The expletive interpretation of the Ethical Dative: a syntactic approach to a non-argumental clitic

2.1 What Ethical Dative is: some core features

Italian ED is attested in the first novel written in Modern Italian, i.e., I Promessi Sposi (Manzoni 1842):

The first core feature of ED is that it does not change the propositional meaning of a sentence, as it does not belong to the thematic grid of the verb (Franco and Huidobro 2008). This can be observed in examples such as (2a) and (5), where the presence of an ED does not affect the truth conditions of the sentence and, can be removed without changing the meaning. Accordingly, I assume that ED does not contribute to the compositional propositional meaning of a sentence. Rather, it adds extra semantic and pragmatic information beyond the propositional meaning as an instance of an expletive/pleonastic phenomenon,^[5] as witnessed by the following sentences with and without ED:

Since ED does not represent any argument of the verb, realizing an expletive function in a sense, it cannot undergo any form of A’-movement (Michelioudakis and Kapogianni 2013), such as wh-fronting (7):^[6]

As indicated above, ED identifies a person who is affected by the event described in a sentence. It is noteworthy that while the 3rd person sing. is not categorically excluded (2), it is less accepted by some scholars (Delbecque and Lamiroy 1996; Roberge and Troberg (2009); see Michelioudakis and Kapogianni 2013 for an overview). According to these scholars, ED is restricted to the speaker and the hearer in the 1st (mi) and 2nd (ti) person. Some Italian cases seem to confirm this preference, as witnessed by the following exclamative sentences with an expletive negation (see the discussion on surprise negation sentence):

Thus, the second peculiarity of ED is a slight preference towards the 1st and the 2nd person singular.^[7]

Another peculiarity of the ED is its obligatory clitic nature (Renzi et al. 2001). As is well known, clitics can also be expressed by means of a corresponding noun, pronoun, or prepositional phrase (D’Alessandro 2017). This can be seen in both core (9a–b) and non-core (9c–d) dative clitics:

In (9c), the presence of gli suggests that Laura has ironed the shirts for the benefit of someone that is not either the hearer or the speaker of the utterance, instantiating a case of a Ben clitic. Benefactives introduce an applicative argument, which is the beneficiary or maleficiary of the action described by the verb (Folli and Harley 2006; see Section 2.2). Typically considered a non-core dative – a type of dative argument that is not directly involved in the core syntactic structure of a verb’s argument schema – it is one of the most common uses of a dative clitic in Italian. However, this possibility is ruled out if the clitic is an ED:

The ungrammaticality of (10b), in which ED ti is realized by an overt PP, stems from such a constraint (see Lo Cascio 1970 for the Italian case). It is worth noting that Italian just misses a clitic form for the 3rd person plural dative – which must be realized by the pronoun loro (‘they’) or PP a loro (‘to them’) – and, as expected, ED is not allowed in these cases (Masini 2012), further showing that ED is strictly dependent on the clitic nature of the pronoun:

The clitic constraint is also attested in other languages belonging to different families, such as, among others, Hebrew (12a–b) (Borer and Grodzinsky 1986), French (12c–d) (Kayne 1975), and Spanish (12d–e) (Cuervo 2003) (see also Boneh and Nash 2012; Jaeggli 1982; Michelioudakis and Kapogianni 2013; Strozer 1976: 145):^[8]

Thus, the third peculiarity of the ED is its obligatory clitic nature.

Another feature of the ED is the non-obligatory co-reference between the ED and the grammatical subject of the sentence. This can be observed in (10), where the 3rd person singular subject of the sentence, Thomas, is not co-referential with the 2nd person singular of the ED ti (‘to you’).^[9] Moreover, ED seems to be banned from co-referring to the subject of the sentence, yielding ungrammaticality in cases such as (13a), unless the auxiliary is changed to ‘be’ (13b):

We will see that not all non-core datives can circumvent the constraint on coreferentiality with the subject of the sentence in this way.

The fifth feature is the possible occurrence of the ED in sentences with ditransitive constructions. Typically, it is impossible to have both a dative clitic and an indirect object in Italian ditransitive constructions (14a) unless they refer to each other (14b). However, ED constitutes an exception to this pattern (14c):

The sixth characteristic is that ED is restricted in its distribution. It cannot be embedded in relative clauses (15a) and it cannot undergo any form of A’-movement (Michelioudakis and Kapogianni 2013; see sentences in (7)). Other types of dative clitics, such as benefactives, are permitted in these contexts (15b):

Finally, an interesting pattern emerges when examining the interaction between passives and dative clitics specifically investigating whether they can appear before the verb and whether the movement of the theme to the preverbal position is influenced by the dative clitic itself. On the one hand, when the theme is left in situ, ED is not as grammatical in passive constructions (Naudé 1997), whereas both core datives and Ben are allowed (Boneh and Nash 2012; Folli and Harley 2006; Rooryck 1988):^[11]

On the other hand, when the theme appears at the beginning of the sentence, i.e., in the preverbal subject position, the sentence results in degradation in benefactive (17a),^[12] whereas it gets better in ED construction (17b):

To summarize the main features of Italian ED, we can state the following: (i) ED does not alter the propositional meaning of a sentence, as it does not belong to the thematic grid of the verb; (ii) ED predominantly appears in the 1st and 2nd person singular, although it also occurs in the 3rd person singular; (iii) ED obligatorily displays the clitic form and therefore cannot occur in the 3rd person plural, as Italian lacks a corresponding clitic for this; (iv) ED is not required to be co-referential with the grammatical subject of the sentence; (v) ED can appear in sentences with ditransitive constructions; (vi) ED does not undergo A’-movement, such as wh-fronting, and, finally (vii) ED can appear in passive structures where the theme moves across the dative clitic to a preverbal subject position, but is not as grammatical when the theme remains in situ. A comprehensive analysis of ED should consider all these features and derive them in a unitary way. Section 3 contains a proposal for an analysis that achieves this. Before we get there, further examination is needed to distinguish ED from other non-core dative clitics, such as Ben and CD, which show the same morphological shapes.

2.2 What Ethical Dative is not: a comparison with benefactive and Co-referential Datives

Italian displays a complex system of non-argumental dative clitics that serve various functions (Renzi et al. 2001; Russi 2008). Among these, the ED, as discussed above, stands out. However, ED often creates confusion due to its morphological and pragmatic similarities with other non-argumental dative clitics, such as Ben and CD. As seen in examples from previous sections, ED has distinctive characteristics. In this section I systematically discuss them, also considering some additional data that support the distinction between ED and other non-argumental datives.

Let us begin with the examples previously examined that distinguish ED from the benefactive. For instance, in sentences (9c–d) – repeated here as (18) – I observed that only ED must be expressed with a clitic, whereas the benefactive can also occur with a full prepositional phrase introduced by ‘a’ (‘to’) or ‘per’ (‘for’) (see Masini 2012 for Italian and Boneh and Nash 2012 for French):

Recall that benefactive clitics introduce an applicative argument representing the beneficiary or maleficiary of the action described by the verb (Folli and Harley 2006). A clitic with a Ben function can represent all singular and plural persons. This can be explained by the intuitive assumption that the action described by a verb can benefit or damage anyone. In fact, the benefactive can also be realized using the 3rd plural pronoun “loro” (‘to them’), which is not possible for ED.

The Ben is also compatible with left or right dislocation (Cecchetto 1999), whereas ED is not:

Moreover, Ben can also appear in relative clauses (21b) and undergo forms of A’-movement (Michelioudakis and Kapogianni 2013), appearing in wh-fronting (21d), unlike ED (21c) (I here repeat the sentences in (15)):

Finally, Ben cannot occur in sentences with ditransitive constructions, even though we can imagine a situation where someone gives a gift to another person for the benefit of a third party. However, this is not grammatically possible, whereas ED can occur in such constructions, as discussed in (14):

The examples given above should be sufficient evidence for considering ED and Ben two distinct phenomena^[13] – I do not repeat the differing behaviors in passive structures already observed in (16) and (17). Let us now consider the differences between ED and other non-argumental dative constructions, specifically CD.

In Co-referential Dative (CD) constructions, the dative clitic refers to the subject of the sentence (Boneh and Nash 2011). Similar to ED and Ben, CDs are a type of non-core dative. One of their primary characteristics is that they do not change the truth conditions of a proposition (I here repeat the sentence (4c) as (23)):

Similar to ED, CD has a restriction on the clitic form, resulting in ungrammaticality when expressed through a corresponding prepositional phrase:

Unlike ED, CD must be co-referential with the grammatical subject. This requirement makes a sentence like (25) ungrammatical, where the CD ti (2nd person singular) is not co-referential with the 3rd person singular subject. In contrast, ED does not exhibit this restriction, as illustrated by the sentence in (10a), repeated here as (25b):

CD also differs from ED in other respects, such as the behavior in combination with ditransitive verbs. While ED is permitted in such contexts (see above), CD is not (I here repeat the sentence (14c) as 26b)):

Finally, CD can occur in relative clauses, while ED cannot:

Based on these differences, I conclude that Co-referential Datives cannot be considered instances of Ethical Datives.

In this section I discussed various types of data that help us to distinguish ED from other types of non-argumental dative clitics. ED exhibits its own distinct characteristics, as shown above. It is important to note that the comparative focus of this section was primarily on identifying ED, rather than providing a comprehensive description of other dative clitics, which display a complexity only briefly touched upon here. Undoubtedly, ED possesses a unique grammatical identity, distinct from any other clitic. In the next section, I will thus present a comprehensive analysis of ED, with the goal of considering all their features and deriving them in a unified manner.

3.1 Ethical Dative and the CP-domain

The assumption that ED is the head of an ApplP generated in the CP domain of the clause can provide a unitary analysis for the numerous features characterizing ED, which are: (i) ED does not alter the compositional meaning of a sentence, as it does not belong to the thematic grid of the verb; (ii) ED predominantly appears in the 1st and 2nd person singular, although it also occurs in the 3rd person singular; (iii) ED obligatorily displays the clitic form and therefore cannot occur in the 3rd person plural form, as Italian lacks a corresponding clitic for this; (iv) ED is not required to be co-referential with the grammatical subject of the sentence; (v) ED can appear in sentences with ditransitive constructions; (vi) ED does not undergo A’-movement, such as wh-fronting, and, finally (vii) ED can appear in passive structures where the theme moves across the dative clitic to a preverbal subject position, but it is not as grammatical when the theme remains in situ. Let us consider these features in turn.

It has been proposed in the literature that non-core datives may be merged outside the thematic domain of the syntactic tree, specifically outside and above the v/V phrases (Boneh and Nash 2011; Wechsler 2020 and the references therein). From there, the ApplP “cannot introduce a new argument, and its function is restricted to assigning the interpretable feature [affectedness]” (Boneh and Nash 2011: 13). I will draw inspiration from this proposal, suggesting that the Italian ED is located even higher, being externally merged in the CP domain, once all predicative relations have already been established (see Moro 2020 and references therein):

Assuming that CP consists of an array of functional heads, as in the cartographic approach (see Rizzi 1997 and subsequent works; Cinque and Rizzi 2010), I propose that ED is merged in a position inside such a complex field, more exactly, in a position between FinP and TP (I consider here a simple version of the CP-domain):^[22]

It is important to note that the theoretical framework adopted does not affect the core essence of this proposal. For instance, within a minimalist approach, there is no impediment to the presence of a functional head that takes the TP as its complement, which is indeed the case in standard transitive affirmative sentences (Chomsky 2001).

Since Ethical Datives are clitics, they inherently function as heads (see Roberts 2010, and references therein). More interestingly, if ED is generated directly outside the TP, it becomes clear why it is not an argument of the verb and, consequently, why it does not affect the propositional meaning of the sentence – behaving like an expletive in this respect. On the other hand, it still plays a role in the interpretation of the sentence, introducing an “affectedness” relation between an individual and an event (see below), since it is merged in the layer where the discourse properties are established (Moro 2020). It should be noted that, assuming the theory of phases as formulated by Chomsky (2001, 2008, 2013), EDs are merged in a different phase than vP. This further explains why they do not affect the thematic core of the verb. As a matter of fact, the expletive interpretation of the dative clitic in the current proposal is due to its syntactic position inside the sentential spine.

Another direct consequence is that ED can co-occur with ditransitive verbs. In fact, if a verb cannot select more than two internal arguments and one external argument (Hale and Keyser 1993), ED can exist in a ditransitive construction only if it does not introduce any argument, since the verb has already saturated its maximal valency.

Assuming that EDs are heads of ApplPs (à la Roberge and Troberg (2009)), their clitic behavior follows (EDs can’t be PPs or overt pronouns) and, at the same time, we can also explain why they get Dat case (McFadden 2004; see below). In Cuervo’s approach, EDs cannot be represented as full dative PPs (or DPs in her analysis) due to their “defective” nature: they are heads that take an argument without projecting a specifier (hence the unavailability of a full dative DP). In my analysis there is no need to introduce such an ad hoc stipulation, since the ED has similarities with some other functional heads populating the left periphery of the sentences (cfr. Rizzi 1997). This is similar to what Jaeggli (1982: 18) proposes on ED, i.e., it represents a category of clitics that does not originate in object position, challenging Kayne’s (1975) movement theory of clitics – where clitics are initially generated in NP position and then moved obligatorily to the verb. Accordingly, ED does not exhibit alternations with any other post-verbal object position, providing compelling evidence that they are generated “by the base in their clitic position” (Jaeggli 1982: 18). The non-argumental status of a dative is also highlighted by the impossibility to be doubled by a full DP (Franco and Huidobro 2008, in line with Jaeggli 1982; Strozer 1976) (I here repeat the sentence (10)):

Moreover, this also takes into account the impossibility of having an ED of the 3rd person plural form, since Italian does not display any clitic of this category, but only the stressed pronouns “loro”.

Treating EDs as heads of high(er) ApplPs also allows us to consider their inability to occur in stative constructions (see sentences in (31)). More specifically, EDs maintain the core property of high applicatives as discussed by Pylkkänen (2002, 2008) – namely, (i) being merged (somewhere) above the VP and (ii) linking an entity to an event by some relation. However, if there is no event to be related to, as in stative constructions, ED cannot appear in such contexts. Assuming that “affectedness” is the semantic relation introduced by ED between an individual – such as the speaker or the hearer of the utterance – and an event (see Berman 1982; Masini 2012; Michelioudakis and Kapogianni 2013; Shibatani 1994 among many others),^[23] ED can be interpreted as follows:

This condition can only be applied if there is an eventive verb phrase complement that ED can take. Following a well-established tradition (Ramchand 2008), I assume that stative predicates do not display such an event and, accordingly, “there is no dynamicity/process/change involved in the predication, but simply a description of a state of affairs” (Ramchand 2008: 33). ED can therefore not select them, as evidenced by copular sentences (Greco et al. 2020; Moro 1997): ED can never occur in either canonical (37a) or inverse copular sentences (37c). In contrast, ED does occur with other raising verbs, such as become (37b–d), which display an eventive predicative structure absent from copular constructions.

This restriction does not apply to predicates denoting activities, accomplishments or achievements, as they do involve some change (Beavers 2011, 2013), which allows ED to be added (see the sentences in (31)).

The hypothesis proposed here also accommodates the behavior involving both the missing coreferentiality requirement on ED-subjects and passivization. Let’s start with the latter, even though the two phenomena are linked. We discussed above that ED appears in passive constructions regardless of the position of the theme, either in situ or in preverbal position, with a better result in the latter condition. It is worth noting that benefactives are ruled out in this condition, whereas the ED is still available (I here repeat the sentences (17) as (38)):^[24]

According to Folli and Harley (2006), when the theme moves beyond the dative clitic in the case of a benefactive, it causes a locality violation because the dative intervenes between the subject position and the theme. This follows from the fact that the base position of the dative in benefactives c-commands the theme’s base position. According to the current analysis, this does not happen with ED, where there is no such crossing violation. This is possible only if the clitic is merged above the subject position. More specifically, two distinct pre-verbal landing sites for Italian subjects have been proposed in the literature in order to accommodate both their structure requirements (ex. EPP) and their discourse properties: one high position linked to specific discourse properties – such as “D-linking” in Pesetsky (1987), “presuppositionality” in Diesing (1992), or as “criterial” in Rizzi’s 1997 framework (yet below the CP field) – and one low position associated with agreement (“AgrS” as per Cardinaletti’s 2004 definition). Leaving aside the disputes regarding these proposals, if ED is merged above the subject positions – particularly above AgrS, or whatever head is responsible for the subject agreement, this implies that ED is not constrained to being coreferential with the grammatical subjects. In that configuration, ED is not c-commanded by the subject and, therefore, an agreement relation is not available, resulting in the absence of a coreferentiality requirement. Moreover, given the discourse conditions usually associated with the preverbal higher subjects, such as the “quasi-topicality” effect noted by Chomsky (2002), this may account for the preference of ED to be associated with the preverbal subject, in contrast to what happens with the postverbal subject, as shown in passive constructions – when the theme is left in situ (I reported the sentences discussed in (16–17)):

This relation with the preverbal subject can be further strengthened by the co-occurrence of ED and topic constructions, such as Clitic Left Dislocation (Cecchetto 1999), where the topicalized element is signaled by the co-reference with a resumptive clitic lo immediately following the ED clitic:

ED is indeed natural in such contexts, further suggesting its involvement in the discourse properties of the sentence. In a similar vein, the analysis proposed above also predicts that ED does not appear in causative constructions, since they exhibit an infinitival complement characterized by an “impoverished functional structure”, i.e., lacking the C-I phase (Roussou and Manzini 2024). This impoverishment also affects pronominal clitics, causing their inability to remain in the subordinate clause (41b) and forcing them to climb to the matrix causative verb (Guasti 1993, 2006) (41c):

ED cannot appear in causative constructions at all (42a), whereas other dative clitics, both argumental (41c) and non-argumental (benefactive) (42b), can:

This patter can be easily taken into account if ED is introduced in the CP domain directly: if the structure is CP-lacking, as in causatives (Roussou and Manzini 2024 and the references therein), there is no space for ED at all, and the movement to the matrix clause is not available as a rescue operation. On the other hand, this does not affect other cliticizations, for example Ben or argumental ones.

The hypothesis explored in this paper aligns with the spirit of Jouitteau and Rezac’s (2008) proposal for French ED, despite the latter’s lack of a formal analysis. Specifically, both studies posit that these constructions originate outside the domains responsible for thematic roles and syntactic movement (Case/A-movement). Pursuing this line, ED cannot control PRO, as Jaeggli (1982: 31) shows for Spanish (a) and Jouitteau and Rezac (2008: 104) for French (b):

Note that other clitics, either dative or accusative, can control PRO. Only ED fails in this respect:

Assuming that ED is outside the thematic domain takes into consideration this fact as well.

The high position of ED also predicts the impossibility of it undergoing any type of A’-movement (Michelioudakis and Kapogianni 2013), such as wh-movement, since the high position of ED does not provide any coherent contexts where locality conditions on movement can be fulfilled (such as selection or similar principles^[25]).

Finally, this is also consistent with the enclitic nature of ED in infinitival clauses:

If ED serves as the head of an ApplP, it may become a suffix of the infinitive verb through Head-to-Head movement, wherein the non-finite verb moves towards the Fin°, preceding the ED: [_ForceP … [_FinP [incontrar_i- [_AppP [ti]] … [_TP t_i]]].

Consider that the “high” position of ED should not be viewed as an isolated phenomenon in Italian. In fact, it has been independently proposed for other expletive functional words, such as (i) negation – in both ‘Surprise’ negative sentences (Greco 2020b; Halm and Huszár 2021) and negative exclamatives (see, among others, Delfitto and Fiorin 2014; Espinal 1997; Villalba 2004; Zanuttini and Portner 2003) – and (ii) coordination structures (Poletto 2005). Specifically, the following Italian surprise negative sentence simultaneously exhibits all these expletive elements (in italics), i.e., coordination, negation, and ED.

The meaning of this sentence can be fully paraphrased as ‘That Maria got off the train is a surprise.’ Despite the presence of the negative marker non, the sentence remains affirmative and negation must therefore be considered expletive – according to the definition I gave in Section 1 – since it does not make a contribution to the truth conditions associated with the sentence. Moreover, the sentence is introduced by a coordination that lacks a left-branching phrase and does not seem to coordinate anything (Poletto 2005), instantiating another case of expletive item since it fails in its propositional role. In Greco’s analysis of surprise negative sentences, negation is externally merged in the CP-domain after the v*P-phase has closed and the entire TP has raised to focalization. This accounts for the sentence’s affirmative meaning, as negation loses its ability to reverse the polarity of the proposition (see the original works for a more detailed discussion). Similarly, Poletto (2005) proposes that the expletive coordination can appear in the fronting position as a topic marker occupying a functional projection in the CP field that is parasitic on focalization. Note that these elements interact with the presence of ED. Specifically, although ED is optional, it makes surprise negation sentences more natural (Greco 2020a) and seems to enforce the expletive reading of negation in negative exclamatives.

The negative exclamative in (47a) is ambiguous between a standard negation reading, where it reverses the polarity of the sentence, and an expletive negation reading, where it does not reverse the polarity but instead provides a universal evaluation of an event (e.g., ‘John walked a lot’). This ambiguity arises because standard and expletive negation in Italian share the same negative morpheme non (‘not’). However, in the negative exclamative in (47b), the presence of ED forces the interpretation towards the expletive negation reading, ruling out the standard one. According to Greco (2021), the different readings in negative exclamatives can be traced back to the twofold derivation of negation: when the negative marker ‘not’ is integrated into the TP-domain – as traditionally assumed in the literature (Zanuttini 2001 and references therein) – it results in a standard negation reading; when it is positioned higher, specifically in the CP-domain, it results in an expletive negation reading, as seen in the surprise negation sentence above.^[26] All in all, the activation of the CP field appears to be crucial for deriving the expletive reading of functional words and the analysis proposed here for ED aligns well with this tendency. Note that dative clitics, negation, and coordination seem to fail in their usual semantic contributions, but they do not lose their syntactic nature.

Overall, analyzing ED as a higher Applicative Phrase has the beneficial outcome of accounting for many of the characteristic features of ED discussed earlier. However, one property of ED does not seem to follow from the syntactic representation in (34), such as their preference to be realized by the 1st and 2nd person singulars (i.e., mi and ti). There is no a priori reason to expect such behavior, as the event described in the sentence can affect any individual, including a third party in addition to the participants in the speech act. In the next section, I will delve into this point.

3.2 Ethical Dative and Speech Act Phrases

According to some scholars (Delbecque and Lamiroy 1996; Roberge and Troberg (2009); see Michelioudakis and Kapogianni 2013 for an overview), the 3rd person singular is not as acceptable as ED as the 1st and 2nd persons. As discussed above, some cases seem to confirm this preference, as evidenced by the following negated sentence expressing surprise (I here repeat the sentence in (8) as (48)):

This preference can be traced back to some independent principles. It can be important to consider the relationship between the speaker and the hearer concerning both attention-seeking and bonding. More specifically, it has been suggested (Speas and Tenny 2003) that certain pragmatic information, such as some markers in languages like West Flemish and Romanian (Haegeman and Hill 2013), are syntactically represented by a functional predicative structure, i.e., a Speech Act Phrase (SAP) (Cinque 1999). SAPs dominate the left periphery of a sentence, instantiating the interface between syntax and conversational pragmatic. This realizes what Kratzer (1999) introduced in her famous paper on the “interpersonal” value of some expressions that indicate the speaker’s attitude or commitment towards the utterance’s content and/or their relationship with the interlocutor. They can be described as “conversational”, presupposing direct contact between the speaker and the interlocutor. Therefore, they would be inappropriate in formal discussions. In their seminal work, Speas and Tenny (2003) propose that a SAP comprises at least three sub-phrases: the speaker, the utterance content, and the hearer. Ethical Datives, in this context, function specifically to highlight a person – either the speaker or the hearer – who is affected by the event described by the sentence. An intuitive explanation for the preference of ED towards the 1st and 2nd person singular is to adopt Speas and Tenny’s hypothesis. One can assume that the speaker and the hearer of an utterance are operators capable of binding a variable realized in the [Head, ApplP] position, which corresponds to our Ethical Dative. If the speaker binds this variable, the ED will exhibit 1st-person agreement; if the hearer binds the variable, the ED will exhibit 2nd-person agreement (see the original work for the binding mechanism underling SAPs).

This syntactic explanation could clarify why the ED shows a preference for the 1st and 2nd persons, and why the 3rd person is less acceptable to some speakers. While the ‘affectedness’ property is always conveyed by the ED, the realization of person is constrained by the SAP. As for the 3rd person singular, it can be proposed (Michelioudakis and Kapogianni 2013) that some logophoric operators are present, ensuring the correct interpretation of 3rd-person pronouns as reported speakers or hearers in indirect speech (see the original works for a more detailed description of these operator–variable relationships).

Notably, in the analysis proposed here, ED is introduced by the ApplP rather than by SAP, which differs from the proposals of other scholars (see, among others, Delbecque and Lamiroy 1996). This approach has the advantage of accounting for all the syntactic and semantic phenomena caused by ED, which might otherwise be difficult to consider only from an SAP point of view. Therefore, it is important to recognize that the reference to SAPs is just one hypothesis that can explain the preference for the 1st and 2nd person singular (see, for example, Ross 1970 for a critique of this idea) and the pragmatic interpretation of ED. This further specifies the role that ED plays beyond grammatical meaning. However, regardless of the theory one assumes, it is incontrovertible that there are crosslinguistic differences between the 1st and 2nd person versus the 3rd person (Bloomfield 1938; Halle 1997). As Speas and Tenny (2003: 330) state, “only the participants in the speech act – the speaker and the addressee, represented by 1st and 2nd person – have true grammatical person”.