It has been widely claimed in the literature that the sentence-final particle ba in Mandarin Chinese is a modal element. This article argues against this claim and shows that ba is an element that has a unified mitigating function with scope over the utterance as a whole. Using the framework of Functional Discourse Grammar (FDG), and more specifically its hierarchical, layered organization, the article provides several arguments that support this new classification of ba. First, ba, like mitigators in general, but unlike modal elements, can occur in sentences with different basic illocutions. Second, ba may co-occur with modal elements of all different subtypes and thus cannot be a modal element itself. Third, ba may occur in sentences in which the speaker is highly confident of the propositional content. Fourth, unlike modal elements, ba may occur in certain types of non-propositional utterances. And fifth, the position that ba occupies with respect to other sentence-final particles shows that it has scope over the utterance as a whole. After thus arguing for the status of ba as a mitigator, we show how the general mitigating function of ba can acquire the more specific mitigating effects that have previously been attributed to it in the literature.
One of the prominent properties of the grammar of Mandarin Chinese is that it has a set of sentence-final particles, the meanings and uses of which have proven hard to define. One of these particles is ba (吧). Its use is illustrated in (1):
|‘Nothing happened.’ (7377.1 )|
Without the particle ba, the sentence in (1) would be a direct statement transmitting certainty of the speaker. With the particle ba, the speaker transmits less certainty and makes it easier for the addressee to disagree.
In view of observations like these, ba has often been characterized in the literature as a modal particle (Wang 1943/1985: 174, Li 1924/2007: 274–6, Hu 1981: 416, Lu 1984: 334, Zhu 1999: 234–41, Zhang 1997: 19, Chu 1998: 139, Zhou 2009: 16–22, Zhao and Sun 2015: 121–32). It has also been characterized in many other ways, for instance, as a particle soliciting agreement (Li and Thompson 1981: 307–11), disturbing the neustic of the sentence (Han 1995: 118), marking degrees (Li 2006: 21–71), expressing estimation (Li 1924/2007: 274–6, Wang 1943/1985: 174), encoding a suggestion (Wiedenhof 2015: 241–2), or undetermined intention (Zhao and Sun 2015: 121–32).
According to Brown and Levinson (1978), face-saving strategies are employed to achieve successful social interactions. Mitigation is one such pragmatic strategy originally defined as reducing the possible unwelcome perlocutionary effects on the addressee (Fraser 1980: 342) and later as modifying the illocutionary force of a speech act (Hengeveld 1989: 140, Hengeveld and Mackenzie 2008: 83, Thaler 2012: 908). There is a wide range of mitigating devices available across languages: hedges, modals, evidentials, the conditional form, person deixis (Schneider 2010: 261–3), diminutives, impersonal subjects, and even understatement (Caffi 1999: 891–906). As revealed in the discussions of four Dutch modal particles by Vismans (1994), particles are also found to have mitigating functions. This article argues that the sentence-final ba in Mandarin is neither modal in nature, nor does it derive its specific functions in discourse from the context in which it occurs; instead, it has a unified mitigating function and the modal readings of the sentence are just one of the mitigating effects that this general function brings about. In order to substantiate our claims, we will make use of the theoretical framework of FDG (Hengeveld and Mackenzie 2008). By using this typologically based framework, we will be able to study the Mandarin data from the perspective of the crosslinguistic generalizations that are the empirical basis of this theory.
In what follows, we will first give a brief outline of some relevant aspects of FDG and further motivate the use of this framework for the purposes of our research. Section 3 then lists the predictions that follow from our claim that ba is a mitigator and describes the methods used to test those predictions. In Section 4, we check the predictions that follow from the fact that we do not treat ba as a modal element. Section 5 further explores the issue of how the more specific interpretations of ba mentioned in the literature can be clustered together and related to its general mitigating function and the contexts in which it is used. We round off with our conclusions in Section 6.
FDG (Hengeveld and Mackenzie 2008) is a typologically based theory of language structure. FDG recognizes four different levels of linguistic organization. The interpersonal (pragmatic), representational (semantic), morphosyntactic, and phonological levels. These levels are organized in a top-down fashion, as shown in Figure 1. This figure shows that the interpersonal level dominates the other three levels, the representational level, the morphosyntactic, and phonological levels, while the morphosyntactic level dominates the phonological level.
Each level has a hierarchical internal organization, in the sense that it contains a series of layers that are in a scopal relationship. In this article, we focus on the interpersonal and representational levels only, as these are the ones that are relevant for the analysis of ba as a mitigator. The layers at these levels and the scope relations between them are shown in Figure 2. In this figure, scope is indicated by the symbols “→” reading from left to right and “↓” reading from top to bottom, which both mean “has scope over”.
The interpersonal level comprises different pragmatic layers, with scope relations between them. The lowest layer relevant here is the ascriptive subact (T), which represents an act of predication. The next layer is the communicated content (C), which encapsulates the message transmitted in an utterance. Then there is the illocution (F), which captures the communicative intention of the speaker; and the highest layer relevant here is the discourse act (A), which represents the basic unit of communicative behavior.
At the representational level, different semantic layers are distinguished, again with scope relations between them: the lowest one is the property (fl) expressed by a lexical element. The next is the configurational property (fc), which consists of the lexical element and its argument(s) and as such provides the basic characterization of a state-of-affairs (SoA). Then comes the SoA (e) itself, which is the situated real or hypothesized situation the speaker is describing. The next layer is the episode (ep), which is a thematically coherent combination of SoAs that are characterized by unity or continuity of time, location, and participants. The highest layer is the propositional content (p), which is a mental construct, such as a belief, piece of knowledge, or wish.
As mentioned above, the levels themselves are also hierarchically related, with the interpersonal level having higher scope than the representational level, as indicated in Figure 1.
FDG systematically distinguishes between relational and non-relational grammatical categories, the former being represented as functions and the latter as operators. Possible realizations of relational categories are adpositions and conjunctions. Non-relational categories include, among others, tense, aspect, mood, evidentiality, and negation. This wide definition of operators opens up an enormous range of grammatical categories. Table 1 gives an overview of all the relevant categories at the clausal level as presented in Hengeveld and Mackenzie (2008), but including the modifications proposed in Hengeveld and Fischer (2018) for aspect; Hengeveld and Dall’Aglio Hatthner (2015) and Hengeveld and Fischer (2018) for evidentiality; Hatther and Hengeveld (2016), Olbertz and Gasparini Bastos (2013), and Olbertz and Honselaar (2017) for modality; Mackenzie (2009) and Hengeveld and Mackenzie (2018) for polarity; and Olbertz (2012) and Hengeveld (2017) for mirativity.
|Interpersonal level||Representational level|
|Discourse act (A)||Illocution (F)||Communicated content (C)||Ascriptive subact (T)||Propositional content (p)||Episode (ep)||State-of-affairs (e)||Configurational property (fc)||Lexical property (f)|
|Mood||Illocutionary modification||Basic illocution; illocutionary modification||Proposition-oriented modality||Episode-oriented modality||Event-oriented modality||Participant-oriented modality|
|Polarity||Rejection||Negative basic illocutions||Denial||Metalinguistic negation||Disagreement||Co-negation||Non-occurrence||Failure||Local negation|
|Tense||Absolute tense||Relative tense|
|Aspect||Event quantification||Qualitative aspect, participant-oriented quantification||Property quantification|
It is important to underline the two important aspects of the system presented in Table 1. First, it is predicted in FDG that the scope of operators will be reflected in the ordering of elements of the clause expressing these operators within the clause. Operators with lower scope are expected to occur closer to the predicate than those with higher scope. Second, the table shows that mood, polarity, evidentiality, tense, aspect, and localization are categories that can be divided into subtypes with different scope properties and thus the category itself cannot be used as a unified tool for analysis. The ordering of the elements expressing these subtypes is again expected to reflect the scope relations within the overall category.
It is outside the scope of this article to give a full motivation of the categorization presented in Table 1, but we provide more details on the categories that play a central role in the remainder of this article, i.e., the ones listed under “mood”. Within this category, a further distinction is made between illocution, which applies at the interpersonal level, and modality, which applies at the representational level.
As seen in Table 1, the category of illocution can be further subdivided into basic illocution and illocutionary modification. Basic illocutions are conventionalized expressions of communicative intentions, and include, e.g., declarative, interrogative, and directive. Illocutionary modifiers express modifications of basic illocutions, such as their mitigation and reinforcement.
Operators expressing illocutionary modification may occur at the layer of the illocution or at the layer of the discourse act. They operate at the layer of the illocution when they are restricted to certain basic illocutions, as in, e.g., A’ingae (Fischer and Hengeveld n.d.), which has a specific expression for the mitigation of directives. They operate at the layer of the Discourse Act when they apply to all basic illocutions that are available in a language, as in the case of Spanish, a language that allows the use of the reinforcing particle que with declarative, interrogative, and directive basic illocutions alike (Hengeveld and Mackenzie 2008: 67).
The category of modality is subdivided into four different categories: proposition-oriented modality, episode-oriented modality, event-oriented modality, and participant-oriented modality. An example containing all four types of modalities, most of them expressed lexically and one grammatically, is given in (2).
|(2)||It is certainly possible that one has to be able to swim in order to gain entrance to the swimming pool.|
Proposition-oriented modality, also known as subjective epistemic modality, expresses the degree of commitment of a speaker with respect to the truth value of a propositional content, a function performed in (2) by the adverb certainly. Episode-oriented modality, which is also known as objective epistemic modality, characterizes episodes in terms of the degree of likelihood of their occurrence. In (2), the modal adjective possible describes such a degree. Event-oriented modality, as expressed by have to in (2), characterizes an SoA as a whole as feasible or mandatory. Finally, participant-oriented modality describes a relation between a participant in an SoA and the realization of that SoA, as illustrated by the modal expression be able to in (2), which ascribes the ability to perform an SoA to a participant.
The fact that the four different modal expressions can be used in a single sentence such as (2) shows that there must indeed be four different subtypes of modalities, which belong to different paradigms. Otherwise, it would be hard to account for the co-occurrence of modal expressions such as certainly and possible in a single sentence. Furthermore, it is important to note that (2) illustrates that the scope of the modal categories determines their order with respect to the predicate, a point we made above. In (2), certainly is furthest away from the predicate, followed by be possible, have to, and be able to, an order that directly reflects the underlying scope properties of each modal element. Note also that any ordering of the modal expressions other than the one in (2) would lead to an ungrammatical sentence.
In this article, we justify our claim that ba is a mitigator rather than a modal element. This claim leads to a number of predictions that we will test in the remainder of this article.
The first prediction is that ba shows grammatical behavior that one would not expect from a modal element. This general prediction leads to a number of more specific predictions. If ba were a modal element, it should not be possible for it to (i) occur in sentences with all possible basic illocutions, more specifically in directives, (ii) co-occur with modal elements of all different subtypes, (iii) occur in sentences in which the speaker is highly confident of the propositional content, (iv) occur in certain types of non-propositional utterances, or (v) occupy a position in which it has scope over the utterance as a whole, including the modal elements that it contains. These five-specific predictions are tested in Section 4.
The second general prediction is that, if it is a general mitigator, ba will receive more specific interpretations that can be explained in terms of the interaction between the mitigating function that ba realizes, the properties of the type of speech act in which it occurs, and the properties of the wider context that this speech act is embedded in. This prediction is tested in Section 5.
In testing the predictions, we make use of corpus data. The corpus used is the Peking Corpus by the Center for Chinese Linguistics (CCL), which contains a sub-corpus of vernacular Mandarin and a sub-corpus of modern Mandarin. The latter, which is the one we used, is made up of 509,913,589 words and covers both spoken and written Mandarin in various text types, including novels, plays, stories, TV programs, movies, newspapers, cyber-text, and translated works. Searches with ba as the key word in the modern corpus resulted in 104,276 instances on 2,086 pages with approximately 50 instances per page. Given the limitations of the CCL corpus as to randomizing automatically, we randomly selected 16 pages, giving 879 instances in total. Of these, 116 occurrences were removed because they were homonyms of ba (as a topic marker, a noun, or an onomatopoeic use),  as well as several instances of repetitions and typos, resulting in 763 instances of the sentence-final use. These constitute the sub-corpus used in this study, all instances of which were analyzed manually.
In this section, we test the predictions concerning the non-modal nature of ba listed in the previous section. The results show that ba can indeed not be a modal element, as it occurs in contexts in which modal elements would not be possible.
Subjective modal elements are generally restricted to declarative sentences (Hengeveld 1989: 138) and to a lesser extent also occur in interrogative sentences. They may occur in these sentence types as these contain a propositional content. This is different for directive speech acts, which do not transmit information but project behavior. Subjective epistemic expressions, therefore, cannot occur in directives (for further discussion see Section 4.5). If ba were a modal element of uncertainty, as claimed in the literature, ba would have to be restricted to declaratives and interrogatives; however, ba may occur in sentences with all basic illocutions encountered in Mandarin Chinese.
The basic illocutions identified in Mandarin are declarative, interrogative, directive, exclamative, and prohibitive, which have distinct intonational features. Declarative is characterized by a flat intonation (Huang and Liao 2011/2015: 99) and a fall at the end of an utterance (Duanmu 2000: 235), while the interrogative intonation usually has an overall higher phrase curve, higher strengths of sentence final tones, and a final-tone-dependent mechanism (Yuan 2006, Liu et al. 2016). Although directives usually have a low tone range similar to declaratives, they enjoy several features distinguishable from declaratives. Directives are restricted to first and second person if explicitly expressed; directives usually have bare verbs; the only aspect marker that can occur in directives is the progressive marker zhe.  In addition, a stronger command is distinguished by an overall forceful tone as well as a short and intense sentence-final tone (He and Jing 1992: 96). Prohibitive requires the use of negative words such as bie, bu, and bu-yao with a stressed intonation, where bie is a negative particle used exclusively in prohibitives (Li and Thompson 1981: 455). Finally, exclamative, in addition to a high frequency of degree adverbs, has a distinct prosodic contour characterized by one or multiple strong stress positions, a wide pitch range, and a low ending boundary (Chen 2007).
In our corpus, ba occurs in all five sentence types, as exemplified in (3). 
|‘You have designed more than 2000 kinds of works.’ (6495.1)|
|‘You understand what I meant?’ (748.1)|
|‘Tell me the truth immediately. How old are you?’ (63.1)|
|d.||(Three tax inspectors are investigating the tax-paying history of the factory. The factory director is calling up the head of the tax department, complaining angrily:)|
|‘I am not saying that you people don’t have the right to investigate, but the vindictive manner in which you did it is so obviously unreasonable as well.’ (9628.1)|
|‘Don’t sling (the rail) anymore.’ (7009.1)|
(3a) is a declarative in which the presence of ba makes the statement sound less assertive and thus leaves more space for the addressee to disagree. (3b) is an interrogative in which the use of ba makes the question sound more consultative. The first clause in (3c) is a directive in which the speaker is pushing a woman to tell the truth. The use of ba reduces the harshness of this command. (3d) is an exclamative in which the speaker is complaining angrily. The presence of ba mitigates the negativity of the speaker’s complaint. (3e) is a prohibitive in which bie indicates that the speaker wants the addressee to stop. By using ba, the speaker makes the prohibition sound less forceful. The sentences above illustrate the combinability of ba with all kinds of illocutions, which shows that ba cannot be an epistemic marker of uncertainty. Besides, in comparison with their counterpart sentences without ba, the illocutionary force of each sentence with ba is mitigated, which brings about the effect that the addressee is given more space either to disagree or to refuse to comply, if necessary.
In order to explore the combinatory status of ba with different illocutions, we manually annotated each instance in our sub-corpus and obtained the absolute frequency and percentage of each illocution combined with ba, as shown in the second column in Table 2. These frequencies and percentages, however, might be correlated with the overall distribution of illocutions within the sub-corpus; therefore, we also investigated the relative percentage of each illocution in the CCL modern Mandarin corpus, regardless of the presence or absence of ba. Due to the accessibility of the text-type data in the CCL corpus, we randomly selected 200 sentences from each of the spoken and literary sub-corpora to uncover the overall distribution of illocutions in Mandarin. We manually annotated those 400 sentences, which resulted in 362 in total after the irrelevant ones such as repetitions and typos were removed. We then calculated the frequencies and percentages per illocution type among these 362 sentences, as seen in column three of Table 2. Finally, this overall frequency per illocution type, regardless of the presence or absence of ba, is divided by the absolute percentage per illocution type with ba in our sub-corpus. The rates and ranks are given in the fourth column.
|Basic illocution||With ba in our sub-corpus||Overall frequency per illocution (with or without the presence of ba)||Rate/Rank|
|Declarative||235 (30.8%)||269 (74.3%)||0.4/4|
|Interrogative||147 (19.3%)||59 (16.3%)||1.18/2|
|Directive||376 (49.3%)||25 (6.9%)||7.14/1|
|Exclamative||4 (0.5%)||7 (1.9%)||0.26/5|
|Prohibitive||1 (0.1%)||2 (0.6%)||0.5/3|
|Total||763 (100%)||362 (100%)|
Table 2 shows important differences in the overall presence of basic illocutions, of which the declarative is the highest in frequency. On an even distribution, it would be expected that ba would most frequently occur with the declarative. However, as shown in the fourth column, the directive use of ba, not the declarative use, is dominant in our sub-corpus, to the extent that ba is used in directives over seven times more often than would be expected on an even distribution. This reveals that the primary use of ba is not modal, as mentioned above, epistemic modality targets a proposition as expressed in a declarative or interrogative sentence, whereas directives are proposals that involve the non-verbal exchange of goods-and-services (objects or actions) that cannot be affirmed or denied (Halliday 1994: 68–71). Hence, in the case of directives, it would be impossible for the speaker to use ba as a modal to show his/her epistemic commitment toward a proposal. Rather, ba is a mitigator as can be seen from the comparison between the directives with and without ba. With the presence of ba, the illocutionary force of directives is softened; without ba, the directives sound more forceful. As the basic function of a directive is to issue an order or a request, which poses a potential threat to the negative face of the addressee (Brown and Levinson 1978), the speaker uses ba as a mitigating strategy to soften this face-threatening effect, thus leaving space for the addressee to refuse to comply.
The second most frequent illocution with which ba combines is the interrogative. There is a small difference between the frequency of interrogatives with ba and the frequency of interrogatives overall. Though less face threatening, interrogatives are somewhat similar to directives in being requests for information and might therefore also be expected to be mitigated. This also holds for prohibitives, the third most frequent illocution, whose strong imposing force tends to be softened by the speaker. As directives, interrogatives and prohibitives are demanding in nature, either for an action or for a piece of information, and it is very likely for them to be mitigated in order to soften the imposition on the addressee. The illocutions that ba combines with least frequently are declaratives and exclamatives. In comparison with other illocutions, declaratives and exclamatives are informative in nature, either conveying a statement or expressing strong emotions. These two illocutions are consequently less face threatening than the other illocutions and thus less in need of mitigation.
Owing to the low frequencies of exclamatives and prohibitives and the lack of available raw data of more text types, what is presented in column five of Table 2 may not reflect the actual rate of ba per illocution. Nevertheless, the substantial differences are still revealing, especially the overwhelmingly higher frequency of ba in directives, which shows that its directive use, not its declarative use, is basic and primary. This is also confirmed by diachronic research conducted by Tantucci (2017), who points out that in recent history ba has witnessed a gradual shift from the directive use to the declarative use. This reveals that the use of the mitigator in directives is the original one and has diachronically expanded to declaratives, functioning to reduce the assertive force of the declarative.
Crosslinguistically, a mitigator can modify a specific illocution or all kinds of illocutions. As argued in Section 2, in FDG a mitigator that co-occurs with all possible illocution types is treated as an operator at the layer of the discourse act, thus modifying the discourse act as a whole. As shown in (3), the presence of ba reduces the force of all types of illocutions. Therefore, ba may be analyzed as operating at the discourse act layer rather than at the layer of the illocution. This stands in contrast to much of the existing literature, in which ba is often regarded as modifying some specific illocutionary force, such as estimation (Wang 1943/1985: 174, Li 1924/2007:274–6), soliciting agreement (Li and Thompson 1981: 307–11), suggestion (Lü 1999: 56–57, Wiedenhof 2015: 241–2), etc. Li (2006: 21–71) comes closest to our position, as she argues that ba has a unified function of marking a low degree of commitment in different sentence types, focusing on “ba’s semantic contributions” (2006: 28). However, she equals the strength of the speaker’s intention to the sentence force. She maintains that ba marks “a low degree of strength with respect to the speaker’s intention to have an action carried out when occurring in imperatives, as well as in various types of questions” (2006: 36). In our view, what’s being weakened by the presence of ba is not the speaker’s intention but the illocutionary force concerned. For instance, in directives, the speaker uses ba not to indicate his/her weak intention to have the action carried out but to indicate his/her openness to negotiate the strong intention of having the action carried out. In other words, what’s being modified is not the intention itself but the strong directive force for the sake of reducing the unwelcome perlocutionary effects that a directive might bring along.
In this section, we have shown that ba behaves differently from modals in two respects in terms of its combinability with different types of illocutions. First, modals have restrictions as to which types of illocutions they can occur with; however, there are no such restrictions on ba as it can combine freely with any type of illocution. Second, modals are expected to occur most frequently in declaratives; however, our data show that the most frequent illocution in which ba occurs is the directive.
In the literature, ba is considered to express the subjective epistemic modality of uncertainty (e.g., Wang 1943/1985: 174, Hu 1981: 416, Zhu 1999: 234–41), which will be argued in this section to be a wrong analysis, as ba cannot be a modal if it itself can co-occur with all kinds of modals, including modals of subjective epistemic modality.
In our sub-corpus, there are 235 declarative sentences with ba, among which 95 co-occur with one or two modal elements, listed in Table 3. Among the four types of modalities distinguished in FDG (see Section 2), subjective epistemic modality and objective epistemic modality are the most difficult ones to classify. We adopt Olbertz and Dall’Aglio Hattnher’s (2018: 139) content question test to distinguish subjective modality from objective modality in our sub-corpus. Subjective modalities occur in propositional contents, which, as such, cannot be located in space or time, while objective modalities occur in episodes, which can be located in space or time. We use content question words of time such as shen-me shi-hou “when” or content words of space such as zai na-er or shen-me di-fang “where” to test each declarative sentence in which an epistemic modal occurs. If the sentence becomes unacceptable after the insertion of the relevant question words, the modal expression is categorized as expressing subjective epistemic modality as in (4).
|‘The happiest state of life … must be being content.’ (517.1)|
|b.||*ren-sheng||zui||xing-fu||de||zhuang-tai…da-gai||shen-me shi-hou/zai na-er|
|‘When/where must the happiest state of life … be being content?’|
|‘In such weather, (you) are likely to miss the hot noodles that your mother made by her own hands a lot.’ (2705.2)|
|‘Where are you likely to miss the hot noodles that your mother made by her own hands a lot?’|
|Type of modality||Modal elements||Frequency||Percentage|
|Subjective epistemic modality||da-gai (大概) “probably” ||21||22.1|
|ye-xu/huo-xu (也许/或许) “perhaps”||19||20.0|
|da-yue (大约) “probably”||5||5.2|
|ke-neng (可能) “possibly”||14||14.7|
|bu-yi-ding (不一定) “uncertainly”||1||1.1|
|Objective epistemic modality||yi-ding (一定) “certainly”||4||4.2|
|ken-ding (肯定) “certainly”||1||1.1|
|Event-oriented modality||gai (应该/该) “should”||11||11.5|
|ke-yi (可以) “may”||1||1.1|
|bu-neng (不能) “cannot”||8||8.4|
|neng (能) “can”||1||1.1|
|Participant-oriented modality||ke (可) “can”||1||1.1|
|hui (会) “would”||7||7.3|
|ke-yi (可以) “can”||1||1.1|
If the sentence is acceptable, the modal word is considered to be an objective epistemic modal, as is the case in (5).
To distinguish between event-oriented, and participant-oriented modality, it is necessary to determine whether the modal expression involves/concerns external circumstances or participants. In (6a), it involves the external circumstances. In this case, a general rule requires the summing up rather than the experience and the lessons themselves; therefore, ying-gai “should” expresses an event-oriented modality. In (6b), the understanding involves the ability of the participant ni-men “you”, so hui “will” expresses a participant-oriented modality.
|‘The experience and lessons should always be summed up.’ (7395.1)|
|‘By then, you will come to understand my reviews and comments on this play.’ (7358.1)|
As Table 3 shows, ba co-occurs with all types of modalities. The most frequent one is subjective epistemic modality, especially with the expressions da-gai, ye-xu, and ke-neng.
As we have shown in example (2), modal elements may co-occur when they belong to different classes, but not when they belong to the same class. What Table 3 shows, however, is that ba may co-occur with any type of modality, so this means that it cannot itself belong to any of these classes.
Examples of combinations of ba with all other types of modalities are shown in (7). Da-gai in (7a) expresses subjective epistemic modality, yi-ding in (7b) expresses objective epistemic modality; gai in (7c) event-oriented modality; and hui in (7d) participant-oriented modality.
|‘Probably it was in the year before last.’ (6474.1)|
|‘Having a son like Henry, you must be very proud of yourself.’ (2746.1)|
|‘You are the ambassador who has stayed the longest time at Beijing. When you leave, you should at least say goodbye (because if you don’t, it would be impolite).’ (7776.1)|
|‘I guess that even the wrong analysis made by the leader of the Wave Theory would comply with the rules of the game in the Wave Theory.’ (2718.1)|
In the sub-corpus, there are five cases of ba occurring with two modal elements in a single declarative sentence. Since each type of modality pertains to a different layer in FDG (Table 1), it is theoretically legitimate for them to co-occur. In (8a), ba co-occurs with the subjective modal da-gai and the event-oriented modal neng, whereas in (8b) it co-occurs with the objective epistemic modal ken-ding and the participant-oriented modal hui.
|‘In your free time, you always sleep or play Ma-Jiang, which probably cannot be considered to be the best choice.’ (8480.1)|
|‘These places certainly won’t be concerned about any Beijing Hu-Kou.’ (1110.1)|
In our corpus, there are no co-occurrences of three modal elements with ba. Nevertheless, the fact that ba can co-occur with all types of modalities and even with more than one type of modality in a single sentence demonstrates that ba cannot itself be a modal, since, if it were a modal, the combination with at least one of these types would lead to ungrammaticality.
There is an even stronger argument to support this observation. If the particle ba and a modal expression are both grammatical, their co-occurrence would even be more severely restricted, as grammatical elements belonging to the same class constitute a paradigm. Among the class of subjective epistemic modals in Mandarin, there are two that are grammatical in nature: da-gai and ye-xu. 
If ba, as a grammatical element, were to express subjective epistemic modality, it would share its categorical grammatical features with da-gai and ye-xu. Therefore, they would fall into the same paradigm and would thus be mutually exclusive in a single sentence. The fact that there are 21 cases of co-occurrence of ba with da-gai and 19 with ye-xu shows that there is no restriction for ba to co-occur with subjective epistemic modality. Examples are given in (9a) and (9b):
|‘Probably this is indeed a big difference between animals and human beings.’ (1137.1)|
|‘Perhaps, this is indeed one big characteristic of the nomadic culture.’ (7783.1)|
In all, we must conclude that ba cannot be a modal marker, due to its co-occurrence with modal elements of all possible types. The uncertainty reading of the sentence in which ba occurs is not due to an inherent modal property of ba but to the mitigating effect that ba brings about when it interacts with the illocutionary force of the sentence. We will come back to this effect in Sections 4.4 and 5.4.
If ba were a modal marker of uncertainty (Hu 1981: 416, Lu 1984: 334, Chu 1998: 139, Zhou 2009: 16–22, Zhao and Sun 2015: 121–32), then a speaker could use ba to show he/she is uncertain about what he/she is asserting. However, ba can occur in sentences in which the speaker displays a high degree of confidence in the truth of the statement (Tantucci 2017). In (10), the village head is highly confident that Chen Yue-Qing is reluctant to sell the orchard because no one would sell something that is very profitable.
|(10)||(Chen Yue-Qing owned an apple orchard. She wanted to sell it because she thought she couldn’t handle this alone after her husband died in a traffic accident. The village head persuaded her not to sell it and lent her some money. After some time, she began to gain profit from the orchard. On a visit to her orchard, the village head said in a happy voice to her,)|
|‘Yet now you are reluctant to sell the orchard.’ (8492.1)|
There are even cases in which the speaker explicitly expresses high confidence by using expressions such as yi-ding and ken-ding, as shown in (7b) above and (11a) below. In these cases, ba cannot encode uncertainty because it would be contradictory for the speaker to explicitly express high confidence in what he/she said and at the same time present it as being uncertain. This can be confirmed by replacing ba in (11a) by an expression of subjective possibility such as ye-xu “perhaps”, which would render the sentence unacceptable, as shown in (11b).
|‘At that night, Dong Wen-Hua must be asleep.’ (3856.1)|
|‘At that night, Dong Wen-Hua must perhaps be asleep.’|
The ungrammaticality of (11b) is due to the clash between the modal expression yi-ding “certainly” and the modal expression ye-xu “perhaps”, which expresses the opposite modal value. The fact that ba does not have such an effect shows that it cannot itself be modal.
However, it is not the case that ba can occur in all declaratives in which the speaker expresses strong confidence. The typical example is a bare assertion concerning the speaker’s own past behavior, as in (12a).
|‘I went to Beijing yesterday.’|
|‘I went to Beijing yesterday.’|
|‘I went to Beijing yesterday?’|
In (12a), the speaker’s going to Beijing is presented as a fact, leaving no space for negotiation. The addition of ba to such a bare assertion makes the sentence unacceptable, as in (12b), because when one shows less than full commitment toward his/her own actions, one calls himself/herself into question. However, if (12b) is changed into a rhetorical question, it becomes acceptable, as in (12c). Similar to (10), ba in (12c) is legitimate despite the fact that the speaker has full confidence in what’s being stated. The difference between (12a) and (12c) is that in (12c), the speaker leaves space for negotiation with the addressee, while this is not the case in (12a).
The fact that the speaker uses ba even when he/she has high confidence in what’s being asserted reveals that ba is not used to encode the speaker’s real uncertainty but instead is used strategically to achieve pragmatic purposes, functioning in the same way as hedges and pragmatic markers across languages. Hence, to express uncertainty is not ba’s inherent nature; rather, when the illocutionary force of an assertion is mitigated, the assertion consequently sounds less certain as this involves a weakening of the speaker’s commitment to the truth of the proposition. To put it more simply, the uncertainty reading of the sentence is just the mitigating effect that ba brings about in an assertion.
As mentioned in Section 4.2, Li (2006: 21–71) argues that ba marks a low degree of the assertive force, which seems to be similar to our argument that ba has a mitigating function in declaratives. However, what she means by “a low degree” is that “with ba the speaker is not wholly certain about the factual status of the proposition and makes a weak assertion” (2006: 60). This shows that her approach reflects a modal perspective rather than one in which ba’s pragmatic mitigating functions are recognized.
Ba not only occurs in sentences with strong epistemic commitment but also in non-propositional sentences. The most frequent co-occurrence is with hao “okay”. For instance,
|(13)||(The son is afraid of swinging. His father is trying to encourage him to have a try by showing him how. Seeing that his father is enjoying himself on the swing, the son says,)|
|‘Alright then, but I don’t want to swing that high.’ (2735.1)|
As mentioned in Section 4.2, human interactions involve two kinds of exchange, an exchange of propositions and an exchange of proposals. The former refers to statements and questions that can be “argued about something that can be affirmed or denied, and also doubted, contradicted, insisted on, accepted with reservation, qualified, tempered, regretted and so on” (Halliday 1994: 70), whereas the latter refers to offers and commands that cannot be affirmed or denied (Halliday 1994: 70). A proposition can be related to epistemic uncertainty but a proposal cannot. For instance, in English, propositional (14) is correct, but it is awkward to say (15).
|(14)||I am not so sure whether we should go or not.|
|(15)||*Let’s probably go.|
In Mandarin, hao is actional since it reacts to a proposal, not to a propositional content. In (13), hao ba is a reaction to the father’s suggestion to swing, rather than an epistemic agreement as to whether the statement is true or not. It is not grammatical to use (16) in response to the father’s encouragement in the context given in (13).
Hengeveld and Mackenzie (2008: 146–9) point out that yes is propositional in nature and can be used to substitute for full propositional contents. Hence, it is used in reaction to a statement or a question but cannot be used to react to a directive because a directive evokes an SoA rather than a propositional content. The reaction to a directive would be an actional one like okay.  This means that it is incorrect for Zhou (2009:16–22), Chu (1998:132–9), and Zhao and Sun (2015:121–32) to claim that ba is used in directives to show that the speaker is unsure of or dubious about the enactment of the speech act. Actually, when ba occurs with actional hao “okay”, it mitigates the willingness to carry out the required action. In (13), without ba, the sentence means “okay”, implying I would like to do that; with ba, it means “alright”, implying I will do that but I am somewhat reluctant.
As introduced in Section 2, FDG recognizes four levels, two of which are relevant in our discussions, namely, the representational level and the interpersonal level. The former is concerned with semantics, while the latter deals with pragmatics. The hierarchical layout presented in Table 1 predicts the actual ordering of linguistic elements pertaining to different layers and levels. If ba is modal, it expresses semantic categories and thus pertains to the representational level. In this section, based on ba’s positioning with respect to other sentence-final particles, we will argue that this cannot be the case.
Mandarin Chinese has a rich inventory of sentence-final particles. Two distinguishing features of Mandarin sentence-final particles are that they have a very high frequency of occurrence in daily conversations and that they can occur in clusters which have a highly restricted linear order. The most basic sentence-final particles are the following six: de (的), le (了), ne (呢), ba (吧), ma (吗), and a (啊). They have different degrees of scope over the content of the sentence and can thus cluster hierarchically at the end of a sentence. We searched all the logically possible permutations of these six particles in the CCL corpus (see Fang and Hengeveld n.d.), finding that the maximal combination of those basic particles is three in a single sentence. The most frequent combination is de le a, which is phonetically fused into de la (的啦) due to the adjacency of two vowels. The second most frequent one is de le ma and the third is de le ba. The third combination is illustrated in (17).
|‘You will certainly be willing to go and hang out with Dad.’|
De in (17) is a modal marker of certainty, which pertains to the layer of the propositional content at the representational level. Le is a mirative marker at the layer of the communicated content at the interpersonal level (Fang 2018). The representational level is lower than the interpersonal level, which is iconically reflected in the ordering of de and le. The sequence de le ba shows that ba has scope over both le and de and thus occupies a layer higher than the layer at the interpersonal level that le pertains to. The fact that ba is located at the interpersonal level, not at the representational level, invalidates the claim that ba is a modal marker; instead, it validates our argument that ba is a pragmatic marker.
Then the question arises as to which layer ba pertains to at the interpersonal level. The two layers higher than the communicated content are the layer of the illocution and the layer of the discourse act, as shown in Table 1. As mentioned in Section 2, the difference between an operator at the layer of the illocution and an operator at the layer of discourse act is that the former accounts for grammatical emphasis and mitigation of a specific illocution or a limited range of illocutions, whereas the latter can combine with all types of illocutions, thus reinforcing or mitigating the discourse act as a whole. Accordingly, ba is an operator at the layer of the discourse act, which can combine with all kinds of illocutions (see Section 4.2) and has a mitigating effect on any of the illocutions that it occurs with.
In the preceding section, we have shown, using a variety of tests, that ba cannot be analyzed as a modal element. In this section, we will show that ba should instead be analyzed as having a generalized mitigating function, which acquires specific effects depending on the context in which it occurs. Since these varying and multiple contextual meanings of ba are specifically dependent on the kind of speech act in which they occur, we organize this section in terms of illocutionary values. For each of these, we show how the type of speech act interacts with the general mitigating function of ba in producing a specific perlocutionary effect. The illocutionary values that we will discuss are directive (Section 5.2), interrogative (Section 5.3), declarative (Section 5.4), exclamative (Section 5.5), and prohibitive (Section 5.6). We summarize our findings in Section 5.7.
Directives in Mandarin very often contain overt first person (wo; wo-men) and second person (ni; ni-men) pronouns, expressing, depending on the nature of the subject, imperative, and hortative illocutions. With a subject that includes the first person, they generally involve actions that the speaker expects to carry out without or with the addressee. The first person singular wo indicates that the speaker offers to do something on his/her own, as in (18a); the first person plural wo-men can indicate that the proposed action is inclusive or exclusive of the addressee. For instance, (18b) could mean that the speaker invites the addressee to do something together with him/her, or that the speaker tells the addressee which action he/she plans to carry out with another person present. With a subject in the second person, singular or plural, directives concern actions that the speaker or addressee(s) is/are expected to carry out, as in (18c) and (18d). Among the examples in (18), (18a), and (18b) are hortative, while (18c) and (18d) are imperative.
|‘Let me go by myself.’ (711.2)|
|‘Let’s go.’ (Li and Thompson 1981: 307)|
|‘Tell us and speak slowly.’ (701.1)|
|‘Do whatever you think is right.’ (925.1)|
For Li and Thompson (1981: 307), ba in (18b) is used to solicit agreement. This, however, is not the correct characterization, as the speaker intends the addressee to carry out the action rather than to agree with a proposition. As argued in Section 4.5, ba for that same reason cannot be a modal here, nor is it a marker to encode the directive mood, as the removal of ba does not affect its directive illocution. The difference between the presence and absence of ba in (18) lies in the strength of the directive force of the utterances. Without ba, the sentences are a neutral order or exhortation, whereas with ba, the directive illocutionary force is softened, so that the speech act becomes negotiable and more suggestive. The fact that ba often co-occurs with expressions of politeness such as nin (an honorific pronoun, similar to vous in French, Qi and Zhu 2005: 62–67) also proves that ba is interpersonally/pragmatically oriented. The basic function of a directive is to order, request, or encourage the speaker or the addressee, sometimes together with the speaker, to perform some action. All these acts are potentially face threatening; therefore, in order to build up solidarity between interlocutors, a language must equip its users with means to be polite, to prevent loss of face, to leave room for the addressee to refuse or disagree, to make the addressee feel more comfortable, etc. (Hengeveld 1989: 131). In a directive speech act, mitigation, as remarked by Thaler (2012: 909), allows for the possibility of refusal and thus meets the addressee’s negative face needs, such as the desire for freedom of action and freedom from constraints imposed by others. Ba in Mandarin serves this very purpose.
In addition to softening a request or a command, ba can mitigate the intensity of negative emotions in directives. Both (19a) and (19b) are directives that involve the strong negative emotions on the part of the speaker. Compared with the directives without ba, the intensity of negative emotions in directives with ba is reduced.
|‘Get out as quickly as possible!’ (Wo Ju/Liu Liu/CCL)|
|‘Alright, shut up!’ (Pi Pi/writer’s name unspecified/CCL)|
In conclusion, we may say that the addition of ba to a directive sentence leads to the mitigation of the directive force of the sentence.
The interrogatives in which ba occurs are often tag questions, which are more consultative than assertions and less inquisitive than polar questions. As observed by Dik (1989:257), in a tag question, the proposition is not being questioned and the nature of a declarative is altered in such a way that it sounds less assertive. Fraser (1980: 342) points out that tag questions are used by the speaker to mitigate the force of a speech act. In the case of polar questions, when their interrogative force is mitigated, they can become tag questions, which sound less inquisitive and more confirmation seeking.
We agree with Li and Thompson (1981: 307) that ba in interrogatives has the effect of soliciting agreement from the addressee, as in (20), but we differ from them in that we consider this effect to be a result of the general mitigating function of ba.
|‘He wouldn’t do such things; wouldn’t you agree?’ (Li & Thompson 1981:307)|
Without ba, (20) would be a straightforward question; with ba, the original interrogative force is mitigated such that the speaker is not asking a question but seeking confirmation from the addressee.
In addition to polar questions, there are three other types of questions in Mandarin, namely, WH questions, alternative questions, and verb-not-verb questions. WH questions request the filling of an information gap by using question words; alternative questions provide an either-or choice for the addressee, usually with the expression hai-shi “or”; and verb-not-verb questions likewise offer two alternatives of an event, but in this case one is the negative counterpart of the other.
When ba occurs in these question types, the interrogative force is reduced and the sentence conveys a more explicit invitation to provide an answer, as shown in the following examples (Zhu 1982: 211):
|‘How much altogether? Please tell me.’|
|‘Please tell me what time you will come.’|
|‘You eat rice or noodles, please tell me.’|
|‘Would you tell me whether you are going or not?’|
(21a) and (21b) are WH questions; (21c) is an alternative question, and (21d) is a verb-not-verb question. The presence of ba creates the perlocutionary effect of inviting the addressee more explicitly to respond to what has been proposed by the speaker.
No such examples are found in our sub-corpus, so we used the question key words italicized in (21) to search the modern corpus of CCL; we found instances of ba in WH questions and verb-not-verb questions such as (22a), (22b), and (22c) but not in alternative questions parallel to (21c), which apparently are somewhat less common.
|‘How much do you want to spend? Please do tell me.’ (Guo De Gang Xiang Sheng Ji/CCL)|
|‘Please do tell me how much you know.’ (Translated Works/Ke Ai De Gu Tou/CCL)|
|‘Would you say that or not? Tell me.’ (Liu Liu/Lie Huo Jin Gang/CCL)|
From the above it can be concluded that the addition of ba to an interrogative sentence other than a polar one creates a more inviting speech act.
When the speaker makes a statement, he/she can present it as a fact as in (23a), leaving no space for negotiation, or add ba to reduce its assertive force as (23b), making it easier for the addressee to disagree (see Section 4.4).
|‘You are joking.’|
|‘You are joking.’ (524.1)|
When ba is added to a declarative with a regular falling intonation, it mitigates the assertiveness of the statement.
When the assertion is negative rather than neutral, as when it contains a criticism, it is expressed with a prolonged duration on the final syllable. When ba is used, this prolonged duration is then expressed on ba itself. The addition of ba reduces the the effect of a negative emotion. In (24a), the mother is blaming her son for not listening to her advice. With ba, the harshness of this criticism is softened. The same is true for (24b).
|(24)||a.||(When the son was leaving for a football match, the mother told him that he should carry a thicker jacket but the son wouldn’t listen and came back coughing and sneezing. Then the mother said,)|
|catch a cold||PERF||MIT|
|‘You’ve caught a cold. (I told you to be careful, you just didn’t listen).’ (Chu and Li 2004: 3) |
|b.||(Someone thought he could fix the computer very easily but it turned out that the computer completely broke down. Another one said,)|
|‘Look at yourself, you bragged too much about yourself. (You shouldn’t have done that).’ (ibid)|
To summarize, the effects of ba in declaratives are as follows: when the sentence has a regular falling tone, ba mitigates the assertiveness of a statement; when the sentence has a falling and prolonged tone, ba mitigates a criticism.
Mandarin exclamatives, expressing the speaker’s emotions (Gao 1986/2011: 584), can be made explicit by having exclamative markers such as demonstrative pronouns (zhe-me “like this”; na-me “like that”), adverbs (zhen “really”; hao “well”; tai “too”) as well as particles (like a), or can be free of any such explicit markers. In the latter case, the exclamative is recognized based on its prosodic contour. Zhao and Sun (2015: 124) claim that ba cannot occur in exclamatives because exclamatives are incompatible with ba’s semantic function of undetermined intention (uncertainty). It is true that epistemic modal expressions cannot occur in exclamatives. The fact that ba does occur, however, once again supports our argument that ba is not modal at all. In our sub-corpus, four exclamative instances with ba were found. In addition to example (3d) in Section 4.2, two more are exemplified in (25) below:
|‘How could grandpa and grandma want to take exclusive possession of it? This is too mean, isn’t it!’ (2709.1)|
|this||do||way||apparently||somewhat||too||go too far||MIR||MIT|
|‘But you have to pay to get the award certificate. That’s taking things too far, surely?’ (7389.1)|
Exclamatives “express the speaker’s affective response to a situation” (Siemund 2015: 702). Both (25a) and (25b) have a specific prosodic contour to express strong negative emotions of anger by complaining and criticizing. According to phonetic experiments by Chen (2007: 50), the explicit marker tai “too” exhibits the highest pitch as well as a longer duration with the word right after it taking a lower pitch. Each sentence ends with a stressed and falling tone on ba. All four instances of ba in exclamatives in our sub-corpus express strong negative emotions of the speaker, as xiao-qi “mean” in (25a) and guo-fen “go too far” in (25b). In both situations, the speaker is expressing anger as he/she regards the situation as illegitimate, unfair, or contrary to what is right. As people have prescriptive norms about the appropriateness of emotions in particular situations, failure to regulate their emotions may lead to face damage and affective impoliteness as displaying anger at someone or doing so publicly may be regarded as being inappropriately or unfairly hurtful, causing an emotional reaction such as embarrassment or anger (Culpeper 2011: 58–59). Therefore, it is understandable for the speaker to use a mitigator to reduce the intensity of an emotion for the purpose of diminishing these undesired perlocutionary effects. In comparison with the absence of ba, the presence of ba softens the strong negative emotions, making them less harsh and offensive.
Prohibitives are used to prohibit an addressee from carrying out the action evoked by the communicated content. In Mandarin, negative words such as bu-yao and bie “don’t” are used to express prohibitive meaning. Prohibitives are similar to directives in that they are both imperatives and can be perceived as challenging the negative face needs of the addressee. In order to make the prohibition less imposing and challenging, the speaker uses ba to mitigate the intensity of the prohibitive illocutionary force.
|‘Please don’t let me down!’ (Hu Xiao Shan Zhuang/CCL)|
|‘First, don’t jump to conclusions.’ (Zhan Dou De Qing Chun/CCL)|
The effect of ba in prohibitions is thus to mitigate the strength of a prohibition.
In this section, we have described in some detail the specific mitigating effects of ba in different contexts. These effects may be summarized as follows:
DIR + ba: mitigate a directive speech act (offer, command, request, etc.)
Y/N INT + ba: solicit confirmation
WH-INT/ALT-INT/VnotV-INT + ba: invite a response
DECL + ba + a falling tone: mitigate assertiveness
DECL + ba with a falling and prolonged tone: mitigate criticism
EXCLAM + ba: reduce the harshness of the expression of a negative emotion
PROH + ba = mitigate the strength of a prohibition
Using the grammatical framework of FDG and authentic data from the CCL Corpus, we have argued on the basis of five criteria that Mandarin ba is not a modal element. First, ba can occur in sentences with all possible basic illocutions; second, ba may co-occur with modal elements of all different subtypes; third, ba may occur in sentences in which the speaker is strongly committed to the propositional content; fourth, unlike modal elements, ba may occur in certain types of non-propositional utterances, and fifth, the position that ba occupies with respect to other sentence-final particles reveals that it has scope over the utterance as a whole. Next, we have shown that ba should be treated as having a unified mitigating function – attaining a higher degree of politeness and leaving more space for negotiation. The general mitigating function accommodates the specific values that the literature has previously attributed to the use of ba occurring under/in various contextual conditions.
Our findings confirm the adequacy of the treatment of mitigation proposed in FDG, the framework we have applied in this article. The fact that generalized mitigation is treated as an operator on the Discourse Act in FDG helps to explain both formal and functional aspects of the particle. From a formal perspective, it accounts for the position of ba with respect to other sentence final particles, as discussed in Section 4.6. From a functional perspective, it explains why the specific contextual uses of ba have to be understood in relation to the specific illocutions with which it combines.
The authors are grateful to the anonymous reviewers for their helpful comments. The authors would also like to thank Hella Olbertz, Lachlan Mackenzie, and Lois Kemp for their comments on earlier versions of this article.
Brown, Penelope, and Stephen C. Levinson. 1978. “Universals in language usage: politeness phenomena.” In Questions and politeness: strategies in social interaction, ed. Esther N. Goody, 56–311. Cambridge: Cambridge University Press. Search in Google Scholar
Caffi, Claudia. 1999. “On mitigation.” Journal of Pragmatics 31: 881–909. Search in Google Scholar
Chen, Hu. 2007. “基于语音库的汉语感叹句与感叹语调研究 (A study of Chinese exclamatory sentence and intonation based on sound corpus).” 汉语学习 (Chinese Language Learning) 5: 45–55. Search in Google Scholar
Chu, C. Chauncey. 1998. A Discourse Grammar of Mandarin Chinese (Berkeley Models of Grammars 6). New York: Peter Lang Publishing, Inc. Search in Google Scholar
Chu, C. Chauncey, and Bin Li. 2004. “论现代汉语句末情态虚词及其英译——以“吧”的语篇功能为例 (Translating Modern Chinese utterance-final particles into English: a case study of the discourse function of BA).” 外语学刊 (Foreign Language Research) 6: 1–10. Search in Google Scholar
Culpeper, Jonathan. 2011. Impoliteness: Using Impoliteness: Using Language to Cause Offence. (Studies in Interactional Sociolinguistics, 28). Cambridge: Cambridge University Press. Search in Google Scholar
Dik, Simon. 1989. The Theory of Functional Grammar: The structure of the clause. Dordrecht: Foris. Search in Google Scholar
Duanmu, San. 2000. The Phonology of Standard Chinese. Oxford: Oxford University Press. Search in Google Scholar
Fang, Hongmei. 2018. “Mirativity in Mandarin: the sentence-final particle le (了).” Open Linguistics 4: 589–607. Search in Google Scholar
Fang, Hongmei, and Kees Hengeveld. n.d. “Sentence-final particles in Mandarin.” subm. Search in Google Scholar
Fischer, Rafael, and Kees Hengeveld. n.d. “A’ingae (Cofán/Kofán).” In Amazonian Languages, An International Handbook, ed. Patience Epps, Lev Michael. Berlin: de Gruyter Mouton. forthc. Search in Google Scholar
Fraser, Bruce. 1980. “Conversational mitigation.” Journal of Pragmatics 4(4): 341–50. Search in Google Scholar
Gao, Mingkai. 1986/2011. 汉语语法论 (On Chinese Grammar). Beijing: The Commercial Press. Search in Google Scholar
Halliday, M. A. K. 1994. An Introduction to Functional Grammar. London: Edward Arnold. Search in Google Scholar
Han, Yang. 1995. “A pragmatic analysis of the BA particle in Mandarin Chinese.” Journal of Chinese Linguistics 23(2): 99–127. Search in Google Scholar
Hatther, Marize Mattos Dall’Aglio, and Kees Hengeveld. 2016. “The grammaticalization of modal verbs in Brazilian Portuguese: a synchronic approach.” Journal of Portuguese Linguistics 15(1): 1–14. Search in Google Scholar
He, Yang, and Song Jing. 1992. “北京话语调的实验探索 (Experimental explorations on Beijing Mandarin intonation).” 语言教学与研究 (Language Teaching and Linguistic Studies) 1992(2): 71–96. Search in Google Scholar
Hengeveld, Kees. 1989. “Layers and operators in functional grammar.” Journal of Linguistics 25(1): 127–57. Search in Google Scholar
Hengeveld, Kees. 2017. “A hierarchical approach to grammaticalization.” In The Grammaticalization of Tense, Aspect, Modality, and Evidentiality: A Functional Perspective, ed. Kees Hengeveld, Heiko Narrog, Hella Olbertz, 13–38. Berlin: de Gruyter Mouton. Search in Google Scholar
Hengeveld, Kees, and Marize Mattos Dall’Aglio Hatthner. 2015. “Four types of evidentiality in the native languages of Brazil.” Linguistics 53(3): 479–524. Search in Google Scholar
Hengeveld, Kees, and Rafael Fischer. 2018. “A’ingae (Cofán/Kofán) operators.” In Systems of tense, aspect, modality, evidentiality and polarity in Functional Discourse Grammar. Open Linguistics, ed. Kees Hengeveld, Hella Olbertz, 328–55, vol. 4. Search in Google Scholar
Hengeveld, Kees, and Lachlan Mackenzie. 2008. Functional Discourse Grammar: A Typologically-based Theory of Language Structure. Oxford: Oxford University Press. Search in Google Scholar
Hengeveld, Kees, and Lachlan Mackenzie. 2018. “Negation in functional discourse grammar.” In Recent developments in Functional Discourse Grammar (Studies in Language Companion Series 205), ed. Evelien Keizer, Hella Olbertz, 17–45. Amsterdam: Benjamins. Search in Google Scholar
Hu, Mingyang. 1981. “北京话的语气助词和叹词 (Mood markers and exclamative markers in Beijing Dialect).” 中国语文 (Studies of the Chinese Language) 6: 416–23. Search in Google Scholar
Huang, Borong, and Xudong Liao. 2011/2015. 现代汉语 (Modern Chinese). Beijing: Higher Education Press. Search in Google Scholar
Keizer, Evelien. 2007. “The lexical-grammatical dichotomy in Functional Discourse Grammar.” In Advances in Functional Discourse Grammar. Special Issue of Alfa: Revista de Lingüstica, ed. Marize Mattos Dall’Aglio Hattnher, Kees Hengeveld, 189–212. vol. 51 (no. 2). Search in Google Scholar
Li Boya. 2006. Chinese Final Particles and the Syntax of the Periphery. Doctoral dissertation. Leiden: Leiden University. Search in Google Scholar
Li, Jinxi. 1924/2007. 新著国语文法 (A New Chinese Grammar). Changsha: Hunan Education Publishing House. Search in Google Scholar
Li, Charles N., and Sandra A. Thompson. 1981. Mandarin Chinese: A Functional Reference Grammar. Berkeley: University of California Press. Search in Google Scholar
Liu, Min, Yiya Chen, and Niels O. Schiller. 2016. “Online processing of tone and intonation in Mandarin: evidence from ERPs.” Neuropsychologia 91: 307–17. Search in Google Scholar
Lu, Jianming. 1984. “关于现代汉语里的疑问语气词 (On interrogative markers in Modern Chinese).” 中国语文 (Studies of the Chinese Language) 5: 330–7. Search in Google Scholar
Lü, Shuxiang. 1999. 现代汉语八百词 (800 Words in Modern Chinese). Beijing: The Commercial Press. Search in Google Scholar
Mackenzie, J. Lachlan. 2009. “English fail to as a periphrastic negative: an FDG account.” Web Papers in Functional Grammar 82: 1–28. Search in Google Scholar
Olbertz, Hella. 2012. “The place of exclamatives and miratives in grammar: a functional discourse grammar view.” Linguística 8(1): 76–98. Search in Google Scholar
Olbertz, Hella, and Marize Mattos Dall’Aglio Hattnher. 2018. “On objective and subjective epistemic modality again: modal verbs.” In Recent developments in Functional Discourse Grammar (Studies in Language Companion Series 205), ed. Evelien Keizer, Hella Olbertz. Amsterdam: Benjamins. Search in Google Scholar
Olbertz, Hella, and Sandra Gasparini Bastos. 2013. “Objective and subjective deontic modality in FDG: evidence from Spanish auxiliary expressions.” In Casebook in functional discourse grammar (Studies in language companion series 137), ed. J. Lachlan Mackenzie, Hella Olbertz, 277–300. Amsterdam: John Benjamins. Search in Google Scholar
Olbertz, Hella, and Wim Honselaar. 2017. “The grammaticalization of Dutch moeten: modal and post-modal meanings.” In The Grammaticalization of Tense, Aspect, Modality and Evidentiality: A Functional Perspective (Trends in Linguistics. Studies and Monographs 311), ed. Kees Hengeveld, Heiko Narrog, Hella Olbertz, 273–300. Berlin: De Gruyter Mouton. Search in Google Scholar
Qi, Huyang, and Ming Zhu. 2005. “现代汉语祈使句句末语气词选择性研究 (On the selectivity of modal particles in imperative sentences in Modern Chinese).” 上海师范大学学报(哲学社科版) (Journal of Shanghai Normal University (Philosophy and Social Sciences)) 34(2): 62–67. Search in Google Scholar
Schneider, Stefan. 2010. “Mitigation.” Interpersonal Pragmatics, ed. M. A. Locher, S. L. Graham, 253–69. Berlin: De Gruyter. Search in Google Scholar
Siemund, Peter. 2015. “Exclamative clauses in English and their relevance for theories of clause types.” Studies in Language 39(3): 697–727. Search in Google Scholar
Tantucci, Vittorio. 2017. “An evolutionary approach to semasiological change: Overt influence attempts through the development of the Mandarin 吧-ba particle.” Journal of Pragmatics 120: 35–53. Search in Google Scholar
Thaler, Verena. 2012. “Mitigation as modification of illocutionary force.” Journal of Pragmatics 44(6–7): 907–19. Search in Google Scholar
Vismans, Roel. 1994. Modal Particles in Dutch Directives: A study in Functional Grammar. Amsterdam: IFOTT. Search in Google Scholar
Wang, Li. 1943/1985. 中国现代语法 (Contemporary Chinese Grammar). Beijing: The Commercial Press. Search in Google Scholar
Wiedenhof, Jeroen. 2015. A Grammar of Mandarin. Amsterdam/Philadelphia: John Benjamins Publishing Company. Search in Google Scholar
Yuan, Jiahong. 2006. “Mechanisms of Question Intonation in Mandarin.” In Chinese Spoken Language Processing (Lecture Notes in Computer Science 4274), ed. Q. Huo, B. Ma, E.S. Cheng, H. Li. Berlin/Heidelberg: Springer. Search in Google Scholar
Zhang, Bojiang. 1997. “认识观的语法表现 (The grammatical representations of epistemology).” 国外语言学 (Foreign Linguistics) 2: 15–19. Search in Google Scholar
Zhao, Chunli, and Li Sun. 2015. “句末助词“吧”的分布验证与语义提取 (On the distributional verification and semantic extraction of the sentence-final particle ba).” 中国语文 (Studies of the Chinese Language) 2: 121–32. Search in Google Scholar
Zhou, Shihong. 2009. ““吧” 的意义功能再议 (Grammatical meaning and discourse function of “ba”).” 语言教学与研究 (Language Teaching and Linguistic Studies) 2: 16–22. Search in Google Scholar
Zhu, Dexi. 1982. 语法讲义 (Notes on Grammar). Beijing: The Commercial Press. Search in Google Scholar
Zhu, Dexi. 1999. 朱德熙文集第一卷 (The Collected Works of Zhu Dexi. Vol. (1)). Beijing: The Commercial Press. Search in Google Scholar
© 2020 Hongmei Fang and Kees Hengeveld, published by De Gruyter
This work is licensed under the Creative Commons Attribution 4.0 International License.