Despite being closely related languages, Norwegian and Swedish display well-known systematic differences in their properties of passives and middles (see e.g. Engdahl, Reference Engdahl1999). In both languages, passives can be formed morphologically (s-passive, as in 1), as well as syntactically (bli-passive, as in 2) (examples from Fábregas & Putman, p. 2):

In Norwegian, the syntactic bli-passive is the default form used to express a range of meanings, while the morphological s-passive is restricted to modal contexts. The morphological passive can also be used to express middles alongside an adjectival construction. In Swedish, the morphological passive is the default, while the syntactic passive is more restricted. The morphological passive form can typically not be used for middles. In their book from 2020, Passives and Middles in Mainland Scandinavian. Microvariation through Exponency, Fábregas and Putman argue that (some of) these differences follow from a difference in so-called exponents (see below) in the two languages. The book consists of six chapters: Chapter 1 is an introduction to the book, presenting the empirical observations and the gist of the proposal. Chapter 2 lays out the theoretical framework. Chapter 3 accounts for properties of passives in the two languages and Chapter 4 proposes syntactic analyses capturing these properties. Chapter 5 deals with middles and Chapter 6 is an extension of the analyses to anticausatives and reciprocals.

The main aim of the book is to explain what the s-morpheme corresponds to structurally in the two languages, accounting for their different properties (restricted to non-episodic, modal meanings in Norwegian but the default form in Swedish). The work is couched within a theoretical framework similar to Distributed Morphology (e.g. Halle & Marantz, Reference Halle, Marantz, Hale and Keyser1993) but posits an additional level of representation, ∑-structure, at which portions of syntactic structure are bundled into exponents which provide instructions for the PF and LF interfaces. They argue that the passive -s morpheme corresponds to different exponents in the two languages: In Swedish, it lexicalizes a pronominal element (see Hedlund Reference Hedlund1992, Julien Reference Julien, Ramchand and Reiss2007), while in Norwegian, it lexicalizes a portion of the extended verbal projection, crucially including a mood projection (cf. Engdahl Reference Engdahl1999). This is the reason why the morphological form can be used to form middles in Norwegian but not in Swedish (I will return to this towards the end). A considerable part of the book is also concerned with bli-passives. In this case, however, exponents play a smaller role and the same analysis is proposed for both languages. As an important part of their analyses of passives, they argue that actives and passives do not differ in their Voice projection, but in what moves into it. Voice, thus, does not come with different features in the active and passive. Instead, Voice is a relational head whose specifier is interpreted as the Figure, relative to the complement, the Ground (see Halliday Reference Halliday1967, Talmy Reference Talmy and Shopen1985).

The book offers an important contribution to the research on passives and middles in general and the area of Scandinavian languages in particular. The analyses presented capture the main empirical observations (many of which are known from before), in a systematic way, and in most parts the line of argumentation is easy to follow and appears to be sound. Having said that, I think the book would have deserved more careful proofreading and references to unpublished work should have been avoided. Below I comment on some parts of the book that I think need further elaboration or clarification.

Most researchers working on passives and middles have probably experienced considerable variation between speakers when it comes to what structures they accept and not. This is also something the authors of this book acknowledge, but, unfortunately, not in a consistent way. The reader is therefore confused when learning that Swedish expletive bli-passives formed from intransitive verbs are ‘impossible’ on p. 59, but (merely) ‘marked’ on p. 76 (based on Teleman et al Reference Teleman, Staffan and Erik1994). In a similar way, middle readings for the s-passive form in Swedish are said to be impossible in some places in the book, but subject to speaker variation in other places. For both these passive forms, it is the stronger claim that is taken up in the structural analyses. I conducted a simple google search to see if bli-passives with intransitive verbs are used at all and found 748 hits for the strings “det blev tränat/blev det tränat” (“it was/was it exercised’) and 3041 hits for the strings “det blev sprunget/blev det sprunget” (‘it was/was it run’). To what extent speakers would judge them as completely well-formed should of course also be investigated, but at least they do seem to exist. For their data on passives, the authors rely on a corpus study (Laanemets Reference Laanemets2012) and while the findings from this study tell us about the relative frequencies of the different constructions and what properties the constructions often have, these data are less useful when it comes to more fine-grained questions of acceptability: to what extent are things not found in corpora at all possible? When it comes to middles, the authors report two studies that they have conducted on Norwegian (18 speakers) and Swedish (12 speakers), respectively. However, few details about the experiments are given (no information about number of experimental items and filler sentences) and at least in the case of the study on Swedish, the number of participants seems too small to say much about different geographical dialects. Both the data on passives and on middles, thus, call for more systematic acceptability studies to be able to say that structures or readings are impossible.

Contra a large number of previous analyses, the authors argue that active and passive structures do not differ in what type of Voice projection they have (in terms of the feature set-up of the Voice head for instance). Voice itself is the same in actives and passives, and the difference instead lies in what appears in the Specifier of Voice and in what appears in the complement of Voice. Voice is a relational head, whose specifier is interpreted as the Figure in relation to the complement, which is the Ground. While I like this part of the analysis in principle, I also think that it needs to be coupled with independent tests for Figure-hood and Ground-hood.

In active structures, the external Agent argument moves into the Specifier of Voice and is thereby interpreted as the Figure, while in passives different situations obtain depending on what type of passive it is. Bli-passives typically have a resultative reading that s-passives lack. In bli-passives, it is the aspectual projection responsible for the resultative reading (originating below little vP and taking VP with the Patient argument as its complement) that moves to Spec,VoiceP in both Norwegian and Swedish. The effect of this is that the resultative reading is profiled and becomes the Figure. Regarding s-passives, Norwegian and Swedish differ. In Swedish it is the s-morpheme, which is a pronominal element, that is merged in Spec,VoiceP and should thereby be the Figure (although the authors do not say much about this). In Norwegian, in contrast, nothing appears in this position, so Norwegian s-passived lack a Figure (again not dwelt much on by the authors).

It is not clear how the authors have arrived at these Figure-Ground configurations in the different structures. In the absence of tests that show what part of the clause is the Figure and what part is the Ground, readers are left with their own intuitions. On my intuition, for instance, what is profiled, and would therefore be the Figure, is the logical subject (typically an Agent) in actives, but the underlying object (typically a Patient) in both bli- and s-passives, the rest of the clause (including aspectual differences) being the Ground. As this is not what the book argues, the analyses presented would be stronger if there were tests to show that the Figure is what they claim in the different structures. For s-passives in Norwegian, thus, it should be shown that they indeed lack a Figure-Ground partition, and from that it would follow that nothing should appear in Spec,VoiceP. As mentioned above, for Swedish, -s is argued to be a pronominal element merged in Spec,VoiceP. Although this analysis builds on previous analyses (Hedlund, Reference Hedlund1992 and Julien, Reference Julien, Ramchand and Reiss2007), more explicit statements as to its nominal nature and how this nominal element relates to the arguments of the verb (in particular the Agent by-phrase) and the Figure-Ground relation should be given.

As said above, bli-passives are argued to contain an aspectual projection below vP. The Asp projection defines external aspect which “does not belong within the verbal domain” (p. 121) and therefore has to move to Spec,VoiceP. One of the arguments for saying that this is external aspect is that the same participles are used to form perfect aspect (p. 112). The authors are not concerned with morphological agreement, but it is nevertheless worth noting that participles used in passives and perfect structures differ morphologically in Swedish. Leaving that issue aside, better motivations for saying why the aspectual projection is merged below the vP in a structural position where it is said not to belong in the first place should be provided. Furthermore, if participles used in perfect structures are the same as those in passives, it would have been useful to see how the former are derived. More specifically, if AspP, by virtue of defining external aspect, has to move from its position below vP into Spec,VoiceP in both passives and active perfect constructions, how come it is the Agent and not the Patient that becomes the subject in perfect constructions? And how are s-passives in the perfect, containing both an s-morpheme and a past participle, (här har det sprungits hela dagen ‘here it has been run all day’) derived?

The book covers a lot and it is only natural that not everything can be dealt with to the same extent. The status of the Agent in passives is discussed and well motivated. What is perhaps lacking is some remarks on why it can be left unexpressed in passives but not in actives, and, as said above, how the Agent relates to the nominal -s argument in Swedish. Future work could also elaborate more on expletive bli-passives in Swedish. As seen above, they are marked but do not seem to be impossible. An account where they are banned for reasons related to morphological agreement might therefore need to be rethought.

The main part of the book is concerned with passive structures but the authors also extend their analyses of these to account for middles in the two languages. In both Swedish and Norwegian, an adjectival participial construction, as in (3), is the most common way of rendering the middle.

The analysis developed in the book builds on Klingvall Reference Klingvall2007 (Klingvall Reference Klingvall2012 is a substantial elaboration of this, not referred to in the book). The adjectival participle has less structure than the participle used in bli-passives and is moreover embedded in an adjectival projection.

In Norwegian, middle readings can also be obtained from a structure using the passive s-form, but that is claimed not to be possible in Swedish. That is, the Swedish correspondence to (4) is said not to receive a middle reading but instead a habitual reading (examples from Fábregas & Putman p. 159–160).

As mentioned above, a larger acceptability study investigating whether the s-form is really excluded or just more restricted in Swedish would be welcome. If it turns out that middle readings of the s-form are possible in Swedish too, which I think they sometimes are, the syntactic structure for these would have to be discussed. I agree with the authors that the structure they propose for s-passives in Swedish cannot directly be recycled in potential middles with an s-form, so it would have to modified to account for such structures. This is not the case for the Norwegian s-form. The s-morpheme in Norwegian is argued to lexicalize Mood, Aspect and Voice, thereby accounting for the obligatory modal reading of s-passives. This structure can easily be extended to middles if, as the authors argue, the modal projection obligatorily present in passives hosts the generic operator associated with middle readings (Lekakou Reference Lekakou2005). The observations that Agents are allowed and modifiers are not obligatory in Norwegian middles can also be explained on the analysis. Being able to account for middles in this way is a very nice extension of the analysis of passives that the authors argue for. Despite the questions I have about some of the empirical claims in the book, I think this work is a very valuable contribution that anyone interested in Agent-demotion constructions, such as passives and middles, should read.