Love as a Term of Address in British English: Micro-diachronic Variation

This paper explores the pragmatic scope of the endearment ‘love’ in contemporary spoken British English. It will be suggested that the function of ‘love’ in interaction can be understood as a ritual framing expression that enables speakers to index certain interpersonal constellations and action contexts in which speakers claim rights and social authority by couching them in affective stance displays. The study is based on the 1994 and 2014 versions of the British National Corpus. The findings show that over the course of twenty years, the use of ‘love’ has become significantly less frequent and has undergone a functional profile shift to index, more centrally than before, other-deprecating evaluation, enacted through joking and performative use in storytelling. Those functions appear to feed off the core semantics and interpersonal constellations of ‘love’ as well as associations with social and linguistic stereotypes.


Introduction
This paper aims to explore the pragmatic scope of the endearment 'love' in contemporary British English. The analysis follows a micro-diachronic, corpusbased, contrastive pragmatic approach to trace shifts in usage between the late 20th and early 21st century. It will be suggested that the function of 'love' in spoken interaction can be understood as a ritual frame indicating expression (RFIE) (Kádár and House, 2020) that enables speakers to index certain interpersonal constellations and action contexts (Levinson, 2013) in which speakers claim rights and social authority by couching them in affective stance displays. 'Love' as an address term with a ritual framing function is the result of an ongoing micro-historical process of enregisterment, whereby "behavioural signs (whether linguistic, non-linguistic or both) are functionally reanalysed as cultural models of action, as behaviours capable of indexing stereotypic characteristics of incumbents of particular interactional roles and relations among them" (Agha, 2007: 55). In ritual communicative contexts ('standard situations' , Kádár and House, 2020), i.e. contexts in which specific behavioural rules apply and participants' rights and obligations are clearly defined, RFIEs, such as politeness markers (e.g. 'please') or honorifics, are used to mark that the interaction is to be understood as de-individualised. That is, they are used to index the situation and a contingent participant role relationship rather than being an expression of specific, ad-hoc interpersonal meaning between speaker and addressee. On this view, 'love' as a RFIE would index a ritual frame for the business at hand for the addressee (and other co-participants and overhearers) that includes a routine expression of an affective stance towards the addressee, which is verbalised because it is conventionally linked to the particular social identity or role claimed by the speaker in the situation (Ochs, 1996), and not necessarily because it corresponds to an actual affective relation between speaker and addressee as individuals. The following analysis will show that speakers use 'love' to instantiate ritual frames within all sorts of communicative contexts in order to achieve a momentary ritual framing of the interaction to support the actions (Schegloff, 2007a) they are currently engaged in. The analysis contributes to a better understanding of the meaning potential and interactional uses of 'love' , and why it can be both claimed as a regional maker of friendly relations and ingroup identity (Beal, 2004;Wales, 2006) and contested as a term that can be used offensively (e.g. Dunkling, 1990;Poynton, 1990).
Endearments such as 'dear' , 'sweetheart' , 'darling' , but in particular 'love' , occupy a special place in the British English address system and usage because of their role in linguistic stereotyping: 'Love' is an intimate form of address that has proliferated into public contexts, where it is stereotypically associated with service encounters and service provider talk. It is also popularly associated with a particular British regional identity (northern England) and non-academic communities in that region (Culpeper and Gillings, 2018). 'Love' is further claimed as a marker of 'northern friendliness' (Culpeper and Gillings, 2018). It is seen as a politeness expression, abundant in conversational routines, that indexes a northern local (e.g. Sheffield) or (pan-)northern English identity characterised by rusticity and rooted in a working-class ethos (Beal, 2004;Wales, 2006).1 Though gendered, in some local northern contexts, it seems to have transcended its gendered semantics and can be used in solidary malemale constellations (Beal, 2004;Dunkling, 1990), suggesting a function as a degendered and ingroup maker, potentially similar to 'mate' in Australian English (Rendle-Short, 2009;Alimoradian, 2014).
On the other hand, other accounts of endearments in Anglo-varieties of English have found additional functions by inspecting the place in the address system that endearments occupy and, against the backdrop of the optionality of nominal address in English, the interactional contexts in which they are used (when, by whom, and to whom). From those perspectives, additional aspects of the functional profile of endearments come to the forefront: namely, their semantic core as lexicalisations of the speaker's affective stance, which metaphorically decreases the social distance between speaker and addressee (Poynton, 1990); their interpersonal core pattern as adult-child address (Braun, 1988) and for intimate speaker-addressee constellations ("close family members, sexual partners, and 'favourite' people" (Leech, 1999: 112)); a genderspecific distribution in terms of being predominantly used by and to women (Leech, 1999); a tendency for non-reciprocal use which coincides with asymmetrical role and power relationships (Wolfson and Manes, 1980); and an entirely optional gendering of the social relationship between speaker and addressee (Poynton, 1990). This research suggests that endearments carry over into non-intimate contexts residual meanings of gendered relationships, interpersonal closeness, and an adult-child talk pattern. As a consequence, they can be used to decrease the social distance between speaker and addressee in potentially inappropriate ways or to project a subordinate child-like position for the addressee through which meanings of triviality and inconsequence can be conveyed. However, only a few empirical studies of endearment use have been conducted (Kramer, 1975;Wolfson and Manes, 1980), and none in the 21st century, for British English, or for 'love' . The pragmatic scope of 'love' remains unclear.
The present analysis of 'love' was carried out on the 1994 and 2014 versions of the British National Corpus. It will show that over the course of twenty years, the use of 'love' has decreased substantially and has undergone a functional profile shift to include, more centrally than before, meanings of addressee and third-party evaluation, enacted through adverseness, joking and self-conscious performative use in storytelling. Those functions appear to feed off the core semantics and interpersonal constellations of the endearment as well as associations with social and linguistic stereotypes. While in the 1994 data, speakers predominantly use 'love' to invoke ritual frames to accomplish actions directly related to the addressee (e.g. 'can you bring in an ashtray love'), in the 2014 data speakers have come to use 'love' for metapragmatic purposes and in overtly theatrical ways as a linguistic trope (Agha, 2007) to momentarily invoke situations, social types and role constellations for self-conscious performative purposes (Derrida, 1972).
The paper is structured as follows. Section 2 briefly outlines Kádár and House's (2020) approach to RFIEs and the role of displaying affective stances for social identity enactment (Ochs, 1996). Section 3 summarises the small research base that exists for endearments. Section 4 presents the corpora under analysis and the micro-diachronic contrastive analysis method. Section 5 present the results of the corpus analysis. Section 6 concludes with a discussion of the main pragmatic scope shift towards performative uses of 'love' .

RFIEs and Affective Stance Displays in Social Identity Enactment
The present study is a form-focused, corpus-driven investigation of 'love' based on Kádár and House's (2020) approach to RFIEs and Ochs' (1996) idea that displaying affective and epistemic stances are central to social identity construction and enactment in interaction. Together they provide a useful lens on 'love' , which invokes quite particular constellations of role relationships in interaction and is an overt expression of the speaker's affective stance. Both approaches start from the indexical value of certain linguistic structures that develops out of frequent co-occurrence of the structure in a social situation. Both emphasise the role of conventional associations between linguistic form and social meaning. From this derives the power of a linguistic form to index its associated social meaning independent of the actual context it is used in (for example, indexing positive affect through 'love' regardless of the actual level of emotional attachment and interpersonal relationship). The core concepts of the two approaches that are relevant to the method of analysis and interpretation of the data are summarised below. Under the ritual framing approach,2 ritual framing takes place in so-called 'standard situations' and is indexed by RFIEs. Standard situations are communicative contexts that "involve participants' rather fixed expectations and perceptions of social role. Role relations are transparent and predetermined […] [T]he participants know where and who they are" (House, 1989: 115 in Kádár and House, 2020: 143). Standard situations include all communicative contexts -institutional and private -in which specific rights and obligations prevail that pre-set specific behavioural roles and role relationships. Participants are expected to follow the rules to maintain face. The indication of a particular ritual frame operative in a standard situation takes place through RFIEs, which index conventionalised roles and role relationships.
RFIEs are established through the process of regular co-occurrence of a linguistic expression in a particular context of use in a community so that, for that community, a conventional association between linguistic form and context is formed. As a result, the linguistic form becomes available as a routine way for speakers to indicate their awareness of the situational context and the interpersonal relationships in that context. RFIEs are used with the goal of projecting a specific ritual frame for the interaction in that situation. Speakers' deployment of a RFIE indicates activation and awareness of the ritual frame. They project expectations of role relationships for the interlocutor and roleconforming behaviour for the speaker themselves.
'Ritual framing' of an interaction is taken to signal a decreased sense of individualistic interactional engagement because the interlocutors' interactional behaviours are constrained by being channelled through the rights and obligations conventionally associated with their role in the indexed frame. In that sense, ritual linguistic behaviour is also communally-oriented, rather than individualistic, interpersonal behaviour. Behavioural and linguistic choices merely index in situ conventionalised forms of behaviour in the relevant community. The individual relationship between interlocutors interacting within a ritual frame is therefore of a second-order nature because it is fed through the set of conventional, sanctioned role relationships invoked by the frame. Kádár and House (2020) identify address terms and politeness markers as source categories for RFIEs and indicators of ritual language use at work.
The present analysis also draws upon Ochs' (1996) understanding of social identities as conventionally linked to the display of affective and epistemic stances in social situations. Under this approach, social situations are seen as multidimensional constructs which include the social identities of participants, the social acts and activities taking place between them, and speakers' expressions of affective and epistemic stances. Community members activate (Stivers, Enfield and Levinson, 2007) and the development and deployment of linguistic markers for that purpose. those situational dimensions through linguistic indexes in order to become "part of the situation" (Ochs, 1996: 410). Och's social identity construct encompasses various indexable aspects of social personae, such as social roles, participant relationship roles, group identity, relative rank and stance displays. Affective stance indexes include realisations of "mood, attitude, feeling and disposition, as well as degrees of emotional intensity vis-à-vis some focus of concern", while epistemic stances are realised through indexes which express "knowledge or belief vis-à-vis some focus of concern, including degrees of certainty of knowledge, degrees of commitment to the truth of propositions and sources of knowledge" (Ochs, 1990: 410).
The enactment of a social identity is contingent on the display of stances that members of a community expect to co-occur with that identity. That is, the realisation of a stance invokes a particular social identity; conversely, a claim to an identity may entail the expression of a social relationship through a particular, conventionally associated stance. It is through projecting specific social identities that members of a community are licensed to engage in particular social activities. Stances underpin the constitution of a social identity and help the speaker to enact that identity in a situation.
Kádár and House's RFIEs and Ochs' interdependence between social identity enactment and affective stance displays allow a preliminary conceptualisation of 'love' as a RFIE with an inbuilt affective stance that indexes a speaker's claim to a participant role relationship infused with positive affect. The focus of the analysis is on the actions that are being undertaken by speakers as they use 'love' to claim specific social relations with their interlocutor in an interactional context.

'Love'
3.1 Endearments Research on endearments in English is sparse and for the most part more than a quarter of a century old. There is no specific research on 'love' . To contextualise 'love' as an address term it is therefore necessary to refer to research on endearments and address in general.
Only two empirical studies of endearments are available. Wolfson and Manes (1980) and Kramer (1975) discuss endearments as part of their investigations into the forms used to address women in the US. They find that endearments addressed to adult women function to introduce an asymmetrical interpersonal relationship between the speaker and the addressee. In their data, endearment use is triggered by the minutest of non-normative behaviours (e.g. hesitations) by the addressee. Apart from those two studies, endearments are mentioned in research on address in general. Endearment use triggered by unconventional customer behaviour was also found in Baumgarten's (in press) study of vocative variation in telephone service encounters in northern England. Poynton (1990) discusses endearments as part of her functional description of the Australian English address system. Her focus is on the systematic difference between male and female address forms in terms of culture-specific power semiotics which stem from the semantics of attitudinals and solidary words that are the lexical sources of endearments and familiarisers, respectively. As a result, women and men receive radically different social positioning in talk addressed to them.
Further studies that include mention of endearments, but do not analyse them in detail, include Braun's (1988) cross-linguistic study of address terms. Braun describes endearments as primarily associated with adult-child address and addressing someone to whom the speaker feels close, but also with Irish males of low social status and service encounters. In Culpeper and Gilling's (2018) analysis of the regional distribution of politeness markers in the British National Corpus, endearments are considered as positive politeness and ingroup identity markers -a perspective first put forth by Brown and Levinson (1987). Brown and Levinson (1987) claim that endearments in English are used to soften face-threatening acts by signalling that the speaker considers the power difference to be small. In Leech's (1999) study of British and American English address terms in the Longman Grammar of Spoken and Written English Corpus, it is mentioned in a footnote that, in British English, 'love' can be used to address strangers "especially in the speech of older women" (p. 111). Dickey (1997: 256), in exemplifying the semantic bleaching of address terms, claims that 'love' has lost its referential meaning and is "a neutral form of address in some parts of England". Finally, Dunkling's (1990) encyclopedia of address terms lists endearments that occur in literary works of fiction.
Of all the English address term types, endearments have received the least systematic attention. This is despite the fact that even the limited research base suggests that endearments intersect in various ways with not only age, social status and social distance, but also gender. Unlike other gendered address forms, however, they target -at least nominally -very specific personality characteristics of the addressee and behavioural expectations about them ('love' , 'dear' , 'pet' , 'darling'), while also being claimed as regional identity markers with a degendered, neutral or positive politeness meaning. At the same time, some of this research (Dunkling, 1990;Poynton, 1990;Leech, 1999) includes anecdotal evidence of controversy around receiving endearments and competing interpretations of their meaning in interaction (friendly v. offensive). These seem to stem from the ambiguous, indistinct meaning of endearments that arises out of indexing intimacy and affect on the one hand, and their ideological value as a local form of 'politic' verbal behaviour (Watts, 2003) on the other.
Although there is no direct correlation between the literal meaning of an endearment and its intended meaning in the practice of addressing an interlocutor, endearments are the result of lexicalisations of affect in intimate relationships. They start being used as forms of address because their referential meaning qualifies them for certain situations and certain types of addressees (Braun, 1988). Over time, referential and social-address meaning become loosened and address meaning becomes independent of its referential meaning. What appears to remain stable, however, is the conventional marking of the kind of social positioning of speaker and addressee through the address term by virtue of its being rooted in a residue of the original semantics and interactional practice. Otherwise there would not continue to be different address categories available to speakers as options. All categories in an address system are sociolinguistically marked by virtue of being categorically different components of the same system (Zwicky, 1974). That means that honorifics, for example, continue to be conventionally associated with indexing speakers' orientation towards status differences and deference, familiarisers ('man' , 'mate' , 'bro') with decreased status difference and ingroup solidarity, and endearments with intimate social distance and positive affect. Speakers' address term choice, therefore, slots both speaker and addressee into a set of culturally available constellations of subject and object positions which are characterised by specific formations of status, power, social distance and affect.

3.2
Pragmaticalisation Because of their intrinsic interpersonal orientation and their association with interactional sequence through signposting the directionality of turns at talk, address terms cannot be clearly distinguished from pragmatic markers, which also have interpersonal and sequence marking functions (Fraser, 1996;Aijmer, 2002). While all address terms have discourse pragmatic functions, some nominal forms of address have been shown to develop into pragmatic markers. High frequency of use makes them vulnerable to pragmaticalisation, in the process of which the address term loses its deictic meaning and addressing function while acquiring affective-emphatic and discourse organisation meanings (Heyd, 2014;Kleinknecht and Souza, 2017;Martínez, 2018).
As a result of pragmaticalisation, the address term can be used with different degrees of deictic anchoring, making the interpretation of whether or not the interlocutor is actually addressed more tenuous. For example, when they are used with an emotive function, such as in exclamations or interjections ('oh dear' , 'oh boy'), their deictic meaning is assumed to be reduced and the referential meaning of the address term is assumed to be semantically bleached (Dickey, 1997). This process has been most extensively described for the group of familiarisers used in male peer-groups for solidary ingroup address (Heyd, 2014;Kleinknecht and Souza, 2017;Martínez, 2018). The degree of pragmaticalisation and semantic bleaching of endearments and their interpersonal and discourse-related functions are less well observed, described and understood.

Method and Data
The micro-diachronic approach to 'love' in contemporary spoken British English follows, with some modifications, Kádár and House's (2020) method of studying RFIEs. The analysis starts from a focus on linguistic forms to identify all occurrences of 'love' in two comparable samples of British English language use and then moves into qualitative analyses of the use of 'love' in interactional contexts to understand its indexical potential in the situations in which it occurs. The analysis includes qualitative coding of interactional characteristics in order to be able to arrive at quantifiable functional descriptions that can be contrasted. The goal of the contrastive analysis is to trace the development of the pragmatic scope of 'love': whether its functional profile has changed over the course of twenty years and whether there are any indications of a pragmaticalisation or a generalised, degendered friendliness meaning of 'love' in interaction.
The analysis was carried out on the spoken components of the 1994 and 2014 editions of the British National Corpus (henceforth BNC1994 and BNC2014). The corpora are representative samples of contemporary spoken British English from speakers across the UK.3 The analysis proceeded along the following steps: First, all occurrences of 'love' in endearment use and its diminutive form 'lovey' were manually extracted from the two corpora to create a 1994 and a 2014 sample of 'love' on which the subsequent steps were carried out. For reasons of space, the analysis presented below will not make a distinction between the different variants of 'love' ('love' , 'lovey' , 'my love') and 'love' will continue to refer to all variants. Second, all occurrences of 'love' were categorised according to the interpersonal scenario in which they occur: dyadic, multiparty or public (accessible to unratified participants). Third, all occurrences were categorised according to situational context (e.g. home, school, work). This information was gleaned from the socio-demographic information in the corpus metadata and through inspection of the extended interactional context. Kádár and House (2020) define as 'standard situations' for a RFIE those that occur in more than 2.5% of the total number of situations present in the sample investigated. The present analysis followed this cut off. Going beyond the Kádár and House model, all situations were analysed in more detail for co-occurring socio-pragmatic frame elements (gender, age, social role relationship). In addition, the interactional context was examined to be able to code the 'actions' -broadly understood as "the main job" that a turn is performing (Levinson, 2013: 107) -being accomplished with the help of 'love' . As with the standard situations, a 2.5% rate of occurrence threshold was applied to distinguish recurrent, 'standard action contexts' from the total number of actions with which 'love' occurs. The final step was the comparison across datasets.

5.1
General Frequencies The use of 'love' in spoken interaction has reduced considerably between BNC1994 and BNC2014 (Table 1). In both corpora it is most frequently used by female speakers and to female addresses ( Table 2). The percentage of male-to-female use is stable at 22.1% and 24.5%, respectively. Female-to-male use rises slightly from 17% to 21%. Male-to-male use is rare at 2.8% and 1.4% in BNC1994 and BNC2014, respectively.
The distribution across interpersonal scenarios is also similar across the datasets. With more than two thirds occurring in multiparty constellations, 'love' appears to be an address form that has a display function oriented to co-participants in the situation. Being generally used in utterance and turn-final position (e.g. 'have a coffee my love'), it does not primarily function to elicit the addressee's attention or to identify them as addressees (Biber et al., 1999), but to claim a specific interpersonal relationship between speaker and addressee. As the addressee's name is available for use in the majority of contexts, other naming options are available in principle. Likewise, the moment in the interaction at which 'love' is used, the directionality of talk is usually    (Clayman, 2011;Rendle-Short, 2010). Speakers, thus, appear to use 'love' to address the intended recipient of their utterance specifically by activating the social relationship meaning of 'love' for others to witness and to support actions pursued in the turn under construction that go beyond addressing the other (Section 5.6 below). The standard situations for 'love' in BNC1994 comprise service encounters, radio phone-in shows, some educational and workplace settings and non-institutional, private encounters (Table 4). The latter by far outweigh the other situations and appear to be the default situational context in which the use of 'love' is licensed. In BNC2014 only non-institutional, private encounters occur. This may be due to the slightly different corpus design parameters of the BNC2014. In a 2018 collection of telephone service encounters (Baumgarten, in press), 'love' was found to occur regularly as part of service provider talk, which suggests that service encounters continue to be a standard context of use for 'love' .
Cross-cutting gender, interpersonal scenarios and standard situations are the action contexts in which 'love' occurs (Table 5). The standard action contexts account for more than 80% of all occurrences of 'love' in BNC1994 and BNC2014, respectively. The examples provided in Table 5 show that 'love' occurs in pre-patterned formulae in conversational routines (e.g. greetings, apologies), but even more often in non-routine, ad-hoc action formulations. This suggest that the application of 'love' in ad-hoc formulations is routinised in order to invoke a specific role relationship between speaker and addressee and to achieve a momentary reconfiguration of the participant framework. This enables a shift in speaker-addressee footing that supports the action to Table 4 Standard situations for 'love' (≥ 2.5% of total occurrences) by interpersonal scenario ( be accomplished. In the two timeframes, speakers are selective regarding the contexts in which they find it useful to activate and display an affective stance through 'love' and to project an interpersonal framework of intimacy in which their actions are framed. This results in different profiles of standard action contexts for 'love' in BNC1994 and BNC2014. 'Love' is routinely associated with only a small number of action contexts. In both BNC1994 and BNC2014 a minority of action contexts are considerably more frequent than the rest. However, these most frequent action contexts are different for the two timeframes, with third-party evaluation standing out in BNC2014. This suggests a recent shift in the pragmatic scope of 'love' . Table 5 shows that while almost all the BNC1994 standard action contexts continue to be standard contexts in BNC2014, those that were most frequent in BNC1994 have become considerably less relevant for the speakers in BNC2014. In BNC1994, 'love' is most typically used in directives, greeting and leavingtaking routines, clarification requests and making assertions -none of which are particularly frequent in the BNC2014 data. The more performative uses of 'love' in joking and storytelling as a means of projecting other-evaluation that already existed in BNC1994 have increased in use, as have more confrontational contexts of direct addressee evaluation and disagreeing. In BNC2014, 'love' occurs most frequently with predominantly negative evaluations of absent third parties and other-deprecating banter and joking.
Overall, the usage patterns suggest somewhat different conventional associations of 'love' with action contexts in the two timeframes. In BNC1994, speakers invoke a 'love'-framework most regularly in action contexts that are speaker-centred and 'speaker agency-rich' . In the context of opening and closing interactions, issuing directions, requesting clarification and managing positions of knowledge vis-à-vis the addressee, the use of 'love' projects an affect-infused, close relationship with the addressee which support the speaker's claim to rights and social authority to manage the interaction with the addressee. While these contexts continue to exist in BNC2014, the 'love'framework appears to have shifted to become more useful than before as a means of projecting deprecating assessments of absent and present parties, often presented in stories or framed as joking. 'Love' in those uses establishes and maintains a communicative focus on the person being addressed or referred to and epitomises something -usually negative or problematicabout them. The examples below exemplify the typical uses of 'love' in the most frequent, non-routine action contexts in BNC1994 and BNC2014.

5.2
Directives, Giving Instructions In BNC1994, 'love'-framed directives occur in parent-child interaction (1), other constellations with an age or authority gap between interlocutors (2) and between partners in a couple (3). Typically formulated in response to or in anticipation of some unwanted behaviour by the addressee, they are speakers' attempts to manage addressees' current or future behaviour.

Carole
No no no come on don't play with that love please pick it up quickly <pause> off the floor <pause> don't walk on it.

5.3
Clarification Requests 'Love'-framed clarification requests occur only in non-institutional settings and commonly as a result of mishearing (5), lack of knowledge (6) or inattentiveness in parent-child and older adult-younger adult constellations (7). They signal attention shift or reorientation to the addressee. In BNC2014, they also act as go-on signals inviting elaboration and mark the speaker's continued attention to the addressee's talk (8) Go on Gordon! <-|-> Come on! Alison <-|-> you know <-|-> <pause> <-|-> garden and that. Gordon <-|-> Well I'll bloody well <-|-> you were fifty! Iris You what love? Alison I just want something with a little bit more garden.

5.4
Making Assertions Making assertions is used as a cover term for information-giving actions, such as explaining and opinion giving through which speakers mark knowledgeability for themselves while projecting assumptions of a corresponding lack of knowledge or full awareness of some information on the part of the addressee. In both BNC1994 and BNC2014, 'love'-framed assertion making achieves a distinctive positioning of the speaker. Speakers claim an unassailable knowledge status and sophistication through elevating typically mundane pieces of information to knowledge that only they possess, and the addressee does not.

Alison
Well why does he make them so busy?
(11) BNC2014; non-institutional, multiparty; parent-adult child; 'property listings' S0521 three-bedroom semi five-bedroom semi S0520 wow S0521 for five hundred S0520 I love it when they write in all caps wow S0521 it's the -UNCLEARWORD S0519 when you look at them love S0520 with a really loud voice S0519 they look so it's not just the number of rooms it's where it is lovey S0520 yeah it's all about location location S0521 I've never heard that before

5.5
Banter, Joking Targeting addressee and third-party characteristics and behaviour for assessment by the speaker constitutes the main usage shift for 'love' from BNC1994 to BNC2014. Banter and joking target the addressee in an overtly non-serious frame by trivialising the addressee's previous contribution (12, 13) or diminishing their status (14).
(12) BNC1994; non-institutional, multiparty; husband-wife Patsy <-|-> And then <-|-> you do other exercises like lying on your back, putting your feet <pause> your <pause> from the knees downwards over the edge of the pool <pause> and then trying to do sit ups. poor love look at him S0331 then we'll just rent the house out -UNCLEARWORD S0330 >>look at him in his long-sleeved T-shirt can't leave him out

5.6
Third-Party Evaluation In assessment actions that target the behaviour and characteristics of absent third parties, 'love' occurs in constructed dialogue (Tannen, 1986) in the form of self-quotations by the current speaker and through voicing (Tannen, 2010) of the third party. Those 'love'-framed assessments often occur in story format and are constructed as utterances that mirror other standard action contexts for 'love' . In other words, speakers draw upon the cultural recognisability of the routine associations of 'love' with action contexts. In (15), for example, Betty voices a greeting and a question to support an assessment of the third party as 'always happy' .

Julie
Which one did you see?
The one with glasses? Betty Yeah. 'Love' is used to activate culturally available inferences about social groups and simultaneously frames those meanings as judgements of aspects of the third party's character. While 'love'-framed assessments are sometimes used to convey a positive evaluation of the third party, such as in (15), by using 'love' as a marker of congeniality in constructed dialogue representing the speech of the third party, they are predominantly used to frame negative assessments. 'Love' occurs in particular with accounts of behaviour by the third party to be judged as breaking the norms of expected behaviour by being recognisably unreasonable and 'foolish' against a framework of normative 'sensible' conduct assumed to be shared by the interlocutors. In those uses, 'love'-framed evaluations refer to specific individuals (16) or occur in generalised assessments of behaviour of a certain type of people (17) in which 'love' functions as a stereotypic social indexical (Agha, 2007) to evoke a social type. When 'love' is voiced as part of constructed speech for the third party, it creates a stark contrast between the positive affect expressed and the behaviour reported on, which upgrades the negative assessment. In BNC2014, the use of 'love' in third-party assessments through constructed dialogue has increased significantly. All assessments in the data are negative. 'Love'-framed assessments are realised as self-quotations of speakers' thoughts about the third party as part of evaluation sequences. They act as metapragmatic 'verdict'-assessment tokens, usually presented at the end of the speaker's account. These self-attributed 'love'-framed assessments create an advantageous position of superiority for the speaker, grounded in greater expertise and knowledge of appropriate behaviour (18-21). Predominantly women are evaluated in this way. Men tend to be evaluated via their use of 'love' in third-party voicings and the accompanying contrast between the affective stance expressed and the reported behaviour (22). In both cases, 'love' functions as a membership categorisation device (Schegloff, 2007b) that assigns the third party to an outgroup characterised by the deficiencies presented in the account. Referring to the third party with 'love' is a tactic to establish dominance by distancing the speaker and their interlocutors from the (type of) person referred to and projecting the assumption of normative common ground between the speaker and the interlocutors.
(18) BNC2014; non-institutional, multiparty; friends, 'teachers discussing pupils in class' S0619 er and then there was another girl who was like dead sassy I was like woah woah woah love you'll get nowhere with that attitude S0618 >>oh was she sat next to her brother who was the muscle man with tattoos?

Conclusion
The analysis of the endearment 'love' in the 1994 and 2014 editions of the BNC suggests a shift in the usage and pragmatic scope of 'love' between the 1990s and 2010s. Most obviously in the data, the frequency of use has halved, while the broad social parameters of use -standard situations, interpersonal scenarios and gender constellations -have remained stable. The standard action contexts in which 'love' is used, however, have changed. In the 1990s data, 'love' is conventionally associated with direct and unmediated management of the interpersonal relationship between the speaker and the addressee of 'love' , for example, to increase the likelihood of compliance with requests and directives and acceptance of claims to superior knowledge status. The 2010s data show a clear preference for 'love' as an assessment token and method of invoking culturally recognisable social personae as part of performative and ludic language use. Crucially, this difference is one of frequency. All standard action contexts that existed in the 1990s continue to exist in the 2010s, showing a continuity in the interactional work that 'love' does, but a diachronic shift in preference for the types of action contexts in which it is put to use. In other words, while the outer conditions of where 'love' is used, who uses it, and to whom remain recognisably the same, it is used significantly less frequently than just 20 years before, and when it is used, it is for purposes that only partially overlap with the earlier conventional use. Overall, the data suggest a new preference for a metapragmatic use of 'love' as a linguistic trope which functions as a stereotypic social indexical (Agha, 2007) and membership categorisation device that can be ascribed to addressees and third parties to make them stand out in discourse in an emblematic way. It is a speaker tactic that turns individuals into culturally recognisable social personae, which are then used as discourse objects in the speakers' construction of social positions and social identities for themselves and their interlocutors.
The status of 'love' as an enregistered feature that encapsulates cultural knowledge of normative behaviour in social interaction can be most clearly seen in its use in multiparty scenarios. In both timeframes, 'love' is most conventionally associated with multiparty interaction. This suggests that 'love'-framed utterances have a communally-oriented display function. They represent an orientation towards norms of interpersonal behaviours and relations in the wider community, rather than being individualistic expressions of affect and intimacy between speaker and addressee at the given moment of use.
An evolving capacity of 'love' to index asymmetrical role relationships and a subordinate position for the addressee is most visible in the performative uses in addressee and third-party evaluation in the 2010s data. These are based on an unequivocal, explicit orientation towards group-specific understandings of rights and obligations in social interaction. 'Love' in self-conscious identity performance is used to show how the person referred to with 'love' or reported to have used 'love' has transgressed those norms and to signal the necessity for that individual to readjust their (most often her) behaviour.
This use of 'love' in interaction contradicts a generalised friendliness or semantically neutral interpretation of the endearment in contemporary British English. Rather, 'love' instantiates interpersonal frames in which speakers are able to claim and occupy positions of social authority, and display that positioning to the group of co-participants in multiparty scenarios. The instantiation of a 'love'-frame for some moment in the interaction is also a marked choice because of the optionality of nominal address in English and the redundancy of marking a participant relationship in a context in which relationship status and participant rights and obligations are known to all parties because they are established by the situation they are in. Previous studies have suggested that 'love'-framing actions is used as a means to mitigate threats to the addressee's face (Brown and Levinson, 1987;Culpeper and Gillings, 2018). The present analysis suggests that while that may have been one function of 'love' in the 1990s, where 'love' is associated with actions that directly affect the addressee (e.g. in requests and directives), this is not borne out by the more recent uses of 'love' . Overt negative assessments, the use of 'love' as a membership categorisation device to refer to people who are characterised as 'out of line' and to invoke social types that can be ridiculed appear to be rooted in an understanding of 'love' as indexing an asymmetrical relationship in which 'love' occupies a subordinate position. It is a means of distancing the speaker from the person addressed or referred to as 'love' .
An alternative analysis is that a 'love'-frame indexes a role relationship between speaker and interlocutors in which actions are deemed permissible and most likely to be uncontested. This interpretation assumes that the meaning of the endearment 'love' carries at least a residue of its original semantic and address meaning. On this view, 'love' continues to be informed by its two core interpersonal constellations: parent-child address -an asymmetrical relation characterised by positive affect and social authority -and its first 'derivation' of intimate address in close or romantic relationships, in which compliance with the speaker's actions is a way of maintaining that relationship. Through uttering 'love' , then, the speaker makes the associated action stand out from the turn under construction and assumes a position of social authority or compliance expectancy as the interpersonal frame in which the action is being pursued.
With regard to 'love' as a RFIE, the analysis suggests that 'love' indexes not so much standard situations, as put forth by Kádár and House (2020), but rather a certain interpersonal constellation defined by positive affect, which translates into a specific role relationship with associated rights for the speaker. 'Love' as a RFIE achieves the projection of an affective, intimate interpersonal relation to support the accomplishment of an identifiable, evolving set of actions in contemporary British English. 'Love' instantiates the role relationship regardless of -or even in strategic contradiction -to an actually held affective stance on the part of the speaker. In sum, 'love' constructs a ritual frame for action formation by indexing a particular interpersonal role relationship which is tied to an overt expression of a positive affective stance. Speakers use 'love'-frames either to address the interlocutor or as a metapragmatic marker and linguistic trope in talk about others. It creates a ritual frame for managing relations with the addressee or for performing evaluation and judgment of others for an audience.

Notes on Contributor
Nicole Baumgarten is a Lecturer in Translation and Intercultural Communication Studies at the University of Sheffield. Her research interests include linguistic inequality, language and intercultural communication, multilingualism/multiculturalism and English as a lingua franca.