3.1Corpus

This study of my as a part of address formulae is based on a corpus of eight British English plays, published between 1899 and 1912. Table 1 gives an overview of the plays and their authors (five men, three women), the years of publication and the size of each play. Corpus size is expressed in the number of speaker turns rather than in the number of words, as the turns are treated as the units of analysis in this study.

This corpus consists of comedies and social realist plays, which have been chosen because they aim to depict contemporary, everyday life. They were published at a time when British theater is increasingly interested in naturalistic drama texts, and favors more natural dialogue and a sincere effort to make the stage the mirror of life (Thorndike 1965Thorndike, Ashley H. 1965 English Comedy. New York: Cooper Square., 561). Still, a corpus of drama texts contains constructed dialogues and has its disadvantages for pragmatic analysis (as pointed out by Austin 1962Austin, John L. 1962 How to Do Things with Words. Oxford: Clarendon. and Taavitsainen and Jucker 2003Taavitsainen, Irma, and Andreas H. Jucker (eds) 2003 Diachronic Perspectives on Address Term Systems. Amsterdam: Benjamins. ). Even though the plays in this corpus have been selected because they aim to depict contemporary life, the social relations in the fictional world of the plays do not necessarily reflect real-life relationships. On the contrary, the drama texts in our corpus often select unusual or strenuous relationships, in order to achieve comic effect or another form of emotional release. However, this does not necessarily affect the pragmatic meaning of address terms: address usage can adjust to the fictional social universe of the plays (Taavitsainen and Jucker 2003Taavitsainen, Irma, and Andreas H. Jucker (eds) 2003 Diachronic Perspectives on Address Term Systems. Amsterdam: Benjamins. ; Buyle and De Smet 2018Buyle, Anouk, and Hendrik De Smet 2018 “Meaning in a Changing Paradigm: The Semantics of you and the Pragmatics of thou .” Language Sciences 68: 42–55. ). Furthermore, drama texts are particularly useful for an analysis of address usage. They give access to many different and easily identifiable speaker-addressee relations. Stage directions too are particularly useful when describing turns in terms of power and solidarity. Finally, as drama texts consist almost entirely of dialogues, nominal address terms occur frequently and enable a quantitative approach.

3.2Annotation procedure

For each turn in the corpus, address terms have been identified, as well as the speaker and the addressee. Each turn has also been annotated for power and solidarity, which have been shown to affect a speaker’s address choices. As a first step, power and solidarity have been interpreted in terms of stable social roles. This means that, typically, all the turns in which one speaker addresses one fixed interlocutor (e.g. all the turns in which Philip addresses Thomas in The Madras House) receive the same value for power (e.g. level) and for solidarity (e.g. close). As a second step, power and solidarity have been interpreted as context-dependent dimensions, and the speech acts with which address terms occur have been identified. The following paragraphs explain how these three determinants, i.e. power, solidarity and context, have been operationalized.

The annotation of speaker-addressee relations along the stable power and solidarity dimensions is primarily based on social role sets, such as parent-child or husband-wife. These role sets have been identified and classified according to their relevance for power (cf. Table 2) and solidarity (cf. Table 3) in reliance on previous research on address usage or politeness (Brown and Gilman 1960Brown, Roger, and Albert Gilman 1960 “The Pronouns of Power and Solidarity.” In Style in Language, ed. by Thomas A. Sebeok, 253–276. Cambridge: MIT Press.; Bates and Benigni 1975Bates, Elizabeth, and Laura Benigni 1975 “Rules of Address in Italy: A Sociological Survey.” Language in Society 4: 271–288. ; Hook 1984Hook, Donald D. 1984 “First Names and Titles as Solidarity and Power Semantics in English.” International Review of Applied Linguistics in Language Teaching 22 (3): 183–189. ; Jaramillo 1996Jaramillo, June 1996 “ Tu and usted .” Hispanic Journal of Behavioral Sciences 18: 522–532. ; Spencer-Oatey 1996Spencer-Oatey, Helen 1996 “Reconsidering Power and Distance.” Journal of Pragmatics 26: 1–24. ; Moreno 2002Moreno, María C. 2002 “The Address System in the Spanish of the Golden Age.” Journal of Pragmatics 34: 15–47. ; Fanego 2005Fanego, Teresa 2005 “ ‘Fare thee well, dame’: Shakespeare’s Forms of Address and Their Socio-Affective Role.” Sederi 15: 23–42.; Hickey and Stewart 2005Hickey, Leo, and Miranda Stewart 2005 Politeness in Europe. Clevedon: Multilingual Matters. ; Clyne et al. 2006Clyne, Michael, Heinz-L. Kretzenbacher, Catrin Norrby, and Doris Schüpbach 2006 “Perceptions of Variation and Change in German and Swedish Address.” Journal of Sociolinguistics 10 (3): 287–319. ; Levshina 2017Levshina, Natalia 2017 “A Multivariate Study of T/V Forms in European Languages Based on a Parallel Corpus of Film Subtitles.” Research in Language 15 (2): 153–172. ). Mainly based on these role sets, speaker-addressee relations in our corpus have received values for the stable power and solidarity dimensions. The power determinant has three possible values: upward, downward and level. Speaker-addressee relations have been marked as upward when the addressee has the ability to influence the speaker, while the speaker does not; they have been marked as downward when the speaker has the ability to influence the addressee, while the addressee does not; and they have been marked as level when there is no obvious power asymmetry. For instance, when the leader of a group addresses a regular member, the speaker has the ability to influence the addressee, which means the speaker-addressee relation is marked as downward. The solidarity determinant has three possible values: distant, close and service. Speaker-addressee relations have been marked as distant when speaker and addressee have never interacted before or had limited contact; and they have been marked as close when speaker and addressee interact regularly, with high degrees of emotional involvement. For instance, if speaker and addressee are immediate colleagues, they can be expected to interact regularly and be involved in each other’s lives, which means the speaker-addressee relation is marked as close. The master-servant role set has been classified separately and labelled service, and includes relations in which speaker and addressee find themselves in a master-servant relationship. The master-servant role set has received a separate value for solidarity because, contrary to other role sets, solidarity can be asymmetrical: servants can be very much involved in their masters’ lives, while masters know little about their servants’ personal affairs.

In most cases, speaker-addressee relations can be described by means of the same role set during the entire play, but this is not always the case. Interlocutors can adopt new roles, and their values for power or solidarity might change. Changes usually occur after time lapses, which coincide with changes of scene. For instance, two young people might meet for the first time in one scene, and be in love in the next scene. In that case, the role set assigned to this speaker-addressee relation changes from ‘strangers’ (distant) to ‘lovers’ (close). Importantly, although role sets are relatively uncontroversial measures of power and solidarity, they do not always translate easily to the power and solidarity dynamics in real life or, for that matter, fiction, which are often more complex. For instance, two interlocutors might be brothers (level), but also boss and employee (upward/downward), and in practice, these role sets are not easily distinguishable. Moreover, as the plays in our corpus often deliberately violate social rules, the translation of role sets to the universes of the plays might become potentially problematic. For instance, while it seems rather unlikely that the boss of a factory would have considered an employee as his equal in the beginning of the 20th century, it is quite plausible in Galsworthy’s Strife, where a strike rearranges traditional hierarchies. For these reasons, we have made adjustments to the annotations based on role sets if the plot gives clear evidence that they are inconsistent with actual power or solidarity dynamics (see above). A final note on the annotation procedure for the power determinant considers the speaker-addressee relations that do not correspond with any previously established role sets (as summarized in Table 2). Those relations have been classified as level, unless the plot gives unambiguous evidence of the contrary.

Context has been interpreted in terms of speech act pragmatics. For each address term under investigation, speech acts have been identified in the communicative units (henceforth C-units) to which the address terms belong. In order to distinguish between different C-units, we follow Leech, who defines the C-unit as “a unit with optimal syntactic independence, in that it is not part of a larger syntactic unit, except by means of coordination” (1999Leech, Geoffrey 1999 “The Distribution and Function of Vocatives in American and British English Conversation.” In Out of Corpora, ed. by Hilde Hasselgård, and Signe Oksefjell, 107–118. Amsterdam: Rodopi., 108). For all instances in our corpus, C-units correspond with the orthographic units divided by full stops, (semi-)colons, exclamation marks or question marks. The identification of speech acts is based on Weigand’s dialogic principle, which claims that individual speech acts are part of a dialogic sequence, including an initiative and a reactive action (2009Weigand Edda 2009 Language as Dialogue. Edited by Sebastian Feller. Amsterdam: Benjamins., 30). In our analysis, the C-unit is always considered as part of such a dialogic sequence: other units within the speaker’s turn, as well as the addressee’s initiative or reactive action within the same dialogic sequence, influence the identification of the C-unit’s speech act. An example of such a dialogic sequence can be found in (3). In the first turn, Maggie takes the initiative and expresses her worries about the future, while in the second turn, Lily reacts by reassuring Maggie. The speech act in the C-unit to which the address term dear belongs can only be properly analyzed as part of the dialogic sequence: Lily’s aim is to reassure Maggie, and not to make a mere statement about the future.

As a next step, speech acts are organized according to Leech’s speech act classification (1983). In the same vein as Brown and Levinson (1987)Brown, Penelope, and Stephen C. Levinson 1987 Politeness: Some Universals in Language Usage. Cambridge: CUP. , Leech classifies speech acts according to how they relate to the social goal of establishing and maintaining comity (1983Leech, Geoffrey 1983 Principles of Pragmatics. London: Longman.). He distinguishes four functional types: competitive, convivial, collaborative and conflictive speech acts (as summarized in Table 4). Leech (1983)Leech, Geoffrey 1983 Principles of Pragmatics. London: Longman. tentatively remarks that with collaborative speech acts, politeness seems largely irrelevant, while with convivial speech acts, positive politeness is most likely. He also notes that with competitive speech acts, negative politeness can be expected, while impoliteness strategies are likely occur with conflictive speech acts. In other words, collaborative speech acts might be considered as face-neutral, convivial speech acts as face-enhancing, and competitive and conflictive speech acts as face-threatening.

Table 4.Leech’s speech act classification (1983Leech, Geoffrey 1983 Principles of Pragmatics. London: Longman.)

Speech act type	Definition and examples
collaborative	the illocutionary goal is indifferent to the social goal; e.g. asserting, reporting, announcing, instructing
convivial	the illocutionary goal coincides with the social goal; e.g. offering, inviting, greeting, thanking, congratulating
competitive	the illocutionary goal competes with the social goal; e.g. ordering, asking, demanding, begging
conflictive	the illocutionary goal conflicts with the social goal; e.g. threatening, accusing, cursing, reprimanding

By exhaustively listing all the speech acts an address term occurs with, we hope to gain more insight into its pragmatic functions.

4.1 First alternation: Dear and my dear

Figures 3 and 4 show how dear (when it is not modified by my), as in (4a), and my dear, as in (4b), are characterized in terms of the stable power and solidarity dimensions. In the final bar of each plot, the number of turns with nominal address terms other than dear or my dear, have been summarized to obtain an additional level of comparison.

With respect to power, both dear and my dear occur less often in turns marked as upward in comparison with other nominal address terms (p < 0.001, Yates’ χ² = 29.142, df = 1 and p < 0.001, Yates’ χ² = 28.72, df = 1), which confirms Anglemark’s analysis of dear (2018Anglemark, Linnéa 2018 “ ‘Heav’n bess you, my Dear’: Using the ESDD Corpus to Investigate Address Terms in Historical Drama Dialogue.” Journal of Historical Pragmatics 19 (2): 186–204. ). While dear is more frequent than other nominal address terms in turns marked as level and less frequent in turns marked as downward, my dear occurs more often in turns marked as downward. Furthermore, a comparison between dear and my dear shows that my dear is more frequent than dear in turns marked as downward (p < 0.001, χ² = 26.082, df = 1) and less frequent in turns marked as level (p < 0.001, χ² = 22.516, df = 1). These findings indicate that, when my modifies an address term with a pronouncedly level profile, the newly combined address term becomes more closely associated with downward interactions. With respect to solidarity, both dear and my dear are more frequent in turns marked as close than other nominal address terms (p < 0.001, Yates’ χ² = 26.096, df = 1 and p < 0.001, Yates’ χ² = 11.023, df = 1). Still, while dear occurs in close turns only, my dear occurs in five turns marked as distant as well (p < 0.05, Yates’ χ² = 4.079, df = 1). These five attestations of distant my dear can be found in two different speaker-addressee relations, both marked as downward (i.e. when an older countess and homeowner addresses a young shop assistant, and when a wealthy client addresses her dressmaker). Also, four out of five attestations occur in disagreements or complaints directed towards the addressee (as in (5)), when exerting power is especially important. A plausible interpretation, then, is that, in these five turns, solidarity is overruled by power, i.e. my dear’s strong association with downward interactions is more relevant in explaining the speaker’s choice for my dear, than its association with close interactions.

In order to refine the analysis for power and solidarity, the next part of the comparison between dear and my dear will focus on the speech acts with which they occur. For each occurrence of dear and my dear, speech acts have been identified and organized according to Leech’s speech act classification (1983Leech, Geoffrey 1983 Principles of Pragmatics. London: Longman.). An overview of the different categories and speech acts attested in our analysis can be found in Table 6. Note that the speech act requesting subsumes a number of different directives, including orders, advice, proposals, invitations and suggestions (Risselada 1993Risselada, Rodie 1993 Imperatives and Other Directive Expressions in Latin: A Study in the Pragmatics of a Dead Language. Amsterdam: Gieben. , 48). The classification in Table 6 is of course not unambiguous, and some speech acts could belong to other categories. Especially some of the conflictive speech acts are often also competitive. When a speaker complains about something the addressee has done, the speaker’s social goal is not only to let the addressee know that the speaker has been wronged in some way, but also to get the addressee to change future behavior. In her analysis of complaints, Trosborg indeed remarks that complaints are representative of Leech’s conflictive function (1995Trosborg, Anna 1995 Interlanguage Pragmatics: Requests, Complaints and Apologies. Berlin: de Gruyter. , 312), but she also demonstrates that when a complaint is issued, a directive act (= competitive speech act) may be implied (1995, 320). Arguably, the same can be said for threats and disagreements. With respect to the collaborative type, it is clear that the illocutionary goals of speech acts such as asking for information, confirmation or clarification are not entirely indifferent to the social goal: the speaker wants something from the addressee, and in that way, they resemble competitive speech acts. However, as the cost for the addressee is quite low, especially in comparison to requests, they have been classified here as collaborative.

Five attestations of dear and five attestations of my dear have been excluded from the speech act analysis. Some attestations are stand-alone address terms (as in (6a)), which means that the speech acts cannot be determined for the C-units in which these address terms occur. Other attestations resemble expletive dear (as in (6b)). For these attestations, it is difficult to determine whether they primarily function as address terms, or as expletives expressing surprise.

Figure 5 gives an overview of the speech act types both dear and my dear occur with. While dear is more frequent with collaborative and convivial speech acts (p < 0.001, Yates’ χ² = 23.712, df = 1 and p < 0.05, χ² = 6.835, df = 1), as in (7) and (8), my dear is more frequent with conflictive speech acts (p < 0.001, χ² = 49.979, df = 1), as in (9). With competitive speech acts (e.g. (10)), both dear and my dear are rather infrequent, and differences are not statistically significant. Figure 6, too, gives an overview of the speech act types for dear and my dear, but only includes attestations in the nine speaker-addressee relations that allow switching between dear and my dear. In those relations especially, the choice for dear or my dear has to be contextually motivated. Importantly, the distribution in Figure 6 is similar to the distribution of all attestations of dear and my dear (cf. Figure 5), which shows that a speech act analysis does indeed make it possible to investigate context-dependent changes in power and solidarity, i.e. observed differences in speech acts do not merely reflect stable power and solidarity dynamics.

Interestingly, most conflictive speech acts my dear occurs with are complaints, disagreements and threats, and all conflictive speech acts dear occurs with are disagreements. Precisely these speech acts are often competitive as well as conflictive. If we also take into account that purely competitive speech acts are rather infrequent with both dear and my dear, we can conclude that, when dear and my dear occur with speech acts that signal that the speaker wants something from the addressee, these speech acts usually also involve an element of conflict. This is not surprising: endearments are most likely to occur as positive politeness strategies. Furthermore, as the overview in Figure 6 shows, especially my dear frequently occurs with these competitive-conflictive speech acts. In (11), for instance, Maggie says she is in no hurry to get married, and her mother responds with a complaint: she thinks Maggie should not have said that (= conflictive), and she also wants to prevent Maggie from talking about marriage in that way in the future (= competitive). The high frequency of my dear with conflictive-competitive speech acts is related to findings for the power determinant: if speakers show that they find the addressee’s behavior or opinions problematic and make clear that they want the addressee to make changes, they are likely to exert power over the addressee. At the same time, my dear’s close association with face-threatening acts (i.e. competitive-conflictive speech acts), in contrast to dear’s preference for face-neutral and face-enhancing speech acts (i.e. collaborative and convivial speech acts), says something about context-dependent solidarity as well. As the results for the stable solidarity determinant have shown, most attestations of both dear and my dear can be found in close interactions. Dear’s association with closeness is clearly reflected in the high frequency of dear with face-enhancing acts, which are expected to occur with positive politeness strategies. With face-enhancing acts, dear mainly strengthens the already solidary act. My dear’s association with closeness, on the other hand, is of a different nature. With face-threatening acts, the use of politeness strategies, such as my dear, does not merely support the speech act involved, but is meant to soften the threat to the addressee’s face. One could thus argue that, while dear reflects an already existing level of intimacy, my dear’s function is to temporarily evoke a degree of solidarity that is not inherent to the context in which it occurs. So, although the analysis for the stable solidarity dimension did not reveal any significant distinctions between dear and my dear, the contextual analysis does show some functional differences in terms of solidarity.

In conclusion, the analysis of dear and my dear for power, solidarity and context shows that (1) dear mainly signals symmetry with respect to the power dimension, and closeness with respect to the solidarity dimension; and that (2) by adding my, the address formula may signal that speakers want to exert power over the addressee, and that they want to increase solidarity with the addressee because they need something that inherently clashes with the addressee’s feelings and wishes. Adding my to an address term that reflects level power dynamics and close solidarity dynamics creates a new address formula that is more closely associated with downward power dynamics and temporarily increased closeness.

Interestingly, the functional differences between dear and my dear are also reflected in gender-related differences. Figure 7 gives an overview of the distribution of dear, my dear and other nominal address terms in the corpus for gender. Turns have been coded for the gender of both speaker and addressee, and labelled as male-male, female-female, male-female and female-male. Figure 7 shows that both dear and my dear are obsolete in male-male interactions and occur almost equally often in female-female interactions. However, dear is significantly more frequent in female-male interactions, while my dear occurs more often in male-female interactions (p < 0.001, Yates’ χ² = 53.765, df = 1). Our findings confirm the claim that endearments are rare in male-male interactions (Leech 1999Leech, Geoffrey 1999 “The Distribution and Function of Vocatives in American and British English Conversation.” In Out of Corpora, ed. by Hilde Hasselgård, and Signe Oksefjell, 107–118. Amsterdam: Rodopi.). Furthermore, my dear, which is more closely associated with exerting power and face-threatening acts, is more often used by male speakers when addressing women. By contrast, dear, which merely reflects power symmetry and closeness, is more often used by female speakers when addressing men. If we assume that gender inequality was an integral part of British society in the period between 1899 and 1912 (Thompson 1990Thompson, Francis M. L. (ed) 1990 The Cambridge Social History of Britain, 1750–1950. Cambridge: CUP.; Mahood 1995Mahood, Linda 1995 Policing Gender, Class and Family: Britain, 1850–1940. London: UCL Press.; Holloway 2005Holloway, Gerry 2005 Women and Work in Britain since 1840. London: Routledge.; Davis 2014Davis, Angela 2014 Modern Motherhood, Women and Family in England, c. 1945–2000. Manchester: Manchester University Press.), it is not surprising that inequality is also reflected in the address system, as a core resource for realizing social deixis.

Recall that for the analysis of the alternation between dear and my dear, only those instances of dear in which it does not modify other address terms have been selected. Where dear or my dear are followed by a nominal address term, the number of comparable collocations is too small. Still, it is possible to investigate which nouns are modified by dear and my dear. The overview in Table 7 shows that dear most often modifies lady + surname, which is less frequently modified by my dear (p < 0.001, Yates’ χ² = 25.839, df = 1). This corresponds with findings for power and solidarity, as lady + surname is most frequent in turns marked as upward and close. Furthermore, given name is more often modified by my dear than by dear (p < 0.02, Yates’ χ² = 6.09, df = 1), which corresponds with findings for power, as given name is attested in turns marked as level (380 instances) as well as down (209 instances).

4.2 Second alternation: My lord/lady and your lordship/ladyship

The next part of the analysis aims to compare the functional profiles of my lord/lady, as in (12a) and your lordship/ladyship, as in (12b). Importantly, this analysis is less straightforward than the comparison between dear and my dear. My honorific rarely if ever alternates with unmodulated forms. However, my honorific does alternate with address formulae of the type your honorific, often within the same speaker-addressee relation. The alternation is complicated because the comparison includes, to some extent, lexical variation (e.g. lord versus lordship) as well as syntactic variation (i.e. in this corpus, my lord/lady is a vocative (Zwicky 1974Zwicky, Arnold 1974 “Hey, Whatsyourname!” In Papers from the 10th Regional Meeting of the Chicago Linguistic Society, ed. by Michael W. La Galy, Robert Fox, and Anthony Brack, 787–801. Chicago: CLS.), while your lordship/ladyship only occurs as an argument of a verb). Nevertheless, the contrast between my honorific and your honorific offers another opportunity to pinpoint the functional contribution of my, as the following analysis shows.

Overall, 75 attestations of my lord/lady and 20 attestations of your lordship/ladyship have been coded for power, solidarity and context. Importantly, 19 out of 20 attestations of your lordship/ladyship occur in speaker-addressee relations in which my lord/lady is also attested. Since the number of turns per speaker-addressee relation can heavily skew the results, a comparison for the stable power and solidarity dimensions at the level of the turn is not revealing. At the level of the speaker-addressee relation, results do show that, for power, your lordship/ladyship occurs only in relations coded as upward, while my lord/lady is attested in speaker-addressee relations coded as upward, and to a lesser extent, as level. Interestingly, when a modifying adjective is included in the address formula (e.g. my good lady, my dear lady), attestations can also be found in speaker-addressee relations marked as downward. With respect to solidarity, results show that both your lordship/ladyship and my lord/lady occur in service, distant as well as close speaker-addressee relations. Tentatively, we can say that your lordship/ladyship is more closely associated with upward power dynamics than my lord/lady, while for solidarity, no notable differences can be observed.

Both my lord/lady and your lordship/ladyship co-occur in four different speaker-addressee relations. In order to investigate how switching is motivated by speech act pragmatics, attestations of both address formulae in these four speaker-addressee relations have been coded for context. Note that one attestation of my lord, which is a stand-alone address formula, has been excluded from the analysis. Figure 8 gives an overview of the speech act types with which we find my lord/lady and your lordship/ladyship.

While your lordship/ladyship is more frequent with competitive speech acts (p < 0.001, Yates’ χ² = 16.866, df = 1) (see, for instance (13)), my lord/lady seems more likely to occur with collaborative, conflictive and convivial speech acts (see, for instance, (14–16)).

In brief, speakers especially opt for my lord/lady with face-neutral and face-enhancing speech acts, as well as with face-threatening acts that are in conflict with the social goal of establishing and maintaining comity. In comparison, your lordship/ladyship is the preferred option when speakers want something from the addressee, and is most likely employed as a negative politeness strategy. As such, findings for speech act pragmatics tie in with our tentative conclusion regarding power: your lordship/ladyship uniquely occurs in upward speaker-addressee relations and is most often used with face-threatening acts that can be mitigated by signaling deference. My lord/lady, on the other hand, occurs in level as well as upward speaker-addressee relations, and is most often used with speech acts that do not require emphasis on existing power dynamics. Admittedly, my lord/lady occurs in a variety of situations, in contrast to your lordship/ladyship, which has a more clearly delineated functional profile. Still, results definitely show that the variation between both address formulae is functionally motivated. Also note that these findings are consistent with the syntactic difference between my lord/lady and your lordship/ladyship. By placing your lordship/ladyship in subject or object position, speakers have the opportunity to avoid direct address by means of the second-person pronoun you, which also mitigates the attack on the addressee’s face.

In previous research on speech acts and politeness, requests especially have received a lot of attention. Interestingly, Blum-Kulka and Olshtain (1984)Blum-Kulka, Shoshana, and Elite Olshtain 1984 “Requests and Apologies: A Cross-Cultural Study of Speech Act Realization Patterns (CCSARP).” Applied Linguistics 5 (3): 196–213. classify requests based on the degree of directness, which influences the degree of imposition involved in the speech act, i.e. the strength of a face-threatening act or the degree to which it interferes with the addressee’s face (Brown and Levinson 1987Brown, Penelope, and Stephen C. Levinson 1987 Politeness: Some Universals in Language Usage. Cambridge: CUP. ). Blum-Kulka and Olshtain (1984)Blum-Kulka, Shoshana, and Elite Olshtain 1984 “Requests and Apologies: A Cross-Cultural Study of Speech Act Realization Patterns (CCSARP).” Applied Linguistics 5 (3): 196–213. argue that the strength of a face-threatening act can be minimized by choosing an indirect realization of requests over a direct realization. They propose three different levels of directness: direct requests, conventionally indirect requests and non-conventionally indirect requests. Direct requests are weaker face-threatening acts than conventionally indirect requests, which are in turn weaker threats to the addressee’s face than non-conventionally indirect requests. If switching between my lord/lady and your lordship/ladyship can be explained by speech act pragmatics, variation between both address formulae might not only be affected by the presence and type of face-threatening acts involved, but also by the strength of the face-threatening acts. If we assume that your lordship/ladyship is the preferred option when conveying negative politeness, your lordship/ladyship can be expected to occur more often with indirect requests than with direct requests. The opposite should be true for my lord/lady. For the present study, we have analyzed the requests based on the classification scheme proposed by Blum-Kulka and Olshtain (1984)Blum-Kulka, Shoshana, and Elite Olshtain 1984 “Requests and Apologies: A Cross-Cultural Study of Speech Act Realization Patterns (CCSARP).” Applied Linguistics 5 (3): 196–213. and Economidou-Kogetsidis (2010)Economidou-Kogetsidis, Maria 2010 “Cross-Cultural and Situational Variation in Requesting Behavior: Perceptions of Social Situations and Strategic Usage of Request Patterns.” Journal of Pragmatics 42: 2262–2281. , which can be found in Table 8. This overview lists the three request types and the different speech act realizations for each type. The realizations of requests in the second column are positioned on a cline from most direct (top) to most indirect (bottom). Examples for each request realization can be found in the third column.

Table 8.Classification of requests according to their level of (in)directness (Blum-Kulka and Olshtain 1984Blum-Kulka, Shoshana, and Elite Olshtain 1984 “Requests and Apologies: A Cross-Cultural Study of Speech Act Realization Patterns (CCSARP).” Applied Linguistics 5 (3): 196–213. ; Economidou-Kogetsidis 2010Economidou-Kogetsidis, Maria 2010 “Cross-Cultural and Situational Variation in Requesting Behavior: Perceptions of Social Situations and Strategic Usage of Request Patterns.” Journal of Pragmatics 42: 2262–2281. )

Level of (in)directness	Realization speech act	Example
direct request	imperative	Leave me alone.
	explicit performative	Please look after the kids for the evening.
	hedged performative	I’m here to ask for an extension.
	obligation statement	Madam, you’ll have to move your car.
	want or need statement	I really wish you’d stop bothering me.
conventionally indirect request	suggestion	How about cleaning up?
	reference to preparatory conditions (e.g. ability, volition or possibility)	Would you mind moving your car, please?
non-conventionally indirect request	strong hint	You’ve left this kitchen in a right mess.
	mild hint	I’m a nun. (in response to a persistent boy)

The overview in Figure 9 shows how the requests with my lord/lady and your lordship/ladyship are classified in terms of directness. While most attestations of my lord/lady occur with direct requests, your lordship/ladyship is most frequent with conventionally, and especially non-conventionally indirect requests. These findings appear to conform with our hypothesis, but the data are obviously too sparse to attain significant differences.

Examples of my lord/lady and your lordship/ladyship with direct requests can be found in (17a) and (17b). All direct requests with my lord/lady contain imperatives, which are positioned at the most direct end of the (in)directness scale (position 1). The direct requests with your lordship/ladyship are less direct than those with my lord/lady: they both contain want statements (position 4). In (18a), your lordship/ladyship occurs with a conventionally indirect request, while the only attestation of my lord/lady with a conventionally indirect request can be found in (18b). The request in (18b) is a suggestion, i.e. the most direct realization of conventionally indirect requests. With non-conventionally indirect requests, your lordship/ladyship is the only option (as in (19)).

In conclusion, the analysis of my lord/lady and your lordship/ladyship for power, solidarity and context shows that (1) your lordship/ladyship is mainly used to signal deference, especially when mitigation is necessary, i.e. when the speaker needs something from the addressee; and that (2) my lord/lady is less often used as a negative politeness strategy than your lordship/ladyship, and when my lord/lady does express deference, reinforcing upward power dynamics is less urgent.