Abstract
When teaching Mandarin Chinese classifiers, teachers usually ask students to memorize ‘classifier + noun’, phrases as collocations. Given that Mandarin Chinese has a vast and complicated system of classifiers, the rote memorization of ‘classifier + noun’ collocations is challenging and monotonous. Therefore, the present study aims to improve that pedagogical method by applying cognitive linguistics. Previous studies did not explicitly demonstrate the cognitive approach to teaching Chinese classifiers, resulting in difficulty in practice. Thus, this study presents a step-by-step introduction to teaching Chinese classifiers with cognitive linguistics. Afterwards, an experiment is presented with children learning Chinese as a heritage language in Germany in order to test the efficiency of the cognitive approach and the traditional method (rote memorization of ‘classifier + noun’, collocations). Results suggest that the cognitive approach facilitated the learning of Chinese classifiers to a great extent. However, due to some limitations of the experiment, the cognitive approach did not outperform the traditional method. Even so, the present study concludes that the cognitive approach is worthwhile and promising for future teaching activities for two reasons. On the one hand, the cognitive approach helped students and was not worse than the traditional approach in this study. On the other hand, previous studies indicate many potentials of the cognitive approach. Finally, several suggestions learned from the limitations of this study are provided for further research.
Keywords
I Introduction
It is nearly unanimous that Chinese 1 classifiers are notoriously challenging for second language (L2) learners (e.g. Hansen & Chen, 2001; N.S.-Y. Liang, 2009; Polio, 1994; J. Zhang & Lu, 2013) and child first language (L1) learners (Erbaugh, 1986). One of the main reasons is that Chinese has a large number of classifiers, which leads to a complex system of rules for the classifiers. Generally, which classifier should be used primarily depends on the noun after the classifier. For instance, the classifier 匹 pǐ is properly used in 一匹狼 yī pǐ láng ‘one CLA 2 wolf’, but not in 一匹楼 yī pǐ lóu ‘one CLA building’, which should be 一栋楼 yī dòng lóu ‘one CLA building’ instead. Nevertheless, the use of classifiers is also affected by lexical taxonomy, linguistic convention, and stylistic creativity (H. Zhang, 2007). For example, the difference between 一个顾客 yī gè gù kè ‘one CLA customer’ and 一位顾客 yī wèi gù kè ‘one CLA customer’ lies in the classifiers, i.e. 个 gè and 位 wèi. The difference between these classifiers indicates stylistic variations. The former is generic, whereas the latter is more polite and respectful.
L2 learners of Chinese whose L1 does not require (many) classifiers, such as English and German, may struggle more with Chinese classifiers, which further challenges their teachers. To illustrate, for German learners a translation approach to teaching Chinese classifiers malfunctions. Because, for instance, the Chinese classifier 把 bǎ does not have an equivalent in German, beginner German learners of Chinese may translate ein Schirm ‘one umbrella’ into Chinese as 一伞 yī sǎn ‘one umbrella’, deviating from the standard form, 一把伞 yī bǎ sǎn ‘one
Therefore, this study aims to improve the efficiency of teaching Chinese classifiers by applying cognitive linguistics. I hypothesize that learners will learn Chinese classifiers more efficiently when they find out and utilize the image-schemata 3 of classifiers. This hypothesis will be tested by an experiment. Another contribution of the present study is that it offers a step-by-step demonstration of how teachers can apply cognitive linguistics to teaching Chinese classifiers in class, which is highly necessary for teaching practice but overlooked by previous studies.
II Literature review
Classifiers are frequently used in Asian languages, such as Chinese, Japanese, Thai, Korean, Vietnamese, and also in hundreds of non-Asian languages, such as Mayan and Australian aboriginal languages, and even in American Sign Language and Cantonese Sign language (Aikenvald, 2000; Craig, 1986; Croft, 1994; Emmorey, 2003; Erbaugh, 2006; Senft, 2000b). As to Chinese, Allan (1977) and Tai and Wang (1990) defined classifiers as nouns identifying prominent and stable properties of substances, in contrast with measure words used to denote the number of substances. However, classifiers also have the function of measure words, i.e. measuring substances. For instance, in the Chinese phrase 一张纸 yī zhāng zhǐ ‘one CLA paper’, 张 zhāng ‘sheet; piece’ is the measurement unit for paper. Meanwhile, 张 zhāng also presents the relatively stable property of paper, as the typical paper we use has been cut into sheets. Given the fuzziness between those two terms, this article uses ‘classifier’ as an umbrella term referring to morphemes that classify and quantify nouns, following other linguists such as Senft (2000a), H. Zhang (2007), N.S.-Y. Liang (2009), J. Zhang and Lu (2013), and S. Jiang (2017).
Based on the nouns collocating with classifiers, scholars categorize classifiers into two groups. One is nominal classifiers (N.S.-Y. Liang, 2009), and the other is verbal classifiers (Liu, 2016). Nominal classifiers classify and measure concrete substances, such as books, animals, and furniture. In the phrase 五本书 wǔ běn shū ‘five CLA book; five books’, 本 běn is a nominal classifier measuring the number of the books. In contrast, verbal classifiers are those classifying and measuring actions. For example, in the phrase 读两遍课文 dú liǎng biàn kè wén ‘read two CLA text; read the text twice’, 遍 biàn is a verbal classifier meaning the times of reading.
According to the countability of nouns co-occurring with classifiers, classifiers are divided into count-noun classifiers and mass-noun classifiers (H. Zhang, 2007, pp. 44–45). Count-noun classifiers are used to classify and measure countable nouns that naturally occur in discrete and countable units, such as tables and utensils. For instance, in the phrase 一个碗 yī gè wǎn ‘one CLA bowl’, 个 gè is a count-noun classifier since bowls are countable with their natural form of existence. Mass-noun classifiers classify and measure mass nouns whose natural forms of existence are uncountable and thus need supplementary measurement methods. For example, in the phrase 一瓶水 yī píng shuǐ ‘one CLA(bottle) water; a bottle of water’, 瓶 píng is a mass-noun classifier because water itself is uncountable.
Classifiers can also be distinguished into general classifiers and specific classifiers (Erbaugh, 1986, p. 403). 个 gè is a typical general classifier, referring to generic items. For example, in the phrase 一个动物 yī gè dòng wù ‘one CLA animal’, 个 gè classifies and measures an unspecified animal. Differently, specific classifiers are used for specified items and build ‘CLA + noun’ collocations, such as 一匹马 yī pǐ mǎ ‘one CLA horse’, 一张床 yī zhāng chuáng ‘one CLA bed’, 一盏台灯 yī zhǎn tái dēng ‘one CLA lamp’, in which 匹 pǐ, 张 zhāng, and 盏 zhǎn are specific classifiers.
Concerning the acquisition of Chinese classifiers, Erbaugh (1986) found that among L1 speakers, adults use classifiers more frequently with more variants (e.g. a goat can be expressed as 一头羊 yī tóu yáng ‘one CLA goat’ and 一只羊 yī zhī yáng ‘one CLA goat’) than children. She also revealed that both adults and children as L1 speakers are inclined to use general (e.g. 个 gè) rather than specific classifiers, although Chinese requires specific classifiers in a considerable number of situations. This finding has been further confirmed by Fang (1985) and Q. Hu (1993). Among L2 learners of Chinese, Hansen and Chen (2001) found that learners, such as English-speaking learners of Chinese, do not use classifiers initially; however, they gain awareness of the critical grammatical roles of Chinese classifiers, and progressively acquire the syntactic and semantic rules for Chinese classifiers. A study by Polio (1994) pointed out that L2 learners of Chinese usually overuse the general classifier 个 gè; meanwhile, their rare uses of specific classifiers are barely correct.
Up to the present day, most attention has been paid to child L1 learners and adult L2 learners of Chinese (M.X. Chen, 2016; Erbaugh, 1986; N.S.-Y. Liang, 2009; L. Zhang & Jiang, 2016), whereas child L2 learners attract little attention. Thus, this study chooses 8- to-10-year-old children learning Chinese as a heritage language (HL) 4 as the target learner group to fill the gap to a certain degree. Moreover, even though Chinese classifiers are often analyzed from a cognitive perspective (M.X. Chen, 2016; H. Zhang, 2007), it is scarcely addressed how the results of those analyses can be applied to teaching Chinese. Therefore, to build a bridge between cognitive linguistics studies on Chinese classifiers and research on Chinese language teaching, the present study will demonstrate a detailed and practical teaching method with cognitive linguistics (see Section IV.3).
According to Niemeier (2017), teaching grammar with cognitive linguistics concentrates on learners’ co-creation of meanings, in strong contrast to teaching grammar traditionally, which emphasizes that learners must follow syntactic rules (p. 61). Therefore, the cognitive approach to teaching Chinese classifiers in the present study does not ask learners to memorize ‘numeral + CLA + noun’ phrases as collocations. Instead, the cognitive approach aims to help learners acquire and utilize the image-schemata of classifiers in meaningful language use. In other words, the cognitive teaching approach provides learners with situations (e.g. daily conversations) where learners are exchanging information with others and simultaneously learning the target language.
In L. Zhang and Jiang (2016), adult English learners of Chinese (mean age: 27) taught by a cognitive teaching approach displayed the superior ability to extend their instructed knowledge to uninstructed items, which eased their learning of Chinese classifiers. Jacobsen (2018), through a three-week teaching experiment, showed that the combination of a cognitive linguistic approach and a task-based language teaching approach (TBLT) was most beneficial for the learners. Niemeier (2017) supported the effectiveness of the combined teaching approach, based on her teaching in six consecutive semesters. Moreover, most of her students were satisfied with the combined approach, because it made learning grammar more interesting and motivating (p. 69). Even though Jacobsen (2018) and Niemeier (2017) did not address teaching Chinese, their results about the successful combination of cognitive linguistics and TBLT have inspired the present study. Therefore, this study employs the combined approach as the cognitive linguistic method for teaching Chinese classifiers.
In sum, previous studies suggest that the acquisition of Chinese classifiers is challenging and time-consuming, and that cognitive linguistics is beneficial for teaching Chinese classifiers to adult L2 learners. Nevertheless, applying cognitive linguistics to teaching Chinese classifiers is still underexplored. For example, studies on teaching Chinese classifiers to child HL learners are still lacking. Moreover, the cognitive approach to teaching Chinese classifiers is too abstract and difficult for teachers to adopt. Thus, the present study aims to answer the research questions below.
Research question 1: How can teachers practically apply cognitive linguistics to teaching Chinese classifiers in class?
Research question 2: Can the cognitive teaching approach help child HL learners significantly gain knowledge of Chinese classifiers?
Research question 3: Can the cognitive approach be more efficient than the traditional method (i.e. rote memorization), regarding teaching Chinese classifiers?
III Analysis of image-schemata of Chinese classifiers
This section demonstrates how Chinese classifiers are analysed through the lens of cognitive linguistics, paving the way for applying cognitive linguistics to teaching Chinese classifiers. However, in this study, only classifiers used for the teaching experiment will be discussed here. The classifiers below are usually linked to nouns (e.g. utensils, vegetables, fruits, and furniture) with which child learners are familiar – thus the reason why those classifiers have been selected for the teaching experiment.
1 张 zhāng
The classifier 张 zhāng is usually connected with nouns denoting flat, thin, and regularly shaped substances, such as paper, photos, and paintings, seen through the phrases 一张纸 yī zhāng zhǐ ‘one CLA(sheet/piece of) paper’, 一张照片 yī zhāng zhào piàn ‘one CLA photo’, and 一张画 yī zhāng huà ‘one CLA(piece of) painting’. It is noteworthy that 张 zhāng can also be connected with nouns such as 床 chuáng ‘bed’ and 桌子 zhuō zi ‘table’. Even though beds and tables are not thin, they share a property, i.e. having a flat surface. It is the surface that shows the primary function of those items. Therefore, in the collocation of 张 zhāng and 床 chuáng ‘bed’ / 桌子 zhuō zi ‘table’, there is a metonymic mechanism for reference, i.e.
2 条 tiáo and 根 gēn
The classifier 条 tiáo often precedes nouns such as 裤子 kù zi ‘trousers’, 蛇 shé ‘snake’, 项链 xiàng liàn ‘necklace’, 鱼 yú ‘fish’, 路 lù ‘road’, and 河 hé ‘river’ sharing the attributes: long, thin/narrow, bendable (H.-J.H. Chen, 1996, p. 55). H. Zhang (2007) added, 条 tiáo can also be followed by some short items, such as underpants, swimming trunks, briefs, and shorts, embodied in the phrases 一条内裤 yī tiáo nèi kù ‘one CLA(pair of) underpants’, 一条游泳裤 yī tiáo yóu yǒng kù ‘one CLA(pair of) swimming trunks’, 一条三角裤 yī tiáo sān jiǎo kù ‘one CLA(pair of) briefs’. He further explained that these short items are categorial members of 裤 kù ‘trousers/pants’, which is preceded by the classifier 条 tiáo. Therefore, these short items share the classifier 条 tiáo as well. This phenomenon is called lexical taxonomy (p. 47).
In contrast to 条 tiáo, the classifier 根 gēn is often linked to nouns such as 火柴 huǒ chái ‘match, a small wooden stick for fire’, 牙签 yá qiān ‘toothpick’, 试管 shì guǎn ‘tube’, 筷子 kuài zi ‘chopstick’, and 黄瓜 huáng guā ‘cucumber’ sharing the attributes: long, thin/narrow, unbendable (H. Zhang, 2007). Compared to nouns following 条 tiáo, nouns after 根 gēn are relatively shorter, therefore, the formulas of the two classifiers are: 条 tiáo = [+longer, +thin/narrow, +bendable]; 根 gēn = [+long, +thin/narrow, –bendable].
3 颗 kē
In most cases, 颗 kē serves as the classifier of small and round substances, such as a tooth, a grape, and a peanut, demonstrated by examples as follows: 一颗牙齿 yī kē yá chǐ ‘one CLA tooth’, 一颗珍珠 yī kē zhēn zhū ‘one CLA pearl’, 一颗葡萄 yī kē pú tao ‘one CLA grape’, 一颗花生 yī kē huā shēng ‘one CLA peanut’, 一颗子弹 yī kē zǐ dàn ‘one CLA bullet’. 颗 kē is also the classifier for nouns such as 星星 xīng xing ‘star’, 行星 xíng xīng ‘planet’, 彗星 huì xīng ‘comet’, and 流星 liú xīng ‘meteor’ which are huge. Nonetheless, celestial bodies are perceived as small items in the sky by naked eyes. As Chinese classifiers have a more than 3,000-year-long history (N.S.-Y. Liang, 2009, pp. 13–15), it is plausible that ancient Chinese people categorized celestial bodies into small items. Consequently, nouns of celestial bodies share the classifier 颗 kē. Moreover, some Chinese nouns denoting modern weapons, such as 炸弹 zhà dàn ‘bomb’, 原子弹 yuán zǐ dàn ‘atom bomb’, and 导弹 dǎo dàn ‘missile’ also have the classifier 颗 kē. H. Zhang (2007) adopted lexical taxonomy to explain this phenomenon, arguing that these nouns have the morpheme 弹 dàn ‘bullet; bomb’ signifying the superordinate level of the category. As 颗 kē functions as the classifier of the general term 弹/子弹 dàn/zǐ dàn ‘bomb/bullet’, its categorial members will inherit the classifier (pp. 46–47). Hence, the prototypical image-schema of 颗 kē can be expressed as the formula: 颗 kē = [−big, +round/–long].
4 把 bǎ
Unlike the classifiers mentioned above that are related to particular shapes of substances, the items denoted by the nouns after the classifier 把 bǎ have a wide range of shapes. As a result, shapes are not the core of the image-schema of 把 bǎ. Given that the objects following classifier 把 bǎ usually have handles, some linguists (Lin, 2004, p. 20; Shao, 2016, p. 49; Yang, 2006, p. 11) claimed that the handles of those objects are the most prominent element, whereas the objects themselves fade into the background, which gives birth to the collocation of those nouns and the classifier 把 bǎ. However, I disagree with their claims above and argue that the most prominent element is the way people use the objects instead of the handles of the objects.
As is known, Chinese is a tone language in which tones usually distinguish grammatical categories and meanings of words. Take 把 ba as an example, when it functions as a verb, referring to grasping or handling something, its phonetic symbol (or pinyin) is bǎ (the third tone) contrasted with 把 bàr (the fourth tone, usually with rhoticity) as a noun denoting a handle of an object. When 把 ba serves as a classifier, its phonetic symbol is bǎ (the third tone). Therefore, the meaning of 把 bǎ (CLA) should be closer to that of 把 bǎ (verb) than to that of 把 bàr (noun). Therefore, 把 bǎ (CLA) should be one of the extended meanings of 把 bǎ (verb) ‘to handle’, rather than deriving from 把 bàr (noun) ‘a handle’. My argument is supported by a dictionary, Handian, which subsumes 把 bǎ (CLA) under the extended meanings of 把 bǎ (verb).
Thus, there is a metonymic mechanism from 把 bǎ (verb) to 把 bǎ (CLA): Within the domain of 把 ba, there are several participants, such as the action of handling something, i.e. 把 bǎ (verb) ‘to handle’; the agent of the action; the recipient of the action, i.e. 把 bàr (noun) ‘a handle’. In this domain, 把 bǎ (verb), standing for the whole domain, gives rise to 把 bǎ (CLA). Thus, the image-schema of 把 bǎ (CLA) can be illustrated as Figure 1.
Image-schema of classifier 把 bǎ.
IV Methodology
1 Participants
The experiment was conducted in scheduled courses of a non-beneficial school teaching Chinese as HL in a German city, as it was the only accessible alternative for us. The students in the school had a 2-hour session each weekend and no sessions during workdays. Each class had only one teacher. For the experiment, two classes and their teachers were recruited. Both teachers were studying in Master’s programs related to Applied Linguistics at the same university. Their teaching experience and ages were not substantially different. One class and their teacher were tagged as Class A, learning/teaching Chinese classifiers with a traditional approach (i.e. rote memorization). In contrast, the other class and their teacher were labelled as Class B, learning/teaching with a cognitive linguistic approach. The latter teacher and his class were assigned to Class B, because he was more familiar with cognitive linguistics than the other teacher.
It was purposefully avoided to change the original teacher of the classes and create a situation where one teacher could teach two classes with the two different approaches. The reasons are as follows: first, if a teacher and a class were utterly unfamiliar with one another, the teaching effectiveness may decrease, especially in 2-hour teaching experiments. Second, the students in the two classes had plenty of inter-class personal interaction at and out of the school. Thus, if the pre-/post-tests of the experiment could not take place in the two classes simultaneously, the confidentiality of the test tasks could not be ensured, which would lessen the reliability of the data.
Nine students (five boys and four girls) went through the experiment in Class A. Their mean age was 9.44 years (SD = 0.88), ranging from 8 to 10. In Class B, 13 students (eight boys and five girls) completed the experiment. Their mean age was 8.38 years (SD = 0.65), ranging from 8 to 10. Almost all of the students were born in Germany. When they were recruited for the experiment, they all studied at local primary schools, using German as the school language. Almost all of them spoke Chinese and German at home. How often the students spoke Chinese at home was unmeasurable. Almost all students exhibited frequent code-switching between Chinese and German while talking with their Chinese parents and Sino-German friends.
The students’ Chinese proficiency was not evaluated by standard tests because they had attended neither Hanyu Shuiping Kaoshi (HSK) nor Youth Chinese Test (YCT). 5 When recruited for the experiment, the students had been learning Chinese in class for around 1.5 years. According to the Chinese school’s evaluation, the Chinese proficiency of the two recruited classes was probably between a beginner and intermediate level. Based on the school’s long-term observation of the two classes and the situation that they used the same Chinese textbooks, 6 their Chinese proficiency should be comparable. Through our 3-month-long observation of the two classes before the experiment, we noticed that all of the students overused the general classifier 个 gè; moreover, their infrequent use of specific classifiers was usually inappropriate, echoing the finding of Polio (1994). It is worthwhile to mention that the students had not learned Chinese classifiers in class. However, due to the language immersion, i.e. their Chinese parents kept speaking Chinese with them, the recruited children had some implicit knowledge of Chinese classifiers.
2 Materials
The students were taught 5 Chinese classifiers, i.e. 张 zhāng, 条 tiáo, 根 gēn, 颗 kē, and 把 bǎ in the experiment. There were no standard sequences for teaching Chinese classifiers. Nevertheless, for the experiment, I chose classifiers that usually co-occur with nouns with which the students should be familiar, such as utensils, furniture, vegetable, fruit, and animals. The teachers used PowerPoints that clarified each teaching step to facilitate the teaching. The PowerPoints were sent to the teachers in advance, allowing them to be well prepared.
3 Procedure
a Teaching Class A with a traditional approach
Session one
Teaching Step 1 – Pre-Test: Students finished a test (see Appendix 1) within 10 minutes. The test consisted of 10 multiple-choice tasks, as this task type was feasible for young students. The teacher could read the tasks to students when necessary. 7
Teaching Step 2 – Introduction: The teacher showed pictures (involving nouns related to the target classifiers) one by one on slides and asked students what they saw in the pictures. For example, students may answer, ‘我看见了一张桌子。’ wǒ kàn jiàn le yī zhāng zhuō zi. ‘I have seen a CLA table.’ This step aimed to invite students to use ‘numeral + CLA + noun’ structured phrases in Chinese.
Teaching Step 3 – Say & Write: (a) The teacher pointed at the pictures randomly and asked students to produce sentences in the structure ‘这/那是一 zhè/nà shì yī + CLA + noun。’ ‘This/That is a + CLA + noun.’ (b) Students practiced writing the target classifiers. (c) Students reviewed the target classifiers by saying, ‘那有一 nà yǒu yī + CLA + noun。’ ‘There is a + CLA + noun.’ with the pictures on slides again. Step 3 was intended to help students become familiar with the target classifiers.
Teaching Step 4 – Activity I: Remember & Say: Students memorized ‘numeral + CLA + noun’ structured phrases on each slide; afterwards, they said the phrases aloud after the phrases disappeared from the slides. In the beginning, there was merely one phrase on each slide. Later, the number of phrases on each slide increased, making the activity progressively more challenging. The teacher used competition mechanisms to motivate students.
Teaching Step 5 – Activity II: Finding a Friend: (a) The teacher separated the class into three groups. (b) Every group received many cards from the teacher. One side of the cards was blank; on the other side, there were Chinese characters: some were nouns, whereas some were numerals plus target classifiers. (c) Students put all cards on a desk, laying the blank side upwards. Later on, students picked up one card; it may show a noun, for instance. Afterwards, students should find a ‘numeral + CLA’ on another card for the card they had picked up, to build a lexico-grammatically correct ‘numeral + CLA + noun’ phrase. When the second card the students chose did not match the first one, they would put the second card back, with its blank side upwards, and choose another card to match the first one. (d) If students correctly built a phrase with two cards, the two cards would be removed. The activity continued until all cards were removed. During the process, the teacher supervised all groups to make sure students matched cards appropriately.
Teaching Step 6 – Activity III: A Growing Snake: Student A said a sentence in the structure ‘我有一 wǒ yǒu yī + CLA + noun。’ ‘I have a + CLA + noun.’ After that, student B repeated A’s sentence and added one sentence with the same structure. Student C repeated A’s and B’s sentences and then added one sentence with the same structure. One round of the activity continued until the teacher asked to stop.
E.g. Student A: 我有一条蛇。 wǒ yǒu yī tiáo shé. ‘I have a CLA snake.’ Student B: 我有一条蛇。 wǒ yǒu yī tiáo shé. ‘I have a CLA snake.’ 我有一根香蕉。 wǒ yǒu yī gēn xiāng jiāo ‘I have a CLA banana.’ Student C: 我有一条蛇。 wǒ yǒu yī tiáo shé. ‘I have a CLA snake.’ 我有一根香蕉。 wǒ yǒu yī gēn xiāng jiāo ‘I have a CLA banana.’ 我有 wǒ yǒu . . . ‘I have . . .’
The classifiers that students used came from the target classifiers. This activity challenged students’ memory but helped them review and entrench collocations of classifiers and nouns. When the teacher noticed students’ serious difficulty in repeating sentences due to the accumulated length, she could stop the round and start a new one.
Teaching Step 7 – Exercises: Fill in blanks: (a) Students filled blanks with the target classifiers in spoken and written forms, e.g. 一 yī ___桌子 zhuō zi ‘one ___(CLA) table’, during which the teacher guided students to check and correct their answers.
Teaching Step 8 – Summary: (a) The teacher guided students to give a brief oral summary of this session before the teacher answered questions from students. (b) The teacher assigned homework that was comprised of several question types and focused on the target classifiers.
Interval: One week
Session two
Teaching Step 9 – Review: (a) The teacher checked students’ homework and helped them to correct their answers. (b) The teacher led students to review the core of Session One. Step 9 lasted about 25 minutes in total.
Teaching Step 10 – Post-Test: Students finished a test (see Appendix 2) composed of 10 multiple-choice tasks within 10 minutes. The teacher could read the tasks to students when necessary.
b Teaching Class B with a cognitive approach plus TBLT
Session one
Teaching Step 1 – Pre-Test: Students finished a test identical to Class A’s pre-test within 10 minutes. The teacher could read out the tasks when necessary.
Teaching Step 2 – Warming Up: (a) The teacher told students a short story where the target classifiers occurred in the ‘numeral + CLA + noun’ structure. The teacher emphasized the target classifiers while telling the story. (b) The teacher helped students retell the story and corrected them when they misused classifiers.
Teaching Step 3 – Task Cycle: Before the start of each task, the teacher helped students to understand the task requirements.
Task one: The teacher gave each student some ‘money’ (props) and asked them to buy as many goods as possible. At the same time, the teacher showed on slides many pictures of goods with corresponding prices. During the task, the situation in a shop was imitated. Students may use the structure ‘你好，我想要 nǐ hǎo, wǒ xiǎng yào numeral + CLA + noun。’ ‘Hello, I’d like numeral + CLA + noun.’
Task two: Students were divided into four groups in which they told their group-mates what they had bought in Task One and why they bought those items. They may use the structure ‘我刚才买了 wǒ gāng cái mǎi le numeral + CLA + noun, 因为 yīn wèi . . .’ ‘Just now, I’ve bought numeral + CLA + noun, because . . .’ Students had opportunities to conduct daily conversations with their peers during this task, where the target classifiers were employed to construct natural and meaningful conversations.
Task three: Student A told classmates what she/he had bought, followed by Student B telling them what A had bought and what she/he had bought for herself/himself. Student C repeated what A and B had bought and added what she/he had bought. One round continued until the teacher asked to stop. This task was the same as Class A’s activity, ‘A Growing Snake’.
Teaching Step 4 – Language Focus: (a) Categorization: The teacher showed lots of objects on slides. After that, students separated them into different groups, mainly according to their shapes, and said their classifiers aloud. (b) The teacher helped students to figure out the image-schemata of the classifiers, based on the categorization in the previous step. Classifier 把 bǎ was unique among the target classifiers, as explained in Section III. Therefore, the teacher spent some time distinguishing 把 bǎ from other target classifiers. (c) Students practiced writing the target classifiers. (d) Students did some exercises about the target classifiers, e.g. filling in blanks (in verbal and written forms).
Teaching Step 5 – Summary: (a) The teacher led students to make a brief oral summary of this session. Afterwards, the teacher answered questions from students. (b) The teacher assigned homework. Class A and B had the same homework.
Interval: One week
Session two
Teaching Step 6 – Review: (a) The teacher checked students’ homework and helped them to correct their answers. (b) The teacher guided students to review the critical content of Session One. Step 6 lasted about 25 minutes.
Teaching Step 7 – Post-Test: Students finished a test identical to Class A’s post-test within 10 minutes. The teacher could read the tasks to students when necessary.
V Results
Tables 1 and 2 present students’ total scores in the pre-and post-tests. In the two tables, ‘increase’ equals the score of the post-test minus that of the pre-test. To protect the students’ privacy, I have replaced their names with IDs. Each ID starts with an A or B, indicating the student’s class, i.e. A refers to Class A, taught by the rote memorization approach; B means Class B, taught by the cognitive approach. The total score of the pre-/post-test was 100 points, consisting of 10 points for each task and ten tasks in total. The pre-/post-test covered five target classifiers in a balanced way, i.e. two scattered tasks tested a target classifier. Each classifier could offer 20 points at most. Tables 3 and 4 present students’ points earned from each target classifier in the pre-and post-tests.
Total scores of Class A.
Total scores of Class B.
Points from each classifier in the pre-and post-tests of Class A.
Points from each classifier in the pre-and post-tests of Class B.
For the judgement about which type of statistic tests (i.e. parametric or non-parametric) should be used to analyse the data above, the students’ total scores in the pre-test, the post-test, and the increases have gone through tests of normality, namely, Kolmogorov–Smirnov and Shapiro–Wilk tests. The results are presented in Table 5. As shown in that table, Class A’s total scores in the pre-test, the post-test, and the increases significantly deviate from normal distributions. 8 In contrast, Class B’s total scores demonstrate normal distributions in the pre-test, the post-test, and the increase. Figure 2 presents Class A’s and Class B’s increases in total scores, in which there are no significant outliers.
Tests of normality: Total scores.
Notes. a Lilliefors Significance Correction. b A lower bound of the true significance. * p < .05. ** p < .01.
Increases in total scores.
Hence, Mann–Whitney U tests will be conducted when the students’ total scores are compared between groups (i.e. Class A vs. Class B). In terms of the within-group comparison (i.e. pre-test vs. post-test) of total scores, Wilcoxon Signed-Rank tests will be implemented for Class A, paired-samples t-tests for Class B.
Regarding each classifier, Shapiro–Wilk tests show that none of the variables (e.g. 根 gēn in the pre-test of Class A, 根 gēn in the pre-test of Class B) is normally distributed, all p < .05. Thus, non-parametric statistic tests will be adopted for both Class A and Class B to compare individual classifiers.
1 Within-group comparisons
Wilcoxon signed-rank tests are implemented for Class A. The results are shown in Table 6. In this table, ‘increase’ equals the score in the post-test minus that in the pre-test. The values in square brackets are 95% confidence intervals (95% CI). According to Table 6, no significant increase occurs in the total score or particular target classifiers. Some medium and large effect sizes are paired with p > .05 and broad 95% CIs covering 0; thus, we cannot reject the null hypothesis that the pre-test median of Class A does not significantly differ from its post-test median.
Wilcoxon signed-rank tests for the pre-/post-tests of Class A.
A paired-samples t-test has been performed for Class B’s total scores. The result suggests that, in general, Class B performed much better in the post-test (M = 83.85, SD = 15.02) than in the pre-test (M = 64.62, SD = 19.84), t(12) = 4.63, p < .001, d = 1.28. Since Cohen’s d = 1.20 is the threshold of very large effect sizes (Sawilowsky, 2009), the value of 1.28 suggests a tremendous increase from the pre-test to the post-test in Class B, which is mainly owed to the effectiveness of the cognitive teaching approach.
Wilcoxon singed-rank tests have been conducted for Class B’s increases in particular target classifiers, of which the results are presented in Table 7. As shown in this table, students in Class B significantly gained knowledge in 条 tiáo and 颗 kē with large effect sizes. It can be further inferred that the cognitive approach strongly facilitated the learning of the classifiers 条 tiáo and 颗 kē.
Wilcoxon signed-rank tests for the pre-/post-tests of Class B.
Note. * p < .05.
2 Between-group comparisons
Mann–Whitney U tests are conducted to compare the total scores and scores from particular classifiers of Classes A and B, of which the results are shown in Table 8. As suggested by that table, only in the knowledge of 条 tiáo in the pre-test did the two classes manifest a significant difference with a large effect size. In other words, students in Class A had a much better knowledge of 条 tiáo than students in Class B before the teaching. The two classes did not outperform each other as to total scores or particular classifiers in the post-test, which does not support the cognitive approach being more effective than the traditional approach. However, the non-significant difference between Classes A and B indicates that the cognitive approach is not worse than the traditional approach.
Mann–Whitney U tests for the comparison between Classes A and B.
Notes. Grouping Variable: Class. All p-values are those of exact significance. ** p < .01.
VI Discussion
Table 4 shows that Class A did not significantly benefit from the traditional teaching approach in general, and in specific classifiers. However, the statistical result cannot deny the efficacy of the traditional approach for two reasons. First, Class A’s total score median in the pre-test was already high, i.e. 90 out of 100. Therefore, it is not surprising that Class A could not engender a statistically significant improvement in the post-test. Second, the standard deviations of the sample means were high, reducing the statistical power. Interestingly, the traditional teaching was quite beneficial to three students in Class A, i.e. A01, A02, and A03. This phenomenon indicates that the traditional teaching approach (i.e. rote memorization) can be helpful to some individuals.
As shown in Table 8, there was no significant difference between Class A and Class B in the pre-test. The non-significant difference remained in the post-test and increases, which contradicts the findings in Section V.1 – Class A did not improve significantly in the post-test, whereas Class B did. Moreover, in the pre-and post-tests, Classes A and B did not outperform one another in total scores and specific target classifiers (except 条 tiáo in the pre-test).
Regarding the non-significant differences between Class A and Class B in the pre-and post-tests, there are several potential reasons why Class B could not outperform Class A in the post-test, seen below.
First, there were a small number of tasks in the pre-and post-tests, which could not comprehensively reflect students’ knowledge of the tested classifiers. Second, the experiment lasted for a short time, i.e. approximately 2.5 hours in total. In such a short time, children could hardly master the usage of image-schemata. Using image-schemata to categorize Chinese classifiers requires several steps, such as extracting the image-schemata from instructed linguistic examples, storing the image-schemata in one’s mind, recognizing the image-schemata of the tested items, matching the image-schemata of the tested items with the stored image-schemata, and deciding the proper classifier for a particular tested item. As S. Jiang (2017) asserted, developing conceptual categorization is a long-term cognitive process; thus, we should not expect learners to internalize the categorization of Chinese classifiers from a cognitive perspective immediately after one or two teaching sessions (p. 187). If learners were well trained and experienced in using image-schemata, they may perform much better.
Third, the experiment prevented the cognitive approach from bringing its potential into full play. To illustrate, a great advantage of a cognitive approach over a traditional approach (i.e. rote memorization) is that learners taught by the former can transfer their instructed knowledge to uninstructed items. In contrast, learners taught by a traditional approach have difficulties linking their instructed knowledge to items they have not encountered (L. Zhang & Jiang, 2016). Thus, learning with the cognitive approach can be less time-consuming and more efficient over a long period. However, rote memorization seems quicker to handle within a short period (e.g. a short experiment). Unfortunately, the experiment in the present study was restricted by some background conditions (e.g. time limits in the Chinese school), which compelled the teaching and tests to be short. Therefore, there was not enough space in the tests to reflect the extent to which learners could transfer their instructed knowledge to uninstructed items.
Fourth, although the mean scores of the two classes were similar in the two tests, Class A had much larger standard deviations than Class B. It means that the two classes’ pre-existing knowledge of the target classifiers differed a lot and thus lacked a good foundation for between-group comparisons. Even so, the large effect sizes in Class B demonstrate the magnitude of the treatment effect on that class.
Fifth, all statistic tests involving Class A were non-parametric, since data from this class were not normally distributed, which leads to less statistical power – in other words, being less able to reject the null hypothesis.
Additionally, the vast majority of the tested classifiers were shape-related classifiers, merely a corner of the panorama of Chinese classifiers. Thus, how a cognitive approach functions on other types of classifiers, such as verbal classifiers, remains unknown. This study also lacks sufficient evidence to explain why Class B significantly gained knowledge of 条 tiáo and 颗 kē, but not of 张 zhāng and 根 gēn, even though those classifiers were shape-related.
Meanwhile, it must be admitted that a cognitive approach to teaching Chinese classifiers has some potential drawbacks. For instance, the teaching approach is based on the existing cognitive linguistic studies on Chinese classifiers. Moreover, Chinese has hundreds of classifiers, yet cognitive linguistics studies have not touched on many of them. Therefore, the underexplored classifiers cannot be easily taught with a cognitive approach at present. Furthermore, Applied cognitive linguistics has not been the mainstream teaching approach, which suggests that it is time-consuming and painstaking to train many teachers with this approach. In addition, language is not only a matter of cognition but also involves other aspects such as stylistics and social conventions, which gives rise to variants of classifiers (H. Zhang, 2007) and might challenge the cognitive linguistic teaching approach.
It is worth mentioning that learners’ characteristics also influence the treatment effect on individual learners. According to He (2006), learners’ identity, motivation, and attitude are significant for learning Chinese as a heritage language. For example, when children are forced by their parents to learn Chinese as a heritage language, they lack the motivation to preserve the linguistic inheritance (F. Liang & Shin, 2021; Mu & Dooley, 2015; D. Zhang & Slaughter-Defoe, 2009). In the present study, learners B03 and B05 were taught by the cognitive approach but showed entirely different treatment effects. In the post-test, B03 had the greatest increase, whereas B05 had the least in Class B. Such a difference is probably related to their linguistic backgrounds and attitudes toward learning Chinese. We learned from informal interviews that B03 was from an immigrant family where both parents immigrated from China to Germany. She only spoke Chinese with the family and diligently learned the family’s language. Differently, B05 came from a German-Chinese family where the father is German while the mother is Chinese. This learner was immersed in a German-speaking environment, even at home. Her parents spoke German with one another; she sometimes spoke Chinese to her mother, even though her mother insisted on speaking Chinese to her. She was sent by her mother to learn Chinese but was reluctant. Those two learners’ willingness/resistance regarding learning Chinese as a heritage language is highly likely to facilitate/hinder the treatment effects. However, more studies are needed to answer how far learners’ attitudes and family languages influence treatment effects.
Another significant issue concerns the selection of target classifiers for teaching and experiments. The pre-tests showed that those students had some basic knowledge of those target classifiers, creating the ceiling effect on knowledge improvement after treatments. However, it does not mean that those classifiers should not appear in the teaching activities and experiments, because teachers should consider learners’ linguistic backgrounds and should not merely teach classifiers that students have never encountered before. Child HL learners have implicit knowledge of classifiers, and instruction can enhance learners’ metalinguistic awareness.
To be specific, Chinese classifiers always occur with nouns, and child HL learners are exposed to native Chinese speakers (i.e. their parents) every day. Thus, those learners are exposed to commonly used classifiers in daily life. Consequently, they have some basic knowledge of those frequently used classifiers, even though they may not master them. Therefore, it might be unrealistic to believe that only classifiers that learners do not know at all are appropriate for teaching and research. However, learners’ pre-existing knowledge of their heritage language can be pretty heterogeneous (e.g. Class A of this study), which should be considered for experiment designs in further research.
On the other hand, given that students often misuse classifiers, there is still a need to include somewhat the conventional usage of classifiers that learners may know. Otherwise, the fossilization of misusing classifiers by learners is likely to occur. Furthermore, teaching language is not solely about passing on thoroughly new linguistic knowledge to students. The development of linguistic competence is assumed to be progressive. Hence, the entrenchment of linguistic norms and enhancement of metalinguistic awareness are crucial for teaching and should not be excluded from language teaching and research.
VII Conclusions
At the onset of the study, three research questions were raised. So far, the first two questions have been answered, leaving the third open for follow-up studies. To be clear, the first question was about how to apply cognitive linguistics to teaching Chinese classifiers practically. Section IV.3 demonstrated the concrete teaching steps and answered the first question, filling the gap that previous studies did not present a replicable and practical approach to teaching Chinese classifiers with cognitive linguistics. Research question 2 concerned whether the cognitive approach could significantly improve learners’ knowledge of Chinese classifiers. The experiment in this study showed that the class taught by the cognitive approach had significant increases in the post-test with enormous effect sizes. Therefore, the answer to research question 2 was quite positive. As to the third question, whether the cognitive approach could outperform the traditional approach, the answer was more or less vague, because non-significant differences were observed between the two tested learner groups. The reasons causing those non-significant differences and the limitations of the experiment have been discussed in detail in Section VI.
However, it should be emphasized that the present study indicates that the cognitive approach was quite beneficial for Chinese classifiers teaching and not worse than the traditional approach. Furthermore, previous studies have suggested the advantage of the cognitive approach over the traditional approach for teaching Chinese classifiers to adult L2 learners (L. Zhang and Jiang, 2016) and for teaching grammar of other languages (Colasacco, 2019; Jacobsen, 2018; Niemeier, 2017). The cognitive approach introduces to learners how native speakers conceptualize the world, offering learners a chance to understand the target language better. Hence, the cognitive approach is highly worth employing to teach HL and non-HL learners Chinese classifiers.
The following are my suggestions for follow-up studies. First, researchers should advance pre-tests to a separate session earlier than the teaching session. The present study shows that mere observation of target students cannot rigorously predict their pre-existing knowledge of classifiers. Moreover, the absence of earlier pre-tests will result in difficulty choosing target classifiers that students need to learn. Second, researchers are encouraged to add an immediate post-test to the session where the teaching is assigned. Because the interval between the teaching and the delayed post-test may cause more uncontrollable factors to the experiment. For instance, some students may attend the teaching session but be absent in the delayed post-test, resulting in data losses. Besides, during the interval, some students may make lots of efforts to deepen their knowledge obtained from the teaching session, while others may thoroughly put learning Chinese aside, which may weaken the comparability of the delayed post-tests. Moreover, the test items in this study were decontextualized, only showing ‘numeral + classifier + noun’ collocations. Since context is crucial for speakers’ language processing (Knoeferle, 2019; Steen-Baker et al., 2017), further research should contextualize test items in pre-and post-tests.
For a wider-scale study with a larger data pool, researchers can conduct a corpus analysis of daily Chinese use to select nouns frequently co-occurring with target classifiers. Researchers should also notice that child HL learners’ ability to write/recognize words in the target language is often inferior to their ability to speak it. It is probably due to their literacy, general competence in the target language, or both. However, the phenomenon may lead to students’ answers on test papers not reflecting their general knowledge of the target language, especially when they cannot recognize the words on the test papers and the teacher does not help. Response time or eye-tracking experiments (Grüter, Lau, & Ling, 2020; Srinivasan, 2010; Was, Sansosti, & Morris, 2017) may be helpful for tests as well, as they can present how students process classifiers before and after the teaching. Therefore, response time or eye-tracking experiments may facilitate discussing which teaching approach helps students access the correct answers more swiftly and accurately.
If Chinese is taught as a heritage language, teachers should pay more attention to child learners who are forced by parents and thus resistant to learning Chinese. Teachers can make teaching sessions more attractive (TBLT can be a good choice), which may help motivate learners. It will be valuable to explore whether the cognitive approach to teaching Chinese classifiers has distinguishable treatment effects on child HL learners coming from different family backgrounds, such as both Chinese parents versus a Chinese parent plus a parent from the country where the family lives. It will also be worthwhile to investigate whether this teaching approach has different treatment effects on HL and non-HL learner groups.
Footnotes
Appendix 1
Appendix 2
Acknowledgements
I give many thanks to the anonymous reviewers for their very insightful and constructive comments on the earlier version of this article. I also thank colleagues in the Chinese language school, without whom the teaching experiment could not be conducted.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
