Since Christopher Potts’ seminal work (Potts Reference Potts2005), Exᴘʀᴇssɪᴠᴇs (or Usᴇ-ᴄᴏɴᴅɪᴛɪᴏɴᴀʟ Iᴛᴇᴍs, as Gutzmann prefers to call them) have garnered a lot of attention in the semantic and pragmatic literature. In his first book, Use-conditional Meaning (Gutzmann Reference Gutzmann2015), Daniel Gutzmann develops his own compositional semantic theory of expressives. In his new book, The Grammar of Expressivity, he turns to their syntactic properties and puts forward a bold hypothesis that expressivity is also a syntactic, not just a semantic phenomenon. More precisely, he claims that a syntactic feature for expressives, [Ex], needs to be postulated (which comes with an interpretable variant [iEx] and an uninterpretable variant [uEx], on a par with other syntactic features). In support of this proposal, Gutzmann presents analyses of the following three linguistic phenomena in terms of [Ex]:

1. 1. Non-local readings of ᴇxᴘʀᴇssɪᴠᴇ ᴀᴅᴊᴇᴄᴛɪᴠᴇs as syntactic agreement involving [Ex]

2. 2. DP-external intensification by ᴇxᴘʀᴇssɪᴠᴇ ɪɴᴛᴇɴsɪғɪᴇʀs in German as movement (and agreement) involving [Ex]

3. 3. Restrictions on ᴇxᴘʀᴇssɪᴠᴇ ᴠᴏᴄᴀᴛɪᴠᴇs as involving syntactic selection for [Ex]

The range of empirical phenomena and theoretical views discussed in this book is quite impressive, clearly demonstrating the author’s broad expertise as an empirically talented linguist as well as a theoretically competent semanticist and syntactician. However, our overall impression is that the book fails to give cogent arguments for the main thesis that [Ex] is absolutely necessary to account for the above three phenomena. In our opinion, there are parts of the book that would benefit from closer examinations of the author’s own analyses with as critical an eye as in other parts of the book where he discusses previous studies. The proposed analyses are certainly very interesting, original, and thought-provoking, but the reader is left with uncertainty as to whether they make correct predictions, and whether alternative analyses are possible. In what follows, we will play the devil’s advocate and attempt to argue for the antithesis that [Ex] is not necessary to analyze the above three phenomena. The possibility of alternative analyses that dispense with [Ex] shows that there is not enough evidence for the existence of [Ex].

In Chapter 4, Gutzmann discusses the readings of sentences like (1).

What is interesting about this sentence is that the negative attitude expressed by damn can be directed towards the cake (the local reading) or towards the event of the dog eating the cake (the non-local reading). Under the local reading, the speaker may not have a negative feeling towards the event itself (e.g. she might be happy that the cake is gone), although it is also compatible with the speaker having a negative attitude towards the event. Likewise, under the non-local reading, the speaker may or may not have a negative feeling towards the cake itself.

Gutzmann proposes that this ambiguity is to be accounted for in terms of syntactic agreement involving [Ex]. Specifically, he assumes that damn carries [uEx], which needs to syntactically agree with an instance of [iEx] that is in a certain syntactic configuration, in order for the derivation to converge. According to Gutzmann, there are (at least) two locations where [iEx] can appear in a sentence like (1): On the local reading, the definite determiner of the damn cake has [iEx], which gives rise to the interpretation that the speaker has a negative attitude towards the referent of the definite description; on the non-local reading, the covert complementizer C has [iEx], which gives rise to the interpretation that the speaker has a negative attitude towards the proposition that the dog ate the cake.

Gutzmann argues that this analysis accounts for locality restrictions on the ‘scope’ of expressive adjectives. In particular, he presents cogent evidence that an expressive adjective in an embedded finite clause in a sentence like (2) cannot receive a non-local reading about the entire sentence. That is, the negative attitude expressed by damn cannot be directed towards Peter’s having said what he said. Let us call this reading a ‘matrix-level reading’.

Gutzmann argues that this observation speaks for his syntactic account, as syntactic operations show locality constraints, and assuming that syntactic agreement across finite clause boundaries is illicit, the unavailability of the matrix-level reading of (2) is explained.

This is an interesting idea, but we think there are some undesirable features as well. Firstly, Gutzmann assumes that an expressive adjective like damn agrees with a determiner or a complementizer. While agreement between an adjective and a determiner is commonly observed across languages, agreement between an adjective and a complementizer, with nothing else in the agreement chain, seems to be very rare at best, and possibly unattested. This casts some doubt on Gutzmann’s agreement account for the non-local reading.

Secondly, and more importantly, it seems to us that there are counterexamples to the generalization that a matrix-level reading across a finite clause boundary is impossible. For instance, let us consider the following pair of sentences. In (3a), damn is in a finite clause, while in (3b), it is in a non-finite clause.

A number of syntactic phenomena show locality restrictions with respect to finite clauses, but not with respect to non-finite clauses, including negative concord in Italian, to which Gutzmann likens the present phenomenon (one complication that Gutzmann ignores is that some negative concord items in Italian such as mai ‘never’ can be licensed by negation in a higher clause). It is then expected that (3b) has a matrix-level reading where the negative attitude is towards Peter’s telling Ann what he told her, while (3a) does not. However, these sentences seem to have the same range of possible readings. In particular, (3a) has a matrix-level reading. Space limitations prevent us from examining more data but similar examples with matrix-level readings do not seem to be difficult to construct.

We therefore think Gutzmann’s syntactic account is not without issues. Certainly, sentences like (2) do not seem to have a matrix-level reading, which Gutzmann convincingly shows with quantitative data, and why (3a) but not (2) allows for a matrix-level reading needs to be understood. We suspect that one of the factors is how much content the matrix material expresses and how likely it is for the speaker to have a negative attitude towards it, but we cannot offer a complete account at this point. What is crucial in the present context is that the availability of the matrix-level reading for (3a) points to the conclusion that the relevant restriction is not syntactic in nature. Then, we should seek for an alternative explanation, perhaps a pragmatic one, instead of the one with [Ex].

We would also like to point out an issue with Gutzmann’s account of what he calls ‘mixed expressive adjectives’ like shitty. These expressive adjectives are so-called because they express negative attitudes like damn but they also behave like ordinary adjectives in contributing to the at-issue meaning. Consequently, they allow for degree modification and can be used predicatively, as Gutzmann discusses with examples in Section 4.7.2 of this book. In the same section, he proposes that mixed expressive adjectives also have [uEx], like damn, but unlike damn, they also contribute to the at-issue meaning of the sentence. Concretely, Gutzmann analyzes the at-issue meaning of (4) to be roughly that the relevant car is bad, and its expressive meaning to be expressing the speaker’s negative attitude towards this proposition.

Note that in this sentence, shitty is used predicatively, and therefore the only element that can agree with its [uEx] is the complementizer. Consequently, the negative expressive meaning is predicted to be about the whole proposition, which Gutzmann says is a correct prediction (p. 122). However, this account makes a wrong prediction for the negated version of (4) that it should also have an expressive meaning.

This example does not seem to be associated with the speaker’s negative attitude towards the proposition that the car is not bad (although it is compatible with it), contrary to Gutzmann’s prediction. A similar issue arises with attributive uses of shitty in sentences like the following. This sentence can be used without necessarily implying that the speaker has a negative attitude towards Paul’s not having a bad car or towards any car.

Therefore, adjectives like shitty simply do not behave like expressive adjectives like damn, and Gutzmann’s analysis that uses [Ex] does not seem to be appropriate for them.

In Chapter 5 of the book, Gutzmann discusses intensification brought about by expressions like sau, voll and total in German (to simplify, we will only consider sau here). An interesting fact about these expressive intensifiers (hereafter EIs) is that they can appear following the determiner, as in (7a), or preceding the determiner, as in (7b).

In both of these examples, sau intensifies the meaning of the adjective coole. While the structure like (7a) is not particularly surprising, the word order of (7b) needs an explanation. Gutzmann analyzes sau to be within the DP in (7a) and outside the DP in (7b), and calls them an internal EI and an external EI, respectively.

Gutzmann observes several peculiarities of external EIs in German. In particular, the following two are important:

1. 1. External EIs are degraded with indefinite articles (with some inter-speaker variation), and with them, the definite article following an external EI can receive an indefinite reading as well as a definite reading.

2. 2. When certain nominal modifiers intervene between an external EI and what it modifies, the expression is ungrammatical, as in (8).

In order to account for these properties of external EIs, Gutzmann proposes that an EI carries [iEx] (or more specifically [iEx: INT] for intensification meaning) and undergoes head movement to D, when D has [uEx]. Thus, for him, sau die is a constituent on the surface in a sentence like (7b). With these syntactic assumptions, Gutzmann claims that the above two properties of external EIs can be accounted for as follows:

1. 1. When sau moves to D, the resulting complex head is spelled out as sau die, regardless of the (interpretable) definiteness feature of D. For some speakers, it could also be spelled out as sau eine, when D carries an (interpretable) indefinite feature.

2. 2. The head movement constraint forbids head movement that skips a head between the original position and the landing position, so when there is an intervening adjective as in (8), sau cannot move across it. (Also, sau cannot move to D, when there is already something else that has moved to D.)

We have several qualms about this analysis. Firstly, head movement out of an adjunct seems to be highly unusual, especially when it is just the head of the adjunct that moves. Specifically, Gutzmann assumes that sau heads DegP, which is an NP adjunct, and the Deg head moves to D as depicted in (9).

We do not seem to find such head movement in the verbal domain, for example. This movement would be comparable to movement of an element like very to T or C in a sentence such as The boy speaks very rapidly. We are not aware of such head movement (unlike phrasal movement of a degree adverbial such as how much). More generally, an element like D seems to be unable to have a direct syntactic relation with a degree adverbial head like very.

Secondly, we are not so sure about the cogency of the analysis of the intervention effects. As Gutzmann correctly points out, head movement is often considered to be strictly local, but it is also known that there are elements that do not participate in the computation of locality for head movement. For instance, in V-to-T movement, adverbials and other adjuncts do not intervene. Now, in an example like (8), the intervening material is an adjective, and it is not immediately clear to us if this adjective should intervene in head movement. The configuration in this case is not exactly identical to the case of V-to-T movement, because what undergoes head movement in Gutzmann’s analysis of external EIs is the head of a complex adjunct, and it is not impossible that other adjuncts intervene for head movement out of adjuncts. But again, we are not aware of other cases of head movement out of adjuncts, and thus we are unable to draw a firm conclusion here.

Thirdly, we wonder if this analysis really requires movement in syntax, as Gutzmann claims. Under his account, the spell-out rules do a lot of work: they spell out a complex head made up of a determiner and sau in a particular way. We think there is a possibility that these spell-out rules could alternatively be understood as purely morphophonological rules applied to linearly adjacent items. There is independent evidence for such morphophonological rules, for example, when want and to are adjacent in the syntactic representation, they can fuse into wanna in the morphophonology (the effects of empty categories in this phenomenon are well known but do not concern us here much, as it is not too farfetched to assume that these rules are sensitive to empty categories). This contraction is considered to not necessarily involve head movement (although the possibility of a head-movement analysis is not excluded). Then we could formulate the relevant rules as follows: (i) an optional metathesis rule that swaps D and sau in the following configuration: [D [_NP [_DepP sau AP] NP]]; (ii) a rule that spells out an indefinite article as a definite one in the same configuration. Moreover, we observe that sau in this configuration receives a prosodic prominence (thanks to Georg Höhn, p.c.). This observation is surely not incompatible with Gutzmann’s syntactic analysis, but one could also see it as what is driving the metathesis rule. Admittedly, these morphophonological rules seem ad hoc, and one might not buy this alternative account, but it should be noticed that the spell-out rules that Gutzmann postulates are equally ad hoc. For these reasons, we are not convinced that external EIs motivate [Ex].

Chapter 6 of the book is devoted to expressive vocatives like you idiot (or du Idiot in German), which Gutzmann claims involves syntactic selection for [Ex]. Specifically, Gutzmann assumes that the determiner use of a second person singular pronoun (in German) always carries a syntactic feature [*Ex, NP*] which requires an NP complement that has [iEx] and has expressive meaning. The second person pronoun denotes the listener, so is of type e, and the NP complement with [iEx], e.g. idiot, denotes a function of type ⟨e, u⟩. Then the entire DP denotes the listener and also has expressive meaning.

We think that Gutzmann’s arguments that [Ex] is necessary to obtain this result are not convincing, because an alternative analysis that dispenses with this feature seems to be possible. Specifically, we could assume that it is the determiner use of a second person pronoun that triggers expressive meaning, rather than that it requires an NP complement with expressive meaning. Gutzmann’s semantic system allows for an analysis where the determiner use of a second person pronoun with a truth-conditional type of ⟨⟨e, t⟩, e⟩ and a use-conditional type of ⟨⟨e, t⟩, u⟩ can combine with an NP of type ⟨e, t⟩ via Mixed Saturation defined in rule (2.55) of the book to give rise to essentially the same result as under his account. Note that an NP like idiot does have a type-⟨e, t⟩ use, as in You are not an idiot (recall the above discussion of mixed expressive adjectives). This also straightforwardly explains why other NPs in this construction, e.g. you linguist, also give rise to expressive meaning.

At this moment we do not have evidence that favors this alternative account over Gutzmann’s, but the possibility of the former implies that expressive vocatives do not necessarily motivate the syntactic feature [Ex].

In conclusion, we think that Gutzmann does not give sufficiently convincing arguments for the main thesis of this book that the expressive feature [Ex] exists. Having said that, it is clear that the phenomena Gutzmann discusses in the book are of theoretical interest, and the book is full of new empirical observations and insightful ideas that could lead to new discoveries. In particular, it is an important contribution of this book that it directs linguists’ attention to the syntactic side of expressives.