Largely due to work by Philippe Schlenker (e.g. Schlenker Reference Schlenker1999, Reference Schlenker2003, Reference Schlenker2004), recent years have seen intensified interest in indexicality and how it filters into contexts of speech and thought representation. For linguists, this has re-raised the question of how to model the division of labor between grammar and pragmatics. Our modest purpose here is to contribute to this debate by reviewing a particular aspect of logophoricity in Icelandic, namely, its licensing within infinitivals.

It has become a matter of general agreement that Icelandic reflexive sig can find an antecedent independently of structural conditions like being a co-argument, a clausemate, or c-commanded, as long as the minimal CP containing sig is a subjunctive CP whose content represents the speech or thought of the antecedent (see Reuland Reference Reuland, Everaert and van Riemsdijk2006, and references cited there). At the same time, Reuland & Sigurjónsdóttir (Reference Reuland, Sigurjónsdóttir, Bennis, Pica and Rooryck1997) have argued that sig, when minimally contained within an infinitival clause, must obey structural binding conditions (see Reuland Reference Reuland, Everaert and van Riemsdijk2006, Thráinsson Reference Thráinsson2007). In fact it was concluded that

The evidence from infinitivals supporting such a view crucially boils down to the following three examples (Reuland & Sigurjónsdóttir Reference Reuland, Sigurjónsdóttir, Bennis, Pica and Rooryck1997:330; Reuland Reference Reuland, Everaert and van Riemsdijk2006:549; see also Thráinsson Reference Thráinsson2007:494).

1. (1)

   

Clearly, ‘[t]hese sentences contain a constituent that is a possible perspective holder, but does not c-command sigʼ (Reuland Reference Reuland, Everaert and van Riemsdijk2006:548). ‘Thus, the sentences in [(1)] show that for sig in infinitival clauses, discourse factors are unable to compensate for the lack of c-commandʼ (Reuland Reference Reuland, Everaert and van Riemsdijk2006:549).

However, as pointed out by Gärtner (Reference Gärtner2009), the examples in (1) violate an important condition on logophoric licensing: the infinitival CPs do not represent (or invite construal as representing) the speech or thought of the putative attitude holder, Jón. They do not express the content of Jónʼs opinion, wish, or belief but, instead, form part of predications over these abstract objects. Via an anonymous reviewer, this critique has found its way (in simplified form) into the recapitulation of these matters by Reuland (Reference Reuland2011:314–323). There it is suggested that ‘ideally one should test structures of the form [(2)]ʼ (p. 320).

1. (2) Jónʼs wish is for SIG to have talent.

Also, it is correctly pointed out that (direct counterparts of) such structures, i.e. for-infinitivals, are absent from Icelandic. At the same time, it is denied somewhat apodictically that ‘any other structure with the required properties . . . exist[s] in Icelandicʼ (ibid.). This is what I would like to take issue with. In fact, quite acceptable counterexamples like the control infinitivals in (3) are not so hard to construct.Footnote ¹

1. (3)

   

The infinitival in (3b), for example, is (construable as) used to report on a first person directive by Jón such as ‘Cite me on every page!’ or ‘I want you to cite me on every page’.

Interestingly, there is some speaker variability concerning such examples (see note 1). However, as became clear to me through comments by Höskuldur Thráinsson (p.c.), the main factor on which to pin this variability is the ease with which speakers tolerate ‘object’ control in control noun configurations. Thus, even for speakers who allow this, a slight blocking effect may result from the fact that a perfectly acceptable finite subjunctive competitor exists (að við styðjum sig _ivið þessar aðstæður ‘that we support him in this situationʼ) that resolves any interpretive ambiguities. Further clarification requires more large scale inquiry into control noun configurations (see e.g. Restle Reference Restle2006), an undertaking that goes beyond the scope of the current remarks.

One may, of course, wonder whether the existence of logophors inside infinitival clauses changes anything for the interaction between grammar and pragmatics. This clearly depends on how that interaction is modeled. Reuland (Reference Reuland, Everaert and van Riemsdijk2006:552) tentatively concludes that ‘[i]nsofar as Icelandic logophoric sig requires a subjunctive . . . the role of the subjunctive should be that of blocking a syntactic connection between sig and its antecedent’. The implementation of this goes back to Reuland (Reference Reuland2001:466–467). Accordingly, the bindable features of the reflexive, φ_sig, get (optionally) attracted (along with I°) by a subjunctive operator in the C-domain. From there, φ_sig is no longer available for A-chain-formation and thus no longer accessible for structural binding. This mechanism accounts for the option of logophoric construal.

Now, if one wants to preserve such a perspective in the light of examples like the ones in (3), it is by no means unattractive to postulate the optional presence of a similar operator in the C-domain of control infinitivals.Footnote ² Formally, that option plausibly does not arise in cases of (counterparts of) clause union, so the standard mechanism of extending the binding domain for reflexives in infinitivals, i.e. ‘verb raising’ at LF according to Reuland & Sigurjónsdóttir (Reference Reuland, Sigurjónsdóttir, Bennis, Pica and Rooryck1997:336 fn.18), remains unaffected there. On the interpretive side, it is striking that Kempchinsky, one of the authors endorsed as sources for the operator analysis of the subjunctive (Reuland Reference Reuland2001:466), suggests ‘that subjunctive complements to verbs of volition, influence, and command are in some sense like embedded imperativesʼ and that licensing involves ‘an imperative operator in the subordinate Cʼ (Quer Reference Quer, Everaert and van Riemsdijk2006:669). This fits nicely with our above observations regarding the examples in (3).Footnote ³

Independent evidence for the necessity of a suitable operator in the left periphery of infinitivals would also come from the licensing of logophoric sig in purpose clauses. Example (4) is a case in point.Footnote ⁴

1. (4)

   

In sum, there are reasons to believe that Icelandic does allow logophoric reflexives inside infinitival clauses. At the same time, there are reasons to believe that this can be treated in ways compatible with a strict division of labor between grammar and pragmatics as envisioned by Reuland & Sigurjónsdóttir (Reference Reuland, Sigurjónsdóttir, Bennis, Pica and Rooryck1997) and Reuland (Reference Reuland2001, Reference Reuland, Everaert and van Riemsdijk2006, Reference Reuland2011).Footnote ⁵