, (4) relating to other factors like story setting or ambiguous explanations.
The results of Experiment 1 showed that the 6-year-olds and 9-year-olds
31
comprehended familiar idioms better than unfamiliar ones with no context; however, in the familiarity condition, the 6-year-old children produced more „Other‟ answers whereas the 9-year-old children produced more „Correct‟ answers. It was found that familiarity played an influential role affecting the children‟s comprehension of idiomatic expressions. The results of Experiment 2 showed that the idiomatic context was a significant factor in the comprehension of idiomatic meanings. For the 6-year-olds, although they chose more literal answers in both contexts, the increasing proportion of idiomatic answers in literal and idiomatic context (literal context: 8.48%, idiomatic context: 25.43%, t (15) = 2.78, p = 0.013 in both familiar and unfamiliar conditions) indicated that the idiomatic context was indeed a useful aid to comprehend idioms. Regarding the familiarity, only the 9-year-olds performed significantly better on the unfamiliar idioms than on the familiar ones (F (1,15) = 7.079, p < .05). Although the results were not significant, the familiarity effect was found at the age of 6 in that the 6-year-olds chose more correct answers to unfamiliar idioms than to familiar ones. As for the meta-pragmatic task, the 6-year-olds explained their reasons mostly related to the context, the 9-year-olds provided explanations related to the context and linguistic conventions, the adults‟ explanations were more related to linguistic conventions, showing that linguistic convention developed significantly at the age of 9.
In brief, the overall results showed that the familiarity effect was found at the age of 6. The context was indeed an important factor in the comprehension of idiomatic meanings at the age of 6, which was evidenced by Gibbs (1991), Abkarian and colleagues (1992), and Levorato and Cacciari (1995, 1999). The results of Experiment 2 showed that all the groups performed better on the idiomatic context condition than on the literal context condition. However, the results are convincing because the
32
design of the literal context is a great interruption which mislead the subjects to choose the literal meanings rather than idiomatic meanings.
2.2.5 Summary
Table 2-3 summarizes the major findings and limitations of the four empirical studies reviewed in this section.
33
Table 2-3 Major Findings and Limitations of the Previous Studies
Major findings Limitations
Levorato and Cacciari (2002)
1. Linguistic aid: The instruction „is like‟
(i.e., comparative instruction) is a crucial cue to motivate the metaphor mappings.
2. Age: A developmental trend is found in the creation of figurative expressions.
3. Type: Transparent metaphors were easier to produce than opague ones.
1. Subjects: Three age groups
(2) A Judgment Task (by adults) Hsieh
(2004)
1. Linguistic aid: The instruction „is like‟
triggers more metaphors.
2. Age: A development trend of ability is found to produce metaphors for
concrete nouns.
3. Type: Concrete nouns were prone to elicit metaphors whereas emotion
1. Context: Metaphors were much easier to comprehend in context than in isolation.
2. Age: The age of 10 was the breakthrough of interpreting the metaphorical meaning.
3. Type: The metaphorical ability was constrained by the categorical type of si.
1. Subjects: Six age groups (Mean ages= 6;4, 7;7, 8;7, 9;7, 10;7,
1. Context: The idiomatic context was important by the age of 6.
2. Familiarity: The familiarity effect was found at the age of 6.
3. Convention: Convention is a significant factor at the age of 9.
1. Subjects: Two age groups (Mean ages= 6;1. 9;6) and adults
2. Materials: 14 plant name idioms 3. Tasks:
(1) A Word-Card Task without context
(2) A Story Picture Selection Task in literal and idiomatic context (3) A Meta-pragmatic Task
34
Generally speaking, in L1 acquisition of metaphorical language, some aids were found to help children in the process of acquisition. First, the word „like‟ was favored in producing metaphorical language (Levorato and Cacciari 2002, Hsieh 2004).
Second, the linguistic context was found to be an influencial clue to help comprehend metaphors (Hsieh 2008, Hsieh and Hsu 2010). Influencing effects have also been discussed thoroughly, such as familiarity (Hsieh and Hsu 2010), categorical types (Hsieh 2004, Hsieh 2008), and age (Hsieh 2008, Hsieh and Hsu 2010).
As for the limitations of these studies, although these researches have studied metaphorical language, their research designs were different. First, the materials were extremely distinct. For example, some investigated the type of metaphors (Levorato and Cacciari 2002, Hsieh 2004), some studied the single word expression (Hsieh 2008), and some looked into the specific type of idioms (Hsieh and Hsu 2010).
Second, different task types were employed in these studies; for example, some used only a comprehension task (Hsieh 2008, Hsieh and Hsu 2010), some employed only a production task (Levorato and Cacciari 2002), and only one of these previous studies employed both tasks (Hsieh 2004). Last but not the least, the age groups recruited in these studies were not the same. With different research methods, diverse types of figurative expressions, and different age groups, the findings varied as well. Yet, the influential factors like context, age, and type effect have not been commonly discussed.
2.3 A New Classification of Lao
In Section 2.1, several studies of the semantic features of lao in different syntactic constructions have been reviewed. It is argued that lao has diverse meanings and it collocates with different types of nouns (Ma 2002, Zuo 2009). Its meanings
35
vary according to the syntactic structures that lao modifies. In addition, the role of the animacy feature has been claimed to determine the interpretation of metaphoricality in Section 2.2 (Hsieh 2008). In this section, a new classification of lao will be proposed.
For the seeming new classification, it actually integrates the contribution from those studies (Ma 2002, Zuo 2009, Hsieh 2008). Lao is reclassified into five subtypes based on its meanings according to the degree of transparency to the core meaning „aged.‟
2.3.1 Lao with a Literal Meaning
According to the definition4 in ancient Chinese, lao is a noun that refers to a seventy-year-old elder. Later, it is conventionally used as an adjective in modern Chinese to describe a person who is far advanced in the lifespan (Ma 2002). This is the first type of lao, which is interpreted literally. This type corresponds to Meaning 2 discussed in Ma‟s (2002) study. When lao is used as an adjective, it often co-occurs with a noun as in (12).
(12) lao xiansheng old man „an old man‟
In (12), lao describes a man who has been living for a long period of time. The meaning of lao entails an object being at an advanced age in the life cycle. As far as the literal interpretation is concerned, lao has an animacy constraint on its collocating object. That is, the collocating noun ought to be animate, i.e., human beings or animals, in order to match with the semantic feature of lao. As indicated in (12), lao collocates with the human referent „man.‟ Specifically, an animate noun which is
4 XuShen (C.E. 58-147), ShuoWenJieZi (說文解字, “Explaining Simple and Analyzing Compound Characters”) was a Chinese dictionary from the Han Dynasty (2nd century CE). It was the first
dictionary to analyze the structure of the Chinese characters and to provide the etymology of the words.
It was also the first to use the principle of organization by the section headers (i.e.,bùshǒu 部首).
36
modified by lao comprise several types: (a) social title terms (e.g. taitai „Mrs.‟, xiansheng „Mr.‟), (b) senior kinship terms (e.g. yeye „grandpa,‟ nainai „grandma,‟), and (c) occupation terms (e.g. jingcha „police officer,‟ yisheng „doctor‟) (Feng 2008).
The animacy constraint on the noun modified by lao can be seen in (13), an ungrammatical sentence if lao is interpreted literally.
* (13) lao yaoshi old key „an old key‟
As shown in (13), the inanimate object has no life cycle; therefore, lao fails to denote the „old‟ condition of it.
2.3.2 Lao with a Non-literal Meaning
In addition to the literal meaning „old‟, lao can be denoted non-literally. Previous studies have shown that the meanings of lao consist of several levels: one of the levels refer to the core meaning (i.e., literal meaning), denoting „old‟; the other level refers to its semantic extensions in which lao is interpreted non-literally (Ma 2002, Feng 2008, Zuo 2009). The non-literal interpretation of lao is re-classified as Type 2 in the present study.
The literal meaning of lao entails the advanced state in a certain timespan, which infers „time continuity‟ as the common semantic feature of lao. The feature „time continuity‟ is „a state that has been long-standing in a certain timespan.‟ Two meanings are derived from this semantic feature: (a) „someone or something that has existed for a long time,‟ and (b) „one who is experienced at something‟ (Ma 2002).
The two extended meanings are regarded as the non-literal meaning of lao (Feng
The two extended meanings are regarded as the non-literal meaning of lao (Feng