1 Introduction

Ever since Lees & Klima (1963) inaugurated the study of English reflexives in Generative Grammar, a popular thread of theoretical development, culminating in Chomsky (1986), has emphasized the role of narrow syntax in the interpretation of reflexive pronouns. The common wisdom expressed in the Binding Theory (Chomsky 1981; 1986) says that reflexives should be structurally c-commanded by their antecedents in a local domain.

However, early work on the topic took notice of a class of phenomena which appear to resist simple syntactic accounts:

(1) a. *Yoko showed John to himself in the mirror.1
  b. *Jane talked about Paul to himself.
  c. *George was pleasing to himself in that concert.

I follow Jackendoff (1969; 1972) in referring to data like (1) – originally discussed by Postal (1968) and Lakoff (1968) – as “unexpectedly bad cases of reflexivization”. Those reflexives are unexpectedly bad precisely because they are not ruled out by any obvious syntactic condition, given that they are fine in (apparently) parallel syntactic environments, such as (2):

(2) a. Yoko showed John himself in the mirror.
  b. Jane talked to Paul about himself.
  c. George was pleased with himself in that concert.

These unexpected reflexivization contrasts (URCs), as l’ll call them, inspired Jackendoff (1969; 1972) and others (Wilkins 1988; Dalrymple 1993; Williams 1994) to pursue semantically based solutions, stated in terms of a thematic hierarchy. Their idea is that there must be an independent principle that ensures reflexives cannot be higher than their antecedents in a suitably arranged hierarchy of thematic roles.

Partly motivated by the assumption that thematic roles are transparently encoded in phrase-structure configurations (Baker 1997), proponents of purely syntactic accounts of anaphora appealed to more abstract properties of narrow syntax to distinguish (1) and (2). This was the strategy adopted by Postal (1968) and Lakoff (1968) themselves, who hypothesized that the strings in (1) were transformationally derived from structures like (2), in violation of the Crossover Principle. More recently, Belletti & Rizzi (1988), Larson (1988; 1990) and Pesetsky (1995) all argued in favor of distinct underlying structures for cases analogous to (1) and (2). These were arranged so as to entail that, at the level where the Binding Theory (Chomsky 1981; 1986) applies, only the reflexives in (2) would be c-commanded by their antecedents.

In this article, I will provide an explanation for URCs like (1)–(2) by pursuing a different kind of approach. The work of Reinhart & Reuland (1993) (R&R), insofar as it does not essentially rely on a thematic hierarchy or on phrase-structure-related notions of licensing, opens the possibility of greatly simplifying the structures that a theory of English reflexives requires from syntax. When augmented with an enriched conception of compositionality (Nunberg 1979; Pustejovsky 1995; Jackendoff 1997; Recanati 2010) and a theory of logophoric discourse roles (Sells 1987; Stirling 1993; Minkoff 2004; Oshima 2007; Charnavel 2019), this version of R&R’s reflexivity-and-chains approach makes it possible to reduce the problem of accounting for URCs to the more tractable problem of logophoric reflexives. The sensitivity of reflexives to perspective-based properties is a typologically robust and reasonably well understood phenomenon. It is, thus, methodologically convenient if URCs can be explained in similar terms.2

The main assumption needed will be a modified version of R&R’s Condition A (CA). This modified CA states that reflexive markers signal that a predicate is reflexive (i.e. that two of its arguments are coindexed at the level of semantic interpretation). This is essentially a correspondence rule – in the sense of Jackendoff (1997) – which encodes the lexical meaning of reflexive pronouns like himself.

My approach to these issues is inspired by Menuzzi (1999; 2004), who reformulates R&R’s theory in important respects. Following him, I argue that logophoric reflexives should not be treated as exempt from (my version of) CA, but as tolerable violations of it. Under adequate circumstances – e.g., when they occur within anti-reflexive predicates (predicates which cannot be interpreted as reflexive) – these violations give rise to logophoric interpretations. If (1)–(2) contain anti-reflexive predicates (as I’ll show), the reflexives therein count as logophoric.

Adapting scattered suggestions by Jackendoff (1992), Safir (2004) and others, I will claim that reflexives in the URCs are not extensionally equivalent to their antecedents: one of them always comes to denote an inanimate proxy (an image or a more abstract representation). A minimal condition on logophoric reflexives is that they should refer back to animate discourse antecedents (cf. Sells 1987; Charnavel & Zlogar 2015; Charnavel & Sportiche 2016). Crucially, that condition is not met for (1) – as Kuno & Kaburaki (1977) first hinted at. The unexpectedly bad cases of reflexivization amount to failures in finding relevant logophoric antecedents. This provides a simple extra-grammatical explanation for URCs.

In Section 2, I present a broader range of URCs and argue against accounts based on the Thematic Hierarchy Condition (THC). In Section 3, I explore some of the semantic peculiarities of URCs and formulate a provisional descriptive generalization that captures the phenomena in a way that bypasses most of the counterexamples to the THC: the unexpectedly bad cases are the ones in which antecedents are interpreted as proxies. In Section 4, I discuss exceptions to this which suggest (along with other data) that reflexives in URCs exhibit logophoric properties. Finally, in Section 5, I present a revised version of R&R’s theory from which the logophoric properties of these reflexives comes out as a natural consequence.

2 Unexpected reflexivization contrasts and the Thematic Hierarchy Condition

In this section, I present a wide range of unexpected reflexivization contrasts (URCs) like (1)–(2) (repeated below) and argue against the Thematic Hierarchy Condition (THC) as an adequate explanation for them. I do this because, even though most would not accept the THC as a theoretical primitive (as Jackendoff (1969) did), the generalization it embodies is still taken to be a descriptively adequate account of URCs. These URCs fall into two groups: syntactic alternations (3)–(8) and semantically related predicates with different linking properties (9)–(13).

Dative alternations (Barss & Lasnik 1986; Larson 1988; Jackendoff 1990a):

(3) a.   Yoko showed John himself in the mirror.
  b. *Yoko showed John to himself in the mirror.

Freely ordered PP complements (Postal 1968; Jackendoff 1990a):

(4) a.   Jane talked to Paul about himself.
  b. *Jane talked about Paul to himself.
(5) a.   Jane heard from Paul about himself.
  b. *Jane heard about Paul from himself.
(6) a.   Jane argued with Paul about himself.
  b. *Jane argued about Paul with himself.
(7) a.   Jane was warned by Paul about himself.
  b. *Jane was warned about Paul by himself.

Passive alternations (Jackendoff 1969; Reinhart & Siloni 2005):

(8) a.   The artist painted herself in a realistic style.
  b. *The artist was painted by herself in a realistic style.

Verbs of communication (Levin 1993; Culicover & Jackendoff 2005):

(9) a.   I told Ringo about himself.
  b. *I mentioned Ringo to himself.
(10) a.   I warned Ringo about himself.
  b. *I denounced Ringo to himself.

Psych-predicates: asymmetries between non-stimulus and stimulus subjects (Postal 1968; Belletti & Rizzi 1988; Grimshaw 1990; Pesetsky 1995):

(11) a.    George was pleased with himself in that concert.
  b. *?George was pleasing to himself in that concert.
(12) a.    Linda was worried about herself when she started to get sick.
  b.  *Linda worries herself when she starts to get sick.
(13) a.    Linda deliberately scared herself (by watching a horror movie).
  b.  *Linda scares herself.3

It is not a priori clear what relevant causal factor on reflexivization co-varies with the shifting judgments in the a–b pairs above. As Reinhart (1983) and Büring (2005) note, contrasts like these are unexpected for theories which assume that licensing conditions for reflexives should be formulated exclusively in terms of phrase-structure relations, since they do not seem to correlate with any obvious syntactic distinction. For instance, all of the reflexives in (3) and (8)–(13) are locally c-commanded by their antecedents – at least insofar as the constituent structure for these sentences is determined according to traditional constituency tests, independently of the binding facts. The reflexives in (4)–(7), on the other hand, are not c-commanded at all, due to the branching structure of PPs. This obviously makes the wrong prediction for (4a), (5a), (6a) and (7a).4

Based on this kind of reasoning, Jackendoff (1969; 1972) and Wilkins (1988) propose that reflexives in English are subject to the following semantic condition:

(14) Thematic Hierarchy Condition (THC) (Jackendoff 1969: 80):
  A reflexive cannot precede its antecedent on the following hierarchy:
  Agent, Experiencer ≺ Location, Source, Goal ≺ Theme

Given usual assumptions about the assignment of thematic roles, the THC adequately describes URCs like (3)–(13): e.g. (3b), (4b), (9b), (10b) are all cases where the reflexives are Goals and their antecedents are Themes, in violation of (14).

The only thing purely syntactic accounts of reflexivization can do in light of these facts is to say the pairs in the URCs are, despite appearances, not syntactically alike after all (Belletti & Rizzi 1988; Pesetsky 1995; Larson 1988; 1990). These approaches try to motivate, at least for a subset of URCs, more abstract underlying structures where the antecedents do not c-command the reflexives in the bad cases of URCs. As Belletti & Rizzi (1988) and Larson (1990) acknowledge, this can be done by syntacticizing the thematic hierarchy, in accordance to some version of the Uniformity of Theta Assignment Hypothesis (UTAH) (Baker 1997).

However, besides being ultimately stipulative and inheriting all of the woes of taxonomic theories of thematic relations (Dowty 1991), the THC – either as a primitive condition on semantic structure or as a descriptive corollary of UTAH coupled with a syntactic Binding Theory – is also empirically misguided. This suggests that neither the abstract structures invoked in UTAH-based approaches nor the independent semantic condition proposed by Jackendoff (1969; 1972) can be motivated by their role in accounting for URCs.5

2.1 Arguments against the Thematic Hierarchy Condition

I will present here three empirical objections to the THC. The first of these exploits the fact that the alleged effects of the THC are not generalizable to forms of bound anaphora other than reflexives. The examples below, inspired by Safir (2004: 142), modify (3) and (12) by replacing the reflexives for reciprocals and bound pronouns:

(15) a. Yoko showed [Ringo and Paul]Goal [each other]Theme
  b. Yoko showed [Ringo and Paul]Theme to [each other]Goal
(16) a. [Every Beatle]Experiencer was worried about [his manager]Theme
  b. [Every Beatle]Theme worries [his manager]Experiencer

The fact that there are no counterparts to URCs for (some instances of) reciprocals and bound pronouns is only problematic for approaches which accept the THC as a generalization under some version of the classical Binding Theory (Chomsky 1981; 1986). These syntactic accounts explicitly purport to subsume all bound anaphors under the same c-command requirement. To be fair, neither Jackendoff (1969; 1972) nor Wilkins (1988) – major proponents of a non-syntactic THC – attempt any such generalization of the THC beyond reflexives. But this is also a limitation of their theories, insofar as they give no principled reason as to why the THC should apply only to a restricted class of lexical items.6

The second and more serious empirical difficulty with the THC is that it literally predicts clear contrasts where there are none. This is manifested in some dative shifts with exchange verbs (17) (Büring 2005: 17), inversions of freely ordered PPs (18) (Kay 1994: 288) and passive alternations (19) (Pollard & Sag 1992: 298).

(17) a. The queen sold [the slave]Goal [himself]Theme
  b. The queen sold [the slave]Theme to [himself]Goal
(18) a. The letter must be addressed by [the applicant]Agent to [himself]Goal
  b. The letter must be addressed to [the applicant]Goal by [himself]Agent
(19) a. The only barber [who]Agent shaves [himself]Theme is Figaro
  b. The only barber [who]Theme was shaved by [himself]Agent was Figaro

According to the THC, (17b), (18b) and (19b) should count as ungrammatical: in all of these examples, as the subscripts for thematic roles make clear, a reflexive corresponds to a thematic role which is higher on the hierarchy in (14) than its antecedent. This outcome is particularly curious since dative alternations, freely ordered PPs and passives were part of the initial motivation for the THC in the first place (Jackendoff 1969). The fact that there are no clear URCs in (17)–(19) means that THC is not an adequate account of the URCs in (3), (4)–(7) and (8) either.

There is another instance of undergeneration by the THC which, to my knowledge, has gone unnoticed in the literature: reflexives in URCs scenarios akin to (3)–(13) are never severely unacceptable with first person antecedents.

(20) a. Yoko showed [me]Goal [myself]Theme in the mirror
  b. Yoko showed [me]Theme to [myself]Goal in the mirror
(21) a. Jane heard from [me]Goal about [myself]Theme
  b. Jane heard about [me]Theme from [myself]Goal
(22) a. Jane was warned by [me]Agent about [myself]Theme
  b. Jane was warned about [me]Theme by [myself]Agent
(23) a. Sam warned [me]Goal about [myself]Theme
  b. “All of this had now been revealed to me by the mouth of José Dias, who had denounced [me]Theme to [myself]Goal
    (Dom Casmurro translation by John Gledson)

What undermines the THC is the fact this change in grammatical person keeps thematic relations constant. There seems to be something special about the first person in that it does not produce URCs where these are expected to arise.

An opposite result obtains when the antecedents in URCs are replaced by inanimate entities, such as institutions. In those cases, even THC-abiding structures are deviant. The examples below come from Kuno & Kaburaki (1977: 652):

(24) a. *John wrote to [Yale]Goal about [itself]Theme (cf. (4a))
  b. *John told [the company]Goal about [itself]Theme (cf. (9a))

Data like (24) suggest that the THC is also faulty of overgeneration.

The third type of counterexample to the THC consists in cases where it fails to predict real existing URCs. Even though, in principle, all NPs can be interpreted as implicit proxies of their literal denotations (e.g. as statues) (cf. Nunberg 1979), Jackendoff (1992) notes that, for (25a) and (26a), the reflexive is deviant when its antecedent has a proxy reading (Nunberg 1995; Hornstein 2001; Safir 2004):

(25) [Context: Ringo and I were drunk in the wax museum. When we got next to the statues of The Beatles, I accidentally bumped into John and…]
  a.   Ringo fell on himself.
  b.   ⟨plain⟩ [Ringo]Theme fell on ⟨statue⟩ [himself]Goal
  c. *⟨statue⟩ [Ringo]Theme fell on ⟨plain⟩ [himself]Goal
(26) [Context: I got mad at Ringo in the wax museum. As soon as we got next to the statues of The Beatles we got into a fight and…]
  a.   I pushed Ringo into himself.
  b.   I pushed ⟨plain⟩ [Ringo]Theme into ⟨statue⟩ [himself]Goal
  c. *I pushed ⟨statue⟩ [Ringo]Theme into ⟨plain⟩ [himself]Goal

It is not the case that (25c) and (26c) are merely contextually implausible interpretations for (25a) and (26a). One can easily devise an example which imposes an interpretation where the antecedent is a wax statue such as (27a). Nevertheless, a reading like (27b) is still impossible:

(27) a.   I melted Ringo over himself.
  b. *I melted ⟨statue⟩ [Ringo]Theme over ⟨plain⟩ [himself]Goal

Thematic roles are too coarse-grained to capture what is going on in these kinds of URCs: the pairs of interpretations in (25)–(26) are thematically identical. Furthermore, they all violate the THC, though only (25c) and (26c) are actually deviant. A different interpretive factor seems to be at work here: these reflexives are only bad when their antecedents are interpreted as a proxies. I will show in the next section that this generalization can be extended for the URCs which motivated the THC and that it avoids some (but not all) of counterexamples to the THC laid out so far.

3 Unexpected reflexivization contrasts and proxy functions

In this section I will begin to sketch a unified account for the patterns observed above. Specifically, the goal will be to formulate a tentative generalization that captures unexpected reflexivization contrasts (URCs) (including (25)–(26)) in a way that bypasses most of the problems posed by (15)–(27). My goal is not to provide a detailed analysis for any of these cases in particular, but simply to state a descriptive regularity which will then motivate a theoretical solution for URCs as a whole. The first step involves recognizing a property that is shared by URCs: all of these cases involve, in their semantic representations, the intervention of proxy functions.7

Proxy functions are functions from entities to their contextually salient proxies – i.e. entities that are “representationally related” to their function’s input (Lidz 2001: 130) (see Jackendoff 1992; Hornstein 2001). For example, picture′ is a proxy function that, when applied to ⟦John⟧, yields picture′(john): the contextually salient picture of John.8 The function story′, when applied to ⟦Sue⟧, yields story′(sue): the story about Sue, an abstract kind of proxy. Proxies can, accordingly, be concrete or abstract, depending on the representational relation they are parasitic on.

A way to detect whether proxy functions have been applied is to exploit their effects on inference – it is mainly because of these that we need to incorporate them into our semantic representations. Given that there is an intimate connection between representations and intensionality (Jackendoff 1975; 1980; Searle 1983), proxy functions should be capable of introducing referentially opaque contexts. This is indeed what we observe in scenarios that give rise to URCs.

Jackendoff (1975: 60–62) offers examples like (28), which are only possible because the two mutually inconsistent properties (having blue vs. brown eyes) are, in fact, attributed to two different entities: one is predicated of a girl and the other of her representational proxy. This duality is a telltale sign of referential opacity.

(28) In John’s painting/poem/mind, the girl with blue eyes had brown eyes.

Similar effects can be found in cases analogous to the implicit statue readings in (25)–(26). The sentence in (29) has a reading in which it is true, even if we take it to be a fact that Ringo Starr has blue eyes. This is so because the property of having brown eyes is attributed to a proxy (i.e. the value of a proxy function) which is distinct from plain Ringo.

(29) In the wax museum, Ringo has brown eyes.

The sentences in (30), which are reminiscent of examples in Russell (1905), are also not contradictory – as they would be if show and paint could not have intensional Themes. This suggests a proxy function (namely, image′) lurking in the semantic representations of URCs (3) and (8).

(30) a. That weird looking mirror showed ⟨image⟩ John taller than he is.
  b. The artist painted ⟨image⟩ herself older than she was.

Moreover, about-complements – but not to, from or by-complements – have de dicto interpretations for which substitution salva veritate and existential generalization are disallowed. These are typical diagnostics for opaque contexts.

Consider the substitution test first. If we accept that Zak Starkey’s father (i.e. Ringo) is, in fact, the sole blue-eyed Beatle, all of the equivalences in (31) follow with the exception of (31a). One simply cannot infer from Don told Sue about Zak Starkey’s father that Don told Sue anything whatsoever regarding the eye-color of a Beatle. That sentence (unlike the one below it) could be true in a scenario where Don does not know that Zak’s dad has blue eyes, or even that he was in The Beatles.

(31) a. Don told Sue about Zak Starkey’s father.
    (≢ Don told Sue about the only Beatle who has blue eyes.)
  b. Don mentioned Sue to Zak Starkey’s father.
    (≡ Don mentioned Sue to the only Beatle who has blue eyes.)
  c. Don heard it from Zak Starkey’s father.
    (≡ Don heard it from the only Beatle who has blue eyes.)
  d. Don was warned about it by Zak Starkey’s father.
    (≡ Don was warned about it by the only Beatle who has blue eyes.)

Note that it is the presence of a de dicto reading (and not the absence of a de re one) which is relevant as a diagnostic for intensionality. The de dicto reading for (31a) asserts a relation between Don, Sue and information about Zak Starkey’s father. Therefore, the particular mode of presentation of the about-complement is truth-conditionally relevant: it matters whether Don described Ringo as Zak’s father or as the blue-eyed Beatle. There is no analogous de dicto reading for (31b)–(31d).

Existential generalization also sets about-complements apart from other P-complements. The only phrase in (32) which could be coherently uttered by an atheist is (32a), as it merely entails the existence of information about a god. (32b)–(32d) all imply that the speaker believes (or has come to believe) that a god exists.

(32) [Context: I met George yesterday. In the middle of our conversation…]
  a. George talked about a god. (|≠ There is a god.)
  b. George talked to a god. (|= There is a god.)
  c. George heard some news from a god. (|= There is a god.)
  d. George was warned about the weather by a god. (|= There is a god.)

I take that the predicate responsible for these opacity effects in about-complements (which appear in (4)–(7), (9a), (10a) and (12a)) is the intensional proxy function information′, which maps any entity x into a piece of information involving x.

Statue readings also block existential generalization over the depicted objects: (29) does not entail that a non-proxy brown-eyed Ringo exists in reality. This is part of what makes fiction possible: one can succeed in referring to a representation of an entity without being committed to the existence of that entity in the real world.9

Though judgements are less clear, stimuli arguments in psych-predicates like (11)–(13) seem to fail existential generalization as well. Contrast (35), where the inference to (33) clearly does go through, with (34):

(33) There was a Nibiru cataclysm.
(34) a. A Nibiru cataclysm was pleasing to sensationalist youtubers. (|≠ (33))
  b. A Nibiru cataclysm worried Bill for a decade. (|≠ (33))
  c. A Nibiru cataclysm scared Sue when she was a child. (|≠ (33))
(35) A Nibiru cataclysm destroyed Mars. (|= (33))

This is expected if stimuli arguments (but not cause arguments such as the one in (35)) are interpreted under a proxy function like information′.10 This is roughly what Grimshaw (1990: Chapter 5) proposes in her analysis of these cases.

Proxy functions have various sources. Three of them will be relevant for URCs: the lexicon, type coercion and (enriched) composition. The first of these consists in cases where functions from entities to proxies are lexically encoded as the meanings of particular words. A likely example is the preposition about that appears in (4)–(7), (9a), (10a) and (12a). Montague (1974: 238), Dowty et al. (1981: 244) and Dalrymple et al. (1997: 264) suggest that this preposition is not semantically empty, but denotes an intensional operator of some sort. I will regard this operator as the aforementioned proxy function information′. If we take lexical items to be tripartite curryesque signs (Curry 1963; Jackendoff 1997; Culicover to appear), a lexical entry for about can be represented in the following attribute-value format (where alphabetical indices indicate correspondences):

    1. (36)

Assuming, with Heim & Kratzer (1998) and Büring (2005), that reflexives correspond to individual variables in semantic representation, the logical syntax of a PP like about himself in (4a) is, thus, information′(x).

Proxy functions can also emerge as a result of type coercion: i.e. “a semantic operation that converts an argument to the type which is expected by a function, where it would otherwise result in a type error.” (Pustejovsky 1995: 111). A possible circumstance for type coercion occurs when predicates place specific contextual restrictions on their arguments. This arguably happens in some of the URCs above.

The predicate denoted by paint in (8), for example, requires that one of its arguments be an animate Agent and, the other, a kind of representation. Following Pustejovsky (2012: 18–19), I will tentatively represent these selectional features as lexically specified typing restrictions on arguments – a proposal reminiscent of McCawley (1968; 1971), who treated selectional restrictions as presuppositions. This requires a lattice of expanded types, where rep and anim are both subtypes of e. The following is a partial lexical entry for the active form of non-locative paint (as it appears in (8a)), employing the same attribute-value format as (36):

    1. (37)

Since a reflexive pronoun does not lexically satisfy the subtyping restriction for being the Theme argument of paint′ (because its semantics is that of a mere e-typed variable), coercion is needed to avoid type underspecification in painted herself (cf. (8a)). The proxy function image′ comes to the rescue and maps the variable to the proper subtype, acting, thus, essentially, as an expression of type ⟨e, rep⟩. Likewise, if paint receives as its Theme argument a non-image individual like Sue, a proxy function has to intervene, yielding λy.paint′ (y, (image′(sue)). I assume something similar applies to the psych-predicates in (11)–(13), which also select instances of the representational subtype as their stimuli arguments.

Most typically, however, proxy functions come from the compositional procedure itself, taking into account how meanings are built up in interaction with their linguistic (and non-linguistic) context. The proxy function picture′ in (38) does not come about to coerce the variable into a subtype required by the verb – it is simply the contribution of the string that picture of to the meaning of the sentence:

(38) Max likes that picture of himself.
  like′ (max1, picture′ (x1))11

If semantic composition can be enriched by inferentially relevant unarticulated constituents (cf. Pustejovsky 1995; Jackendoff 1997; Recanati 2010), the same can be said for statue scenarios like (25)–(26). In (39), for instance, even though there is no specific string in narrow syntax which encodes a proxy function – nor is a proxy function required to satisfy the subtyping restrictions on the arguments of admire′ – an unarticulated constituent corresponding to the proxy function modifies the variable and maps its referent into a statue, yielding (39b):

(39) [Context: When we got next to the statues of The Beatles in the museum…]
  a. ⟨plain⟩ Ringo began to admire ⟨statue⟩ himself.
  b. begin′ (admire′ (ringo3, statue′ (x3)))

This also seems to be an adequate description for (3) (i.e. Yoko showed John himself in the mirror) and other cases involving transitive perception verbs (e.g. John saw himself in the mirror). As a reviewer notes, the Themes in these perception predicates are only circumstantially interpreted as images (i.e. as image′ (john)). This suggests that proxy functions are not lexically required by these predicates, but, in fact, emerge as optional unarticulated constituents, giving rise to intensional effects such as the one observed in (30a).

These “unarticulated proxy functions” which appear in (3), (25)–(26) and (39) are licensed by a construction which shifts the usual reference of an NP to an entity that is related to the NP’s literal denotation (cf. Nunberg 1979; 1995). What this construction does, in effect, is to embed the independent semantics of an NP into the scope of a proxy function P . This general proxy construction, which licenses the proxy function in (39), is stated as follows:

    1. (40)
    1. The implicit proxy construction:12

On this account, the proxy construction is licensed for any entity-denoting NP. In this respect, I differ from Hornstein (2001) and Lidz (2001), who take proxy functions in cases like (39) to be explicitly encoded by the -self morpheme. In my view, these proposals have the very counter-intuitive consequence of dissociating proxy readings like (39) from virtually identical readings which are licensed for non-reflexive NPs, such as pronouns and proper names: e.g. I saw Ringo in the wax museum. He looked very lifelike (see also Safir 2004: 113).

One interesting piece of evidence for this treatment of unarticulated proxy functions as parts of an NP construction is that controlled subjects and null objects do not allow proxy readings, as the contrasts in (41) and (42) show (cf. Lidz & Idsardi 1998: 122). This is expected if we take proxy readings to be the result of a proxy construction which is only licensed for overtly realized NPs.

(41) [Context: Ringo and I were in the wax museum. When I accidentally bumped against the statues of The Beatles …]
  a.   Ringo expected to fall.
    *⟨plain⟩ Ringo expected ⟨statue⟩ ∅ to fall
  b.   Ringo expected himself to fall.
      ⟨plain⟩ Ringo expected ⟨statue⟩ himself to fall
(42) [Context: Ringo and I were in the wax museum. All of a sudden…]
  a.   Ringo began to undress.
    *⟨plain⟩ Ringo began to undress ⟨statue⟩ ∅
  b.   Ringo began to undress himself.
      ⟨plain⟩ Ringo began to undress ⟨statue⟩ himself

Since both reflexives and their antecedents are NPs whose semantic type is e, they can always, if necessary, supply their meaning as an argument to a proxy function.

Proxy functions are, thus, a common property of the sentence pairs that participate in URCs. Regardless of where these functions come from (the lexicon, type coercion or semantic composition), the resulting predicate is always an instantiation of the schema in (43), where R is a meta-variable for n-ary predicates (for any n ≥ 2), P is a meta-variable over proxy functions, n is a meta-variable over indices and τ and υ are meta-variables over terms (individual constants or variables):

(43) R(τn, P(υn))

Given what I said about (39), it is clear how this works for the URCs in (3) and (25)–(26), repeated below:

(44) a.   Yoko showed ⟨plain⟩ John ⟨image⟩ himself in the mirror.
      show′ (yoko, john2, image′ (x2))
  b. *Yoko showed ⟨image⟩ John to ⟨plain⟩ himself in the mirror.
      show′ (yoko, x6, image′ (john6))
(45) a.   ⟨plain⟩ Ringo fell on ⟨statue⟩ himself
      fall′ (ringo3, statue′ (x3))
  b. *⟨statue⟩ Ringo fell on ⟨plain⟩ himself
      fall′ (statue′ (ringo1), x1)
(46) a.   I pushed ⟨plain⟩ Ringo into ⟨statue⟩ himself
      push′ (me, ringo19, statue′ (x19))
  b. *I pushed ⟨statue⟩ Ringo into ⟨plain⟩ himself
      push′ (me, statue′ (ringo25), x25)

Though less obvious, I contend that the lexical URCs in (4)–(12) (leaving (13) aside from the moment) are also instantiations of (43). For these cases, the proxy functions are either driven by type coercion or by simple functional application of information′, which, as I argued, is a proxy function lexically encoded by about. The sample representations below are based on proposals by Kuno & Kaburaki (1977), Grimshaw (1990), Jackendoff (1992), Pustejovsky (1995) and Safir (2004):

(47) a.     Jane talked to Paul about himself.
        talk′ (jane, paul35, information′ (x35))
  b.   *Jane talked about Paul to himself.
        talk′ (jane, x43, information′ (paul43))
(48) a.     Jane heard from Paul about himself.
        hear′ (jane, paul10, information′ (x10))
  b.   *Jane heard about Paul from himself.
        hear′ (jane, x3, information′ (paul3))
(49) a.     Jane was warned by Paul about himself.
        warn′ (paul11, jane, information′ (x11))
  b.   *Jane was warned about Paul by himself.
        warn′ (x8, jane, information′ (paul8))
(50) a.     The artist painted herself (in a realistic style).
        paint′ (the-artist5, image′ (x5))
  b.   *The artist was painted by herself (in a realistic style).
        paint′ (x6, image′ (the-artist6))
(51) a.     I told Ringo about himself.
        tell′ (me, ringo3, information′ (x3))
  b.   *I mentioned Ringo to himself.
        mention′ (me, x9, information′ (ringo9))
(52) a.     George was pleased with himself in that concert.
        please′ (george5, performance′ (x5))
  b. *?George was pleasing to himself in that concert.
        please′ (x11, performance′ (george11))

As suggested by the opacity tests in (30)–(34), the verbs in these URCs require that their Theme arguments belong to representational subtypes of e – hence the pervasiveness of proxy functions. In (50), (51b) and (52) the proxy functions come from type-coercion. In the other cases, the proxy functions are simply the result of applying the semantics of about (i.e. information′) to the P-complements.

As for (13), note that even though (53a) does not contain a proxy function in its semantics (since its subject is an extensional cause), (53b) does, because stimuli arguments are, as suggested by (34), construed as abstract kinds of proxies.

(53) a.   Linda deliberately scared herself (by watching a horror movie).
      scare′ (linda18, x18)
  b. *Linda scares herself.
      scare′ (information′ (linda1), x1)

If URCs all involve proxy functions, we can formulate a more adequate generalization to replace the THC (cf. Jackendoff (1992) and Safir (2004) for similar ideas):

(54) Generalization:
  Proxy functions cannot be applied to the antecedents of reflexives.

This generalization fully accounts for (44)–(53) and avoids many of the counterexamples mentioned in Section 2. No contrast is predicted for (17)–(19) (repeated as (55)–(57)), simply because the predicates in question have no proxy functions in their semantic representations, given the absence of opacity effects.

(55) a. The queen sold the slave himself.
    sell′ (the-queen, the-slave32, x32)
  b. The queen sold the slave to himself.
    sell′ (the-queen, x9, the-slave9)
(56) a. The letter must be addressed by the applicant to himself.
    address′ (the-applicant20, x20, the-letter)
  b. The letter must be addressed to the applicant by himself.
    address′ (x13, the-applicant13, the-letter)
(57) a. The only barber who shaves himself is Figaro.
    shave′ (the-barber8, x8)
  b. The only barber who was shaved by himself was Figaro.
    shave′ (x16, the-barber16)

Even though (57b) shows that passives in general do not partake in URCs (and this militates against the THC), passives like (50b) do, since they involve proxy functions. The contrast here is between the passives of predicates with fully coextensional coarguments (cf. (58)), which are reasonably acceptable (perhaps requiring special accenting), and the passives of predicates whose Theme arguments are representational (cf. (59)), which are unequivocally bad:

(58) a. ?John was washed by himself.
      wash′ (x22, john22)
  b. ?John was killed by himself.
      kill′ (x14, john14)
(59) a. *John was seen by himself on TV.
      see′ (x51, image′ (john51))
  b. *John was sculpted by himself in wax.
      sculpt′ (x9, statue′ (john9))

This may be at the heart of what Lees & Klima (1963), Langacker (1969), Jackendoff (1969) and other early work perceived as problematic with reflexives within by-phrases. The problem is, however, more fine-grained than what the vocabulary of thematic roles can express: the passive alternation, which turns Themes into antecedents, does not yield universally bad results, given (57) and (58). URCs only emerge when these passivizable Themes are embedded within (intensional) proxy functions, as in (59). This is exactly what the generalization in (54) predicts.

But (54) is not without problems. First of all, the reciprocal and bound pronominals in (15)–(16) are still somewhat of a puzzle. Since (54) merely replaces the THC as a descriptive statement, it does not offer any compelling reason for why only reflexives fall under its domain. Moreover, it also fails to predict the acceptability profile of sentences parallel to URCs with first person and inanimate antecedents like (20)–(24). These problems can only be solved when (54) is dropped in favor of a more explanatory framework which is capable of retrieving the effects that (54) gets right. This is the task I’ll undertake in the remainder of this article.

4 Logophoric properties in unexpected reflexivization contrasts

In this section, reflexives in unexpected reflexivization contrasts (URCs) will be shown to share crucial properties with the class of so-called logophoric reflexives: namely, an animacy restriction, a generalized acceptability in the first person and an allowance of split antecedents. Recognizing these common traits, along with the role of proxy functions discussed in Section 3, will be decisive for motivating a general explanation for URCs that overcomes the flaws of generalization (54). The proposal in Section 5 entails that reflexives in URCs are, in fact, logophoric.

Recall that generalization (54) leaves out two of the scenarios that I raised, in Section 2, as objections against the Thematic Hierarchy Condition: one involving first person antecedents and the other involving inanimates. Contrasts between these problematic cases and the data that fall under (54) are rearranged below.

(60) a. *Yoko showed John to himself in the mirror.
  b.   Yoko showed me to myself in the mirror.
(61) a. *Jane heard about Paul from himself.
  b.   Jane heard about me from myself.
(62) a. *Jane was warned about Paul by himself.
  b.   Jane was warned about me by myself.
(63) a. *I denounced Ringo to himself.
  b.   José Dias denounced me to myself.
(64) a.   Linda wrote to Paul about himself
  b. *John wrote to Yale about itself.
(65) a.   I told Ringo about himself.
  b. *John told the company about itself.

What is puzzling about (60)–(65) is that the acceptability contrasts are not due to proxy functions at all. Thus, (60)–(65) actually falsify generalization (54). In (60)–(63), the proxy functions all apply to antecedents, whereas in (64)–(65), they apply to reflexives. The only variable that changes within each pair is the nature of the antecedent: when the antecedent is the first person, the result is fine and, when it is inanimate, the result is bad, regardless of where the proxy function goes.13

The contrast between third and first person reflexives is reminiscent of data Ross (1970) presented as a motivation for his performative analysis. His examples explored contexts in which reflexives were only acceptable in the first person:

(66) a. *Physicists like himself don’t often make mistakes.
  b.   Physicists like myself don’t often make mistakes.
(67) a. *As for herself, she won’t be invited.
  b.   As for myself, I won’t be invited.
(68) a. *This is a picture of themselves.
  b.   This is a picture of myself.

Many have also observed that these same contexts also yield contrasts between inanimate and animate antecedents for reflexives (Kuno & Kaburaki 1977; Golde 1999; Minkoff 2004; Postal 2006; Charnavel & Sportiche 2016):

(69) a.   Winston Felix claimed that physicists like himself were a godsend.
  b. *The Nature of It All claimed that books like itself were a godsend.
(70) a.   Winston Felix claimed that as for himself, he would not grant the point in dispute.
  b. *The Nature of It All claimed that as for itself, it would not grant the point in dispute.
(71) a.   That picture of himself made folks think they should avoid Bill.
  b. *Those pictures of itself made folks think they should avoid the cave.

Though judgments are less sharp, Postal (2006) also mentions focalized reflexives (72) and topicalized reflexives (73) as subject to the animacy constraint. The as for x-self reflexives in (67) and (70) are probably related to topics as well. Relevant contrasts are provided below (small caps indicate strong stress):

(72) a.     It was HIMSELF that Winston Felix claimed ordinary people could never understand.
  b. ??It was ITSELF that The Nature of It All claimed ordinary people could never understand.
(73) a.     Winston Felix claimed that HIMSELF, ordinary people could never understand.
  b. ??The Nature of It All claimed that ITSELF, ordinary people could never understand.

What has generally been assumed is that the reflexives in these contexts – like x-self, as for x-self, picture-NPs and focus/topic – are conditioned by the perspective or empathy assumed in the discourse (Lebeaux 1985; Kuno 1987; Zribi-Hertz 1989; Pollard & Sag 1992; Charnavel & Zlogar 2015 i.a.). This is why they are called, somewhat loosely, logophoric reflexives (Reinhart & Reuland 1993), in analogy to the logophoric pronouns of African languages (see Clements 1975). These reflexives must refer to entities whose viewpoints are represented within their discourse contexts. The following contrast, based on Pollard & Sag (1992: 274), makes this perspective-sensitive nature particularly clear.

(74) a.     John was going to get even with Mary. That picture of himself in the paper would really annoy her, as would the other stunts he planned.
  b. *?Mary was quite taken aback by the publicity John was receiving. That picture of himself in the paper had really annoyed her.

The animacy constraint follows from the fact that bearing a perspective role in discourse – like John does in (74a) but not in (74b) – minimally entails animacy (see Minkoff 2004; Postal 2006; Charnavel 2019). The general acceptability in the first person follows from the speaker’s inalienable status as a possible – in fact, the default – perspective-bearer for every discourse (Sells 1987; Stirling 1993).

The two properties which were problematic for generalization (54) suggest, therefore, an observational correlation between logophoricity and the URCs data examined here. But the similarities between URCs and logophoric contexts do not stop there. Like the typical logophoric reflexives in (66)–(68), about complement reflexives are also (relatively) acceptable in the first person even in the absence of an overt antecedent (Reinhart & Reuland 1993: 715, Büring 2005: 233):14

(75) a. *Can you talk with myself about Lucie?
  b. ?Can you talk with Lucie about myself?

Another relevant trait is the possibility of having split antecedents. It is known since Lebeaux (1985) that, unlike normal coargument reflexives, typical logophoric reflexives have this property:

(76) a. *John protected Mary from themselves. (Keenan 1988: 220)
  b.   John protected Mary from mobsters like themselves.  

Judgments like (77), also reported by Büring (2005: 233) and Charnavel & Sportiche (2016: 58), suggest that some of the reflexives in URCs might be similar to logophoric reflexives in this respect too – though Wasow (1979: 25) and Hornstein (2001: 155) report a case like (77b) as bad.

(77) a.   Mary and Sue were sisters, and they looked exactly alike. Mary was the only one who claimed not to be able to see the resemblance. So one day, Sue showed Mary themselves in the mirror so that she could see their faces together and could compare them.                                                  (Okada 1998: 69)
  b.   John talks with Mary about important political ideas, while Bill only talks with Sue about themselves.                                                  (Okada 1998: 68)
  c.   Mary is John’s illegitimate child. John has been supporting her financially and emotionally since she was a very small girl, but their relation has been kept a secret up until now. Now that Mary has become a grown-up, the time will soon come when he must tell her about themselves, i.e., about their secret kinship.                                                  (Okada 1998: 68)
  d. ?Paul Simon and Art Garfunkel went to the wax museum. When they passed through the statue of Elvis and reached their statues, Paul made a rude remark about Art’s hair. They got into a big fight in which Art pushed Paul into Elvis and Paul pushed Art into ⟨statue⟩ themselves.

Reflexives in URCs share, thus, a number of crucial observable traits with logophoric reflexives. Since patterns in data invite the inference to a common natural kind, this already constitutes a weak motivation for saying that reflexives in URCs are logophoric. Charnavel (2019), however, argues for a stronger claim: given that animacy is a necessary condition for a reflexive to be logophoric (due to the nature of perspective), inanimate reflexives are always non-logophoric. From this it follows, conversely, that animate reflexives that cannot be replaced by inanimate ones are necessarily logophoric. What this means is that a restriction to animate antecedents can function effectively as test for identifying logophoric reflexives.

Therefore, according to Charnavel (2019), the properties mentioned above – in particular, the animacy restriction – are enough to count reflexives in URCs as logophoric. If this is true, a general solution to URCs comes for free: Since the antecedents of the bad cases in (44)–(53) are mapped into inanimate proxies (e.g. images, statues, information), they are not genuine perspective bearers. They can’t therefore, be linked to the types of discourse roles that license logophoric reflexives (see Sells 1987; Zribi-Hertz 1989; Golde 1999; Minkoff 2004; Charnavel 2019). These sentences would be bad for the same reason as (69)–(73).

The general acceptability of first person reflexives in URCs would also be predicted on these grounds, because the speaker is always independently available as a logophoric antecedent. This entails that the speaker does not lose its capacity to license a logophoric reflexive, even when a pronoun which refers to it is embedded within a proxy function which maps its referent into an inanimate proxy. The speaker’s omnipresence makes it immune to the effects of generalization (54).

But how can we make sense of this theoretically? Taking note of the phenomenal identity between URCs and logophoric contexts (as evidenced by their common properties discussed in this section) does not replace a theory of when and how logophoric readings emerge. In this respect, it seems puzzling to think of reflexives in URCs as instances of logophoric reflexives, because these are usually conceptualized as non-local, or as exempt from the locality embodied in CA (see Pollard & Sag 1992; Reinhart & Reuland 1993, i.a.).

However, this conception of logophoric reflexives is fraught with problems, as others have also argued (Dalrymple 1993; Menuzzi 1999; Charnavel & Sportiche 2016).15 In Section 5, we will see that these issues can be overcome if we treat logophoric reflexives not as exempt from CA but as but as tolerable violations of it. A similar theory was proposed, for independent reasons, by Menuzzi (1999). Besides accounting for the data, this view of logophoric reflexives enables a simplification of CA, which, in fact, turns it into a basic lexical property of reflexives. Insofar as Section 5 succeeds in presenting an independently motivated theoretical characterization of logophors which includes URCs, the logophoric properties laid out in this section will be explained (or, at least, be made less mysterious).

5 A theory of logophoric reflexives for unexpected reflexivization contrasts

This section is divided in two parts. In the first, I review some of the rationale that supports a reconceptualization of logophoric reflexives along the lines of Menuzzi (1999; 2004). The resulting alternative falls within the reflexivity-and-chains approach (Reinhart & Reuland 1993), but differs from its previous implementations in crucial respects. In the second part, I proceed to explain how this new view of logophoric reflexives extends to unexpected reflexivization contrasts (URCs), which were not hitherto addressed within the reflexivity-and-chains framework.

5.1 A reconceptualization of logophoric reflexives

In Reinhart & Reuland (1993) (R&R), logophoric reflexives like (66)–(73) are treated as exempt from the grammatical requirement embodied in Condition A (CA).16 This is meant to capture the fact that logophoric reflexives are subject to less stringent constraints than coargument reflexives. In particular, like ordinary pronouns, they need not have local antecedents (i.e. they can be free in the domain of the closest subject). In order to capture this property, R&R’s formulation of CA defines two circumstances in which reflexives can be exempt:

(78) A reflexive is exempt from R&R’s CA (Menuzzi 1999: 160):
  a.   when its predicate does not have a subject (in which case the predicate does not qualify as a syntactic one), or
  b.   when the reflexive itself is not a syntactic argument of a predicate (in which case it does not reflexive-mark anything).

Condition (78a) arguably covers reflexives within picture-NPs and like x-self constructions (assuming these are subjectless predicates).17 (78b) describes reflexives within as for phrases (which are plausibly topics) as well other kinds of A′-reflexives (e.g. (72)–(73)), none of which need be locally bound. This includes focus reflexives, which are overtly in argument positions but covertly move to A′-positions at LF, where CA is taken to apply. Example (79a) comes from R&R and (79b) comes from Safir (1992: 41):

(79) a. This letter was addressed only to myself.
  b. Did Jim tell Sue that Jane would kiss Bill in the moonlight?
    – No, Jim told Sue that Jane would kiss HIMSELF in the moonlight.

However, as Menuzzi (1999; 2004) notes, the theoretical characterization in (78) does not adequately circumscribe the entire class of reflexives that can have non-local antecedents, mainly because of two kinds of outliers:

(80) Reflexives within sentences with expletive subjects (Kuno 1987: 99):
  a. They made sure [that it was clear to themselves that this needed to be done].
  b. Paul claimed that [it would be good for himself if Mary left].
(81) Reflexives within contrastive predicates (Safir 1992; Golde 1999):
  a. Mary considered [her brother similar to herself].
  b. The veterans thought that [the new recruits would be more qualified than themselves].

The reflexives in (80) and (81) are all blatant violations of R&R’s CA: they reflexive-mark their predicates, which are nonetheless not reflexive. This means these reflexives cannot be treated as logophoric by R&R’s standards.18

But even though they do not fall under R&R’s formal definition of logophoric reflexives in (78), they do conform to their descriptive characterization of this category: they are locally free while not producing strong unacceptability like most cases of CA violations. Moreover, they also share with other logophoric occurrences an animacy restriction on their antecedents (Postal 2006: 12):

(82) a. *That book claimed that [there would always be itself for people to count on].
  b. *The Nature of it All considered [all other books worse than itself].

If R&R want to account for the relative acceptability of non-local antecedents, (80) and (81) should fall under the same rubric as their logophoric reflexives. Since the reflexives in (80) and (81) are syntactically indistinguishable from CA abiding ones, logophoric reflexives are better seen as tolerable violations of CA which are rendered acceptable under particular discourse conditions, as in Menuzzi (1999).

This makes it possible to strip away some of the complications that would need to be loaded into CA with the purpose of excluding non-local reflexives like (80)–(81) from its domain of application. In other words, if we no longer need to build into CA the adequate conditions for exemption, CA itself can be simplified. Modifying the formulation in Menuzzi (1999: 173) – in particular, abandoning the reference to syntactic reflexivity he inherits from R&R – I propose the following:

(83) Condition A (CA):
  A reflexive marker M (e.g. himself) must be the argument of a syntactic predicate which corresponds to a reflexive semantic predicate, where:
  a. a syntactic predicate P is formed of a head H and all of the syntactic arguments of P (including the subject of P, if there is one);
  b. a constituent X is a syntactic argument of P iff H subcategorizes or semantically selects X;
  c. a semantic predicate is reflexive iff at least two of its arguments (including the one that corresponds to M) are coindexed.

This CA is clearly violated in (80)–(81), as well as in the standard cases of logophoric reflexives subsumed by R&R’s (78), typical examples of which are given in (84):

(84) a. Bill felt that nobody appreciated that picture of himself.
  b. Marie said that Ernest hates physicists like herself.
  c. Sue promised that, as for herself, she would do all that she could.

The reflexive markers in (80)–(81) are syntactic arguments of their predicates, as per (83b). Nonetheless, none of these predicates map to reflexive semantic predicates. Since having a subject is not a requirement for being a syntactic predicate in (83), the picture NP and like x-self reflexives (84a)–(84b) violate (83) in the same manner: they are arguments of the syntactic predicates headed by picture and like, which, as we will see below, are actually infelicitous under reflexive interpretations. A′-reflexives like (84c), on the other hand, violate the CA in (83) simply because they are not syntactic arguments – the same reason why they were exempt for R&R.

The main innovation here is that, unlike R&R’s CA, (83) is not a condition on predicates, but on reflexive markers. This essentially turns CA into a lexical property of (English) reflexives – much like Chomsky (1981) did in his classical formulation of CA. Since this lexical property is, basically, a statement of the meaning of reflexive markers as reflexivizers of predicates, there is no need to interpret (83) as part of a dedicated Binding Theory or Reflexivity module of grammar. Moreover, for this same reason, the violation of (83) in logophoric readings is no more suspicious than coercion and other kinds of standardized non-literality (Bach & Harnish 1979; Levinson 2000), all of which can be seen as “violations” (or avoidances) of the lexically encoded meanings of expressions.

As it stands, the account here states that violating CA (instead of exemption) is a necessary condition for logophoricity. In order to make this work, however, we must determine what distinguishes acceptable violations of CA in logophoric readings (including (80)–(81)) from genuinely ungrammatical violations like (85):

(85) a. *Mary said that John loves herself.
  b. *A woman who traveled with Brad kissed himself.
  c. *The men shaved herself.

This is easier for the logophoric reflexives that R&R subsumed under (78b) (e.g. (84c)): these are distinguished from ungrammatical violations of CA in virtue of not being arguments. It is for logophoric reflexives in argument positions – i.e. those that fall under (78a) and the outliers (80)–(81) – that the problem becomes more stringent. I propose that, in all of these cases, reflexive markers are arguments of syntactic predicates which correspond anti-reflexive semantic predicates:

(86) A semantic predicate Π is anti-reflexive (with respect to an indexing i of its arguments) iff Π being reflexive (under i) is infelicitous: i.e. it results in a tautological, contradictory or undefined interpretation.

Like the similar notion of implied non-coreference discussed by Safir (1992), anti-reflexivity is, at bottom, an interpretive property that results from the interaction between pragmatic expectations of informativeness and the semantics of certain predicates. Particular illustrations of the concept are discussed below.

My hypothesis is, thus, that only the CA violations which occur with non-argument reflexives or within anti-reflexive predicates result in logophoric readings. What both of these scenarios have in common is that they are contexts where the CA, as stated in (83), has no way of being (felicitously) fulfilled. What this means is that violations of CA trigger logophoric interpretations if and only if compliance to CA is impossible or disfavored. I call this the Logophoric Strategy:

(87) Logophoric Strategy (LS): When a reflexive marker cannot be the argument of a syntactic predicate which is semantically reflexive, the variable to which it corresponds must refer to an entity that bears a logophoric discourse role – i.e. to some kind of perspective bearer.19

The reasoning is simple: CA states that the lexical duty of reflexive markers is to indicate that syntactic predicates are semantically reflexive. When they cannot do so (either because they are not arguments of a syntactic predicate or because of anti-reflexivity), they must determine their reference by means of some other strategy. This is where logophoric readings kick in, as a kind of non-literal interpretation akin to the generation of an implicature (Grice 1989).20

The LS clearly does not apply to the ungrammatical cases in (85), as there is nothing that disfavors a reflexive interpretation for the semantic predicates love′, kiss′ and shave′. All of these have perfectly plausible and felicitous reflexive interpretations. Therefore, their reflexive arguments in (85) cannot be logophoric.

The same cannot be said for the genuinely logophoric reflexives in (80) and (81). Reflexive interpretations are undefined, and, therefore, infelicitous for (80) because the semantic predicates therein (namely, clear′ and good′) have, as arguments, an experiencer and an event corresponding to the extraposed clause (e.g. Mary left in (80b)). These could not, in principle, correspond to coindexed terms in semantic representation, due to a clash in semantic types.21

Anti-reflexivity also adequately identifies reflexives in (81) as logophoric, but for different reasons. Reflexive interpretations are disfavored there because they would result either in tautology (#Mary is similar to herself) or in contradiction (#The recruits are more qualified than themselves) – both of which would be infelicitous. Note that, when the predicates in (81) are replaced by predicates which are not anti-reflexive, the sentences become bad, as expected (Menuzzi 2004: 138):

(88) a. *Mary considered [her brother hostile to herself].
  b. *The veterans thought that [the new recruits would be nice to themselves].

The LS unifies cases like (81) to the more usual examples of logophoric reflexives in argument positions (e.g. (84a) and (84b)). Picture-NP reflexives, for instance, are syntactic arguments of a nominal head whose semantics is the relational counterpart of the picture′ proxy function: i.e. λx.λy.is-a-picture′ (y, ˆx). As a representational relation, this predicate is anti-reflexive (more on this below). Reflexives in like x-self constructions are fed as semantic arguments to the similarity predicate λx.λy. like′ (y, x). This predicate is anti-reflexive for the same reason as similar′ in (81): #John is like himself is tautological, and, hence, infelicitous.

As observed by a reviewer of this paper, anti-reflexivity also plausibly explains contrasts such as the one between (89) and (90).

(89) a.     John quit the company founded by himself.
  b.     John recommended a book written by himself to me.
(90) a. *?John quit the room destroyed by himself.
  b. *?John recommended a barber admired by himself to me.

Note that the predicates found′ and write′ (but not destroy′ and admire′) are anti-reflexive: there is no coherent interpretation for a coindexation between their necessarily animate Agents and inanimate Themes. Therefore, the reflexive markers of the syntactic predicates in (89) are predicted to be logophoric.22

5.2 Anti-reflexivity and unexpected reflexivization contrasts

Anti-reflexivity also correctly predicts the otherwise puzzling logophoric properties of the reflexives in URCs. Recall from Section 3 that all of the syntactic predicates in URCs (except (53a)), due to the ubiquitous presence of proxy functions, instantiate the schema (91) (cf. (43)) in their semantic representations.

(91) R(τn, P(υn))

Any semantic predicate whose overall shape is (91) will be non-reflexive – i.e. it will entail a violation of CA for its syntactic predicate.23 This is so because, due to the proxy function, the two relevant arguments of R in (91) are not coindexed, even though two of the terms contained in it are. But the fact that proxy functions map their arguments into representations also makes (91) anti-reflexive: P (υn) simply could not be coindexed with τn because such indexing would result in a undefined interpretation in which an entity is construed as identical to its representation.

Saying that (91) could be reflexive is entails that there is an admissible indexing n such that, for an arbitrary assignment function g, the proxy of g(n) (e.g. the statue of Ringo) can be identical to g(n) itself (e.g. Ringo), since both would be coindexed. But such indexing would, in fact, yield an undefined interpretation, because proxy functions are never identity functions. This is a corollary of the fact that proxies are construed as representations and that representational relations are never reflexive (i.e. if x is a representation of y, x and y are necessarily distinct).24

When mustered with R&R’s Chain Condition in (92) (Reuland & Reinhart 1995: 255) – which might itself be reducible to simple economy principles, as Reuland (2001; 2011) suggests – this provides, along the lines hinted in Section 4, a plausible theoretical foundation for a general account of URCs like (93).

(92) Chain Condition (CC): A maximal A-chain (α1, …, αn) contains exactly one link – α1 – which is [+R] (i.e. referentially independent).25
(93) a.   Jane talked to Paul about himself.
      talk′ (jane, paul35, information′ (x35))
  b. *Jane talked about Paul to himself.
      talk′ (jane, x43, information′ (paul43))

Reflexives that correspond to υ in schema (91) (cf. (93a)) are generally acceptable because, despite the CA violation, they retain the LS as a viable interpretive option. Unless their antecedents are inanimate, these reflexives can be coindexed to a perspective bearer – i.e. to an antecedent that bears some kind of logophoric discourse role in the sense of Sells (1987). Reflexives that correspond to τ in (91) (cf. (93b)), on the other hand, are deviant because, in addition to violating the CA, they also fail the LS: since their antecedents are mapped by proxy functions into inanimate representations, they can’t bear the logophoric roles that the LS depends on. As there is no other way to interpret them, the result is invariably bad.

Because of the CC, CA violations in URCs cannot be avoided by replacing reflexives with pronouns. Given their possible use as deictics, pronouns are classified as referentially independent NPs (i.e. as [+R]). Thus, if a pronoun was used in (93), it would give rise to an ill-formed A-chain comprised of it and its antecedent.

Unlike usual instances of logophoric reflexives, which are for the most part long-distance, reflexives in URCs are fully local – i.e. technically, they are tails of A-chains headed by their antecedents. If we assume, following Reuland & Reinhart (1995), that reflexives are the only (overt) anaphoric forms (in English) which are [-R] (since they cannot be used as deictics), they would be the only option allowed by the grammar in contexts like URCs.26 Therefore, we do not have in URCs the competition with pronouns that yields the “markedness” of most logophoric reflexives (see Menuzzi 1999: Chapter 3).

In my account, therefore, reflexives in URCs are predicted to be both logophoric and in complementary distribution with pronouns – the latter property due to the CC. On the basis of these considerations, the sample of judgements below follow:

(94) a. Yoko showed John himself/*him in the mirror.
    show′ (yoko, john2, image′ (x2))
  b. Yoko showed John to *himself/*him in the mirror.
    show′ (yoko, x6, image′ (john6))
(95) a. Jane was warned by Paul about himself/*him.
    warn′ (paul11, jane, information′ (x11))
  b. Jane was warned about Paul by *himself/*him.
    warn′ (x8, jane, information’ (paul8))
(96) a. The artist painted herself/*her (in a realistic style).
    paint′ (the-artist5, image′ (x5))
  b. The artist was painted by *herself /*her (in a realistic style).
    paint′ (x6, image′ (the-artist6))
(97) a. I told Ringo about himself/*him.
    tell′ (me, ringo3, information′ (x3))
  b. I mentioned Ringo to *himself/*him.
    mention′ (me, x9, information′ (ringo9))
(98) a. George was pleased with himself/*him in that concert.
    please′ (george5, performance′ (x5))
  b. George was pleasing to *?himself/*him in that concert.
    please′ (x11, performance′ (george11))

What goes wrong in the “unexpectedly bad cases” such as (94b), (95b), (96b), (97b) and (98b) is that, in light of an inevitable violation of CA which cannot be remedied by the use of a pronoun (as pronouns would violate the CC), the LS is also not available, because the antecedents are mapped, via proxy functions, into entities that cannot bear any logophoric roles: statues, images, information and performances.27

The CC also aids in the explanation for the otherwise problematic set of judgements below:

(99) a. ??Brad swears that [Janet was talking about himself].
  b. ??Frank told Ava that [his daughter looks like himself].

The syntactic predicates highlighted above plausibly translate into anti-reflexive predicates in semantics: talk about and look like have intensional proxy arguments, which yield semantic structures analogous to (91). According to the LS, this would make their reflexive complements logophoric. On the face of it, this could lead us to expect the sentences (99) to be fully acceptable, contrary to fact.

This is not a problem though because pronouns, when they do not incur in CC violations, are preferred over CA-violating structures like (99). Reflexives, in those contexts, are marked and can only be “saved” by a logophoric interpretation. It is a known fact that logophoric interpretations in English – unlike, for example, in Japanese (Yashima 2015: 137–141) – are subject to intervention effects: i.e. logophoric reflexives resist to finding their antecedents across other potential bearers of logophoric roles (cf. Zribi-Hertz 1989; Pollard & Sag 1992). The inherent markedness of long-distance logophoric reflexives and the presence of interveners (Janet in (99a) and Ava in (99b)) is enough to account for the anomaly of (99).

Note that, precisely because anti-reflexivity does trigger logophoric readings, the data in (99) are, surprisingly, not as bad as the following:

(100) a. *Brad swears that Janet was talking to himself.
  b. *Frank told Ava that his daughter looked at himself.

The sentences in (100) are worse than (99) because the reflexives in (100) violate CA in environments where their syntactic predicates could be semantically reflexive. This means that the LS does not emerge as an option for (100).

In short, these considerations strongly suggest that, to properly accommodate URCs, a revision of the CA is in order. The main puzzle consists in the fact that the reflexives which partake in URCs exhibit logophoric properties – the chief of which is their resistance to inanimate antecedents – without displaying the typical syntactic appearance of logophoric reflexives (e.g. non-locality).

Of the previously available formulations in the literature, Menuzzi (1999; 2004) is the one which most easily makes theoretical sense of the logophoric nature of URCs. The alternative I explored, following him, makes the condition for the emergence of logophoric readings extra-grammatical, which is just what we need for URCs. My hypothesis is that the defining property of logophoric reflexives in argument positions (such as the ones in URCs) is not their syntactic configuration but the fact that they occur within syntactic predicates which disallow semantic reflexivity – i.e. they not only violate CA, but, in a sense, inevitably violate it. Since CA (which regulates the interpretation of non-logophoric reflexives) is violated, the only way to interpret these reflexives is to resort to the LS stated in (87).

The general acceptability of first person reflexives and the general constraint against inanimate antecedents in contexts which give rise to URCs follows without stipulations (cf. the examples (60)–(65)). As we saw in Section 4, under standard assumptions, the speaker is always a putative perspective-bearer for an utterance and inanimate objects never are (see Sells 1987; Stirling 1993; Postal 2006; Charnavel 2019 i.a.). What this means is that, regardless of whether the proxy function applies to the reflexive or to its antecedent in URCs, first person reflexives will always manage to satisfy the LS by finding a suitable logophoric antecedent in the context (namely, the speaker). The opposite will be true of inanimate reflexives in URCs. For these cases, the LS is never a viable interpretive option because their referents are not possible perspective-bearers in any discourse context.

I should note lastly that the LS, like the THC discussed in Section 2, applies specifically to reflexives. This limitation is natural, because the LS is simply a non-literal interpretive strategy that arises as a result of a violation of a lexical requirement on reflexives: CA. This explains without stipulation the fact that reciprocals and bound pronouns are not subject to the animacy restriction on antecedents that emerges when reflexives interact with proxy functions (cf. (15)–(16)).

6 Conclusion

This article begun by exploring a set of unexpected reflexivization contrasts (URCs) which have been problematic throughout the history of Generative Grammar. A popular proposal based on a (semantically primitive or syntactically derived) thematic hierarchy was reviewed and dismissed in virtue of some empirical drawbacks. Given these results, the alleged effects of the thematic hierarchy must be treated as epiphenomena of a more fine-grained interpretive factor.

I followed the thread initiated by Grimshaw (1990), Jackendoff (1992) and Safir (2004) in claiming that URCs have a peculiar semantic property: they all involve predicates which, for some reason or other, have proxy functions applied to one of their semantic arguments. I argued that this constitutes a logophoric context (i.e. a context in which CA cannot be fulfilled). Reflexives are nonetheless mandatory, because pronouns in these configurations violate the CC. But, since CA fails, the only way to interpret reflexives in URCs is to resort to a non-literal interpretive strategy – i.e. to what I called the Logophoric Strategy (LS). This, however, is only possible for cases in which the antecedent is a perspective bearer – that is, when it is the reflexive (and not the antecedent) which is linked to a proxy. When the antecedent is mapped into an inanimate proxy, the LS fails to provide an interpretation for the reflexive and the result is invariably bad.

There is still some vagueness in my formulation of LS. I did not provide, for example, a fully explicit characterization of anti-reflexive predicates or of how logophoric roles are attributed to discourse referents. Most of these loose ends can only be overcome by embedding the analysis developed here into a more formally rigorous framework. I have tried, as much as possible, to make my proposal neutral with respect to many details of formal implementation. For instance, I have not addressed here whether logophoric roles should be represented in narrow syntax (Ross 1970; Speas 2004; Sundaresan 2018; Charnavel 2019) or in separate Discourse Representation Structures (Sells 1987; Stirling 1993; Huang 2000).

Regardless of these issues, if my analysis ultimately works, it derives the conflicting judgements in URCs from more principled and independently motivated notions than previous approaches. If reflexives in URCs are logophoric – as many reflexives in many languages are –, URCs are simply explained on the basis of what counts as a possible logophoric antecedent. Moreover, this account embeds this explanation within a robust overall theory of anaphora: the reflexivity-and-chains approach of Reinhart & Reuland (1993).28


CA = Condition A; CC = Chain Condition; LS = Logophoric Strategy; R&R = Reinhart & Reuland (1993); THC = Thematic Hierarchy Condition; URCs = Unexpected Reflexivization Contrasts; UTAH = Uniformity of Theta Assignment Hypothesis


  1. I adopt the notation in Langacker (1969), signaling semantic covaluation between expressions with boldface. As usual when it comes to studies of anaphora, judgments refer to string-interpretation pairs (and not merely to strings themselves). Most of these judgments were gathered from the fifty-year literature on anaphora in Generative Grammar, but others (which were not previously mentioned in the literature) were informally collected by me from native English speakers. Unless otherwise indicated, reported judgments were consistently agreed upon by all of my informants. [^]
  2. Following the terminological practice of R&R, I use the term logophoric (and its cognates) in a broad sense for any kind of perspective-orientation found in language. There is also a narrower use of the term (cf. Culy 1997; Schlenker 2003; Oshima 2007), which dissociates strict logophoricity from the kind of perspective-sensitivity found in English reflexives. The latter is plausibly related to Kuno’s (1987) notion of empathy. Logophors in the narrow sense, on the other hand, are essentially counterparts of first person pronouns whose reference can be shifted under attitude contexts. [^]
  3. The relevant reading here, as well as in (12b), requires a devolitionized subject, as in The finitude of life worries/scares Linda. [^]
  4. One could, as Chomsky (1981) hypothesized, appeal to a reanalysis operation, which turns VP strings into constituents and eliminates the PP node in order to get the antecedent to c-command the reflexive. But this move creates more problems than it solves. Unless a specific motivation is given for why to but not about can be reanalyzed with the verb, the problem is simply inverted: (4b), (5b), (6b) and (7b) now become unexplained. Moreover, as Chomsky (1981) observed, cases like (i), where there is an intervening adverb between V and P, are hard to reconcile with reanalysis:
    (i) John spoke angrily to Cynthia about herself.
    Even if such issues could be overcome, a rule of reanalysis is dubious in itself (Baltin & Postal 1996). Since it involves the elimination of a phrasal projection as a sub-operation (an instance of Pruning, in the sense of Ross (1969)), it is not structure preserving – in current terms, it violates the No-Tampering Condition (Chomsky 2005). An equally stipulative and empirically flawed alternative would be to devise a new condition on binding so that PP nodes are disregarded in the computation of c-command, like the notion of everything-but-PP-command defined in Pesetsky (1995: 173). [^]
  5. The problem is, thus, not with UTAH in itself, but with URCs used as an argument for UTAH, insofar as this argument assumes the THC in (14), which seems to be a wrong generalization for English reflexives (as I’ll show in the next subsection). It may well be that UTAH-based structures are needed for other reasons: e.g. to handle quantifier scope or syntactic alternations (Larson 1988) – but see Culicover & Jackendoff (2005) for a general critique. It is also possible that there are purely syntactic accounts for URCs (based on UTAH) that are not committed to the THC as stated in (14). Since I am not aware of any such proposal, I will not explore this alternative here. [^]
  6. Of course, one can (and, in my view, should) take (15) and (16) as evidence that there is something peculiar about reflexives. R&R implicitly acknowledge this by reformulating Condition A (CA) as a condition on reflexive predicates. Menuzzi (1999) goes further and treats CA as a lexical property of reflexives themselves. I will exploit this in Section 5 and attempt to derive from it an explanation for URCs. It might also be possible to save the THC from (15)–(16) by abandoning a central tenet of the Binding Theory: the c-command condition on binding – as proposed (on other grounds) by Barker (2012). [^]
  7. By “semantic representation” I mean roughly what Reuland (2011: 34) calls logical syntax: an autonomous formal representation which supports inference (i.e. that has “the degree of detail required by the inference system”) and can be subject to a model-theoretic interpretation. My use of the notion also bears resemblance to what Jackendoff (1983; 1990b) calls Conceptual Structure, especially because semantic representations, in my story, need to incorporate inferentially relevant constructional meanings and unarticulated constituents. The most important point, for me, is this: regardless of how one defines semantic representation, it is not a level of narrow syntax. [^]
  8. This is a simplification. Since the semantic details of proxy functions are somewhat orthogonal to my concerns, I tentatively treat proxy functions as an abridged way of talking about choice functions over sets of representations of an entity. More precisely, I assume the contextually salient proxy of an entity x ( P (x)) is the result of applying the choice function ι⟨⟨e, te⟩ to the set of entities that stand in a particular representational relation  R to x’s intension. (ia) states this general equivalence, which will be pressupposed henceforth, and (ib) is states it for the example given above.
    (i) a. P (x) ⟺ ιy( R (y, x̂))
      b. picture’(john) ⟺ ιy(is-a-picture′ (y, ̂john))
    Since λ-abstraction over x yields, for both of the formulae in (ia), functions of type ⟨e, e⟩, the simplification is harmless. In order to properly capture the intensional flavor of proxy functions, (i) takes representational relations to be relations between entities (of type e) and individual concepts (typed ⟨s, e⟩ and signaled by the Montagovian cap (ˆ) operator (Montague 1974)), following a proposal suggested in Partee (1986). I assume, thus, that representational relations are never reflexive: i.e. no entity stands in a representational relation to itself. This point will be relevant later, because it entails (contra Reuland & Winter (2009)) that the identity function can never be a proxy function. [^]
  9. As a reviewer notes, the application of the existential generalization test over depicted objects in fiction and in cases like (29) is questionable if one denies that non-proxies are part of the semantic representation of proxy readings. But this denial would be misguided because it would incur into the trap of sense enumeration accounts of polysemy (Pustejovsky 1995): it would favor a theory that stipulates n separate word meanings for each n-ways an NP can be interpreted as a proxy. So Ringo would be listed as meaning Ringo, statue of Ringo, painting of Ringo, etc. Likewise for all other NPs. This misses the generalization that proxy meanings are derived from non-proxy ones. [^]
  10. Existential generalization is possible over cause-like psych-subjects (as in A Yeti deliberately scared Sue, which is analogous to (13a)). In such cases, there is no evidence for proxy functions. [^]
  11. A few comments about the notation here. Since semantic representation is not a level of narrow syntax, the reference to indices in (38) (and in my other representations for reflexives) is fully compatible with the Inclusiveness Condition (Chomsky 1995), which only constrains narrow syntactic objects. I assume referential indices are interpreted via assignment functions, which map integers onto entities (Heim & Kratzer 1998; Büring 2005). A second point is that, although I will textually mention thematic role labels such as Agent, Theme etc., I will not represent them explicitly, neither by means of role predicates (as in Parsons (1990)), nor indirectly through predicate decompositions (as in Jackendoff (1990b)). Given the skepticism Dowty (1991) convincingly raised with respect to these approaches, I wish to remain neutral on whether roles have reality in semantic representations. I will, however, make my semantic argument lists vaguely ordered to reflect some kind of hierarchy of event roles such that, e.g., for predicates that participate in lexical alternations, participants which trigger the same sorts of inferences occupy the same “position” in argument lists. [^]
  12. The implicit proxy construction is technically a constructional idiom (see Goldberg 1995), given that there is a specific part of its semantics – namely, the proxy function P – that does not come from narrow syntax, but from the construction itself (see Jackendoff (1992) for arguments in favor of this view). Moreover, what I am calling here the proxy-construction might be more like a family of different constructions related in virtue of being instantiations of the schema (40). This “family of constructions” rendition appears to be quite plausible if one contemplates the highly conventionalized character of proxy functions that can be expressed in the absence of explicit syntactic cues (Nunberg 1979; Recanati 2010). Only a small part of possible representational predicates can, in fact, replace the P meta-variable in (40): statue′ and image′ are some of them. [^]
  13. Given the discussion of implicit proxy functions in Section 3, the following example shows the same point as (64b) and (65b). It contains an unequivocal instance of a proxy function in its semantics (i.e. replica′) and, since the non-proxy antecedent is also inanimate, none of the interpretations work:
    (i) a. *While robbing the Louvre, I exchanged ⟨plain⟩ the Mona Lisa for ⟨replica⟩ itself.
          exchange′ (me, mona lisa3, replica′ (x3))
      b. *While robbing the Louvre, I exchanged ⟨replica⟩ the Mona Lisa for ⟨real⟩ itself.
          exchange′ (me,replica′ (mona lisa3), x3)
  14. R&R and Büring (2005) take this as evidence that about PPs (but not with or to PPs) are adjuncts and constitute binding domains on their own. Since the properties mentioned here (improvement in the first person and animacy restriction) are not exclusive to about complements, this account is too limited. Moreover, the stipulated syntactic difference between about PPs and to PPs seems to be unmotivated. Just to mention one problem, the former do not seem to pattern with respect to passivization with PPs that are incontrovertibly adjuncts (e.g. John was talked about/*near/*before). [^]
  15. A problem with this view (which I will not address here) is that it entails that logophoric and non-logophoric reflexives are mere homophones, despite the fact that the formal coincidence between markers of reflexivity and logophoricity is found in typologically unrelated languages (Charnavel 2019). [^]
  16. Reinhart & Reuland (1993: 678) formulate their version of CA as follows: “if a syntactic predicate is reflexive-marked, then it is reflexive”, where: (i) a syntactic predicate is formed of a head P, all syntactic arguments of P, and a subject of P; (ii) syntactic argument of P is a constituent assigned θ-role or Case by P; (ii) a predicate is reflexive if two of its arguments are coindexed; (iv) a predicate is reflexive-marked if one of P’s arguments is a self anaphor (i.e. a reflexive like himself). Another predicate-based theory, with similar empirical consequences, is proposed by Pollard & Sag (1992). However, as a reviewer of this paper notes, it is possible that the latter actually avoids some of the problems that will be mentioned in connection to R&R’s alternative. [^]
  17. R&R actually analyze reflexives within like x-self constructions (e.g. (76b)) as instances of (78b). Though these reflexives are indeed not syntactic arguments of verbs, they are syntactic arguments of like. It seems, thus, more plausible to subsume them under (78a), at least in R&R’s system. [^]
  18. For the argument to work, it must be shown that these examples are not reducible to focus, which is the only way R&R can analyze locally free reflexives in argument positions. No focus seems to be necessary in (80)–(81), though more research on the topic is required. Consider the variants of (80b) and (81a) in (i), where focal stress (signaled by caps) does not fall on the reflexive:
    (i) a. ?Though he knew that her departure would be for the best, Paul never claimed that it would be EASY for himself if Mary left.
      b. ?Does Mary consider her FATHER similar to herself?
        – No, she considers her BROTHER similar to herself.
    The judgments here are particularly murky, but a similar argument built on similar examples is sketched in Safir (1992: 41–42). Additional support comes from contrasts such as the one below:
    (ii) a. ?Mary always thought her brother, but not her father, was similar to herself.
      b. *Mary always thought her brother, but not her father, was mad at herself.
    What (ii) suggests is that there is a real difference with respect to non-local binding between non-focused reflexives within contrastive predicates (iia) and other kinds of predicates (iib). [^]
  19. There are various subtly different taxonomies of logophoric discourse roles and tests to identify them available in the literature (e.g. Sells 1987; Stirling 1993; Golde 1999; Oshima 2007; Charnavel 2019). Sells (1987), for example, posits the following: SOURCE (the intentional agent responsible for communicating the proposition), SELF (the subject whose mind the proposition reports) and PIVOT (the individual in terms of whose space-time location the proposition is evaluated). Since animacy is a minimal condition on all discourse roles in all taxonomies, I will not be concerned here with classifying the particular logophoric discourse role that the LS depends on. [^]
  20. The idea that logophoric readings are derived as implicatures is suggested in Menuzzi (1999: 171): “If the speaker S uses a reflexive-marker M contained in a syntactic predicate P that is incompatible with reflexivity, S is probably not using M to reflexive-mark P; hence, S must be using M with an implicated non-literal function [i.e. logophorically], where the literal function of a reflexive-marker is to mark reflexivity […].” For a more explicit account along these lines, see Menuzzi (2004). [^]
  21. A similar effect is predicted (uniquely by my approach, as far as I’m aware) for the non-extraposed variant of (80b) (i.e. Paul claimed that Mary’s leaving would be good for himself). Since the judgments I gathered for cases like these were not stable, I refrain from discussing them here. [^]
  22. Though the LS accounts for more cases of logophors in argument positions than R&R’s (78), another reviewer questions whether data like the following can be analyzed in similar terms:
    (i) a. But Rupert was not unduly worried about Peter’s opinion of himself.  
      b. And that was exactly it, he thought, he really did not care too much what happened to himself. (Zribi-Hertz 1989: 709)
    (ii)   Each candidate hopes the convention will nominate himself. (Helke 1979: 110)
    (iii) a. If Cassandra has filled my bed with fleas, I am sure they must bite herself.  
      b. […] and let his net that he hath hid catch himself. (Baker 1995: 68)
    If we construe opinion′ and happen′ in (i) as anti-reflexive predicates – the former being tantamount to a proxy function (with the reflexive as an argument and Peter’s as a modifier) and the latter taking an event and an experiencer as arguments –, the logophoric nature of the reflexives therein would follow from the LS. The reflexives in (ii)–(iii) are more problematic. Even though it is technically feasible to assign the status of foci to these reflexives (for which the LS would then predict logophoric readings), this does not seem to be the most insightful analysis (especially for (iii)). I take (ii)–(iii) to reflect a literary or stylized variant of English – one in which CA does not hold in the same way as in standard English. See Golde (1999) for an account along these lines. [^]
  23. The only exception to this among URCs is (53a) (i.e. Mary deliberately scared herself), whose semantics is scare′ (mary13, x13). Since the semantic predicate there is genuinely reflexive, the reflexive is not logophoric and the sentence is correctly predicted to be acceptable by the CA in (83). [^]
  24. Even if this were not the case, the simple fact that, for most of URCs, Themes are required to be inanimate representations and the represented entities are lexically required to be animate (e.g. experiencers, agents, etc) would make the predicates anti-reflexive. Coindexing an inanimate P (υi) with an animate τi would inevitably yield an undefined interpretation. [^]
  25. The CC is a way to restore the intuition that A-movement and A-binding are intimately related. In order to properly implement this, R&R resort to a generalized definition of A-chains, which includes not only chains formed by movement, but also purely representational chains formed by way of anaphoric dependencies. The unification of A-movement and A-binding under a generalized conception of chains is also proposed, within a different framework, by Hornstein (2001). [^]
  26. Following Bouchard (1984), R&R propose that [R] – which is treated as a primitive property in many theories (e.g. Chomsky 1981; Burzio 1998; Safir 2004) – is determined by the φ-feature specification of NPs. Prima facie, English reflexives present a challenge for this view: they are [-R], but not clearly φ-deficient (like the Dutch analogue zichzelf). Reuland & Reinhart (1995) argue that, since English reflexives lack a case contrast (e.g. there is no form heself), they are unspecified for case. φ-deficiency follows if case counts as a φ-feature, as in Chomsky (1981). A more promising line of argument is that the φ-deficiency of English reflexives derives, not from their inherent features, but from their morphosyntactic composition. Anagnostopoulou & Everaert (1999) and Reuland (2011) claim English reflexives are compounds, whose parts are equally prominent: this makes them ambiguous with respect to whose φ-features project. Menuzzi (1999), on the other hand, building on Helke (1979), argues that English reflexive forms are headed by (and, thus, inherit their φ-deficiency from) the self morpheme. Regardless of details in implementation, the point is that even though the pronominal part of the reflexive may be specified, the whole composite structure is not. The reduction of [R] to φ-specification, though crucial for the mechanism of chain formation proposed in Reuland (2011), is actually inessential for the account developed here. Any principle that bars local binding of referentially autonomous NPs is enough to derive the desired effects. [^]
  27. An alternative account for reflexives in raising to object structures – which have been a source of difficulty for R&R’s approach – might follow by an almost identical line of reasoning:
    (i) a. [Brad expects himself] to love Janet. ⟺ expect′ (brad5,love′ (x5, janet))
      b. [Brad believes himself] to be in love with Janet. ⟺ believe′ (brad1,love′ (x1, janet))
      c. [Brad envisioned himself] marrying Janet. ⟺ envision′ (brad2,marry′ (x2, janet))
    The syntactic predicates of the reflexives in (i) do not correspond to semantically reflexive predicates: like URCs, therefore, these cases constitute violations of CA as formulated in (83). However, the semantic predicates that correspond to the matrix syntactic predicates enclosed in brackets are also anti-reflexive: they contain as their arguments an animate entity (⟦Brad⟧) and a proposition of some sort. Therefore, the LS is technically an option for these cases. If this account is right, the only difference with respect to URCs is that the LS never fails when reflexives occur as raised objects: the predicates which allow raising to object invariably offer perspective bearers as possible antecedents. The fact that sentences in (i) do not allow pronouns is, as in R&R, explained by the CC. [^]
  28. It remains to be seen whether these facts are compatible with other approaches. What appears to be minimally required, aside from a violable predicate-based CA such as (83), is a purely syntactic constraint ruling out local binding of referentially autonomous NPs – i.e. an analogue of the Condition B in Chomsky (1981), such as the CC in (92). What this means is that the present approach is actually incompatible with pure competition-based theories (e.g Burzio 1998; Safir 2004; Levinson 2000), that attempt to derive the availability of pronouns from violations of constraints on reflexives. This is so because, in URCs, the violation of CA crucially does not make pronouns readily available. [^]


I am grateful to Peter W. Culicover for generously sharing his feedback on many previous versions of this work. I would also like thank Isabelle Charnavel, Ray Jackendoff and Richard Larson for conversations on topics related to the ones I explore here. Bob Levine, Ashwini Deo, Rafaela Miliorini, Jordan Needle, Symon Stevens-Guille, the audiences at Synners/OSU and three anonymous reviewers provided very helpful comments as well. All remaining errors are mine.

Funding Information

This research is funded by a CAPES-PROEX scholarship 88882.344838/2019-01.

Competing Interests

The author has no competing interests to declare.


Anagnostopoulou, Elena & Martin Everaert. 1999. Toward a more complete typology of anaphoric expressions. Linguistic Inquiry 30(1). 97–119. DOI:  http://doi.org/10.1162/002438999553977

Bach, Kent & Robert Harnish. 1979. Linguistic communication and speech acts. Cambridge, MA: MIT Press.

Baker, C. Lee. 1995. Contrast, discourse prominence, and intensification, with special reference to locally free reflexives in British English. Language 71(1). 63–101. DOI:  http://doi.org/10.2307/415963

Baker, Mark. 1997. Thematic roles and syntactic structure. In Liliane Haegeman (ed.), Elements of grammar, 73–137. Dordrecht: Kluwer Academic Publishers. DOI:  http://doi.org/10.1007/978-94-011-5420-8_2

Baltin, Mark & Paul M. Postal. 1996. More on reanalysis hypothesis. Linguistic Inquiry 27(1). 127–145.

Barker, Chris. 2012. Quantificational binding does not require c-command. Linguistic Inquiry 43(4). 614–633. DOI:  http://doi.org/10.1162/ling_a_00108

Barss, Andrew & Howard Lasnik. 1986. A note on anaphora and double objects. Linguistic Inquiry 17(2). 347–354.

Belletti, Adriana & Luigi Rizzi. 1988. Psych-verbs and θ-theory. Natural Language and Linguistic Theory 6(3). 291–352. DOI:  http://doi.org/10.1007/BF00133902

Bouchard, Denis. 1984. On the content of Empty Categories. Dordecht: Foris. DOI:  http://doi.org/10.1515/9783110850000

Büring, Daniel. 2005. Binding theory. Cambridge: Cambridge University Press. DOI:  http://doi.org/10.1017/CBO9780511802669

Burzio, Luigi. 1998. Anaphora and soft constraints. In Pilar Barbosa, Danny Fox, Paul Hagstrom, Martha McGinnis & David Pesetsky (eds.), Is the best good enough?, 93–114. Cambridge, MA: MIT Press.

Charnavel, Isabelle. 2019. Locality and logophoricity: A theory of exempt anaphora. New York, NY: Oxford University Press. DOI:  http://doi.org/10.1093/oso/9780190902100.001.0001

Charnavel, Isabelle Carole & Christina Diane Zlogar. 2015. English reflexive logophors. In Proceedings of the 51st annual meeting of the Chicago Linguistic Society. Chicago, IL.

Charnavel, Isabelle & Dominique Sportiche. 2016. Anaphor binding: What french inanimate anaphors show. Linguistic Inquiry 47(1). 35–87. DOI:  http://doi.org/10.1162/LING_a_00204

Chomsky, Noam. 1981. Lectures on government and binding. Dordecht: Foris.

Chomsky, Noam. 1986. Knowledge of language. New York, NY: Praeger Publishers.

Chomsky, Noam. 1995. The minimalist program. Cambridge, MA: MIT Press.

Chomsky, Noam. 2005. Three factors in language design. Linguistic Inquiry 36(1). 1–22. DOI:  http://doi.org/10.1162/0024389052993655

Clements, George N. 1975. The logophoric pronoun in Ewe: Its role in discourse. Journal of West African Languages 10(2). 141–177.

Culicover, Peter W. to appear. Language universals, variation and change: A constructional approach. New York, NY: Oxford University Press.

Culicover, Peter W. & Ray Jackendoff. 2005. Simpler syntax. New York, NY: Oxford University Press. DOI:  http://doi.org/10.1093/acprof:oso/9780199271092.001.0001

Culy, Christopher. 1997. Logophoric pronouns and point of view. Linguistics 35(5). 845–860. DOI:  http://doi.org/10.1515/ling.1997.35.5.845

Curry, Haskell B. 1963. Some logical aspects of grammatical structure. In Roman Jacobson (ed.), Structure of language and its mathematical aspects: Proceedings of the twelfth symposium in applied mathematics, 56–68. American Mathematical Society. DOI:  http://doi.org/10.1090/psapm/012/9981

Dalrymple, Mary. 1993. The syntax of anaphoric binding. Stanford, CA: Center for the Study of Language (CSLI).

Dalrymple, Mary, John Lamping, Fernando Pereira & Vijay Saraswat. 1997. Quantifiers, anaphora, and intensionality. Journal of Logic, Language and Information 6(3). 219–273. DOI:  http://doi.org/10.1023/A:1008224124336

Dowty, David. 1991. Thematic proto-roles and argument selection. Language 67(3). 547–619. DOI:  http://doi.org/10.1353/lan.1991.0021

Dowty, David, Robert Wall & Stanley Peters. 1981. Introduction to Montague Semantics. Dordecht: Reidel Publishing Company. DOI:  http://doi.org/10.1007/978-94-009-9065-4

Goldberg, Adele E. 1995. Constructions: A construction grammar approach to argument structure. Chicago, IL: University of Chicago Press.

Golde, Karin E. 1999. Binding theory and beyond: An investigation into the English pronominal system. Columbus, OH: The Ohio State University dissertation.

Grice, Paul. 1989. Studies in the way of words. Cambridge, MA: Harvard University Press.

Grimshaw, Jane B. 1990. Argument structure. Cambridge, MA: MIT Press.

Heim, Irene & Angelika Kratzer. 1998. Semantics in generative grammar. Malden, MA: Blackwell.

Helke, Michael. 1979. The grammar of English reflexives. New York, NY: Garland.

Hornstein, Norbert. 2001. Move: A minimalist theory of construal. Malden, MA: Blackwell.

Huang, Yan. 2000. Anaphora: A cross-linguistic approach. New York, NY: Oxford University Press.

Jackendoff, Ray. 1969. Some rules of semantic interpretation for English. Cambridge, MA: Massachusetts Institute of Technology dissertation.

Jackendoff, Ray. 1972. Semantic interpretation in generative grammar. Cambridge, MA: MIT Press.

Jackendoff, Ray. 1975. On belief-contexts. Linguistic Inquiry 6(1). 53–94.

Jackendoff, Ray. 1980. Belief-contexts revisited. Linguistic Inquiry 11(2). 395–413.

Jackendoff, Ray. 1983. Semantics and Cognition. Cambridge, MA: MIT Press.

Jackendoff, Ray. 1990a. On Larson’s treatment of the double object construction. Linguistic Inquiry 21(3). 427–455.

Jackendoff, Ray. 1990b. Semantic Structures. Cambridge, MA: MIT Press.

Jackendoff, Ray. 1992. Mme. Tussaud meets the Binding Theory. Natural Language and Linguistic Theory 10(1). 1–32. DOI:  http://doi.org/10.1007/BF00135357

Jackendoff, Ray. 1997. The architecture of the language faculty. Cambridge, MA: MIT Press.

Kay, Paul. 1994. Anaphoric binding in construction grammar. In Annual meeting of the Berkeley Linguistics Society 20. 283–299. Berkeley, CA. DOI:  http://doi.org/10.3765/bls.v20i1.1441

Keenan, Edward L. 1988. Complex anaphors and bind alpha. In Papers from the 24th anual regional meeting of the Chicago Linguistic Society, 216–232. Chicago, IL.

Kuno, Susumu. 1987. Functional syntax: Anaphora, discourse and empathy. Chicago, IL: University of Chicago Press.

Kuno, Susumu & Etsuko Kaburaki. 1977. Empathy and syntax. Linguistic Inquiry 8(4). 627–672.

Lakoff, George. 1968. Pronouns and reference. Unpublished manuscript.

Langacker, Ronald W. 1969. On pronominalization and the chain of command. In David A. Reibel & Sandford A. Schane (eds.), Modern studies in English, 160–186. Englewood Cliffs, NJ: Prentice Hall.

Larson, Richard. 1990. Double objects revisited: Reply to Jackendoff. Linguistic Inquiry 21(4). 589–632.

Larson, Richard. 1988. On the double object construction. Linguistic Inquiry 19(3). 335–392.

Lebeaux, David. 1985. Locality and anaphoric binding. The Linguistic Review 4(4). 343–363. DOI:  http://doi.org/10.1515/tlir.1985.4.4.343

Lees, Robert B., & Edward S. Klima. 1963. Rules for English pronominalization. Language 39(1). 17–28. DOI:  http://doi.org/10.2307/410759

Levin, Beth. 1993. English verb classes and alternations. Chicago, IL: University of Chicago Press.

Levinson, Stephen C. 2000. Presumptive meanings: The theory of generalized conversational implicature. Cambridge, MA: MIT Press. DOI:  http://doi.org/10.7551/mitpress/5526.001.0001

Lidz, Jeffrey. 2001. Condition R. Linguistic Inquiry 32(1). 123–140. DOI:  http://doi.org/10.1162/002438901554603

Lidz, Jeffrey & William J. Idsardi. 1998. Chains and phono-logical form. University of Pennsylvania Working Papers in Linguistics 5(1). 109–125.

McCawley, James D. 1968. Concerning the base component of a transformational grammar. Foundations of Language 4(3). 243–269.

McCawley, James D. 1971. Interpretative semantics meets Frankenstein. Foundations of Language 7(2). 285–296.

Menuzzi, Sergio. 1999. Binding theory and pronominal anaphora in Brazilian Portuguese. Utrecht: Holland Academic Graphics.

Menuzzi, Sergio. 2004. Non-conflicting violations of grammatical constraints: Logophoric reflexives, peculiar passives, and Gricean implicatures. Ilha do Desterro 1(47). 129–168.

Minkoff, Seth A. 2004. Consciousness, backward coreference, and logophoricity. Linguistic Inquiry 35(3). 485–494. DOI:  http://doi.org/10.1162/ling.2004.35.3.485

Montague, Richard. 1974. Formal Philosophy. New Haven, CT: Yale University Press.

Nunberg, Geoffrey. 1979. The non-uniqueness of semantic solutions: Polysemy. Linguistics and Philosophy 3(2). 143–184. DOI:  http://doi.org/10.1007/BF00126509

Nunberg, Geoffrey. 1995. Transfers of meaning. Journal of Semantics 12(2). 109–132. DOI:  http://doi.org/10.1093/jos/12.2.109

Okada, Sadayuki. 1998. Reflexive pronouns with split antecedents. Journal of Pragmatics 30(1). 59–84. DOI:  http://doi.org/10.1016/S0378-2166(97)00088-X

Oshima, David. 2007. On empathic and logophoric binding. Research on Language and Computation 5(1). 19–35. DOI:  http://doi.org/10.1007/s11168-006-9020-0

Parsons, Terence. 1990. Events in the semantics of English: A study in subatomic semantics. Cambridge, MA: MIT Press.

Partee, Barbara H. 1986. Noun phrase interpretation and type-shifting principles. In Jeroen Groenendijk, Dick de Jongh & Martin Stokhof (eds.), Studies in discourse representation theory and the theory of generalized quantifiers, 115–143. Dordecht: Foris.

Pesetsky, David. 1995. Zero syntax: Experiencers and cascades. Cambridge, MA: MIT Press.

Pollard, Carl & Ivan A. Sag. 1992. Anaphors in English and the scope of binding theory. Linguistic Inquiry 23(2). 261–304.

Postal, Paul M. 1968. Crossover phenomena. Yorktown Heights, NY: IBM Corporation.

Postal, Paul M. 2006. Remarks on English long-distance anaphora. Style 40(1–2). 7–18.

Pustejovsky, James. 1995. The generative lexicon. Cambridge, MA: MIT Press.

Pustejovsky, James. 2012. Type theory and lexical decomposition. In James Pustejovsky, Pierrette Bouillon, Hitoshi Isahara, Kyoko Kanzaki & Chungmin Lee (eds.), Advances in generative lexicon theory, 9–38. New York, NY: Springer. DOI:  http://doi.org/10.1007/978-94-007-5189-7_2

Recanati, François. 2010. Truth-conditional pragmatics. Oxford: Oxford University Press. DOI:  http://doi.org/10.1093/acprof:oso/9780199226993.001.0001

Reinhart, Tanya. 1983. Anaphora and semantic interpretation. Chicago, IL: University of Chicago Press.

Reinhart, Tanya & Eric Reuland. 1993. Reflexivity. Linguistic Inquiry 24(4). 657–720.

Reinhart, Tanya & Tal Siloni. 2005. The lexicon-syntax parameter: Reflexivization and other arity operations. Linguistic Inquiry 36(3). 389–436. DOI:  http://doi.org/10.1162/0024389054396881

Reuland, Eric. 2001. Primitives of binding. Linguistic Inquiry 32(3). 439–492. DOI:  http://doi.org/10.1162/002438901750372522

Reuland, Eric. 2011. Anaphora and language design. Cambridge, MA: MIT Press. DOI:  http://doi.org/10.1093/obo/9780199772810-0050

Reuland, Eric & Tanya Reinhart. 1995. Pronouns, anaphors and case. In Hubert Haider, Susan Olsen & Sten Vikner (eds.), Studies in comparative Germanic syntax, 241–268. Dordrecht: Kluwer Academic Publishers. DOI:  http://doi.org/10.1007/978-94-015-8416-6_11

Reuland, Eric & Yoad Winter. 2009. Binding without identity: Towards a unified semantics for bound and exempt anaphors. In Sobha Lalitha Devi, Antonio Branco & Ruslan Mitkov (eds.), Discourse anaphora and anaphor resolution colloquium, 69–79. New York, NY: Springer. DOI:  http://doi.org/10.1007/978-3-642-04975-0_6

Ross, John R. 1969. A proposed rule of tree-pruning. In Sandford A. Reibel & David A. Schane (ed.), Modern studies in English: Readings in transformational grammar, 288–299. Englewood Cliffs, NJ: Prentice-Hall.

Ross, John R. 1970. On declarative sentences. In Roderick A. Jacobs & Peter S. Rosenbaum (eds.), Readings in English transformational grammar, 222–272. Waltham, MA: Ginn-Blaisdell.

Russell, Bertrand. 1905. On denoting. Mind 14(56). 479–493. DOI:  http://doi.org/10.1093/mind/XIV.4.479

Safir, Ken. 1992. Implied non-coreference and the pattern of anaphora. Linguistics and Philosophy 15(1). 1–52. DOI:  http://doi.org/10.1007/BF00635831

Safir, Ken. 2004. The syntax of anaphora. New York, NY: Oxford University Press. DOI:  http://doi.org/10.1093/acprof:oso/9780195166132.001.0001

Schlenker, Philippe. 2003. A plea for monsters. Linguistics and Philosophy 26(1). 29–120. DOI:  http://doi.org/10.1023/A:1022225203544

Searle, John R. 1983. Intentionality: An essay in the philosophy of mind. Cambridge: Cambridge Univeristy Press. DOI:  http://doi.org/10.1017/CBO9781139173452

Sells, Peter. 1987. Aspects of logophoricity. Linguistic Inquiry 18(3). 445–480.

Speas, Margaret. 2004. Evidentiality, logophoricity and the syntactic representation of pragmatic features. Lingua 114(3). 255–276. DOI:  http://doi.org/10.1016/S0024-3841(03)00030-5

Stirling, Lesley. 1993. Switch-reference and discourse representation. Cambridge: Cambridge University Press. DOI:  http://doi.org/10.1017/CBO9780511597886

Sundaresan, Sandhya. 2018. Perspective is syntactic: Evidence from anaphora. Glossa 3(1). 1–40. DOI:  http://doi.org/10.5334/gjgl.81

Wasow, Thomas. 1979. Anaphora in generative grammar. Philadelphia, PA: John Benjamins Publishing Company. DOI:  http://doi.org/10.1075/sigla.2

Wilkins, Wendy. 1988. Thematic structure and reflexivization. In Wendy Wilkins (ed.), Syntax and semantics: Thematic relations, 191–213. San Diego, CA: Academic Press. DOI:  http://doi.org/10.1163/9789004373211_012

Williams, Edwin. 1994. Thematic structure in syntax. Cambridge, MA: MIT Press.

Yashima, Jun. 2015. Antilogophoricity: In conspiracy with the binding theory. Los Angeles, CA: University of California dissertation.

Zribi-Hertz, Anne. 1989. Anaphor binding and narrative point of view: English reflexive pronouns in sentence and discourse. Language 65(4). 695–727. DOI:  http://doi.org/10.2307/414931