This book deals with two related topics: the difference between English compound nouns and phrases, and the relation between lexicon and syntax in a lexicalist approach to word formation.

As to the first topic, Heinz Giegerich raises the question whether the difference between tóy shop (a shop where toys are sold) and toy shóp (a shop which is a toy) can be characterized as the distinction between compound and phrase, parallel to a dáncing instructor (someone who teaches dancing, a compound) versus a dancing instrúctor (an instructor who dances, a phrase). The relation to the second topic is that, if certain word combinations are phrases but have at the same time lexical status, the lexicalist position that syntax and lexicon form different modules of grammar has to be weakened.

In the first chapter, a number of useful semantic categories are introduced for the characterization of the semantic relation between modifier and head in N+N and in Adj+N combinations. The most important classification for the argumentation of this monograph is the distinction between ascriptive and associative attribution. For instance, in beautiful house there is ascriptive attribution, as the property of being beautiful is ascribed to the head noun house. In the Adj+N combination dental decay, on the other hand, we see associative attribution: the decay is not qualified as dental, but the decay is related to the referent of dental, namely teeth. This semantic distinction is well known from the linguistic literature, often under another label, the distinction between qualitative and relational adjectives (Booij Reference Booij2002b: 108–9).

Chapter 2 focuses on the morphology and lexical semantics of associative adjectives. Whereas ascriptive adjectives can occur in predicative position, and be modified, as in a very beautiful house, the house is beautiful, this does not hold for associative adjectives: *very dental decay, *the decay is dental. Hence, one may conclude that ascriptive adjectives project adjectival phrases, but associative adjectives do not. Giegerich argues that the Adj+N combinations with associative adjectives might therefore be considered to be lexical in nature (p. 34). However, a test with pro-one-anaphora reveals that they are not subject to Lexical Integrity, as we can have sentences like (p. 36):

1. (1) Is he a rural policeman or an urban one?

in which the part policeman is accessible for pro-one-replacement. Giegerich points out that the location of main stress cannot be used to decide if a word combination is a compound or a phrase: not all English compounds have forestress, and many Adj+N combinations of the type polar bear have forestress.

A first comment on Giegerich's reasoning is necessary here. From related Germanic languages such as German and Dutch we can learn that Adj+N combinations with associative adjectives are phrases, because they exhibit agreement between the adjective and the head. For instance, given the assumption of Lexical Integrity (‘syntactic rules cannot manipulate pieces of words’), the Dutch construction de financiël-e adviseur ‘the financial advisor’ must be assumed to contain an independent adjective financiëel, as it is inflected by means of the suffix-e. Hence, such Adj+N combinations cannot be compounds (Booij Reference Booij2002a, Reference Booij2009; Schlücker Reference Schlücker2014). English does not provide this evidence, as English adjectives are not inflected. However, the corresponding data in related Germanic languages make clear that the notions ‘word’ and ‘lexical item’ should not be identified: all words are lexical items, but not all lexical items are words. The phrasal status of the English Adj+N combinations with associative adjectives is confirmed by the pro-one-anaphora test.

Before taking up this issue again, Giegerich makes two detours. In chapter 3 he provides a lengthy and very informative analysis of the stress pattern of English compounds, with a convincing conclusion: the location of main stress cannot be used as a decisive criterion for determining if a word combination is a compound or a phrase. In particular, compounds may have either forestress or final stress, and Giegerich speaks of ‘the stress mess of compounds’ (p. 75).

The second detour, in chapter 4, is entitled: ‘Interlude: the porous nature of lexical stratification’. In this chapter, Giegerich discusses the theory of lexical strata, in particular the idea that the English lexicon consists of two strata. This topic has been dealt with extensively in the linguistic literature, and the empirical and conceptual problems for this theory are well known (see Booij Reference Booij, Cheng and Sybesma2000 and Bauer et al. Reference Bauer, Lieber and Plag2013: chapter 27). Giegerich is aware of these problems, but wants to stick to the stratal theory. What he concludes is that the two strata partially overlap, and that some morphological processes can apply in both strata. This leads Giegerich to a more general conclusion concerning the organization of grammars: modules may partially overlap. Just as strata 1 and 2 of the English lexicon can partially overlap, so can stratum 2 and syntax: word combinations like dental decay are lexical but at the same time phrasal in nature.

Giegerich's discussion of the stratal overlap between strata 1 and 2 is not always clear or convincing. An example of stratal overlap is the use of the noun-forming suffix -er in words like fiddler and bottler. According to Giegerich, fiddler is bisyllabic, but bottler (presumably) trisyllabic with a syllabic /l/ (but phonetic forms of these words are not given, so the reader is kept uncertain about the data). This difference would follow from fiddler being formed on stratum 1, where syllabicity of the /l/ is lost, and bottler on stratum 2, where the syllabicity of the /l/ of the base verb is preserved. Giegerich argues that this corresponds with a difference in degree of semantic idiosyncrasy of the resulting nouns, but these differences remain unclear (p. 85).

Chapter 5 returns to the main issue of this book, the distinction between compounds and phrases, by broaching the issue of Lexical Integrity. One concern often related to Lexically Integrity is whether phrases can function as building blocks of words. First Giegerich seems to accept the (empirically wrong) idea that only lexicalized phrases can appear within compounds. Later on, he modifies this view by claiming that ‘English compounds can breach lexical integrity in the sense of phrase-inside-compound if they are nonce formations, provided that information about the precise nature of the semantic association of the head with the embedded phrase is given by the preceding context’ (p. 115). Giegerich's example to illustrate this restriction is the compound wooden door policy to refer to a (fictitious) ‘policy not to fit wooden doors in premises licensed for the storage of inflammable materials’ (p. 114). It is not clear to me whether this contextual requirement is a real restriction, since the use of new complex words usually requires an appropriate context, and hence this restriction may be vacuous after all. Anyway, Giegerich does acknowledge that phrases can be used as building blocks of compounds. He therefore concludes that modules can overlap, at least in English (p. 120). Giegerich proposes that word combinations of the type dental decay are ‘generated in an “area” of the modularized derivation which is simultaneously part both of (stratum 2 of) the lexicon and of the syntax . . .’ (p. 122). Ascriptive adjectival attribution on the other hand only takes place in syntax, and hence ascriptive adjectives will be ordered before associative adjectives in noun phrases, as in terrible dental decay.

My problem with this conclusion is that it does not provide us with a clear and principled view of the relation between syntax and lexicon, but only says that there is some overlap. I have formulated my own view of this issue in a monograph on Construction Morphology (Booij Reference Booij2010), and refer the reader to this study. As far as Adj+N combinations of the associative type are concerned, we could think of treating them as tight syntactic combinations, in which an Adjective is Chomsky-adjoined to a head Noun, resulting in a syntactic compound. The crucial property of associative adjectives is that they are non-projecting, i.e. they do not project a phrase of their own. Arguments for this analysis are given in Booij (Reference Booij2009).

In sum, Giegerich's monograph raises interesting questions, gives a careful description of semantic patterns of interpretation, and the complicated stress facts concerning English compounds, but keeps us still in the dark as to how precisely we have to conceive of the relation between syntax and lexicon. It is also a pity that in discussing this issue, Giegerich makes use of English only, whereas his conclusions pertain to the theory of the architecture of grammar in general. A comparison with related Germanic languages such as German could have made the analysis stronger.