Xin Sennrich's monograph The Many Faces of English -ing comprises an Introduction, Conclusions and seven substantive chapters. Chapter 2 (pp. 13–17) outlines the main assumptions. Word classes are defined in terms of HPSG-style inheritance hierarchies. Gerunds have the external distribution of nouns but the internal syntax (dependents, modifiers) of verbs. However, the noun class as such does not provide any specific set of dependents that must be inherited by gerunds, so that a gerund can be defined as just a rather idiosyncratic type of noun (p. 14). Participles are purely adjectives (again with the internal syntax of verbs). However, in section 2.2 gerunds are argued to be mixed categories, following Malouf (Reference Malouf2000), inheriting from the type noun and a supertype relational, which subsumes verbs and adjectives as well as gerunds.

Chapter 3 (pp. 18–28) summarises the lexical, syntactic and semantic properties of the adjective category. Relational adjectives (derived from/motivated by nouns) denote concrete or abstract nominal entities: financial (advisor), dental (decay), from finance, tooth. Participles are event-denoting adjectives. They can be transitive, just like the (true) adjectives worth, near, like.

Chapter 4 (pp. 29–83) argues that even aspectual participles are (just) adjectives. Hence, The boy is playing (on) the piano has the same syntactic structure as The boy is near (to) the window. In The prisoners have escaped, escaped is the predicative complement of have. This verb does not take prototypical adjectives because it only selects event denotations: *She has nice/tired. Participles function as pre/postmodifiers. This includes the predicative complement of a perception verb with a controlled object: I saw [the boy][_____ smoking in the class room]. Where the participle is an attributive postmodifier, I saw [the [boy smoking in the classroom]], V-ing does not realise progressive aspect: ‘progressive aspect is not expressed by the present participle itself, but is realised as the composition of the predicate verb be and the event-denoting semantics of the present participle’ (p. 38). In examples such as Sennrich's (17) the participle functions as a free adjunct: (17) Standing on the chair, Tom can touch the ceiling.

Section 4.2 argues that participles have the same external distribution as adjectives and so they must be (simply) adjectives. The meaning of participial adjectives, interesting, charming, etc., cannot be derived from the verb semantics (p. 52), but it is the only thing which distinguishes them from the participles they are derived/converted from. Participial adjectives denote properties, while true participles denote events and hence are unable to serve as the complement of a verb that selects a property-denoting predicate such as seem, look, sound, become. There is a contrast between postmodifying present participles (the girl reading a book) and what Sennrich calls appositive postmodification (You'll have great fun choosing a name for your duck, example (72a)).

Chapter 5 (pp. 84–95) examines the consequences of treating participles as (just) adjectives. Sennrich argues that there would be problems if we assume that participles and adjectives belong to different categories in that it would entail one of the following: that participial adjectives are (i) derived from verbs via -ing/-ed suffixation; (ii) converted from the participles; (iii) diachronically lexicalised as adjectives. Analysis (i) fails to explain adjectives derived from irregular past participles (drunk, broken, …). Analysis (ii) would require a derivational process, conversion, to be performed over an inflected form, the participle, in contravention of a putative universal principle of (English) morphology which states that regular inflection cannot appear inside derivation. Analysis (iii) has the problem that it is more productive than lexicalisation usually is, and is morphologically and semantically more transparent than uncontroversial cases of historical lexicalisation such as cunning, gruelling.

Section 5.1.4 presents Sennrich's account of participial adjectives: an event-denoting participle only has to undergo semantic shift to become a property-denoting term like a prototypical adjective (p. 90): the boy is very charming (*the audience).

Sennrich devotes section 5.2 (pp. 92–5) to aspect and voice. In the progressive aspect construction the participle is not a verb (pp. 93, 95), rather, in her example (6b) He is smoking, smoking expresses a stage-level predication, as opposed to the individual-level predication expressed in (6a) He smokes. The apparent grammatical categories of perfect aspect and passive voice are analysed ‘as a knock-on effect for [sic] other areas of English morphosyntax’. The ‘perfect aspect’ is no more than the composition of the adjectival ‘perfect participle’ as the predicative complement to have, while passive voice is just the composition of the (homophonous) ‘passive participle’ with the predicative complement to be.

Chapter 6 (pp. 96–129) considers the categorial status of gerunds. They contrast with the associated nominals, the building of the bridge, which are entirely nominal. Gerunds have the external distribution of nouns, but ‘phrases headed by gerunds have the internal structure of verb phrases’, and this ‘contrast leads to the explanation for the categorial status of gerunds: the gerund is a mixed category’ (p. 97). However, associated nominals, like gerunds, denote events (‘eventualities’, p. 99), presumably dynamic as opposed to stative situation types. Thus, a stative verb such as know does not allow an associated nominal: ??John's knowing of calculus (cf. John's knowledge of calculus).

Subsection 6.2.2 demonstrates that gerunds have the external distribution of noun phrases: subject, subject complement to be, direct object, complement of preposition. This includes indirect or first objects in double object constructions: Mary gives playing the piano all her energy and time [sic]. (Cf. also Mary gives all her time and energy to playing the piano.) (This appears to be the only distribution that is never available to non-nominal phrases, e.g. finite or to-infinitival clauses.) One place where gerunds are disallowed is complements to certain verbs, such as know, promise: Mary promised *writing/to write the letter. Sennrich suggests that this is because of high-frequency competition from to-infinitives (p. 105), though it is hard to know how that would explain the distribution, or why both types are available to other verbs: Mary continued writing/to write the letter. One construction which permits gerunds is extraposition with certain adjectives: (22b) It is pointless buying so much food (also possible with the infinitive), while a genuine noun is impossible in this position: (23b) *It is pointless the purchase of so much food. It is not clear how this advances Sennrich's claims, however, if gerunds are supposed to be nouns (and if infinitives are verbs).

The contrast between gerunds and to-infinitival phrases is addressed in subsection 6.2.3, where Sennrich points out that infinitival clauses can also function as subjects, subject complements, direct objects, though not as indirect objects or complements to prepositions.

Section 6.3 explores the verbal internal syntax of gerunds, contrasting this with the nominal internal syntax of associated V-ing nominals. One obvious difference is discussed under heading F, p. 117, ‘aspect’, which I cite (almost) in full:

Gerunds can also be passivised – Tom's regularly/*regular being helped by his colleagues … – and can license various types of double object/complement construction – John's giving Mary his car, her hammering the sheet flat, her expecting/persuading John to see the doctor – as well as particle/prepositional verbs – his looking up the information / looking the information up vs *his looking up of the information / looking of the information up.

After a very short historical survey of the development of the single V-ing form, chapter 7 (pp. 130–59) is devoted to establishing that gerunds and present participles are distinct parts of speech. This runs contrary to the position advanced in The Cambridge Grammar of the English Language (Huddleston & Pullum et al. Reference Huddleston and Pullum2002, CGEL henceforth), where the two usages are subsumed under a single category of ‘gerund-participle’, whose distribution and interpretation is dependent on the morphosyntactic context, not on intrinsic properties of the lexical form. Sennrich enumerates a set of gerund-only contexts and participle-only contexts so as to establish complementary distribution (much of this largely repeats what was said in the previous chapter). She then develops a series of grammatical tests based on the idea that gerunds and participles will behave differently when they follow a verb (section 7.3). Since a gerund is a noun, it should be able to serve as the complement of a verb, and undergo noun phrase oriented processes. A participle, being an adjective, should fail to undergo such processes. Sennrich contrasts gerund complements, as in They discussed [visiting the museum] with what for her is a participial complement to the verb keep: They kept [visiting the museum]. She then looks at sequences of V-ing V-ing, listing four possible combinations: (i) gerund + gerund, (ii) gerund + participle, (iii) participle + gerund, (iv) participle + participle. Of these, all are claimed to be possible except for (iv): (i) We enjoy celebrating winning the competition; (ii) Keeping practising regularly is important; (iii) We were celebrating winning the competition; (iv) *We were keeping travelling to Europe. She concedes that case (ii) is ‘controversial’, in that many speakers reject such examples. This includes me. For speakers such as me the generalisation would be that a participle cannot serve as the complement to any V-ing form, though little seems to hinge on this.

In section 7.4 Sennrich considers constructions of the form Verb + NP + V-ing. What for Sennrich are gerund complements to verbs such as dislike, appreciate, remember appear in two forms: as Acc-ing constructions (31a) or as Poss-ing constructions (31b) (Sennrich does not use this terminology):

With other types of verb only the Acc-ing construction is possible (CGEL: 1238, 3Cii):

With verbs like catch, keep, etc. we are dealing with a raised object or exceptional case marking construction: the pronoun him/them is in the object form as though it were a complement of the main verb, but it also functions as the subject of the V-ing form. With a genitive (his/their) the pronoun can only be in the subject position of the V-ing clause (specifically, the determiner of the V-ing nominal) (pp. 148–53 takes us through the standard arguments from the 1970s for distinguishing these cases). The following section then summarises Sennrich's arguments against the unitary gerund-participle category. CGEL takes the gerund and participle uses to be in complementary distribution. Sennrich disputes this on the grounds that both can be the complement of be: My hobby/son is playing the piano. But note that the authors of CGEL, like most grammarians, assume that copular be and auxiliary verb be are distinct elements and set up distinct grammatical contexts, so for those grammarians the two instances really are in complementary distribution. Sennrich (p. 156) cites some of the examples discussed in CGEL (p. 1221), in their critique of the gerund/present participle distinction.

She argues that seem, become, etc. select property-denoting complements and the present participles are event-denoting. This leaves the examples with the main verb keep. Sennrich says of these (p. 156, essentially following CGEL): ‘Both They kept staring at them and They kept calm are grammatical, but their semantics are different. The present participle staring denotes the event that the subject is involved in, whereas the prototypical adjective calm denotes a property of the subject.’

Sennrich also takes issue with the traditional assumption (also found in CGEL) that the gerund-participle V-ing is an inflectional suffix. She claims that -ing is always a derivational suffix. As ever, the principal argument is external distribution: only derivational morphology can change the part of speech from verb to noun or adjective. A final alleged problem for the CGEL approach, in which the participles are all inflected forms of the verb, arises when they converted to true adjectives: boring, tired. Sennrich claims that such conversion violates a general principle of (English) grammar, which says that derivation cannot apply to inflected forms.

Chapter 8 (pp. 160–81) looks at constructions in which the V-ing form premodifies a noun in compounds and noun phrases. In drinking water and similar compounds the V-ing form has to be an associated V-ing nominal, and not a gerund. The semantic relation between the head and modifier is just as varied as that in any noun–noun compound. Sennrich claims (p. 174) that no ascriptive interpretation is possible, comparable to that found in boy actor, luxury flats, steel bridge. (However, it seems to me that you can get examples of coordinate compounding such as the writing editing process, an acting directing role. Sennrich does not discuss such cases.) The dancing girl example has two stress patterns, the typical compound stress, dáncing girl, and a phrasal stress pattern: dancing gírl. For Sennrich this means that the first is a compound, while the second is a case of a participle serving as an attributive modifier to the head noun.

This monograph presents a useful summary of a good many of the facts relating to V-ing forms. None of the data are new (and in some cases they are facts which have been discussed for the past sixty years). For the most part the discussion is easy to follow, though it is rather repetitious in places. The analysis claims to be based on Malouf's multiple inheritance model of gerunds, but there is virtually no discussion of the HPSG background and the HPSG technology is not really utilised, so this cannot be regarded as an HPSG analysis or even an analysis in terms of multiple (or orthogonal, or default) inheritance.

What is novel is Sennrich's theoretical proposals:

1. 1. No V-ing forms are actually verbs: they are either (mixed category) nouns or adjectives.

2. 2. There is no category of aspect.

3. 3. Participles are adjectives that can have exactly the same complementation/ modification properties as verbs and can denote (dynamic) events.

4. 4. There is no unitary gerund-participle category. Gerunds and participles are not in complementary distribution.

Point 2 presupposes that there are no aspectual auxiliaries, only lexical be/have.

Sennrich's account leaves unanswered a number of rather difficult questions:

• • Why do no other adjectives have the same verb-like properties as participles?

• • How precisely does the copular be + present participle come to acquire the semantics/grammatical function of the progressive aspect?

• • How can perfect aspect be computed? Is the perfect participle really a kind of passive participle? How can have take an adjective phrase complement?

• • If the gerund-participle is two distinct categories, despite being formally identical, is this also true of the perfect and passive participles?

Perhaps the most important question is this: if gerunds are mixed categories because of their VP-like internal syntax, why are participles equally not mixed categories? Without a convincing answer to all of these questions, Sennrich's proposals lose much of their appeal.

It seems that the account in CGEL and that of Sennrich can be reconciled to a large degree if we recognise that gerunds and participles are examples of transpositions (Spencer Reference Spencer2013) or category-changing inflection (Haspelmath Reference Haspelmath, Booij and van Marle1996). This is reflected in the fact that they are mixed categories. It also helps if we distinguish two different types of paradigm, a form paradigm, which lists all the forms of a lexeme and the morphosyntactic properties they are associated with, and a content paradigm, which lists all and only the morphosyntactic and morphosemantic properties that are accessible to syntax/semantics (Stump Reference Stump2016). The content paradigm would distinguish English verbal categories such as bare/to-infinitive, imperative, non-3sg present indicative, but the form paradigm would only list the single base form as the form-correspondent to all of those disparate categories (syncretism). We can then say, with CGEL, that the gerund-participle, or ‘V-ing form’, is single member of the form paradigm, but also say, with Sennrich, that it realises several properties at the content paradigm level, including the associated V-ing noun (noun), gerund (mixed noun–verb), attributive modifier (true participle, mixed adjective–verb) and converb to the progressive auxiliary be (non-finite verb form). Transpositions are frequently converted into other parts of speech, either by the grammar or by lexicalisation processes.

I noticed very few typos: the predictive complement for the predicative complement (p. 95); subject, complement for subject complement (p. 107); *There is a good change for *There is a good chance (p. 108); prediction for predication (p. 122); and on p. 154 the first line accidentally repeats the last line of the previous page.