Out of Africa: Logophoric pronouns and reported discourse in Finnish and High Latvian dialects

Acta Linguistica Lithuanica LV (2006), 55 - 87 Out of Africa: Logophoric pronouns and reported discourse in Finnish and High Latvian dialects Nicote Nau Uniwersytet im. Adama Mickiewicza In this paper I will discuss properties of logophoric constructions and their place within a typology of reported speech. The data from Finnish and Latvian dialects prove that the phenomenon is not restricted to Africa and support the view expressed by Culy (1994; 1997) and others that logophoric pronouns, in contrast to long-distance reflexives, are rooted in reported discourse and cannot be explained in syntactic terms. Bhat’s (2004) claim that their primary function is to distinguish the participants of a reported speech act from the actual speaker and addressee leads to the opposition of logophoric constructions to direct speech, rather than indirect speech, and to a functional explanation of the distribution of types of reported discourse. 1. INTRODUCTION A logophoric marker is a device used in reported speech, referring to the author whose words are reported, while in such a context the use of an anaphoric third person pronoun signals reference to another person. Consider the following minimal pair from colloquial Finnish, cited after Saukkonen (1967): (1) Se, sano, ettei han, voi tulla. PRO say:pRs:3 COMP:NEG:3, LOG can come:INF ‘He/she, says that he/she, won’t be able to come.’ (reported utterance: ‘I won’t be able to come’) 55 Nicote Nau (2) Se, sano, ettei Sport) VOI tulla. PRO say:PRS:3 comp:NEG:3 PRO can come:INF ‘He/she, says that he/she,,., ,, won’t be able to come.’ (reported utterance: ‘He/she won’t be able to come’) The phenomenon of logophoricity has been widely discussed in linguistics within the last thirty years, since the term was introduced by Claude Hagége in his seminal paper (Hagége 1974). Overviews are provided by Roncador (1988; 2006) and Stirling (1994). The data I will discuss in this paper support the view that the use of logophoric pronouns cannot be explained by syntactic parameters, nor in purely semantic or pragmatic terms. Instead, they are tightly connected to reported discourse. Formal features which are frequently found in constructions with logophoric pronouns follow from the nature of reported speech, but they are not conditions for the use of the pronouns. Similarly, the marking of point of view is not at the basis of the logophoric pronouns in Finnish and High Latvian, but in some of their varieties developed as a secondary function. While the Finnish and Latvian facts presented here have long been known by linguists of the respective countries, outside of their homelands they are still largely ignored, due to the fact that most of the literature is difficult to access by non-specialists of the languages. Finnish linguists have only recently started to describe the Finnish logophoric pronoun from a more general point of view and to publish their results in English (see especially Laitinen 2002; 2005). The Latvian data are still a well kept secret, but should remain so no longer. Because of the lack of accessible data, I will give two longer examples in the appendix. In section 2, I will briefly comment on the data used in this paper. Most of the discussion of properties of logophoric constructions in the linguistic literature has been based on data from languages of West and Central Africa. It has even been claimed that the phenomenon is restricted to this continent (Culy 1994: 1059; Roncador 2006: 314)'. One aim of this paper is to challenge this view by showing that Finnish and Latvian dialects have logophoric pronouns which are very similar to those of African languages. In section 3, I will discuss their properties in detail. ! Giildemann (2003) is more cautious, and Kibrik (2001: 1133) explicitly mentions other parts of the world. 56 Logophoric pronouns in Finnish and High Latvian Speech reports with logophoric pronouns do not fit into the traditional division of direct and indirect speech (cf. Roncador 1988). Since one of the main features distinguishing these two types of reported discourse is the treatment of the reported speaker as either first or third person, it is obvious that with logophoric pronouns, we are dealing with a third type. Furthermore, they provide a unique way to disambiguate the reference to persons when a current speaker reports the words of another speaker. It is important to recognize that there are two different kinds of possible referential ambiguity in reported speech: (i) reference to the reported speaker as opposed to another non-participant of the current speech situation; (ii) reference to the reported speaker as opposed to the current speaker. In languages like English, the first type of ambiguity appears in indirect speech reports (as in the translations of (1) and (2)), while the second type arises when it is not clear whether the report is direct or indirect, as in the sentence She says I know. The fact that logophoric pronouns provide an effective way to deal with this second type of ambiguity has not been given enough consideration. In section 4, I will discuss this question in more detail and develop a typology of reported speech based on the opposition of logophoric constructions to both direct and indirect speech. Logophoric pronouns as understood in this paper should be distinguished from reflexive or other pronouns which may have a logophoric use as a secondary function. Examples of the latter, most often discussed under the name ‘non-clause-bounded reflexive’ (NCBR) or ‘long-distance reflexive’ (LDR), are found in languages of East Asia, like Japanese, Chinese and Korean, as well as in Europe, the most famous here being Icelandic (see Sells 1987; Stirling 1994; Huang 2002). An important difference between the two types of pronouns is that in languages with a logophoric use of reflexives, the use of an anaphoric pronoun in a logophoric context does not exclude coreference, nor is the LDR unambiguous, as it may still have a reflexive meaning. Languages with such pronouns are sometimes called ‘mixed logophoric languages’ (Culy 1994; 1997; Huang 2006: 236). In contrast, where the logophoric function is the primary function of a pronoun, the marking of coreference in speech reports is (ideally) unambiguous and obligatory; it is grammaticalized. However, grammaticalization often is a matter of degree. Logophoricity may start as a secondary function used occasionally, and 57 NicoLe Nau in the course of time become primary and obligatory. On the other hand an erstwhile ‘pure’ logophoric pronoun may acquire other secondary func tions, which in turn may become central. We should therefore not suppos¢ the distinction between ‘pure’ and ‘mixed’ logophoric languages as clear-cu as Culy (1994) proposed (see also Giildemann 2003 and Bhat 2004: 69 70 on this point). Varieties of Finnish and Latvian provide examples for ¢ whole range of variation from clearly grammaticalized logophoric pronoun: (which will be be dealt with in the current paper) to occasional logophori: uses of a primarily anaphoric pronoun. 2. DATA 2.1. Pronoun hdn in Finnish The Finnish pronoun hdn, plural he, is originally an anaphoric third persor pronoun (see Itkonen 1992 for etymology and cognates in other Finno-Ugric languages). It is used as a logophoric marker in all dialects of Finnish a: well as in non-dialectal spoken varieties (Hakulinen et al. 2004: 708, 1368 1408; Laitinen 2005). The anaphoric third person pronoun in these varietie: is se, plural ne. The degree of grammaticalization of the logophoric functior of hdn and the range of other functions of this pronoun vary. In the Eastert dialects it seems to be most grammaticalized (unambiguous and obligatory in a logophoric construction), while in Western and especially Southwestert dialects and in non-dialectal colloquial styles its use in a logophoric contex is not obligatory and a third person pronoun does not exclude coreferenct with a reported author. In these varieties, hdn also has a wider range 0 functions. In Standard Finnish Adn is a third person pronoun for anaphoric reference to humans, while se is a demonstrative and used for anaphoric reference to non-human referents. In this paper, I will consider only dialec data and focus on those dialects in which the logophoric function is mos grammaticalized. The dictionary of Finnish subdialects (SMS) provides a good overviev of the uses of this pronoun, presenting its functions and quoting example: in two separate entries for the singular hdn (Vol. 4, pp. 418-423) and th: plural he (Vol. 4, pp. 1-5). Vilppula (1989) provides an overview of the use: 58 Logophoric pronouns 1n Finnish and High Latvian of hdn, he outside of reported speech. A number of Finnish scholars have carried out research on individual subdialects?. Ylikahri (1996) describes the subdialect of Siikainen in Western Finland, where han is basically a logophoric pronoun but its use may be extended. Kuiri (1984) presents a monographic treatment of reported speech in two neighboring dialects of East Finland, Kainuu and Northern Karelian, with a section devoted to the use of hdn. Without using the term ‘logophoric’, she is probably the first to draw attention to the similarities between Finnish han, the Japanese longdistance reflexive zibun and the logophoric pronoun neh in Mabila. Ikola (1960) explores grammatical features of reported speech in Finnish from both a synchronic and a diachronic point of view. Curiously, he does not discuss hdn explicitly, but his book is a good source for examples, as he presents data from many dialects. 2.2. High Latvian sys Latvian dialects are divided into three groups: Tamian or Livonian, Central, and High Latvian. In opposition to High Latvian, Tamian and Central dialects, which cover the western and central part of Latvia, can be classed together as Low Latvian’. High Latvian is spoken in the Eastern part of the country, that is, in Latgalia and some neighboring territories. It is also referred to as Latgalian. The question whether Latgalian/High Latvian is a dialect of Latvian or a separate language has been discussed controversially, but is of no importance for the current purpose. A grammaticalized logophoric pronoun is found in many subdialects of High Latvian. It is the pronoun Sys, stemming from a demonstrative of speaker deixis (‘this’; for etymology and cognates, see Euler 1993). The demonstrative function of this pronoun has been lost in most subdialects or is retained only where the pronoun is used as a determiner. The usual anaphoric third person pronoun is jis. The demonstrative tys ‘that’ is also used ? Unfortunately, unpublished theses I found quoted in the literature have not been available to me * Information on Latvian dialects can be found in Balode & Holvoet (2001) and Gaters (1977). 59 NicoLe Nau anaphorically. The following example illustrates the use of jis (third person), tys (demonstrative) and Sys (logophoric): High Latvian, speaker Anna (AA 70) (3) tys Broks, jd, tys, ci pordzeivuojs beja. pet Broks yes pro very suffer:pap:M.sG AUX:PST Jis, teice, ka Sys, grybiejs niiSautis PRO Say:PST:3 COMP —LOG want:Pap:m.sG shoot:1NF:REFL ‘this Broks,, well, he, had really suffered. He, said that he, wanted to shoot himself Some High Latvian subdialects, all Low Latvian dialects, and Standard Latvian have lost the anaphoric pronoun jis and replaced it with vins. In Standard Latvian, vins is used only for animated referents, a curious parallel to Standard Finnish, where, as noted above, the use of hdn vs. se is also based on animacy. Standard Latvian and some Low Latvian subdialects still make use of Sis (the Low Latvian and Standard form of the pronoun) as a demonstrative, but more often, Sis is used as a second, functionally marked anaphoric pronoun (cf. MLLVG-I, 517). It may also be used as a logophoric marker, but this function is not grammaticalized in Low Latvian dialects as it is in High Latvian, and the use of the anaphoric vins in reported speech does not exclude coreference. The High Latvian data presented here are taken from two different sources. Most valuable for this research was the recording of an interview from the Latvian Oral History collection of life stories. The narrator, Anna, born in 1918, comes from the village Vilani in Central Latgalia. She is interviewed about her life by her granddaughter in 1993. For the transcription of fragments of this text, I tried to use the new Latgalian standard orthography without (too much) standardizing of the words. This task would have been impossible for me without the help of Lidija Leikuma, a dialectologist from the University of Latvia and native speaker of High Latvian, whom I acknowledge most gratefully. The other kind of source used are dialect texts published by Latvian dialectologists. To make the examples more readable and to facilitate the editing of this paper, I have (again with the kind help of Lidija Leikuma) simplified the often very sophisticated phonetic transcriptions of the sources, that is, | have deleted the markers of pitch accent, reduced the variety of vowel allophones, marked palatalization only where 60 Logophoric pronouns in Finnish and High Latvian it does not follow from the position of a consonant, and noted semivowels as <j> and <v>. I have also consulted grammatical descriptions of subdialects, but unfortunately they usually present only very short examples out of context, which makes it difficult to interpret the construction, or no examples at all. Still, these sources, as well as personal communication with Anna Stafecka and Lidija Leikuma, convinced me that the logophoric use of Sys is quite widespread in High Latvian. In the narrative of Anna, the logophoric Sys is clearly grammaticalized and the use of the anaphoric jis in a speech report points to another person than the reported speaker: High Latvian, speaker Anna (4) Tagad jis, suoka runuot, tys bruolans,, lai es now pro. Start:pst:3 talk:ivr per cousin comp 1sg precejiis ar ja,. marry: pa:r.sc with pro:acc ‘Now he, started to say, this cousin,, that I should marry him, [= the uncle]’ (AA 55) (not 1) A larger fragment of Anna’s narrative, showing the regular use of the logophoric pronoun, is given in the appendix. 3. FORMAL PROPERTIES OF THE LOGOPHORIC MARKER AND THE LOGOPHORIC CONSTRUCTION In this section, I will discuss the formal features of the logophoric construction in Finnish and High Latvian dialects. The same features have been discussed for other logophoric languages, and I will mainly follow the typology of Stirling (1994). 3.1. Formal properties of the logophoric marker The logophoric pronoun in Finnish and in High Latvian has the same inflectional categories as demonstrative and third person pronouns in the respective language: in Finnish, it is inflected for number and case, in High Latvian for number, case and gender. When used as a subject, it triggers third person agreement. 61 NicoLe Nau There seem to be no restrictions as to possible case forms and syntactic functions. Most often the logophoric pronoun is used as a subject or an object, it also appears as genitive attribute or complement of an adposition. Laitinen (2005: 77-78) draws attention to the fact that Finnish hdn shares two morphosyntactic properties with (singular) first and second person pronouns. First, only personal pronouns (but not se) and the interrogative kuka ‘who’ have a special accusative ending. Interestingly, this feature has entered the standard language from the Eastern dialects, where, as noted above, the logophoric use of hdn is more grammaticalized than in the West. The second feature regards the use of the possessive suffix, which I will not discuss further, as there is a lot of dialectal variation in this area. A more striking parallel to first and second person pronouns is found in both languages regarding the assignment of number, which is not a matter of agreement with the antecedent. This becomes evident in cases where a plural logophoric pronoun is used in the report of a single speaker, as in the following examples: Finnish, Siikainen [Western Satakunta; Hime group] (Ylikahri 1996: 186) (5) Kalle, sano ettd he... ,, otti sen aina —_veneeseen Kalle say:prs:3 comp oG:pL _take:pst:3 pro:acc always boat: ko he, lahti, kalastelemaan ni kissa sai ruakaa. ‘(inel. 1) when toc:p. leave:pst:3 fishing:inr:1LL so cat get:pst:3 meal:par ‘Kalle, said they,,.., ,, always took the cat into the boat, when they,,,.1, 1) left fishing, so it got a meal.’ (reported utterance: “we always take the cat...”) High Latvian, speaker Anna (AA 99) (6) jis, teice, ka, vot, Si, jau ka karuojusi [...] PRO Say:PsT:3 COMP PTC LOG:PL PTC when fight_in_war:PAP:M.PL ‘He, said that, so, when they ua 1 Were in the war [...]’ (reported utterance: “when we were in the war”) This phenomenon is well known from the logophoric languages of Africa and is usually described in a way stating that a singular antecedent may ‘trigger’ a plural logophoric pronoun if its referent is included in the set denoted by the pronoun, while a plural antecedent never triggers a sin62 Logophonic pronouns in Finnish and High Latvian gular logophor (cf. Sells 1987: 449; Roncador 2006: 313). However, the use of the word ‘trigger’ seems somewhat misleading to me. In my eyes, number assignment simply follows from the functioning of logophoric pronouns within reported discourse and reflects their closeness to first person pronouns: a plural logophor refers to the reported speaker and other persons, just as a first person plural pronoun refers to the current speaker and others. In other words, a plural anaphoric pronoun replaces a ‘we’ of the reported utterance. And while it is natural for a single speaker to speak of himself and others as ‘we’, it would be odd for a group of speakers to refer to one of them as ‘T’. In both cases (first person plural and plural logophoric pronouns) we may semantically distinguish between inclusive and exclusive, but this distinction need not be formally reflected in the language. In fact, in Finnish and High Latvian, there is no such distinction in neither personal nor logophoric pronouns. The examples above both illustrated an exclusive ‘we’ (reported speaker and non-participants of the reported speech act), while the following is an example of an inclusive ‘we’ (reported speaker and addressee of the reported utterance): High Latvian, Varaklani (Jokubauska 1988: 139) (7) sokeite,,,, nu niu piéc Sus... ,) dzandas pakal. Say:PAP:F.SG PTC NOW after LoG:acc.M.pL chase:pst:3 after ‘[she,] said, now [they] were chasing after them, > (= ‘after us, you and me’) (nel, 1) Similarly, the gender of the pronoun in High Latvian is assigned according to the sex of the referent. As a result, there usually is gender agreement between the antecedent and the logophoric pronoun. In cases where the antecedent is a single female speaker but the logophor refers to her and at least one male person, the logophoric pronoun is marked plural masculine (see ex. (7)). Again, we find the same kind of gender assignment as in personal pronouns, while it makes little sense to speak of a feminine antecedent “triggering” a masculine logophoric pronoun. 63 Nicote Nau for canonical logophoric constructions with a verb ‘to think’ or ‘remember’: Jaaski-Kirvu [Southern Karelian; Southeastern group] (Ikola 1960: 169, after Sirelius 1894) (15) Pekka, muist sitt, jott onha hdnell, Pekka remember:pst:3 then comp be:prs:3:pTc —LOG:ADE toinekii vel’. second:prc brother ‘At that moment Pekka, remembered that he, had yet a second brother.’ Laitinen (2005: 88-90) claims that Finnish hdn is used with predicates denoting various kinds of mental states as well as perception and even cause. Unfortunately, she does not specify the varieties which allow these uses, and her article deals not only with dialects. Her example (22) (Laitinen 2005: 88), with a verb ‘to hear’, seems to be from Standard Finnish. Judging from the examples in SMS, in Finnish dialects only predicates of communication and thought are widespread in logophoric constructions, while predicates further to the right-hand side of the hierarchy are found only in subdialects where the logophoric pronoun is also used in various non-logophoric functions. In High Latvian, the restriction to reported speech proper is still more pronounced. In my sources, there is no example in which a predicate denoting something other than a speech act introduces a logophoric context. The lack of examples, of course, is not proof that they are impossible. However, in the only sentence reporting a thought in a canonical construction, speaker Anna uses the anaphoric pronoun, not the logophoric: High Latvian, speaker Anna (AA 131) (16) A, jis, grybé, jis, dimuo, ka jam, inuokums byus PTC PRO want:pst:3 pro think:pst:3 comp pro:paT income be:rur:3 lyls, ka te is cylvaki, jam, moksuos. big:m.sc comp here go:rur:3 people pro:pat pay:FuT:3 “A, he, wanted, he, thought that he, would have a big income, that people would come and pay him,.’ Note that in this sentence, the verb is in indicative mood, not reportative. This example suggests that in High Latvian, the use of logophoric pronouns 70 Logophoric pronouns in Finnish and High Latvian is restricted to speech reports proper, in other words, that thoughts and mental states are not treated as communication. However, in constructions without an introductory verb it is not always evident that what is reported are spoken words, rather than thoughts. Consider the following example from fairy tale, where the hero sits alone in despair, because he has been given the task to pick out poppy seeds from the ashes: High Latvian, Varaklani (Jokubauska 1988: 138) (17) siesé D raudois, ka Sys var izlaseit. sit:pa:mM.sc and cry:pa:m.sc what:acc LoG can:prs:3 pick_out:inr ‘[He,] sat and cried, [thinking/saying?] how could he, pick them out.’ I suppose that in High Latvian, a semantic extension of the logophoric construction towards reported thoughts or feelings appears only outside of the canonical logophoric construction or in varieties in which the logophoric function is less grammaticalized. Further research is needed here. Summarizing the findings of this section, the following may be stated: First, there are several formal features that characterize the typical logophoric construction. However, it is not possible to define the use of logophoric pronouns in Finnish and High Latvian in formal terms. Most important in this respect is the assignment of number and gender, where logophoric pronouns behave like first person pronouns, their use in independent clauses, and their rather loose relation to an antecedent. Second, formal and semantic characteristics of logophoric constructions derive from the fact that they represent reported discourse: the regular presence of a report-opening element (a speech act verb with a tendency to fossilize in one form, a complementizer and/or particle), the restriction to predicates. of communication (in High Latvian) or communication and thought (in Finnish). Third, the formal properties of the logophoric pronouns in Finnish and High Latvian are not peculiarities of these two languages, but are paralleled in the logophoric languages of Africa, to which they are neither genetically nor geographically related. It is therefore safe to conclude that these are indeed the characteristics of logophoric constructions. A more detailed comparison of individual languages may reveal further parallels as well as differences, but this is beyond the scope of the present article. Instead, I will now turn to the place of logophoric constructions within a typology of reported speech. 71 Nico.e Nau 4. TYPES OF REPORTED DISCOURSE AND PERSON ASSIGNMENT The traditional binary classification of instances of reported discourse into direct and indirect speech is based on data from written standard varieties of European languages, where several grammatical and lexical features tend to cluster at two poles®. It has long since been noted that in spoken varieties, most notably colloquial speech and dialects, these features may combine in very different ways, making it difficult or even impossible to distinguish between direct and indirect speech (for Finnish dialects, see Ikola 1960: 235-237). As becomes evident from the discussion in section 3, logophoric constructions in Finnish and High Latvian dialects may be more similar to either of the two ideal types. In the canonical logophoric construction, they probably resemble indirect speech more, while in constructions with (more) independent clauses, absolute tense, indicative mood, and discourse particles, a speech report with logophoric pronouns is closer to direct speech. In these cases, it is only the pronoun that distinguishes a logophoric context from direct speech. Consider the following example, where the use of quotation marks is an indicator of how the linguist transcribing the text interpreted the utterance: Urjala [Northern Hime; Hame group] (Ikola 1960: 236, citing Kannisto 1902) (18) Sano “kylld te mennds saatte”, kylld hdl lapsen kattoo. say:prs:3 prc 2PL go:INF_ may:2pL prc L0G child:acc look:prs:3 ‘[she,] said: “you may leave”, she, would look after the child.” Note the exact parallel of the two coordinated clauses: both contain a discourse particle and neither is formally dependent. I suppose that it was only the use of pronouns and not a significant difference in intonation that led to the use of quotation marks for the first clause as opposed to the second. © There is vast literature on reported discourse; for a comprehensive bibliography see Giildemann 2002, for features associated with the distinction of direct and indirect speech, see contributions to Coulmas (ed.) 1986 and Ginthner 2000. ? Translation in analogy to Ikola’s German translation: ‘Sagte “ihr konnt schon gehen”, sie werde schon aufs Kind aufpassen’. This translation proves the interpretation of the logophoric context as indirect speech. It also illustrates the problem of translating spontaneous spoken language — the German translation reflects a rather formal register. 72 Logophoric pronouns in Finnish and High Latvian In a High Latvian dialect text I found the following (almost) minimal pair of reported utterances, which differ in that the original speaker is referred to by a logophoric pronoun and a first person pronoun respectively. In the transcription, both are graphically treated as direct speech: High Latvian, Varaklani (Jokubauska 1988: 138 and 139) (9) nu i sokute: prc and say:PA:F.sG —ki Suos tdus teu aizlyka Sinakt —struoduot? what:acc Loc:r.cen father 2sa:par order:pst:3 tonight work:nF ‘And [she,] said: What did LOG, father make you do tonight?’ (20) — nu, ki teu aizlyka munc tdus struoduot? prc what:acc 2sc:par order:pst:3 my:m.sc father work:inr ‘Now what did my father make you do?’ The rendering of the original speaker in a speech report may be taken as crucial for the distinction of different types of reported discourse and for the understanding of the nature of logophoric pronouns. This point has been neglected by those researchers who have focused on the opposition between logophoric and third person pronouns. In such accounts, emphasis is placed on the fact that logophoric pronouns as markers of coreference solve the ambiguity of third person reference that arises in indirect speech reports, where the reported speaker is treated as a third person. However, as noted in the introduction, there is another kind of potential ambiguity for which logophoric pronouns provide a solution, namely, ambiguity between the reported and the actual speaker, both of which are referred to by a first person pronoun in direct speech. This was explicitly pointed out by Stirling (1993: 257), as well as, more recently and with special emphasis, by Bhat (2004: 58-74), whose unorthodox account of logophoricity I found very inspiring. According to Bhat, the primary function of logophoric pronouns is to differentiate between the participants of the reported speech act and those of the current speech act (endophoric and exophoric speech act participants, in his terminology). According to how reference to the reported speaker (S.) is made in a speech report, we may thus distinguish three basic types: (i) S, referred to by a first person pronoun (direct speech), 73 NicoLe Nau (ii) S, referred to by a third person (anaphoric) pronoun (indirect speech), ii) S, referred to by a logophoric pronoun distinct from @ and (ii) (logophoric context). To my knowledge, only the first type is found in all languages. The second and third, therefore, are options that contrast with direct speech. In varieties where the logophoric pronoun is obligatory in a logophoric context, indirect speech as defined here is excluded. This may be an argument for Bhat’s above cited claim, as both (ii) and (ii) allow the distinction between endophoric and exophoric speaker. If there is a grammaticalized logophoric pronoun, there is no need for indirect speech. On the other hand, where there is a neat distinction between direct and indirect speech, as in European written standard languages, there is no need for logophoric pronouns. Instead, languages may develop means to solve the ambiguity that arises between several third persons in indirect speech. There are also varieties where all three types are attested, that is, apart from direct speech, a reported speaker may be referred to by either a logophoric pronoun (which in this case is not fully grammaticalized) or a third person pronoun. The next step towards a typology of reported speech along these lines is to look at the options available for reference to the addressee of the reported utterance (A,). The picture becomes more complicated here, though there are clear preferences. In Finnish and High Latvian logophoric constructions, the addressee of the reported speech act most often is referred to by asecond Person pronoun, as it is in direct speech. The same phenomenon is reported for logophoric constructions in African languages (Roncador 1988: 290). Most remarkably in my eyes is the fact that, as in typical direct speech, in a logophoric context the deixis of the original message often overrides the deixis of the actual message: a reported second person remains second person even when it refers to the current speaker or includes her’, In the following example from High Latvian, the speaker reports an utterance made to herself: “It is, however, also possible to use a first person pronoun in cases where the reported addressee is (or includes) the same Person as the actual speaker. 74 Logophoric pronouns in Finnish and High Latvian High Latvian, speaker Anna (AA 35) (21) [...] reita atit tdvs i muote, i jau tur morning:.oc come.prs.3 father and mother and prc_ there taida lauku such:F.sc country vecine it leidza jai, nu i atnas women go.prs.3 with pro:pat.F prc and carry.prs:3 mozu barnu small:acc child:acc i lik man, ka lai and let:prs:3 1sg:dar comp comp —vot Si atnazusi bruoleiti tev. prc LoG:p. bring:pap:p. brother:pim:acc Beotoar ‘in the morning, father and mother are coming, and such an old village woman goes with her, [she/they] carries/carry a small child and lets/let me, — [interrupts] - there, LOG brought you, a little brother.” In an English translation of this last clause, we have to choose between the models of direct and indirect speech: ‘we brought you a little brother’ or ‘they had brought me a little brother’. The following example from Finnish shows the same principle: Finnish, Ristijarvi [Kainu; Savo group] (Kuiri 1984: 119) (22) (mies,) sano jotta hdn, ei ldhej jotta man say:prs:3 comp L0G NEG:3. leave comp sind —_ossoat ite 2sc be_able:prs:2sc self ‘(the man,) said that he, would not leave, that I, could do it myself” or: ‘(the man,) ale “I, won't leave, you, may do it yourself”.’ There are two more options for marking the reported addressee in a logophoric context. First, as in indirect speech, it may be marked with a third ° It is not clear who is the reported speaker in this text and the subject of the verbs ‘carry’ and ‘let’: the mother, the village woman, or the group of the three adults. 75 Nico.te Nav Person pronoun. In the following example this option is chosen in the first clause, while in the second clause a second person pronoun is used. This example thus shows a switch between two models of person assignment (such switches are far from uncommon). High Latvian, Varaklani (Jokubauska 1988: 138) (23) lai jis, izlosa par nakti, a ka ne comp pro pick_out:prs:3 for night:acc and if not teu, reita golva byus nist. 2sc:par_morning:.oc head be:rut:3 down ‘[they, said] he, should pick out [the Poppy seeds] over night, and if not, in the morning he, would (literally ‘you will’) lose [his] head’ In my sources, a reported addressee in a logophoric context is referred to by a third person pronoun only in cases where the reported addressee is not referred to by a logophoric pronoun, that is, either reference to the speaker is omitted (as in example (23)), or the reported speaker is identical to the actual speaker and therefore referred to by a first person pronoun. Second, in some subdialects and non-dialectal varieties of Finnish and Latvian it is possible to use the logophoric pronoun (han, Sys) also for reference to the addressee of reported speech. I will call this option the ‘logophoric addressee construction’. Consider the following examples: Finnish, Lavansaari [Southern Karelian; Southeastern group] (SMS sub hdn) (24) Mie mdnen sanomaa_ Annil jos Isc go:prs:Is¢ sayzinr:uL Annizau if hdnel olliis aikaa tulla meil _ kohvil. LoG:apE be:cnp time:par come:iNF 1pLiau. coffee:Aut ‘I,’m going to ask Anni,, whether she, would have time to come to us for coffee.’ Finnish, Vihti [Western Uusimaa; Southwestern group] (SMS sub hdn) (25) se, kysyi isalt, sit osaak hdr, — ruattii. Pro ask:pst:3 father:as. then know:9 Loc, Swedish ‘Then he, asked father, whether he, knew Swedish.’ 76 Logophoric pronouns in Finnish and High Latvian It may seem odd that the same marker can be used with two different meanings, marking either the source or the receiver of a speech act. However, there are several factors that prevent ambiguity. First, in example (24), the reported speaker is a first person and the reported addressee is the only possible antecedent for the logophoric pronoun. Second, most often logophoric marking of an addressee is found where the reported speech act is a question or a command, that is, in a context where the addressee is more important for the speech act than with statements. It is also not common for a speaker to ask someone else questions about himself — for example, in (25) the speaker surely knows whether he himself speaks Swedish! Thirdly, this option, like the marking of the reported addressee by a third person pronoun, apparently is available only if there is no reference to the reported speaker in the same clause, or if the original speaker is identical to the current speaker and referred to by first person pronoun. That means it is used in contexts where it is only necessary to distinguish between endophoric and exophoric addressee, while there is no ambiguity problem for reported and actual speaker. Again, this may be considered an argument for the view that the primary function of logophoric pronouns is to solve this kind of ambiguity. I am not sure of the spread of the logophoric addressee construction in the two languages under consideration, but it is definitely much rarer than the use of hdn and Sys for a reported speaker and implies this use. Kuiri (1984: 122-123) states that it is not found in the dialects of her investigation (Kainuu and Northern Karelian, both in East Finland). In High Latvian, I found it only in varieties where the use of Sys to mark the reported speaker is not fully grammaticalized. The three options to mark a reported addressee (A_) (second person, third person, or logophoric) are thus dependent on the marking of the reported speaker (S.) and its relation to the actual speaker (S,). Taking this into account, the Finnish and High Latvian data may be summarized as follows: (i) A, is referred to by a second person pronoun, if S_ is marked as first person (direct speech), or S, is marked by logophoric pronoun (logophoric construction), or S_ is not expressed; (ii) A, is referred to by a third person pronoun, if Sis marked as third person (indirect speech), 77 Nicotre Nau or S, is marked as first person and S, = S., or S, is not expressed; (iii) A, is referred to by a logophoric pronoun (logophoric addressee construction) if S, is marked as first person and S, = S., or S, is not expressed. As can be seen, the first option is least restricted, and it is also by far the most frequent in the dialects with a logophoric pronoun. It is also attested in several logophoric languages of Africa. I will only briefly comment on further types of person marking in reported speech involving logophoric pronouns. First, some languages are less restrictive with the combination of logophors with third person pronouns: in Babungo, the reported addressee may be marked for third person when the reported speaker is referred to by a logophoric pronoun (Heath 2004: 1006); in Pero, logophoric reference to the reported addressee is compatible with third person reference for the reported speaker (Bhat 2004: 65). Second, a few languages are known to have two types of logophoric pronouns, distinguishing between reported speaker (LOG-S) and reported addressee (LOG-A). An example is Mupun (Culy 1997: 855). With these data in mind, we may derive the following typology of reported speech based on the marking of reported speaker and reported addressee. Given that for each reported participant there are three options, nine types are logically possible. Of these, three are well attested, three are marginal, and three are probably impossible. 78 Logophoric pronouns in Finnish and High Latvian Table 1. Marking of reported speaker (S,) and reported addressee (A,) (in cases where both are distinct from actual speaker and actual addressee) Preferred types (attested in several unrelated languages) Type Marking Schematic example “@) 8,=1st,A,=2nd | He, said to her,:I, love you,. a Direct speech (2) | 8, = 3rd, A, = 3rd He, said to her, that he, loves her,. Indirect speech | (3) $, = LOG, A, = 2nd | He, said to her, (that) LOG, loves you,. | Logophoric construction Marginal types: Variations of the logophoric construction (4) | S, = LOG-S, A, = LOG-A He, said to her, (that) LOG-S, loves LOG-A, - (Mupun) “(&) |S,=L0G,A,=3rd _| He, said to her, (that) LOG, loves her, (Babungo) (6) |S,=3rd,A,=LOG He, said to her, (that) he, loves LOG,. | @ero) Unlikely types, maybe impossible ”) | S, = 1st, A, = LOG He, said to her, (that) I, love LOG,. “(8) |S,=1st,A,=3rd __| He, said to her, (that) I, love her, (9) | 8, = 3rd, A, = 2nd | He, said to her, (that) he, loves you, A functional explanation for the distribution of these types is easy to find. Given that the first type, direct speech, is basic, all other types may be considered as dealing with the reference problem that it poses, namely, the distinction between reported (endophoric) and actual (exophoric) speech act participants. In this respect, types (2) and (3) are both successful and simple, while (4) may be considered redundant. Types (7), (8) and (9) do not solve the ambiguity, but rather increase it. 79 Nicote Nau a jis vinis vuordils pateics, ka Sys Vuoceja and Pro one:1oc.pl. word:10c.Pt, say:PAP.M.SG COMP LOG Germany:1oc tik bejs i nikuo Sys nazyns nikur. just be:pap.m.sc and nothing:acc Loc _NeG:know:Pa.m.s nowhere Nu tod paturiejusi, palaidusi jau ji prijjom. prc then keep:pap:m.p.. let:pap:M.PL PTC PRO:ACC away Tikai jis prasiejs, ka dit Sam— nu, dokumentu only pro ask:pap:M.sc comp give:inr LoG:paT prc document:acc kaidu, some:acc lai Suo naizturat cela. Nu jis i atguo kuojom, COMP Loc:acc NeG.detain:par way:Loc prc pro and come:pst:3 on_foot nu i tod jis prasiejs, lai didit sam dorbu. prc and then pro ask:pap:m.sc comp give:par Loc:paT work Tod ideve Jersika jam Par me&sorgu. then give.pst:3 Jersika:.oc pro:pat as forester:acc ‘He (pro) left Preili and went to Daugavpils, announced himself. Well, there they took him (pro), beat [him] and arrested [him], but he (pro) told them simply that he (log) had only been to Germany and that he (log) didn't know anything. Well, so they kept [him] for a while and then set him (pro) free. He (pro) only asked that they give him (log) — well, some kind of document, so that he (log) wouldn’t be detained on his way. So then he (pro) came here on foot, and then he (pro) asked that they give him (log) work. So they gave him (pro) [a job] as a forester in Jersika.’ (A2) Speaker Jékabs Stafeckis, born 1907 (text recorded in 1981) (Stafecka 1988: 150) This text shows the different kinds of functions of the logophoric pronoun and the demonstrative pronouns. The demonstrative itys, discontinuous form tys ... ite, is used for reference tracking, to distinguish one referent from the other. The logophoric 3ys, on the other hand, is not a reference tracking device. It is not used to show coreference with one of the noun phrases as such, but indicates coreference to the reported speaker — exactly which of the two gentlemen this speaker is, is not indicated at all, the listener infers this only through the logic of the story. 86 Logophoric pronouns in Finnish and High Latvian The text tells the family legend of how the Stafecki family had come from Poland to Latgalia. It was recorded and transcribed by the narrator’s daughter-in-law, dialectologist Anna Stafecka, who also kindly helped me to understand this intricate Passage, for which I am very grateful. Masalski_beja tepat. tdvainc sacé, ka senejiis _laikiis, Masalski was right_here father say.pst:3 comp old:.oc.pt time:1oc.pL. ka tys kunks ite,, vot, ir beis, adbraucs i comp per lord here prc aux:prs:3 be:pap.M.sc come:PAP.M.sG PTC nu Piilejys vince kunks, iz iti -kungu,, from Poland:cen one lord to pet:acc lord:acc a itam _ kungam, beis sunc, lops _ beis, and bet:par lord:par be:par dog good_be:pap.m.so soks, ka _ soka, atdit Sam, til suni, SaY:PA.M.SG COMP Say:pRs:3. give:INF LoG:DAT DET!ACC dog:acc Sys, teu, idiiikys vinu —— gimini atmeit, Loc 2sG give:FAP:M.sc one:acc family:acc exchange i tys, ar miru. to, vot, atvec a suni, and pro with peace:acc pre pre bring:pap:m.sc per:acc dog:acc addeus, give:PaP:M.sc acsyuties vinu _gimini, to Stapecki beisi. Sent:pap:m.s¢ one:acc family:acc prc Stafeckis:pL be:pap:m.pL ‘Masalski (a manor) was right here. My father said, in the old times, that this gentleman here,, well, a. gentleman, had come from Poland to this gentleman, (= of Masalski). And this gentleman, had a dog, a good one. Now (he, (= the guest from Poland)] said that [the other] should give him, this dog, he, would give him, a family (of bondsmen) in exchange, and he, agreed. So, {he,] brought the dog, gave it [to him,], sent a family, and these were the Stafeckis.’ Nicole Nau Uniwersytet im. Adama Mickiewicza Zaktad Battologii al. Niepodlegtosci 4, PL 61-876 Poznan, Polska naunicol@amu. edu. pl 87