Diachronic change of rapport orientation and sentence-periphery in Mandarin

Facework and rapport management

Over the last decades, research centred on modality has been increasingly concerned with ‘interactional, textual and rhetorical functions, such as persuading, manipulating, challenging, confronting, accepting, encouraging the flow of the conversation and creating cohesive texts’ (cf. Cornillie and Pietrandrea, 2012: 2109; for example, see also Englebretson, 2007; Simon-Vandenbergen and Aijmer, 2007; Tantucci, 2016d, 2017b). Somewhat surprisingly, to date very few studies addressed facework as an important variable of modalised propositions (e.g. Heritage, 2012 for an approach based on conversation analysis).

The concept of face is first considered by Goffman (1967) as the ‘the traffic rules of social interaction’ (p. 12). In Brown and Levinson’s (1987) influential research on (im-)politeness, face is addressed as a universal concern which can refer to two wants of the individual. A so-called positive face that necessitates approval by others and a negative face requiring one’s actions or thoughts to be unimpeded by others. Brown and Levinson’s (1987) approach has often been challenged for neglecting the interpersonal or social perspective on face, while placing too much emphasis on the Western ideal of individual freedom and autonomy. In turn, Matsumoto (1988), Ide (1989) and Mao (1994) all stress the importance of ‘social identity’ as a concept in Japanese and Chinese societies.

A second important matter of debate in facework studies regards the relationship between language use and identity during and beyond interaction (e.g. Culpeper, 2011; Spencer-Oatey and Ruhi, 2007). There is a strand of research where identity is indeed viewed as a by-product of interaction (Benwell and Stokoe, 2006; Hecht et al., 2005; Heritage, 2001: 48). In other cases, facework is tackled as a dimension that endures across interactions unless otherwise challenged (Spencer-Oatey, 2005: 102–103).

In this article, we will be looking at face from a usage-based perspective and a corpus-based framework. The theoretical foundation of our analysis will be consistent with a conceptualisation of face as interactional, necessarily involving ‘evaluation by others, which in turn presupposes that interaction has indeed taken place’ (Haugh and Bargiela-Chiappini, 2010: 2074). This approach draws on the so-called ‘co-constituting model of communication’ (Arundale, 2010), which places special emphasis on the relationship that is achieved interactionally between two or more personas, rather than a long-term person-centred construct such as Goffman’s (1955) claimed self-image/social identity, or Brown and Levinson’s (1987) social wants. We are thus interested in observing face when emerging ‘as a relational and interactional phenomenon that arises in every day talk/conduct’ (Arundale, 2010: 2079).

In Brown and Levinson’s (1987) framework, there is a clear stress on intentionality and the face-threatening potential of any speech act. A more multifaceted taxonomy is proposed by Spencer-Oatey (2008), as she suggests that people can hold four different types of rapport orientation:

1. Rapport enhancement orientation: a desire to strengthen or enhance harmonious relations between the interlocutors.

2. Rapport maintenance orientation: a desire to maintain or protect harmonious relations between the interlocutors.

3. Rapport neglect orientation: a lack of concern or interest in the quality of relations between the interlocutors (perhaps because of a focus on self).

4. Rapport challenge orientation: a desire to challenge or impair harmonious relations between the interlocutors.

(Spencer-Oatey, 2008: 32)

Despite not having been designed for usage-based purposes, Spencer-Oatey’s taxonomy can operationally inform an annotating scheme tackling S’s overt attempts to maintain, enhance or challenge his or her ongoing rapport with H.

We are aware of potential downsides of a corpus-based approach to facework and rapport management, as a distinctive focus on interaction cannot capture the long-term construing of personas’ face and their perception within a social group from an ethnomethodological angle (e.g. Samra-Fredericks, 2010). However, it is also worth noting that corpus-based analysis has the advantage of operationally tackling facework when it occurs as an overtly codified phenomenon, unveiling large-scale patterns of interactionally ‘overt’ rapport management. In addition, facework seen as overtly marked device can be statistically analysed and compared cross-culturally and diachronically. In fact, in the case studies from section ‘Change in the rapport orientation of Mandarin evaluations’, we look diachronically at rapport-management in written interaction where S evaluates some state of affairs. In particular, we pay attention to rapport-oriented elements that are added to the proposition as a surplus of meaning that is additional to a mere speech act of information transmission.

Sentence-periphery and rapport management

From a usage-based perspective, rapport maintenance (RM) can be often identified as expressed through pragmatic marking. Pragmatic markers (here-forth PMs) act as procedural instructions or ‘linguistic “road-signs” to intended meaning’ during linguistic exchanges (Hansen, 1998: 199; Waltereit, 2001). Clause periphery is widely acknowledged to be a formal diagnostic for identifying PMs, as they take scope over the whole clause and intersect with speech act-oriented modality and intersubjectivity (e.g. Narrog, 2012; Sweetser, 1990).

Traugott’s (2003) working definition of intersubjectivity felicitously captures the speaker/writer’s overt attempt to codify his or her awareness of his interactional rapport with the addressee as it regards ‘the locutionary agent’s expression of his or her awareness of the [addressee/reader]’s attitudes and beliefs, most specially their “face” or “self-image”’ (p. 128). Functions of language that are most likely to mark the speaker/writer’s attention to the intersubjective face of the interlocutors are often related to politeness and meta-discursive functions such as turn-giving, agreement-seeking or elicitation of response (Traugott, 2012: 10). Concerning the identification of clause-peripheral PMs, Traugott (2012) argues that all contextual variables being equal, when meta-discursive and peripheral ‘uptake by another interlocutor appears on a regular basis in [the same genre of] texts, then the marker is being used intersubjectively’ (p. 10). Research on peripheral usages of clearly and no doubt reveal a process of intersubjectification between the 16th and the 18th century (Traugott, 2012: 11). Turn-taking devices and question tags soliciting a response by the hearer also result from intersubjective reanalysis such is the case of clause final right? and is it not?, isn’t it? and similar ones (Tottie and Hoffmann, 2006). A process of intersubjectification has been similarly observed through sentence peripheral reanalysis of Mandarin connectives (Wang and Huang, 2006) and CFP (Rhee, 2012; Tantucci, 2017b; Tantucci and Wang, 2018).

Mandarin sentence-periphery

Spencer-Oatey’s notion of rapport maintenance (RM) clearly intersects with Traugottian intersubjectivity, as procedural pragmatic markers constitute a codified attempt to monitor the on-going interaction with the addressee and his or her potential reactions to the utterance. This interactional side of RM is compatible with Goffman’s original emphasis on the communicated awareness of face coded as a spontaneous mechanism with ‘involvement in the face of others that is as immediate and spontaneous as the involvement [the speaker] has in his or others’ face’ (Goffman, 1967:6). Overtly codified attempts of RM indicate when the speaker finds necessary to encode his or her awareness of the addressee as a distinctive effort or ‘surplus’ over mere propositional meaning and ‘politic’ behaviour (e.g. Culpeper, 2011; Gouldner, 1960; Kasper, 1990; Tantucci, 2016b, 2017a, 2017b, 2018; Tantucci et al., 2018; Tantucci and Wang, 2018; Watts, 2003), for example, Actually, I’m tired now vs I’m tired now (cf. Tantucci, 2017b; Traugott and Dasher, 2002 on the intersubjective functions of the discourse markers actually).

In Mandarin and most Sinitic Languages, clause-final particles (语气词 yǔqìcí) – here-forth CFP – correspond to a highly grammaticalised form of pragmatic markers, as they are inherently characterised by procedural function, highly intersubjectified modal meaning and peripheral usage (cf. Chappell and Peyarube, 2016). Typologically, modal particles are often considered as markers of evaluations (Doherty, 1987). As a grammaticalised sub-class of PMs, they have scope over the whole clause, they do not carry stress, they are not used to form sentences in isolation or cannot be coordinated (e.g. Hansen, 1998: 42–44; Waltereit, 2001).

Despite not being obligatory (cf. Bisang, 1996: 535 on the issue of non-obligatoriness in the Chinese grammatical system), Mandarin CFP are added as a procedural ‘surplus’ of meaning at the end of the clause to ‘facilitate conversational interaction and collaborative “production,” coding emotions as varied as surprise, exasperation, indignation, and impatience, not to mention conveying the desired or perceived role relationship between speaker and addressee’ (Chappell and Peyarube, 2016: 321). Consider the usage of 吧 ba and 啊 a in (2) as follows:

[referring to China Airlines]

(2) A: 它们更便宜。

  tāmen gèng piányì

  they more cheap

  ‘They are even cheaper’.

 B: 不会吧, 华航还是很贵啊。

  bùhuì ba, huáháng háishi hěn huì a

  not being-the-case BA, China Airlines still very expensive A

  ‘It can’t be come on, China Airlines are pretty expensive actually’.

(Adapted from Wu, 2004: 26)

The particle 吧 ba is used to mitigate B’s disagreement with A. 吧 ba is often ‘used to code suggestions’ (Chappell and Peyarube, 2016: 323) or invite H to take part to a physical or ‘epistemic’ co-action (Tantucci, 2017b), namely, engaging the addressee in a shared activity or shared belief. The intersubjective nature of 吧 ba depends on its function to check or confirm that the addressee accepts the validity of the given proposition and is often rendered with tag-questions in English, for example, don’t you think so? or wouldn’t you agree? (cf. Li and Thompson, 1981: 307). Similarly, the CFP 啊 a (or 呀 ya) is characterised by ‘a hortatory use in prompting or urging the addressee to carry out the action desired by the speaker’ (cf. Chappell and Peyarube, 2016: 323). When it is employed epistemically, 啊 a emphasises S’s subjective certainty (e.g. Xu, 2007) while expecting the addressee’s acknowledgement of the state of affairs of p.

Both statements uttered by B in (2) are epistemically in opposition with what is said by A. Despite the difference in meaning between 吧 ba and 啊 a, in both cases there is a marked effort to acknowledge the addressee’s potential reactions to his or her disagreement and the pre-emptive attempt to ‘save’ their interactional rapport. Put simply, in (2B) there is an overt intention to ‘protect harmonious relations between the interlocutors’ (Spencer-Oatey, 2008: 32), which in Mandarin is often grammatically (albeit not obligatorily) encoded in clause final position.

There is an evident mismatch between particles in modern and pre-modern and classical literary Chinese, 文言文 wényánwén. The latter used to be the register for almost all formal writing in China until the early 20th century. As an illustration, 矣 yǐ and 也 yě are two most frequently used CFP in Classical Chinese. The aspectual meaning of former is roughly equivalent to the CFP indicating current relevance 了 le in Modern Mandarin (cf. Edwin and Pulleyblank, 2010: 166), while at the modal level it often underpins the speaker’s epistemic reasoning and conjecturing (Xu, 2002: 193). On the contrary, 也 yě in dialogic contexts expresses confirmation of some state of affairs, with a similar usage as clause final 啊 a (Xu, 2002: 183). In turn, it is often acknowledged that Modern Mandarin CFP such as 呢 ne and 吧 ba do not have any comparable particle in Classical Chinese (e.g. Guo et al., 1999, Lü, 2002; Smith, 1991; Wei, 2015).

As it will be discussed in section ‘Data preparation and annotation’, our annotation scheme includes a categorical variable referring to CFP appearing at the end of the clause. The presence of CFP in dialogical evaluations is an overt indicator of the speaker’s RM intentions, as she or he overtly opts for a marked ‘surplus’ of meaning (CFP are not obligatory), being expressed specifically to acknowledge the addressee’s potential reactions to the utterance.

Data retrieval and normalisation

We retrieved our data from the Peking diachronic corpus of Mandarin Chinese CCL. We specifically looked at subcorpora encompassing 清 Qīng dynasty (1644–1911), 民国 Mínguó period (1911–1949) and the 现代 Xiàndài (Modern) period (1949-present). We centred our enquiry on the fictional section of each subcorpus due to two reasons. First, the fictional section is the only one that is constant across the three subcorpora. Second, as we aimed at capturing whether and how rapport orientation changed from the pre-modern to the modern era, the fictional section provides a controlled environment of dialogic exchanges that could be representative of the formal and pragmatic features at stake in each sub-period. Crucially, data from trials, plays, conversation in novels and letters is representative of language relatively close to speech, and constitutes a precious resource for diachronic investigation (cf. Culpeper and Kytö, 2010; Tantucci, 2017a, 2017b). Written language is widely acknowledged to be an important window to investigate language change (Traugott and Dasher, 2002: 46) as ‘text provides a mode of speech’ (Olson, 1994: xviii). Diachronies of words and constructions display well-attested reflexes in contemporary spoken data (Biber, 1988) and gradient changes of meaning that can be attested in diachronic re-analysis unveil sedimentation of spontaneous innovation (cf. Traugott and Trousdale, 2013).

All collocates that have been included in our survey are corpus-driven and correspond to evaluative utterances including a modal auxiliary. We first retrieved a random sample of 200 collocates based on the three most frequent auxiliaries occurring in each period, we then annotated each usage by looking at nine variables: period, modal auxiliary, presence of speaking subject, presence of syntactic subject, presence of CFP, type of CFP, modal meaning of the auxiliary, rapport and propositional face (i.e. whether S’s evaluation targeted H’s persona). All occurrences from our data set have been through manual double-blind annotation (two annotators), with a matching rate of 86% (see 5.2 for an illustration of our annotation scheme). All remaining collocates have been disambiguated after a third round of annotation from a third annotator.

At this point, we normalised our observations based on the size of the fictional section of each corpus, respectively, 45,229,510 words for the Qing dynasty (1644–1911), 35,371,339 words for the Minguo Period (1911–1949) and 14,052,591 words for the Modern period (1949-present). We calculated the per-million-word rapport of each set of observations based on the size of each fictional sub-section, respectively, 4.4, 5.6 and 14.2. We then multiplied our observations per each rapport minus one (i.e. the actual sample): 3.4, 4.6 and 13.2. This ‘post-annotation’ method of normalisation allowed us to control the proportion and the inner relationship among the nine variables from each sample by also taking into account the size of each period of the CCL. Normalised data have been used for association tests in section ‘Sentence-periphery and increased subject-hood marking’.

Finally, it is important to acknowledge that from a strictly historical pragmatic angle, it cannot be taken for granted that past evaluations can be analysed with the same accuracy as contemporary evaluations. Temporally and contextually situated socio-pragmatic mismatches are always at stake in the analysis of historical data. At the same time, research on semasiological change is an example where illocutional modification and invited inferencing trigger semantic reanalysis (e.g. Brinton, 2017; Traugott, 2012, 2016; Traugott and Dasher, 2002). While this project is centred on speech acts, rather than specific lexemes, the underlying methodology is fairly similar, as the manual annotation of all the evaluations in our data set has taken into account information about immediate context and the nature of each text where evaluations were realised.

Annotation and usage-based operationalisation of the criteria

We opted for a two layered annotation model to analyse rapport orientation through dialogic interaction in each fictional subcorpus. The first layer underpins rapport, that is, whether S aims at maintaining (RM), enhancing (RE), challenging (RC), or whether she or he simply neglects (RN) his or her rapport with the addressee while making an evaluation.

The second layer regards whether the speaker overtly expresses a positive or negative evaluation of the addressee. We define this phenomenon as propositional facework (Tantucci and Wang, 2018), in order to shed light on whether there is a diachronic increase of cases where S overtly says what she or he thinks about H, either in the form of a propositional face-threatening act (pFTA) or propositional face-enhancing act (pFEA). Our corpus-driven annotation scheme was based on the following usage-based taxonomy, specifically intending to capture overt signs of rapport-management in interaction.

This approach aligns with a functional strand of corpus-based approaches to (im-)politeness (e.g. Culpeper and Archer, 2008; Del Lungo Camiciotti, 2008; Honegger, 2000; Jucker, 2000). It is thus important to emphasise the usage-based nature of this study, as our annotation depends on interactionally marked elements informed by the taxonomy in Table 1. This approach is not based on ‘interpreting’ the feelings of the single interactants, but rather on identifying overtly marked signs of rapport management. We do not aim at uncovering individuals’ specific understanding of each other’s ‘face’ as such throughout interaction, as this would reach far beyond the scope and the empirical testing of this project. Rather, we are interested in operationalising interactional ways in which elements of rapport orientation overtly emerge from spontaneous speech acts of evaluation in temporally and culturally situated contexts of fictional interaction.

OPEN IN VIEWER

Table 1.

Criteria for the usage-based identification of six different layers of rapport management.

Tag	Rapport-orientation
RE	S makes the attempt to improve his or her rapport with H. She or he says something that is advantageous for H (e.g. proposes something that may be beneficial for him or her: ‘you could do p, p would be good for you’).
RE + pFEA	S overtly says something that boosts H personal/social image (e.g. appraisals, positive comments).
RM	S overtly codifies his or her awareness of H’s potential reactions to the utterance. This intersects formally with presence of clause-periphery intersubjective markers (PMs) and peripheral periphrastic formulae.
RN	S makes an evaluation without any overt element of RE or RM.
RC	S utters something that is disadvantageous to H (e.g. gives an overt order, accuses or exerts some form of power over H).
RC + pFTA	S overtly says something about H that downgrades his or her personal/social image.

RE: rapport enhancing; RM: rapport maintenance; RN: rapport neglecting; RC: rapport challenging; PM: pragmatic marker.

An operational annotation of rapport management and propositional facework in interaction

This section illustrates the criteria for the annotation scheme in section ‘Introduction’. Importantly, our annotation is based on contextually informed dialogic exchanges, and controls for book types in which they occur (all comparable fictional works meant to be representative of each period) as well as large contextual spans where evaluations take place. Labels of each category are given in chevron (<>) at the top of each example (or group of examples). In the following are first given two cases of rapport enhancement RE:

<RE>

(3) ‘我家只有老母一人, 你若不能携带妻子同去赴任, 可以让她先住在我家,

 wǒ jiā zhǐ yǒu lǎomǔ yī rén, nǐ ruò bù néng xiédài qīzi tóngqù fùrèn,

 kěyǐ ràng tā xiān zhùzài wǒ jiā

 ‘My home only there-is old-mother one person, you if cannot take-along  

 wife together go to-duty,can let her first stay my home

 等您上任后安排好了再来接她。’

 děng nín shàngrèn hòu ānpái hǎo le zài lái jiē tā

 wait you be-at-office organise then come to pick-up her.

 ‘There is just my mother at home, if you cannot take your wife to your post, you can let her

 stay in my home. Once you will be done, you can come and pick her up’.

CCL/Minguo (1911–1949)/Gujin Qinghai

<RE>

(4) ‘那东西是我拿的, 那东西！我不能连累你们, 我去自首！我去自首！’

 nà dōngxi shì wǒ náde, nà dōngxi！wǒ bù néng liánlèi nǐmen, wǒ qù zìshǒu! wǒ qùzìshǒu

 that thing is I take DE, that thing! Wo cannot involve you, I go give-up I go give-up

 ‘It was me the one who took that thing! That thing! I cannot get you into trouble, I am going

 to give myself up, I am going!’

CCL/Xiandai (1949-present)/Xiayan

In (3) above, S overtly makes an attempt to enhance his or her rapport with H: she or he makes an evaluation with 可以 kěyǐ ‘can, be-allowed’ to make a suggestion that is meant to be beneficial to H. Something similar occurs with 不能 bù néng ‘cannot’ in (4) whereby S evaluates a possibility that would ultimately enhance his or her rapport with H. In our scheme, RE utterances are valid when they are potentially compatible with following evaluations stressing a positive conditional outcome from H’s perspective, for example, 这对你会有帮助/好处 zhè duì nǐ huì yǒu bāngzhù hǎochù ‘that would be good for you’:

<RE+pFEA>

(5) ‘你真会用香水, 闻起来 [. . .] 这么清淡, 而又这么幽远。’

 nǐ zhēn huì yòng xiāngshuǐ, wénqǐlai zhème qīngdàn ér zhème yōuyuǎn

 ni really can use perfume, smell-start so mild, and also so remote

 ‘You really know how to choose your perfume, it smells so fresh and distant at the same

 time’.

CCL/Xiandai (1949-present)/Caoyu Richu

<RE+pFEA>

(6) ‘可是最能唱这歌的, 不是咱们的密斯杨吗？’

 kěshì zuì néng chàng zhè gē de, bù shì zánmen de mìsīyáng ma？

 however most can sing this song DE, not is out DE Mi Siyang MA

 ‘However, who would disagree that Mi Siyang is the one who can sing best this song?’

CCL/Xiandai (1949-present)/Xiayan

In (5) and (6) above, S enhances the personal/social image of H as she or he positively evaluates and praises his or her skills, intellectual abilities or emotional/moral qualities at the propositional level. In our data set, such usages are given as both RE and pFTA. They can be identified when they are potentially compatible with subsequent evaluations further boosting H’s personal/social image, for example, you are pretty good, 你很棒/厉害/好 nǐ hěn bàng/lìhai/hǎo ‘you are amazing/a good person’ and the like. They differ from bare RE (e.g. (3) and (4) above), as they are not compatible with comments entailing an advantageous outcome for H, such as * that would be good for you ³ :

<RM>

(7) ‘今文字之厄若此, 谁复能漠然哉！’

 jīn wénzài zhī è ruò cǐ, shéi fù néng mòrán zāi

 today script ZHI disaster as-such, who again can indifferent ZAI

 ‘How can we again ignore the disastrous state of today’s writing!’

CCL/Qing (1644–1911)/Liaozhai Zhiyi

<RM>

(8) 医生：‘嗯？啊唷！这个怎么能吃呢？赶快拿去倒掉它。’

 yīsheng：‘“En? Ayo!” zhè ge zenme néng chī ne？gǎnkuài ná qù dàodiào tā’

 Doctor：‘EN? AYO!’ This CLASS how can eat NE? Quickly take go throw-away it

 ‘The doctor: “Mm? Oh no! How can this be eaten come on, throw that away

 immediately”’.

CCL/Xiandai (1949-present)/Xiayan

While making an evaluation in (7), S employs the grammaticalised exclamative CFP 哉 zāi (cf. Xu, 2002: 1999) to make overt his or her intention to account for H’s stance while expecting him or her to finally agree with p. Similarly, in (8) despite the presence of a rhetorical question, S conveys a clear evaluation about what H is about to eat. The evaluative force of the utterance is mitigated with the CFP 呢 ne, which is employed as intersubjective marker to invite H to take part to the evaluation. In both (7) and (8), S does not merely qualify the evaluation modally, but also overtly conveys awareness of his or her on-going rapport with H with the clause peripheral PMs 哉 zāi and 呢 ne, through which she or he negotiates the ‘common sense’ of his or her statement (see Tantucci, 2013, 2017b, 2018 about immediate vs extended construals of intersubjectivity; cf. Tree and Schrock, 1999, 2002; for detailed accounts of the interactional function of intersubjective PMs in conversation). Evaluative utterances that we labelled as RM formally include CFPs as peripheral PMs of intersubjectivity. They are also not compatible with ensuing comments impinging on either H’s advantageous conditions, or to his or her social/personal image, such is the case of, respectively, RE ((3) and (4)) and RE + pFEA in ((5) and (6)):

<RN>

(9)  ‘黑夜之间, 焉能一战成功。’

 hēiyè zhījiān, yān néng yīzhàn chénggōng

 night between, how can one battle sucess

 ‘There is no way we can win a battle in one night’.

CCL/Qing (1644–1911)/Qijia Wuyi

<RN>

(10) ‘咱们立刻起程, 不要伤心。’

 zánmen lìkè qǐchéng, bù yào shāngxīn

 we immediately set-out, not must sad.

 ‘Let’s leave immediately, don’t be sad’.

CCL/Minguo (1911–1949)/Gujin Qinghai

All the modal evaluations that do not include any overt codification of H as an interactional persona have been marked as RN. Such usages occur without conventionalised CFPs of intersubjectivity, nor they overtly express any evaluation that is somewhat connected to H’s personal or social image. Cases of RN are given in (9) and (10):

<RC>

(11) ‘你给我搬出去, 我这个屋子不能让你住！’

 nǐ gěi wǒ bānchūqù, wǒ zhè ge wūzi bù néng ràng nǐ zhù

 you to me move-out, I this CLAS house not can let you live

 ‘Move out, I cannot let you stay in this house anymore!’

CCL/Xiandai (1949-present)/Caoyu Richu

<RC>

(12) 打手们：‘对, 我们得见见八爷。’

 dǎshòumen：‘duì, wǒmen děi jiàn jiàn bā yé’

 goons: ‘yes, we must see see Ba master’

 ‘The goons: “yes, we must go and pay a visit to master Ba”’.

 陈白露：‘不成, 你不能见。’

 chén báilù：‘bùchéng, nǐ bùnéng jiàn’

 Chen Bailu: ‘no work, you no can see’

 ‘Chen Bailu: “No, you cannot see him”’.

CCL/Xiandai (1949-present)/Caoyu Richu

We coded as RC all cases where S makes an evaluation that may have a negative impact on his or her rapport with H. As an illustration, in (11) the deontic usage of 不能 bùnéng ‘you cannot’ clearly intersects with the attempt to impair the on-going rapport with H. Similarly, in (12) Chen Bailu overtly rejects a request made by H. He makes an evaluation (你不能见 nǐ bùnéng jiàn ‘you cannot see’) that directly affects H’s negative face and inhibits his or her freedom of action and autonomy. Utterances labelled as RC are compatible with preventive apologies or ad hoc formulae preparing H to hear something more or less discomforting such as, frankly, I am sorry but p, I have to tell you that p, 对不起, 可是 duì bù qǐ, kěshì ‘I am sorry, but’, 我跟你说 wǒ gēn nǐ shuō ‘let me tell you’. This type of mitigating formulae (which can also be anaphoric) are distinctive of RC usages as they are clearly not compatible with REs, RE + pFEAs or bare RMs:

<RC+pFTA>

(13) ‘我与你没有缘分, 不要靠近我。’

 wǒ yú nǐ méiyǒu yuánfèn, bù yào kàojìn wǒ.

 I and you not have destiny, not must be-close me

 ‘You and I are not meant to be together, you should’t get close to me’.

CCL/Minguo (1911–1949)/Gujin Qinghai

<RC+pFTA>

(14) 潘月亭：（恼怒）‘你怎么能把我的信拆开!’

 pān yuè tíng: (nǎonù) ‘nǐ zénme néng bǎ wǒ de xìn chāikāi！’

 Pan Yueting: (upset) ‘you how can BA I DE letter take apart’

 ‘Pan Yueting: (with an upset tone) “How could you unseal my letter!”’

CCL/Xiandai (1949-present)/Caoyu Richu 

In (13–14) above are finally given two cases of RC + pFTA, as S’s evaluation directly targets and challenges H’s face at the propositional level.

All in all, most of the functions discussed in this section are marked for (im-)politeness, as the speaker overtly intervenes on his or her on-going rapport with the addressee. The only cases where (im-)politeness is not overtly at stake are utterances of rapport neglecting (RN). From this taxonomy, (im-)politeness intersects with evaluative speech acts as a gradient dimension, which can range from overt appraisals (RE + pFEA) or criticisms (RC + pFTA) of the addressee’s persona, to evaluations aimed at boosting (RE) or challenging (RC) S and H rapport, to finally cases of overt orientation to monitor the harmonious interaction among interlocutors (RM). Our annotation aims at unveiling IC, emerging from the intersection of formal and pragmatic variables that contribute to the encoding of evaluations in different periods of written Mandarin. With this in mind, beyond the above classification of rapport-management we also took into account:

All the columns of our spreadsheet are illustrated as follows, with one sampled collocate of could (corresponding to (11) above) out of all the 600 annotated evaluative occurrences (Table 2).

OPEN IN VIEWER

Table 2.

Sampled row of annotation from our data set.

lexeme	speak_subj	synt_subj	sent_p_PM	PM	polarity	modal_m	rapport	p_facework
neng	yes	yes	no	absent	neg	deo	RC	absent

RC: rapport challenging; PM: pragmatic marker; sent_p_PM: pragmatic marker occurring at sentence periphery; modal_m : modal marker; p_facework: propositional facework.

All collocates that were not evaluative speech acts were manually excluded from our samples. Missing observations of any of the six lexemes we queried were then replaced from a randomised sample from our annotated data set, with a final spreadsheet counting 200 observations for each lexeme in each period.

Rapport and CFPs

One important hypothesis of this study was the strong relationship between CFPs and significant changes of interactional orientation involving rapport management. We thus started by looking holistically at the similarity of evaluations across the three periods by focusing on CFPs as a dependent variable. We plotted a hierarchical clustering model (Steinbach et al., 2000) of our annotated samples with the aim of unveiling significant classification of usages based on similarity of distribution among all the variables that we listed in section ‘An operational annotation of rapport management and propositional facework in interaction’. In cognitive linguistics, this method is called behavioural profile analysis (BPA) and relies on multivariate exploratory statistics to differentiate senses and forms of a lexeme (cf. Gries, 2010; Jansegers and Gries, 2017: 3).

The left-hand side plot above is called a dendrogram. Each object (which in our case is an evaluative utterance type) represents its own cluster, or a ‘leaf’. Subsequently, most similar objects (the ones for which the distance between the objects is the smallest) are merged. This procedure is then repeated until all leaves and branches are merged into one tree-like representation, leading to each CFP appearing at the bottom of the plot (i.e. the left-hand side of Figure 1). The greater the height of each cluster, the greater the behavioural distance among particles, thus the stronger the difference in terms of formal and pragmatic features, for example, rapport, presence of syntactic/speaking subject, polarity, modal meaning and so on. In red is given the optimal number of significant clusters (two in this case) that determine the behavioural classification of CFP across the tree periods: Qing dynasty (1644–1911), Minguo (1911–1949) and Xiandai (1949-present). The second plot at the right-hand side of Figure 1 provides the same results in a more intuitive visualisation, with differences among the two main clusters that are also represented by distance.

OPEN IN VIEWER

Figure 1.

Clustering based on formal and pragmatic similarity of CFP.

From the above, we can immediately notice a strong differentiation between one cluster including comparatively older particles 耶 yé, 哉 zāi, 耳 ěr, 也 yě, 矣 yǐ, 乎 hū, 焉 yān and a second cluster combining absence of particles together with newer ones 啦 la, 的 de, 喏 nuò, 啊 a, 呐 na, 了 le, 呢 ne, 嘛 ma, 吧 ba. Intuitively, this partition reflects evaluative usages marked at sentence periphery, respectively, before and after the 20th century, with a substantial replacement and reduced frequency of CFP (as absent is attracted to the ‘recent’ cluster). The inner relationship among variables leading to the classifications above, can be unveiled with a so-called ‘snake-plot’, which represents the effect-size differences between the average values in both clusters (cf. Levshina, 2015: 313) and positions them in relation to either the first or the second cluster.

Figure 2 above includes the behavioural profiles of all the annotated evaluations of our data set, which, to different degrees, are, respectively, more attracted to cluster 1 (right-hand side) or cluster 2 (left-hand side). It is based on the same multifactorial calculation of correspondences among variables that we see in Figure 1. Yet, in this case we can see which covariants concur with one another to the partition of the two clusters emerging from Figure 1, for example, which form of rapport orientation tends to occur in which period in connection with which particle, with which modal meaning and so on. As for Figure 1, the closer the distance among covariants, the closer the relationship among them.

OPEN IN VIEWER

Figure 2.

Inner formal and pragmatic variables contributing to the classification of CFP.

At the two poles of the plot appear, respectively, the Xiandai period (1949–present) at the bottom-left corner, and the Qing dynasty (1644–1912) at the top-right one. Evaluations of the Xiandai period are closely associated with utterances (marked in red) oriented towards RE, face-enhancing propositional face (p_face_FEA), RN and RC. This entails a fundamental result:

Evaluative speech acts during the Modern period tend to be comparatively more marked for (im-)politeness, with compositional linguistic acts more overtly impairing or strengthening the harmonious interaction with the addressee.

Not surprisingly, all these profiles are also closely associated with the Minguo period (1911–1949), still appearing at the bottom-left part of the plot and thus sharing similar forms of evaluations as for the Xiandai stage. Quite differently, the Qing dynasty (1644–1912) – at the top right corner – is strongly associated with more fixed RM orientation and absence of evaluations impinging on propositional face (p_face.absent). This indicates another crucial finding:

Evaluations from the Qing period tend to be more structurally fixed, thus marked with clause final particles (CFP) at sentence periphery and with a preference for more conventional and idiomatic formulae to maintain the harmonious relations with the addressee.

Evaluations during the modern period shift towards absence of CFP is qualitatively exemplified below. In (15), the modalised evaluation is from the Qing dynasty and is structurally marked with clause-final 也 yě, occurring as a fixed RM device. As Xu (2002: 183) points out, the clause-final 也 yě that characterises the written Chinese language in use from the end of the Han Dynasty (220 AD) to the early 20th century includes a comparable range of usages as to the clause-final 啊 a (see example (2). In (16), 不可能 bù kěnéng ‘it cannot be possible’ is unmarked at sentence periphery and occurs as a rapport neglecting (RN) utterance, namely, without overt coding of the addressee’s potential reaction to the evaluation.

<RM>

(15)  安老爷道：‘是不能也, 非不许也’

  ān lǎoye dào：‘shì bù néng yě, fēi bù xǔ yě’

  An master say: ‘is not can YE, not not allowed YE’

  ‘An master said: “In fact, this simply cannot be done, it is not a matter of permission”’.

CCL/Qing (1644–1911)/Xianü Qiyuan

<RN>

(16) A: 大哥, 芳蜜来了！

  dàgē, fāngmì lái le

  brother, Fangmi arrive now

 B: 徐芳蜜, 那不可能, 不可能！

  xúfāngmì, nà bù kěnéng, bùkěnéng

  Xu Fangmi, that not is-possible, not is-possible

  ‘A: Brother, Fangmi is here!’

  ‘B: Xu Fangmi? That can’t be possible, can’t be possible!’

 CCL/Xiandai (1911-present)/Laoshe Xiju

IC of rapport and sentence periphery

The interaction between the change of CFP and rapport orientation from the Qing dynasty to Minguo (1911–1949) and Xiandai (1949–present) periods can be captured on a two-dimensional space with a multiple correspondence analysis (MCA; Nenadic and Greenacre, 2007). The latter allows to model associations among variables by calculating the chi-square distance between different categories of the variables and between observations. These associations are then represented graphically as a map, which eases the interpretation of the structures in the data, the closer the distance between variables, the stronger the statistical correspondence.

In the plot above the two dimensions represent 84.1% of variation, which is considered a good approximation for MCA visualisation. The strong separation of particles before and after the 20th century is quite striking. In fact, it is possible to notice how the same particles (given in blue) – as well as absence of any of them – from the same cluster in Figure 1, all appear around the two modern periods Minguo and Xiandai (in red) at the left-hand side of the map. Similarly, we can find confirmation of the strong correspondence of the second cluster of comparatively older particles with the Qing dynasty (Dim1: 2.2; Dim2: 0.2), bottom right. What is also most revealing is the distribution of rapport orientation (in green), with less compositional and less structurally fixed utterances, all closely associated with Minguo and Xiandai periods: RC, RE and RN. Conversely, the Qing dynasty has a strong correspondence with RM usages.

Temporal- and context-bound intersections of formal and illocutionary dimensions of this kind are what we call IC. IC encompass converging factors at various levels of verbal experience that contribute both locally (i.e. at the morphosyntactic level) and peripherally (i.e. at the illocutionary level) to the encoding of contextually and temporally situated speech acts (e.g. Mey, 2001; Tantucci, 2016d). In the case of our study, we have been focusing on IC of contextually and temporally situated evaluations and obtained a holistic representation of how the rapport management of evaluations have been changing in Mandarin written interaction. Most relevant IC from the Qing period have been the intersection of RM with presence of structurally coded CFP at sentence periphery. This a fundamental result, as it unveils how facework as a by-product of language in use (e.g. Arundale, 2010; Haugh and Bargiela-Chiappini, 2010: 2074) is formally and pragmatically bound to interactional conventions that may change or decrease diachronically. One specific IC regards the close relationship during the 20th century between 吧 ba and RE orientation, such is the case of (17):

<RE>

(17) ‘大概很辛苦了吧？那你可以去休息一会吧！’

 dàgài hěn xīnkǔ le ba？na nǐ kěyǐ qù xiūxi yì huì ba

 broadly very hard-working LE BA? then you can go rest a-bit BA

 ‘You have been working quite hard isn’t it? Then you can definitely go to rest a bit!’

 CCL/Xiandai (1911-present)/Shijiamounizhuan

As illustrated in (17), 吧 ba is most attracted to evaluations that boost the rapport among the interlocutors, namely, S suggesting H to rest. As discussed in section ‘An operational annotation of rapport management and propositional facework in interaction’, RE utterances further evaluations stressing a positive conditional outcome from H’s perspective, for example, 这对你会有帮助/好处 zhè duì nǐ huì yǒu bāngzhù hǎochù ‘that would be good for you’. During the same period, we can also notice one IC from Figure 3 underpinning absence of CFPs and RN utterances, see (18):

<RN>

(18) ‘咱们应该看到一个更大更好的家庭。’

 zánmen yīnggāi kàndào yī ge gèng dà gèng hǎo de jiātǐng

 we shoudl see one CLAS more big more good DE household

 ‘We should look at a bigger and nicer household’.

 CCL/Xiandai (1911-present)/Laoshe Xiju

OPEN IN VIEWER

Figure 3.

Two-dimensional correspondence of rapport, CFP and periods.

RN are utterances where the rapport with the addressee is not overtly taken into account, this is made evident by the absence of CFPs as well as interactional forms that aim at overly challenging or enhancing the rapport with the addressee.

All in all, it is possible to conclude that evaluative language has been changing significantly in written Mandarin, with fictional interaction originally being more structured and marked at sentence periphery. Quite differently, 20th-century written interaction has become less compositional, less mitigated at the structural level and more marked for (im-)politeness, with increased usages of both RE and RC evaluations.

Sentence-periphery and increased subject-hood marking

Albeit modern Mandarin is still characterised by frequent employment CFPs (e.g. Chappell and Peyarube, 2016), it should now be clear how evaluations in the written language have become increasingly less structurally marked at sentence periphery. The two plots in Figure 4 account for the normalised frequency from our annotation (see section ‘Data retrieval and normalisation’) of presence versus absence of CFP in evaluative speech acts. The bar plot on the left-hand side provides the normalised frequencies of CFP before and after the 20th century. The visualisation on the right shows the Pearson residuals based on the chi-square difference between observed and predicted frequencies, with a highly significant mismatch (<2.22e–16) between the two periods. Namely, the darker the blue colour, the more significant the ‘positive’ presence of CFP, while increasingly dark red bars indicate significant absence of it (cf. Tantucci and Di Cristofaro, 2019 on Pearson residuals and diachronic data).

OPEN IN VIEWER

Figure 4.

Bar-plots and significance of the decrease of CFP after 1911.

From Figure 4 above, it is possible to note a significant decrease of CFP after 1911, as process of ‘de-peripherization’, which indeed support the main points of our discussion.

Another crucial tendency encompassing the transition from the Qing dynasty to the 20th century has to do with subject-hood, as both syntactic and speaking subject (e.g. 我 wǒ ‘I’, 我们 wǒmen ‘we’) ² have become increasingly frequent after 1911. Figures 5 and 6, respectively, indicate a significant increase of the overt encoding of the subject in the evaluative language during the 20th century. Particularly revealing is the bar-plot in Figure 6, showing how during the Qing dynasty speaking subject (the evaluator) used to be structurally almost absent from evaluative utterances, while it then increased significantly during the 20th century.

OPEN IN VIEWER

Figure 5.

Bar-plots and significance of the increase of syntactic-subjects after 1911.

OPEN IN VIEWER

Figure 6.

Bar-plots and significance of the increase of speaking-subjects after 1911.

Crucially, we need to bear in mind that all the changes that we have discussed in these last sections (i.e. evaluations increasing marked for (im-)politeness, substantial change of sentence periphery marking, increased syntactic and speaking subject-hood marking) are both the result of a natural development of the Chinese language, together with socio-cultural elements of prescriptivism that contributed to an abrupt transition of form and usage in between the end of the Qing dynasty and the Minguo period. The most important of those is probably the 五四运动 wǔsìyùndòng ‘May the 4th movement’.

The ‘May the 4th movement’

A final yet fundamental note thus needs to be made about the role of prescriptivism in the change of the Chinese written language and the pragmatics of evaluations represented in fictional texts.

Before the 1911, the official style of the written language in China used to be the so-called literary Chinese (文言文 wényánwén). It corresponds to the written Chinese language in use from the end of the Han Dynasty (220 CE) to the early 20th century. Due to its official register, the written style of 文言文 wényánwén remained prescriptively stable over the years, thus increasingly diverging from the natural development of the spoken language (Pulleyblank, 1995). Our data from the Qing period are mainly from fictional texts produced during the so-called ‘literati era’ from 1723 to 1840, which fell within the Qing dynasty (1644–1912). These narratives can be divided into classical-language fictions and vernacular fictions, both of which exhibit hybrid features in terms of language and narration (Wang, 2010). The 文言文 wényánwén was classically regarded as the socially more ‘important’ language and by the late imperial period it ‘was considerably different from colloquial (e.g. it applied a monosyllabic lexicon in contrast with the colloquial polysyllabic one), while the vernacular imitated spoken style and thus was closer to colloquial’ (cf. Pan and Kádár, 2011: 26).

Abrupt formal and pragmatic changes in the written language in between the end of the 19th and the beginning of the 20th century (sections ‘Rapport and CFP’, ‘IC of rapport and sentence periphery’ and ‘Sentence-periphery and increased subject-hood marking’) can be probably traced back to the 五四运动 wǔsìyùndòng ‘May the 4th movement’. This was a political, cultural and anti-imperialist movement that affected the Chinese system of values, together with the language use per se. Students and scholars demanded political and cultural reforms inspired by Western-style democracy. This led to the birth of a new anti-traditionalist intellectual class that criticised core elements of traditional Chinese culture and the Confucian ideology. Intellectuals from ‘May the 4th Movement’ agreed that 文言文 wényánwén style was a ‘dead language’ and claimed that literature should now be written in vernacular Chinese. The famous writer Hu Shi (1891–1962) was the ideological father of this literary revolution and introduced the terminology 白话文 báihuàwén ‘written vernacular Chinese’ to address the new writing style born from the 五四运动 wǔsìyùndòng. He gave the guidelines to create a new form of literature aiming to avoid classical allusions, discard stale and outworn literary phrases, and adopt vernacular words and expressions (Wang, 2010) together with major orthographic innovations, such as the standardisation of the use of de particles in 1920 (Airaksinen, 2014; Dluhošová, 2008; Gunn, 1991; Latham, 2007).