ИСТИНА

Archi (Lezgic, East Caucasian) has a system of speech reporting typical of the family. Speech reporting is structurally very different from subordination, being the only construction where, in what appears to be a dependent clause, a finite form may be used. In reporting, person reference is subject to various changes. First of all, to refer to the reported speaker within the RC (reporting clause), there seems to be a strong preference to use a pronoun related to (identical to the stem) of the reflexive pronoun. <Ex.1> Similar systems of other East Caucasian languages have been described as having a logophoric pronoun (e.g. Ljutikova 2001). They then qualify as ‘mixed’ rather than ‘pure’ logophoric systems in Culy (1994) terms, and are very much in line with the use of ‘self’ pronouns in logophoric contexts outside Africa (Huang 2000). Another class of situations that feature a change from the original speech act deixis is the reference to the current speech act participants within the RC. In the latter case, personal pronouns are preferred independently from their speech act roles in the reported speech act. (For a similar preference to use second person pronoun for the current addressee, cf. the phenomenon of “second person magnetism” in Evans 2013). Person reference within the RC is adjusted to, or, better, overridden by, person reference in the current speech act. The meaning ‘He said that Ii am a fool’ is not usually expressed as ‘He said: Hei is a fool’, as would be expected in direct reporting. Whether such shifts are to be considered as categorical indications of indirect reporting is controversial. Indeed, they are not necessarily connected to other similarly systematic shifts in temporal or spatial deixis. Following Nikitina’s suggestion (2012), Daniel (2015) argues that this is a special mechanism applied to fulfil pragmatic preference in referring to current speech act participants by personal pronouns. The approach complies with (Spronck, Nikitina in prep.) who view indirectness not as a (necessarily) binary opposition to direct reporting but as a cumulative effect of change of perspective that may or may not involve various deictic categories including personal, temporal or spatial deixis, depending on the language-specific ‘grammar of reporting’; in other words, partial indirectivization rather than indirect speech. For Evans (2013), Archi does not have a canonical indirect speech but only constructions that deviate from the canonical direct speech. In other words, for Archi, I consider the phenomena observed in personal reference domain to be a special ‘person adjustment’ mechanism without positing indirect reporting as a strategy per se; this account is an alternative to the approaches based on the notion of semidirectness per (Aikhenvald 2008) or biperspectival speech per (Evans 2013). Capitalizing on these observations, in this paper I intend to focus on a less conspicuous and, to the best of my knowledge, a typologically under-documented phenomenon of deictic shift in reported commands. As various other East Caucasian languages, Archi allows for the use of commands in speech reporting. This is not surprising, because, as said above, Archi reporting is direct. While in languages like English volitionals cannot usually occur in indirect reporting, cf. English <ex. 2> and Archi <ex. 3>. However, cf. the following complication <ex. 4>. The situational context is as follows: the actual speaker conveys to the actual addressee their mother’s command, issued when the addressee was not present. The ergative on ‘mom’ is licensed by the quotative clitic and marks the agent of speech. The whole construction is the most natural - if not the only - way to translate English Mom told you to come. Reporting in <4> is formidable in the following way. In the original speech act, the imperative form could not have been used, because, in Archi, an imperative cannot be addressed to anyone but the addressee, and Zumzum was not present when the request was issued. As confirmed by our native consultant, who ‘reconstructed’ the original speech act, a special special jussive (third person imperative - I follow the terminology in Auwera et al. 2013) form must have been used instead. The use of the imperative thus shows this is not (intended as) a literal reproduction of the speech act being reported; note also the use of the second person pronoun ‘you.sg’. On the other hand, the reporting speaker does not subscribe to the use of the imperative. She only reports a command issued by someone else, as shown by the use of the quotative marker. I suggest that, in such cases, a person adjustment mechanism parallel to but different from that observed with personal pronouns is at work. In reported command in <4>, the jussive (a command to a non - speech act participant) shifts to the imperative (a command to the addressee). In other words, in RC, the moment that the intended performer of the reported command is the current addressee, the person category of the command change. The reported jussive (third person command) addressed to the actual addressee is ‘person-adjusted’ to the imperative (second person command). Unlike the case of personal pronouns, the change cannot be explained by a shift to the perspective of the current speaker - on the contrary, she clearly conveys someone else’s illocution - but to the perspective of the current addressee (cf. reinterpretation of pragmatic categories in RC as discussed in Spronck & Nikitina in prep.) It is the person category of the performer, not of the issuer that is at stake. This is not surprising, as commands revolve around the person of the performer, the subject of the action to be carried out (Dobrushina 2012).