The Gutnish-specific si-passive combines be or become with a participle, directly followed by the element si. Unlike regular periphrastic passives, si-passives focus on the process rather than the result, opening up the construction for unergatives, which are unattested in the regular type. However, si-passives are quite limited when it comes to the subject. Internal arguments can only become subjects if they strand a preposition or a particle. Otherwise, the subject is expletive. I argue that si is part of the participle ending in Aspo, where its phi-features block agreement with the internal argument, trapping it in situ and depriving Aspo of a link to the result of the event. Originally a reflexive, si was reanalysed as a marker of the participle rather than the infinitive in contexts where conjugational changes had made them ambiguous. These changes never affected Fårö, where the si-passive is thus correctly predicted to be absent.

I argue that reflexive clitics are not pronominal, but verbal. Cross-linguistically, reflexive clitics can realize either an unaccusative or an unergative Voice head, both of which allow anaphoric interpretations (as suggested by the work of Reinhart and Siloni 2004, 2005). I contrast the anaphoric Voice analysis with two well-known pronominal analyses of reflexive clitics: one, proposed for French, postulating an anaphoric external argument (McGinnis 1998, Sportiche 1998), and another, proposed for Icelandic figure reflexives, postulating an expletive argument in [Spec, pP] (Wood 2014, 2015; Wood and Marantz 2017). Evidence against the external-argument analysis for French includes: a language-internal contrast between unergative and unaccusative anaphoric clauses (Labelle 2008); the absence of a c-command requirement on the licensing of anaphoric Voice; the absence of a lethal ambiguity effect with anaphoric Voice (McGinnis 1998, 2004); and the interpretation of focus constructions with seul ‘only’ (Sportiche 2014, Haiden 2019). Evidence against the Icelandic expletive-argument analysis includes: the observation that not all figure reflexives have a pP, or allow an impersonal passive (Moser 2021); and the difficulty of extending the analysis to other languages with reflexive clitics – in particular, the difficulty of accounting for the widespread observation that anaphoric clitics are restricted to referential dependencies involving the external argument.

There are conflicting reports in the literature concerning whether the Scandinavian languages use prepositional passives as in English. Maling & Zaenen (1985) showed that Icelandic does not have the construction; instead the Icelandic data should be analyzed as topicalization of the complement of a preposition in impersonal passives. They suggested that the same account would be appropriate for Danish and Swedish, whereas Norwegian is reported to have a rather productive prepositional passive (Lødrup 1991). In order to find out to what extent and in what ways prepositional passives are actually used, we carried out a series of investigations in Danish, Norwegian and Swedish contemporary text corpora, analyzing over 3600 potential prepositional passives, with a balance of periphrastic and morphological passive forms. We have found that prepositional passives are indeed used in all three languages, but rather infrequently, ranging from 3.4 per million words (3.4/mw) in Swedish, 5/mw in Danish to 16/mw in Norwegian. The majority of the prepositional passives are periphrastic bli(ve)-passives. The passive subject is typically animate, a person or an animal, who is psychologically affected by the action, or the lack of action, expressed by the participle. The notion of affectedness that is relevant for these languages thus differs from what has been described for English. Prepositional s-passives are found in coordinated structures and in infinitival complements of modal verbs, a context known to favour s-passive.