The Personal Dative in Appalachian English as a Reflexive Pronoun ANASTASIA M. CONROY Abstract In this paper, we investigate the Personal Dative construction found in Appalachian English and Southern American English. In this construction, a pronoun is permitted to be coreferential with a local antecedent. We argue that the pronouns in the Personal Dative construction are a type of anaphor, despite their phonological similarity to full pronouns. We analyze the Personal Dative in light of the approach to SE anaphors taken by Reuland (2001), suggesting these are bound pronouns that do not take a theta role. This work suggests that the Personal Dative construction is not unique, but is another instance of a phenomenon found cross-linguistically. Introduction The Personal Dative construction is attested to in Southern American English and Appalachian English (Wolfram and Christian, 1976), and is shown in (1). 1. I got me some sandwiches The construction in (1) has recently been analyzed as an idiom (Webelhuth & Dannenberg, 2006) due to its apparently unsystematic distribution of properties. Furthermore, it has been noted for its curious properties with respect to Principle B (Fodor, 1992). It appears that the Personal Dative construction violates Principle B, as it allows a pronoun to be coreferential with a local antecedent. This requires either a novel analysis of the Personal Dative, or a revision of the generalizations concerning pronouns. Finally, although the Personal Dative appears to behave like the reflexive –self in many ways, the Personal Dative is not acceptable with some verbs that allow the selection of the standard reflexive, as in (2). 2. a. * I hurt me b. I hurt myself This leaves the exact distribution and characterization of these pronouns a mystery. In this paper, we investigate the properties of the Personal Dative construction, and argue it reflects a more general phenomenon found across languages. Specifically, we claim that although the pronoun in the Personal Dative construction appears to be a standard pronoun, it is actually a reflexive, demonstrating properties of one type of reflexive pronouns. Furthermore, we are able to capture its distribution in a systematic way, undermining the evidence this construction is idiomatic. Cross-linguistically, there are two different types of anaphors: simplex anaphors (SE anaphors) and SELF anaphors (Reinhart and Reuland, 1993). SELF anaphors (complex reflexives) are reflexives such as himself in English or zichzelf in Dutch. SE anaphors are reflexives1 are morphologically simplex, such as zich in Dutch. We argue that the Personal Dative is a case of an SE anaphor in Appalachian English. 1 In this paper, we use the terms reflexive and anaphor interchangeably. Furthermore, we use data from the Personal Dative in Appalachian English to reflect on a current debate on the treatment of SE anaphors. The attempt to explain the distribution of SE anaphors has led to two competing theories of theta-role requirements of SE anaphors. One account, proposed by Lidz (2001), claims that SE anaphors, when accompanied by the proper verbal morphology, take their own theta role, equal to that of their antecedent. A second account, proposed by Reuland (2001), claims that SE anaphors are dependent upon their antecedent, and therefore, take no theta role of their own. Appalachian English, because it allows the Personal Dative in a non-theta-role assigning position, uniquely decides between these two hypotheses, in support of Reuland’s theory. The paper proceeds as follows. In section one, we provide a description of some key properties of the Personal Dative. In section two, we demonstrate evidence that the Personal Dative found in AE is a reflexive. We show that the PD is a bound variable, and patterns much like reflexives in other languages. In section three, we show that the Personal Dative patterns like SE anaphors (like zich in Dutch) in a number of ways, including meaning and distribution differences the SELF anaphor, himself. In section four, we review the two accounts of theta role assignment for SE anaphors, proposed by Lidz (2001) and Reuland (2001). In section five, we use the data concerning the Personal Dative in AE to argue that SE anaphors do not take a theta role, supporting the analysis by Reuland. Finally, in section six, we explore some remaining issues concerning case assignment and SE anaphors. 1. Description of the Personal Dative In this section, we will outline some data concerning the Personal Dative construction, and some mysteries surrounding the construction. The Personal Dative2 construction refers to a construction in which a pronoun is used in an object position, marked with accusative case, to be coreferential with its antecedent, as in as in (3a). In this paper, we will use the term ‘Personal Dative’ (PD) to refer to the pronoun in this position. These sentences have similar meaning to a sentence with the reflexive, as in (3b). 3. a. I bought me some sunglasses b. I bought myself some sunglasses This construction is used widely in Southern American English, however, we will primarily investigate the dialect specific to Appalachian English (AE), not because of dialectal differences, but because of the availability of documentation on the topic (Wolfram and Christian, 1976; Wolfram and Schilling-Estes, 1998). The Personal Dative is not restricted to first person pronouns. It is available in all persons and numbers, as shown in (4) (Webelhuth & Dannenburg, 2006). 4. a. I got me some candy i i 2 This construction is also referred to as the ‘ethical dative’ (Fodor, 1992), although it appears this term is used for historical purposes rather than for descriptive reasons. Ethical datives, a pronoun in dative form accompanied by a preposition, are found in other languages, such as Greek and German (Malone, 1988); however, it is not clear the phenomenon we discuss here is identical. b. We got us some candy i i c. You got you some candy i i d. He got him some candy i i e. They got them some candy i i The Personal Dative construction has a myriad of interesting properties that we will describe below, and return to in the course of this paper. First, the PD appears to be a violation of Principle B (Chomsky, 1981), as they appear to be pronouns that are coreferential with a local antecedent. Second, the PD occurs with a wide variety of verbs, but not others, leaving behind a curious distribution. Finally, the PD, although used in much the same way as a reflexive (5b), differs slightly in meaning (Fodor, 1992). Next, let us review each of these properties in turn. 1.1 Principle B Principle B (Chomsky, 1981) states that a pronoun may not be coreferential with an antecedent in its local domain. In General American (GA), this is the binding constraint that rules out interpretations like (5), because him can not be coreferential with an antecedent, Fred, in the same clause. 5. a. *Fred likes him i i b. *Fred gave him a headache i i The Personal Dative is of interest with respect to Principle B (Fodor, 1992). As we have seen, the Personal Dative object pronoun is coreferential with a locally c- commanding antecedent, as in (6). This appears to be a Principle B violation, because the c-commanding antecedent should not permit a local pronoun to corefer. Notice, this cannot be a property specific to constructions with indirect objects, as (5b) is of the same form as (6). 6. I bought me some sunglasses The Personal Dative is of the same phonological form as a full pronoun3, and in (6), is coreferential with a local antecedent. Therefore, one mystery is how Personal Dative construction is permitted to violate Principle B. There are three possibilities we examine. One possibility is that Appalachian English does not obey Principle B at all. The second possibility is that Personal Datives are pronouns that are exempt from Principle B. Finally, it is possible that Personal Datives are not pronouns at all. We support the latter, suggesting that the Personal Dative is an anaphor with the same morphological form as the pronoun, giving the appearance of a Principle B violation. In the next section, we will review some data concerning the Personal Dative and its distribution with respect to different verbs. 1.2 Verbal Selection Interestingly, the PD appears in the same distribution as he SELF anaphor, himself, for many verbs (7) - (9) (examples from Webelhuth & Dannenberg, 2006 and Wolfram and Christian, 1976). 3 We will return to the pronunciation of pronouns in AE in section 5.2. 7. a. She went into the store to get her a pair of shoes i i b. She went into the store to get herself a pair of shoes i i 8. a. I shot me a pheasant b. I shot myself a pheasant 9. a. I finally did buy me a coffee pot b. I finally did buy myself a coffee pot However, it is not the case that the Personal Dative is simply an altered pronunciation of the SELF anaphor. There is a second class of verbs in AE that permit SELF anaphors, but not the Personal Dative. In AE, the Personal Dative cannot be used as objects of the verbs in (10a) and (11a), whereas SELF anaphors are permitted in this position (10b) and (11b) (Examples from Wolfram and Christian, 1976). 10. a. * I hurt me b. I hurt myself 11. a. * We could see us in the mirror b. We could see ourselves in the mirror The examples in (10a) and (11a) also show that it can not be the case that AE is devoid of Principle B. If it were the case that Appalachian English had no constraint on reference of pronouns with local antecedents, then we would predict that these examples would be acceptable. Furthermore, AE contains verbs that permit the PD, but not SELF anaphors, as shown in (12) and (13). (Examples from Wolfram and Christian, 1976). 12. a. I’m gonna write me a letter to the president b. * I’m gonna write myself a letter to the president 13. a. I only need to sell me a dozen more toothbrushes b. # I only need to sell myself a dozen more toothbrushes These examples serve two purposes. One, they show that the Personal Dative is not simply an altered form of SELF anaphor4. Second, they show that the PD has its own distribution, which needs to be described. In the next section, we will turn to some meaning differences between the Personal Dative and the SELF anaphor that have been observed. 1.3 Meaning Differences We have just shown that the Personal Dative’s distribution cannot be described by the distribution of SELF anaphors. This suggests that the Personal Dative, at least in some way, is different from the SELF anaphor. There is a further difference between the two types of pronouns, and that is meaning. Fodor (1992) observed that the Personal Dative has a slightly different interpretation than the SELF anaphor. The example with the Personal Dative (14b) means that the whittling was for my own benefit, as in, I did it for the enjoyment of whittling. 4 It is not clear if SELF anaphor is used in AE to the same degree as it is used in General American, or if it is used as an emphatic. We put aside this issue here, but believe this is an issue of interest. However, (14a) does not have this interpretation, it only means that the stick went to me in the end, requiring a transfer of possession to the direct object. 14. a. I whittled myself a stick b. I whittled me a stick 15. I whittled Bob a stick Notice that the SELF anaphor, while requiring a transfer of possession, behaves identically to a full NP, as in (15). Therefore, we have an observation that the Personal Dative has a benefactive interpretation, which does not require the transfer of possession that is required in a typical double object construction. This interpretation is not available with either full NPs or SELF anaphors. To this point, we have eliminated the possibility that AE is a language that does not respect Principle B. This leaves two potential explanations as to the apparent Principle B violations. Either the Personal Dative is a pronoun that is exempt due to some special status, or the PD is not a pronoun at all. In the next section, we argue for the latter possibility and review evidence that the Personal Dative is an anaphor. 2. Evidence for the Personal Dative as a Reflexive The starting point of investigation of the Personal Dative is its behavior with respect to the SELF anaphor. We have seen that Personal Datives and the SELF anaphor have differing distribution and meanings, suggesting that, at least, the Personal Dative and the SELF anaphor are not fully interchangeable. Languages have at least two different types of anaphors: SE anaphors, such as zich in Dutch, and SELF anaphors, such as zichzelf in Dutch and himself in English (Reinhart and Reuland, 1993). We claim the PD is actually an SE anaphor, predicting its differences in distribution from the English SELF anaphor. First, we will review the evidence PD has reflexive properties. In this section, we begin by reviewing evidence (from Wolfram and Christian, 1976) that the Personal Dative cannot be obtained by a transformation from the benefactive construction. Then we show that the PD is a bound variable, and patterns with respect to possession like anaphors in other languages. This will lead us to conclude that the Personal Dative is an anaphor. In the next section, we investigate the Personal Dative’s status as an anaphor with respect to the two different types of anaphors we have just covered. We claim that the distinction found in AE between the PD and SELF anaphors (himself) patterns like the SE/SELF anaphor distinction noted across other languages (Reinhart and Reuland, 1993), suggesting the PD is an SE anaphor. Let us now turn to the investigation of the PD as an anaphor. 2.1 Against PD as a transformation from the for-Construction In the last section, we showed that the Personal Dative is not simply a replacement for the SELF anaphor in English. However, from the meaning differences between the SELF anaphor and the Personal Dative, we found that the Personal Dative has a benefactive meaning, that is, it does not require the transfer of possession. Therefore, before arguing that the Personal Dative is an anaphor, we must show that the PD can not be obtained by a transformation of a benefactive construction. As noted in the previous section, the PD has a benefactive interpretation, perhaps suggesting that it is a transformation from the benefactive for-construction, as (16a) has roughly the same meaning as (16b). 16. a. I bought me some sunglasses b. I bought some sunglasses for me The pair in (16) have similar meaning, and thus, are likely candidates to be transformations of one another. Notice that if the Personal Dative construction is derived via a transformation, this may be an indication as to why these pronouns have special status with respect to Principle B. Wolfram and Christian (1976) found that it is not the case that the Personal Dative occurs in the same distribution as the benefactive, as shown in (17). 17. a. I’m gonna write me a letter to the president b. I’m gonna write a letter to the president for Fred c. * I’m gonna write Fred a letter to the president Notice in (17a) that the Personal Dative can be used in this construction. Furthermore, the for-construction can be used, as in (17b)5. However, the for-construction can not be directly translated into the indirect object construction, as shown in (17c). Therefore, there is not a one to one mapping between the for-construction and the indirect object Personal Dative. Although the Personal Dative frequently mirrors the distribution of for- constructions, this is an inadequate description of its distribution. This suggests that the PD is not derived from some other construction. Furthermore, this leaves the PD’s exemption from Principle B unexplained. We will review the evidence for the Personal Dative’s status as an anaphor next. 2.2 The Personal Dative’s Status as an Anaphor In this section, we will investigate some evidence for treating the Personal Dative as an anaphor. 2.2.1 Bound Variable We have seen that the Personal Dative is not able to be achieved by a transformation from the for-construction, leaving unexplained how the PD is a pronoun that is not subject to Principle B. One possible analysis is that these are regular pronouns that are exempt from Principle B by some special status. Under the binding theory formulated by Grodzinsky and Reinhart (1993), only bound variables are subject to binding principles. Therefore, it is conceivable that the Personal Dative avoids being subject to the restrictions of binding by strictly being coreferential with its antecedent. In this section, we show that the PD is a bound 5 One may question whether in (8b) a letter to the president is acting as one NP, in which case, the structure would be V NP PP. However, the relevant contrast is (8a) and (8c), which holds as a contrast, suggesting a difference between a full NP and the Personal Dative. variable, behaving just like an anaphor. This suggests that these are not pronouns with some special status with respect to Principle B, but are actually anaphors. Let us explore the possibility that the Personal Dative is a standard pronoun, but is somehow exempt from Principle B. Because only bound variables are subject to binding constraints, a pronoun can be coreferential to a locally c-commanding antecedent and evade Principle B6. The coreferential interpretation can be seen under ellipsis. In (18a), the pronoun can be taken to refer to Mary, and thus, under ellipsis, the interpretation of the conjunct clause is Mary likes Mary, called the strict interpretation. 18. Jon likes Mary, Susan likes her, and Mary does too! a. OK Mary likes Mary [strict interpretation, coreferential] b. * Mary likes herself [sloppy interpretation, bound variable] The coreferential interpretation yields a strict interpretation under ellipsis, whereas the bound variable interpretation yields a sloppy interpretation. If the Personal Dative were exempt from Principle B because it is not a bound variable, and falls under an exception, then we would expect the bound variable interpretation to be unavailable. That is, under ellipsis, we would not expect sloppy interpretations, because bound variables are subject to Principle B. This expectation is not borne out, as shown in (19). 19. I bought me a pair of shoes, and Nell did too OK Nell bought herself a pair [sloppy interpretation] The Personal Dative is permitted with the sloppy interpretation7, suggesting it is a bound pronoun. This is irreconcilable with the idea that the Personal Dative is a pronoun with a special status with respect to Principle B. We suggest that the Personal Dative is an anaphor. In this case, we would expect the PD to be locally bound. This is exactly what we see in General American, where anaphors must be locally bound. With the SELF anaphor, the sloppy interpretation is obtained, as in (20). 20. Bill shaved himself, and Frank did too OK Frank shaved himself [sloppy interpretation] Reflexives require a bound variable interpretation, giving rise to sloppy interpretations under ellipsis. The fact that bound variable interpretations are permitted with the Personal Dative suggests that it is an anaphor, and not a pronoun with an exception with respect to Principle B. We will present one more piece of 6 Of course, there are complications concerning how, exactly, a pronoun avoids Principle B by coreference, but we abstract away from that point here. Showing that the Personal Dative is a bound variable obviates the need for this depth of discussion. 7 We do not have data concerning if the strict interpretation is available as well. However, the interpretation of such results depends on the interpretation of ‘identity under ellipsis’. Because the Personal Dative is identical in morphological form to the pronoun, one may expect the strict interpretation with the elided pronoun available. evidence that the Personal Dative behaves like a reflexive. Then, we will refine our notion of reflexive to show that the Personal Dative behaves like a subclass of anaphors, namely the SE anaphors, observed in other languages. 2.2.2 Possession Recall, we discussed that the Personal Dative does not require transfer of possession. Interestingly, this same observation occurs with anaphors in benefactive constructions in other languages. In Kannada, a sentence in (21), containing the benefactive dative, can have two interpretations (Lidz and Williams, 2004). 21. a. Rashmi tann-age keek-annu suTTu-koND-aLu Rashmi self-DAT cake-ACC prepare-VRM.PST-3SF ‘Rashmi made a cake for herself...’ b. adare ad-annu tann-a taayi-ge koTT-aLu but it-ACC she-GEN mother-DAT give.PST-3SF ‘...but she gave it to her mother.’ That is, the interpretation of (21a) is available that the cake is for Rashmi. However, the interpretation that the cake was given to her mother, for the benefit of Rashmi is also available, as shown in (21b). These two interpretations are not available with full NPs as dative objects. 22. a. Nannu Rashmi-ge keek-annu suTT-u-koTT-e I rashmi-DAT cake-ACC prepare-PP-BEN.PST-1S ‘I made Rashmi a cake...’ b. ?? adare ad-annu nann-a taayi-ge koTT-e but it-ACC I-GEN mother-DAT give.PST-1S ‘...but I gave it to my mother.’ In (22), the only reading available is that Rashmi received the cake. The interpretation where the mother is the owner, for Rashmi’s benefit is unavailable. Just as we have seen in English, full NPs in a benefactive construction in Kannada require transfer of possession. However, the SE anaphor does not. Recall, that the Personal Dative also does not require the transfer of possession. This is further evidence that the Personal Dative is not a full NP, but is an anaphor. We also find further examples where the PD does not require transfer of possession. In (23), the friend is the owner of the present, and the benefit is to the person in the indirect object (Example from (Wolfram and Christian, 1976)). 23. He went to the store to buy him a present for his friend That is, with the Personal Dative, the object in the benefactive construction can be the possessor, permitting the indirect object (the Personal Dative) to be the benefactor. This is not available with full NP indirect objects. In this way, we see that the Personal Dative in AE patterns with regard to possession like the SE anaphor in Kannada. We see a similar phenomenon demonstrated in (24), where the purchaser is unspecified, but the benefit is to the person in the indirect object. 24. I only need to sell me a dozen more toothbrushes Clearly, in (24), if the speaker’s goal is to earn money, then the speaker can’t also be the purchaser. Therefore, there is no transfer of possession to the direct object, and the Personal Dative patterns like a reflexive in other languages. The evidence from interpretations under ellipsis and possession interpretations suggests that the Personal Dative behaves as an anaphor. By morphological form alone, it seems obvious that the Personal Dative, me or him must be an SE anaphor. We have also seen that with respect to transfer of possession, the PD hints at being an SE anaphor. However, in the next section, we show that the PD shares many additional distribution similarities with SE anaphors in other languages. This lends support to our claim that the Personal Dative in AE is a genuine SE anaphor. Finally, this resolves the Principle B mystery: Personal Datives appear in violation to Principle B because they are actually anaphors. 3. Comparison with other SE anaphors We have shown that the Personal Dative is an anaphor, not a pronoun. In this section, we show by distribution and meaning differences that the Personal Dative is a SE anaphor, and patterns exactly as we would expect with respect to the SELF anaphor, himself. In this section, we will look at Personal Datives, as compared with other SE anaphors. We suggest that the distributions of these two types of pronouns closely mirror one another, and argue that the Personal Dative is an instance of an SE anaphor in Appalachian English (AE). 3.1 Distribution with verbs Recall, the Personal Dative does not occur in the same distribution as the SELF anaphor. That is, although there are verbs which permit both, there are also verbs which only allow either the PD or the SELF anaphor. This pattern, too, is repeated when we turn to languages, like Dutch, that have both types of anaphors. Just as in AE, Dutch has verbs which permit both the SE anaphor, zich and the SELF anaphor, zichzelf, as shown in (25). (All Dutch examples from Reuland, 2001). 25. a. Oscar wast zich Oscar washes zich b. Oscar wast zichzelf Oscar washes himself Furthermore, in Dutch, we see there are cases where the SELF anaphor can be used (26), but the SE anaphor cannot be (27). 26. Oscar haat zichzelf Oscar hates himself 27. *Oscar haat zich Oscar hates zich Therefore, it appears that there are cases where the SE and SELF anaphors have differing distributions, just as the Personal Dative and SELF anaphors have differing distributions. We will wait until section (5) to give an account of the Personal Dative’s exact distribution of occurrence. For now, we simply notice that the pattern appears to closely follow that of other SE anaphors. Next, we will continue to examine the similarities between the Personal Dative and SE anaphors as support for our claim that the Personal Dative is an instance of an SE anaphor. 3.2 Meaning differences Recall that a meaning distinction in AE exists between the Personal Dative and SELF anaphors. This difference, too, is seen cross-linguistically. First, we will review the Dutch facts, then return to AE. In ‘Madame Tussaud’ contexts, we see that the SE and SELF anaphors permit different interpretations. Imagine the following situation. Ringo Starr walks into Madame Tussaud’s Wax Museum and sees a statue of himself in wax and a mirror. Ringo decides he needs a shave. 28. Ringo shaved himself It has been observed that the pronoun in object position used to describe the act of shaving alters the available interpretations (Jackendoff, 1992). There are two possible interpretations of (28). Either Ringo shaved his statue (which is a representation of himself) or Ringo shaved his own body. In both Kannada (Lidz, 2001) and Dutch (Reuland, 2001), which have SE and SELF-anaphors, it has been observed that a sentence with a SELF anaphor, tann-annu-taane in (29), is ambiguous. It can either mean Hari saw his own person, or that he saw his statue. 29. Hari tann-annu-taane nood-id-a Hari self-ACC-self see-PST-3SM ‘Hari saw himself (= Hari or statue)’ 30. Hari tann-annu nood-i-kond-a Hari self-ACC see-PP-REFL.PST-3SM ‘Hari saw him (= Hari, *statue)’ However, a sentence with a SE anaphor, tann-annu in (30), can only mean that Hari saw his own body, and can not have the statue interpretation. SE anaphors require exact identity with their antecedent (i.e. the interpretation where Ringo must shave his own body), where complex antecedents need only some function of identity to its antecedent (i.e. a statue is a representation of Ringo). Said differently, there appears to be generalization that SELF anaphors can have both the statue and identity interpretations, while SE anaphors are restricted to only the identity interpretation.