Pedagogical issues related to the adverb dou [都] in Mandarin Chinese

3.1 Formal analyses of dou

As presented in Sections 1 and 2, many previous studies have focused on different aspects of theoretical analyses of dou. Although the present study focuses on the pedagogical part, a brief review of these studies of the adverb would give us a better understanding of the grammar. However, due to the large number of previous studies on this topic, the review will be selective rather than exhaustive.

Cheng (1995) claimed that dou has a dual function. First, as a distributor, dou quantifies over plural entities. It can also function as a binder, providing quantificational force for wh-polarity items that do not have inherent quantificational force. Cheng also argued that there exist some locality restrictions, and these restrictions associated with dou vary depending on the element with which it is associated.

Further, Lin (1998) argued that dou is a generalized distributivity marker distributing over the members of a plurality cover. Lin also claimed that dou is in the head of a functional phrase, the distribute phrase (DistP), which is projected above VP in the syntax. Observe the two examples in (9) below.

As (9a) shows, dou is obligatory when the sentence contains a universal quantifier like mei ‘every’; in contrast, the adverb is optional when the noun phrase it quantifies over is a simple definite noun, as (9b) illustrates. As dou is present in (9a), the DistP is projected in the syntax given that dou is the overt realization of the head of this functional phrase. Lin claimed that noun phrases containing a universal quantifier such as mei ‘every’ have a strong quantificational feature that must be checked against dou locally before being spelled-out. As a result, the noun phrase must overtly move to the specifier of DistP, which explains the word order of [every + NP + dou]. To further account for the syntax and semantics of (9b), Lin argued that a definite noun phrase na xie xuesheng ‘those students’ only optionally bears a quantificational feature. Therefore, when dou is present, the DistP is projected, and the noun phrase with the plural form moves to the specifier of DistP before it moves to check the EPP (extended projection principle) feature. However, when dou is not present, the DistP is not projected, and the plural noun phrase moves directly out of its merged place to check the EPP feature.

The syntactic difference proposed by Lin (1998) nicely explains the semantic differences for the plural noun phrases in (9b): when dou is present, the sentence has a distributive reading due to the projection of the DisP; in contrast, when dou is not present, the functional phrase is not projected, and there is no distributive reading detected in the sentence. Nevertheless, the study did not include a discussion of some other usages of dou, such as those in (8), where it seems to have a meaning equivalent to already.

Xiang (2008) provided a unified account of dou in auguring that dou is simply a maximality operator that operates on a plural set and outputs a maximal plural individual; furthermore, Xiang showed that the seemingly different uses of dou are because of the context sensitivity of this plural set. One advantage of Xiang’s analyses is that dou in a sentence such as (8a) and (8b), which has been argued to have an equivalent meaning to already in English, also contributes to a scalar meaning. According to Xiang, dou in this type of sentence also adds a tone of even sense to the interpretation, like those examples in (6). The examples in (10) are from Xiang (2008: 238).

As illustrated, dou seems to have the meaning of already in both examples. However, Xiang indicated that the adverb yijing ‘already’ can be explicitly added to both sentences. For example, the adverb can appear immediately after dou in (10a). Xiang also argued that the first utterance of (10a) is more than a question, and asking the question is not the speaker’s main point. In contrast, the speaker simply tries to convey that it is quite late at the time of speech, and it is unexpected to see the listener not in bed. Similarly, by adding dou in (10b), the speaker expresses surprise about how rapidly the time has passed, and that the children have all grown up. As a result, Xiang (2008) argued that dou in the examples in (10) adds a strong sense of maximal surprise or unexpectedness from the speaker’s point of view. The meaning of ‘already’ in this type of sentence is simply a byproduct rather than the meaning of dou.

Recall that dou can also be used in a sentence with a de-resultative construction, as (7a) and (7b) illustrate. This type of sentence has been argued by previous studies, such as Sybesma (1999), to convey some sort of degree. Therefore, a sentence such as Lisi qi-de dou ku le can also be interpreted as ‘Lisi got so mad that he started crying.’ According to Xiang (2008), there is a scale defined by the gradable predicate in this sentence type. In (7a), for example, there is a scale of angriness, and the degree of the predicate is open to many alternative possibilities.

Moreover, recall that in (7a) and (7b), dou was optional. Xiang claimed that if dou is added to a sentence, it emphasizes the upper-bound of the maximal degree. Compare the examples in (11).

According to Xiang, the presence of dou in the de-resultative construction creates some subtle differences in meaning. A sentence like (11a), without dou, simply describes the degree of the predicate: being sad to the degree that one cries is not uncommon under regular circumstances. However, in a sentence such as (11b), which contains dou, it emphasizes the upper-bound of the maximal degree. There exists an implicature in the sentence that crying in this context is actually a sign of deep grief, and a more accurate translation for this sentence is to add the adverb even. Thus, Xiang’s (2008) study provides a unified analysis of the different usages of dou, and it offers the present study valuable insights.

3.2 Pedagogical issues

In addition to the formal linguistic analyses of dou, many previous studies have discussed CSL learners’ acquisition and classroom instruction of the adverb (Li 2012, 2013; Wang 1999; Wu and Tao 2016; Zhou and Wang 2007). Zhou and Wang (2007), for example, listed several linguistic environments in which dou can generally be used, and stated that in these environments, dou is sometimes obligatory and sometimes optional. This inconsistency in the morphosyntax of dou has caused difficulties for CSL learners. To address these issues, the authors provided an analysis of the levels of difficulty of these environments and argued that it is relatively easier for CSL learners to acquire a sentence in which dou is obligatory. According to them, the environments where dou appears can be further divided into three groups based on their difficulties for acquisition. A detailed description of each category and examples are provided in (12).

As (12a) illustrates, when the appearance of dou in a sentence is obligatory, the difficulty level is the lowest. For example, the noun phrase that dou quantifies over is a wh-word. In this sentence, the appearance of dou is obligatory.^[6] The authors further indicated that the level of difficulty of a sentence in which dou is optional is in the middle. For example, dou in (12b) is optional, although with the appearance of dou, a distributive meaning is salient. Finally (12c) represents the type of dou-sentence that has the highest level of difficulty for CSL learners. The noun phrase that dou quantifies over (i.e., na ge pingguo ‘that apple’) does not have a plural form; rather, dou is used to emphasize the entireness of the noun.

Zhou and Wang (2007) provided possible explanations for why there is such a hierarchy. Although the explanations are sound, their study seems descriptive, and there is no empirical data supporting their arguments. In addition, their study did not address other patterns of dou, such as a sentence containing the phrase yidianr ‘a bit’ and a negation word, such as in (5). Nevertheless, several other studies on dou have provided empirical support for the different levels of difficulty of dou-sentences for CSL learners. For example, Li (2012) conducted a grammaticality judgment study, and the results showed that the subjects have difficulty accepting sentences in which dou quantifies the objects (e.g., the examples provided in [1]), and those in which dou is obligatory (e.g., the examples provided in [4]). This result was not predicted by Zhou and Wang’s (2007) study, as they stated that it is relatively easier for language learners to acquire dou when it is obligatory. In a follow-up study, Li (2013) examined the dou production data of CSL learners and similar results were found; even advanced CSL learners still have some difficulties in producing syntactically obligatory dou in a native-like manner. The author then suggested that more emphasis in the classroom should be placed on clarifying the syntactic constraints of dou.

Among the studies discussing CSL learners’ acquisition of dou, Wu and Tao’s (2016) study is one of the few that has focused on the ways that dou is used in conversational discourses. Based on their findings, the authors provided several pedagogical suggestions regarding material design, classroom activities, and assessment methods. The data collected in this study came from a corpus of natural conversations as well as TV shows with a total number of just over a half million words.

Using the 20,395 instances of dou identified in the data, the authors randomly selected parts of the data for further coding. They found that 50% of the instances containing dou were the pattern that contained a plural noun phrase before dou, as illustrated by the examples in (1). However, the authors suggested that the main function of dou is beyond the purely quantification function and concluded that dou is used as a pragmatic device. Furthermore, there is a gap between the usages of dou in daily life and some CSL textbooks. One of the significant contributions of Wu and Tao’s (2016) study is that it demonstrated that natural discourse-inspired pedagogy is possible in many aspects of the language teaching and learning process. For example, to narrow the gap between L2 pedagogy and natural discourse, it was proposed that authentic conversational discourse be included in instructional materials.

Wu and Tao’s (2016) much more functional approach to dou provides the present study with valuable insight. First, by analyzing the use of dou in conversational discourses, we attain a clearer picture of the ways in which dou is used in native speakers’ daily lives. This meets one of the goals of many foreign language classes, which is to prepare students for target language use with more authentic materials. Second, their study provides several pedagogical suggestions, including but not limited to classroom activities, material design, and homework assignments that could advance learners’ proficiency levels.

3.3 Corpus linguistics

A corpus is commonly used to refer to a relatively large collection of representative samples of natural texts. According to Sinclair (1991), empirical data is needed to prove the abstract rules in grammar. As a result, corpora are used to provide objective, quantitative, and empirical evidence, which has started to serve as an alternative tool for data analysis in formal linguistics. Similarly, Barlow (1996: 32) stated that ‘the results of a corpus-based investigation can serve as a firm basis for both linguistic description and, on the applied side, as input for language learning’. This indicates that corpora can greatly affect teaching materials and syllabus design (Hymes 1992). There are several advantages to using corpus data in a foreign language classroom. For example, the items that frequently occur in corpus data are the most likely to be encountered in real-life communicative situations. Furthermore, corpus data provide holistic information about language structures. Due to the statistical software used in corpus linguistics, statistical information about different structures can therefore be obtained. For example, if we want to study the usages of dou, corpus data can supply frequencies of use and the environments in which the adverb is most likely to occur. Based on the results, we can then deduce some rules about the adverb.

There are different applications of corpora in L2 pedagogy. According to Römer (2011), for instance, applications of corpora in L2 teaching include both the use of the actual text collections and software packages for corpus access and corpus methods (i.e., the analytic techniques that are used when we work with corpus data). The first type of application, also called direct application, is an approach that mainly affects how something is taught and learned. It involves the teacher and learner in the process of working with corpora. For example, students are instructors themselves and directly work with corpora and concordance tools to find out about language patterns and the behavior of words and phrases. In other words, they learn how to use different types of tools to search for the grammatical patterns of words or phrases that can be found in the corpus data. On the other hand, the second approach, also called indirect application, can help language instructors make decisions about what to teach and when to teach the material. Thus, it has an effect on the teaching syllabus and the design of class materials. In addition, Römer (2011) also pointed out that previous studies have discovered that there are many mismatches between naturally occurring target languages, such as German and English, and the type of target language that is used as a model in the teaching materials for these languages. Given that corpus data generally reflect real-life communicative situations, foreign language learners are able to encounter and experience these situations when the class materials are designed accordingly. The present study will place its focus on the indirect applications discussed by Römer (2011). In particular, it analyzes some specialized corpus data from the Academia Sinica Corpus of Balanced Chinese, which captures a particular genre that can influence syllabus design for CSL courses.

3.4 Summary

I have given a review of several previous analyses focusing on dou’s semantics and syntax. As illustrated, the usages of dou can be quite inconsistent, which may cause CSL’s learners’ difficulty in acquiring the adverb. In particular, while some examples show that the noun phrases that dou quantifies over must be morpho-syntactically plural, others must be semantically plural.

I have also reviewed some studies that discussed CSL learners’ acquisition of the adverb and studies about applications of corpora in L2 pedagogy. Given that corpus data generally reflect real-life communicative situations, foreign language learners can encounter and experience these situations when the class materials are designed accordingly. However, as Römer (2011) suggested, there are substantial mismatches between naturally occurring target languages and the type of target languages put forward as a model in the examined teaching materials. Therefore, the main goal of the present study is to see whether there is a gap between the instruction of dou in a CSL textbook and some corpus data. A comparison of these two types of discourses will help us better prepare CSL students for target language usage.

4.1 Textbook analysis

4.2 Results of the corpus data

The corpora data used in the present study is from Academia Sinica’s Balanced Corpus of Modern Chinese (中研院現代漢語平衡語料庫). The corpus includes six different genres: literature; life; science; arts; philosophy; and society. The genre selected in this study is life, which includes a total of 2,253,102 word tokens and 3,547,925 character tokens. A total of 770 tokens of dou were found in this genre. One of the reasons I selected this genre is because it also includes some conversational discourses. According to Nunan (1991: 39), ‘to most people, mastering the art of speaking is the single most important aspect of learning a second or foreign language, and success is measured in terms of the ability to carry out a conversation in the language’. Also, as suggested by Tao (2005, 2011, it is important that we apply conversation-based findings to second language teaching. Therefore, this genre was selected.

Recall that Li (2012, 2013 found that CSL learners had some difficulty acquiring sentences in which dou is obligatory. To see how these constructions are used in daily lives, I focused my analysis on the following constructions in which dou is obligatory: (a) universal quantifiers; (b) wh-words without an interrogative meaning; (c) the lian … dou … construction;^[7] and (d) the yi ‘one’ … dou … negation construction. I will also include a discussion of some interesting findings about the ‘semantic plurality’ of those noun phrases that are quantified by dou. As we will see, these noun phrases do not have any morpho-syntactical clues that signal their plurality; nevertheless, they can be quantified by dou.

Table 2 below gives a summary of the frequencies and percentages of different types of dou-constructions found in the corpus.

The category of ‘others’ in Table 2 represents the most frequent usage of dou. In fact, 80% of the usages of dou belong to this category. In these cases, dou quantifies over a noun phrase appearing before it. They can be further divided into several subgroups. The first subgroup includes noun phrases (either functioning as the subject or object) that have some morphosyntactical clues signaling their plurality. The second subgroup includes noun phrases that do not have a morphosyntactic clue signaling its plurality. Also, this category includes dou, which does not seem to have a quantificational function.

The table also shows that there is a total number of 106 (13.8%) dou-sentences that have a quasi-universal quantifier. They are: 58 tokens of mei ‘every’; 14 tokens of suoyou(de) ‘all of’; four tokens of yiqie ‘all’; two tokens of quan ‘the entire part of’; and nine tokens of renhe ‘any.’ I also included the pronoun dajia ‘everyone’ in this group, and 19 tokens were found in the data. However, note that mei ‘every’ can be used to modify either the noun phrase before dou or a frequency adverbial.

There are 33 tokens of the lian ‘even’ … dou construction found in the corpus. One of the semantic significances of this construction is that some presuppositions must exist in the discourse. I will discuss them in detail in Section 4.3.

Another type of sentence in which dou can be used is a sentence containing a wh-word. In general, the semantic meaning of wh-words in this type of construction is not interrogative. Additionally, there is a syntactic requirement when the wh-word is in the object position; it must move before dou, just like those objects that are quantified by dou in a regular dou-sentence (e.g., [1c]). As can be seen in Table 2 there are 12 tokens that are found in the corpus. Compared to other constructions related to dou, the frequency is relatively low. A total of 12 tokens were found in the corpus, and the wh-words found in the data include three tokens of shei ‘who,’ six tokens of shenme ‘what,’ two tokens of zenme ‘how (come),’ and finally, one na ‘which.’

The last construction related to dou found in the corpus is the use of the numeral yi ‘one’ and a negation word, such as mei(you). This type of sentence is special in that the noun phrase appearing before dou contains the numeral yi ‘one,’ and there is a negation word in the sentence. Nevertheless, as Table 2 shows, there are only three tokens found in the corpus.

4.3 Discussion

5.1 On the movement of plural objects

Although the use of dou quantifying over direct objects frequently occurs in the corpus and is introduced early in the textbook, the instruction may need further details. Specifically, the objects are required to move to a place before the adverb when they are being quantified over by dou. Without movement, the sentence is ungrammatical. Given that there is no such requirement in a language such as English, L1 transfer may likely occur. One activity that we can use in the classroom is for learners to identify grammatical patterns on their own by giving them both grammatical and ungrammatical sentences. Providing ungrammatical sentences to learners increases their awareness of the incorrect usage of the adverb.

The handout that can be used for this activity is presented in Appendix A. The activity was inspired by researchers such as Hawkins (1984), who argued for the importance of second/foreign language learners’ awareness of the grammar of the target language. The handout includes two parts. The first part contains the sample sentences (both grammatical and ungrammatical ones), and the second part contains sample questions for group discussion. Students can be divided into small groups and have a discussion to determine the grammatical patterns on their own. As can be seen in Appendix A, the subjects in sentences (a) and (b) have a plural form and are both grammatical. In contrast, the subject in sentence (c) is singular, and dou cannot quantify over it. The ungrammaticality of (c) helps students identify this property. The following three questions are designed for students to identify the semantic property of the direct objects in a dou-sentence when they are being quantified over by dou. The contrast between sentences (d) and (f) and sentences (e) and (g) will help them recognize the fact that the direct objects must move to a position before dou.^[9] After going over the seven sentences listed in Part I of the handout, students can begin answering the questions in Part II. Finally, the last step of the exercise is for all the groups to present their answers.

5.2 On the element of being quantified—the semantic plural

Many noun phrases that dou quantifies over do not have morphosyntactic clues signaling their plurality. As discussed, these noun phrases contain some ‘semantic plurality.’ It would benefit learners if we could design an activity that introduces this usage of dou. In the following activity, I will present two exercises, the details of which are included in Appendix B. The main purpose of the activity is to introduce the ‘semantic plurality’ of the noun phrases that are quantified by dou in these different constructions. However, given that these grammatical constructions are introduced at a more advanced level, the activity is suitable for CSL learners who have studied Chinese for at least a year. The constructions that will be included in the activity are summarized in (22).^[10]

As Appendix B illustrates, the main goal of Exercise 1 is to introduce (and review) the properties of ‘semantic plurality.’ The exercise includes eight sentences that contain a subject with a plural form. The purpose of asking these questions is for students to identify different types of plurality of the noun phrases being quantified by dou. For example, sentences (1–3) have morphosyntactic clues signaling the plurality of the subject. However, the subjects in (4–8) do not have any morphosyntactic clues; they only contain a semantic plurality.

The main goal of Exercise 2 is to focus on the grammatical patterns of the constructions in (22). As illustrated, sentences (a–c) contain short sentences that test students’ comprehensive knowledge regarding their understanding of some of the usages of dou. As illustrated, the questions can be asked in either Chinese or English. In (d), a short conversational discourse is presented, and a series of comprehensive questions are asked. Finally, in (e), both grammatical and ungrammatical sentences are presented, and the students are asked to figure out the correct usage of the construction that contains dou and a negation word. The exercise tests the grammar of dou in different types of constructions indirectly. For example, the question in (e) does not seek students’ judgment of the grammaticality of the four sentences; rather, it asks students to come up with their own grammatical rules based on their observations of the sample sentences. Moreover, the short conversational discourse in (d) demonstrates how the construction is properly used in the context.