Relative clauses: Structure and variation in everyday English is a sophisticated new monograph by Andrew Radford, which sheds fresh light on a much-studied aspect of morphosyntax, namely relative clauses. Radford's attempt to unravel the source and structure of non-standard relatives which ‘raise challenging descriptive, typological and theoretical questions about the nature of relative clauses’ (p. 5) represents an innovative theoretical advance in the study of relativisation in English. The aim of the book is to raise awareness of three sets of non-canonical relative structures found in colloquial English which challenge the traditional filler–gap analysis of canonical structures: resumptive relatives, prepositional relatives and gapless relatives. Though non-canonical in construction, their use is widespread in the spontaneous spoken English of the live, unscripted radio and TV broadcasts used by Radford in his analysis.
This is a pioneering book which, like the author's previous monograph, Colloquial English: Structure and variation (Reference Radford2018), forces us to reconsider the nature of spoken language in formal grammars. Radford's comprehensive theoretical analysis shows that syntax alone is not enough to account for occurrences extracted from spontaneous, colloquial language, and that these are better explained as the result of a complex interplay between syntactic, pragmatic, sociolinguistic and psycholinguistic factors. Language is an instrument of communication and should, therefore, be studied as such. The book examines a vast number of examples, including many which challenge previous theoretical assumptions. The author obligingly attempts to provide an explanation for all of them, using (or adapting) standard rules and operations where possible, and improvising new ones when traditional theoretical approaches fail. Despite the author's rich and extensive knowledge of the formal and non-formal literature on the subject, he himself admits that an accurate explanation of certain structures is not always possible without an ‘awful lot of (essentially ad hoc) spellout magic’ (p. 119) or ‘some unusual (and patently) ad hoc spellout rules’ (p. 116), solutions which at times feel overly laboured and only partly successful (as in the section entitled ‘A fronted preposition analysis’).
The monograph comprises a prologue, four chapters and an epilogue. Chapter 1 presents an overview of standard relative clauses in everyday English, while chapters 2, 3 and 4 focus on three types of non-standard relative clauses found in colloquial English. Chapter 2 looks at resumptive relatives, i.e. relative clauses whose relative pronoun is ‘reprised by a resumptive pronoun lower down in the clause’ (e.g. Supermarkets are now making a big thing about selling wonky vegetables, which years ago they would just have been discarded (p. 2)). Chapter 3 deals with non-standard prepositional relatives, in which the preposition is repeated, either identically (‘preposition doubling’, e.g. It's the world in which we live in (p. 2)), or with ‘mismatching prepositions’ (e.g. The freedom in which we played with (p. 3)). Finally, chapter 4 focuses on gapless relatives, i.e. relative clauses with no filler–gap dependency (e.g. The main target was to finish ahead of Ferrari, which we've extended our lead by 4 points (p. 3)). The epilogue offers a brief summary of the book's purpose and findings, and concludes that non-standard relative clauses have a complex syntactic structure of their own, which may be syntactic or pragmatic in nature, or the result of hypercorrection or performance/processing errors. The book also provides a glossary of rules, systems and approaches, which will be especially useful for students and linguists less familiar with this area.
In the prologue the author describes how the data were collected, and justifies the exclusion of ‘structures produced by non-native speakers’ (p. 4). Here it would have been interesting to know the author's definition of a native speaker in today's multilingual world, since this has been a matter of some debate over the past decade (see Mesthrie Reference Mesthrie2010). A definition of ‘canonical’ would also be useful, particularly in view of some non-canonical examples where the non-standard structure is the only possible option to avoid violating certain grammatical constraints, as in the case of ‘structures in which a resumptive is used to relativise an inaccessible subject’ (p. 67) (e.g. She has a gash on her leg, which they're not sure if it's going to heal over (p. 68)).
Chapter 1 provides the background to relative clauses, and presents an excellent review of formal research into the clause periphery, from early works by Chomsky (government and binding, barriers, minimalism) to Rizzi's ‘map of the left periphery’ to account for recursion as Specifiers of CP (Reference Rizzi and Shlonsky2015). Radford proposes a new projection, RELP, as a way to achieve a unitary characterisation ‘under which all relative clauses contain a RELP projection’ (p. 31), in order to facilitate the analysis of relative clauses. He admits, however, that it is difficult to establish a uniform analysis to account for all of the examples and that, in some cases, it is necessary to resort to a ‘purely ad hoc descriptive artifice with no explanatory power’ (p. 44). Although the idea of a new projection that would be generally applicable to all relative clauses is a very appealing one, the difficulty acknowledged by Radford exposes the fundamental staticness of the model and its inability to account for existing and emergent non-canonical examples, as shown in the chapters that follow.
In chapter 1, Radford provides a description of the main types of relative clauses found in standard registers and varieties of English (traditional ‘filler–gap’ analysis). He challenges the traditional classification of relative clauses into restrictive and appositive by identifying a third category, labelled ‘kind relatives’ (following Prince Reference Prince, Guy, Baugh, Schiffrin and Feagin1995), which represents a ‘sub-type of restrictive relatives, but which shares some properties with appositives, and has unique syntactic and semantic characteristics of its own’ (p. 10). The traditional classification of relative clauses into restrictive and non-restrictive is problematic and it is not always easy to decide whether a relative clause is restrictive or not (see Denison & Hundt Reference Denison and Hundt2013 for a summary and an additional proposal for a four-fold classification of relative clauses). The ‘kind relatives’ category proposed by Radford represents one of these ‘problematic’ relative clause types, because although they represent a sub-type of restrictive relatives, they share certain properties with appositives and, more importantly, have syntactic and semantic characteristics of their own (p. 10). Even the examples Radford uses to illustrate this third category do not always support his claims: in the sentence The top speed, which please don't try to reach, is 220 miles an hour (p. 12), for example, it is unclear how the author's ‘kind relative’ differs from a traditional appositive relative clause.
Chapter 2 deals with ‘resumptive relatives’, the first type of non-canonical relative clauses analysed in the monograph. Resumptive relatives are defined as relative clauses whose antecedent is ‘reprised’ by a resumptive element (pronoun or noun phrase). Although considered ungrammatical, resumptive relatives are frequent in English (Radford reports 444 examples in his corpus (p. 83)).
The author demonstrates that these examples are the product of a merge (rather than move) derivation, in which the antecedent, relativiser and resumptive are independently generated in separate positions (p. 122). These constructions have been present in the English language throughout its history (Fischer Reference Fischer and Blake1992; Traugott Reference Traugott and Hogg1992; Rissanen Reference Rissanen and Lass1999), and are commonly said ‘to relativise constituents in inaccessible positions’ (p. 55) in order to satisfy the inaccessibility hypothesis. The examples provided show that in many cases this hypothesis holds true, especially in contexts where syntactic constraints rule out the use of gaps (e.g. frozen subject cases, contact relatives, examples of constraint on extraction domain, locality violations, cases of relativisation of objects of unstrandable particles, and examples of direct speech quotations) (pp. 67–79). The issue here is that all of these examples represent different kinds of structures and it is not clear whether ‘cases of contact relatives’ (e.g. He's one of those players Ø he likes to feel bat on ball) (p. 71) or ‘direct speech quotations’ (e.g. They've got a transfer committee that people say: ‘Well, why should you have a transfer committee?’) (p. 73) should be classified as resumptive relatives. In the case of contact clauses, the fact that we are dealing with spontaneous conversation may justify the repetition of the pronoun he; while in the case of direct speech quotations, the clause hosting the resumptive is an independent clause with an autonomous speech domain and an intonation boundary marked by punctuation marks, which would justify the repetition of an element from the preceding clause. Most of the examples of resumptive relatives (57.4%), however, contain resumptives in gap contexts (i.e. contexts where gaps are admissible) (p. 75), thus undermining the inaccessibility hypothesis.
Another hypothesis that is tested is the length effect, since it is frequently said that the longer the string of constituents between the relativiser and the gap, the higher the chances of inserting a resumptive element to facilitate processing (‘processing constraints’). Radford's analysis of ‘Intervening String Length’ (ISL) shows no clear evidence of any length effect on the use of resumptive pronouns, with ISL values of zero found in the most represented category (18.6%). This analysis (table 42) is based on all the examples classified as ‘resumptive relatives’. However, it would be interesting to compare ISL in examples of optional resumptives as well; that is, in examples in which gaps are admissible. The inclusion of examples where syntactic constraints rule out the use of gaps (e.g. frozen subject constraint) is another issue with this analysis, as their presence may distort the findings. For instance, there are 78 examples (table 34, p. 77) of frozen subject constraint (the most frequent type of resumptives) and 78 examples of 0 ISL (table 42, p. 81). Despite these minor issues, the chapter presents a very interesting challenge to traditional views that resumptives are only used in inaccessible positions and when the relativiser and the gap are separated by a large number of constituents.
Chapter 3 looks at prepositional relatives. Standard prepositional relatives occur in two different ways: pied-piping and preposition stranding. In Radford's analysis, special emphasis is placed on two forms of non-canonical prepositional relatives: ‘preposition doubling’, where the clause contains a pied-piped preposition and a copy of the same preposition (p. 133); and ‘preposition intrusion’, a less frequent phenomenon found in clauses which ‘contain a fronted relative pronoun preceded by a (seemingly spurious) preposition which the relative pronoun appears not to be the complement of’ (p. 134). Cases of double prepositions are challenging from a structural point of view. As acknowledged by the author, the main problem is that ‘spelling out an in situ copy of the preposition appears to be inconsistent with the general assumption made about spellout that only the highest copy of a moved constituent is overtly spelled out, and that lower copies obligatorily receive a null spellout unless they require an overt spellout for independent reasons’ (p. 138).
Based on the examples presented, preposition doubling appears to occur more frequently with short, commonly used prepositions such as to, in, of, on and with. Missing from Radford's approach, however, is a comparison of canonical and non-canonical examples in order to observe the overall frequency and behaviour of the phenomenon in relation to the overall frequency of relative clauses, including relatives with ISL of 7 or more. Since copy prepositions are often attributed to memory lapse, this alternative analysis might also help to shed light on what motivates copy prepositions in short relative clauses (as counterexamples to the long-distance tendency of copy prepositions). A length effect has already been demonstrated for double prepositions in earlier English (Yáñez-Bouza Reference Yáñez-Bouza2007: 52–4), but in Present-day English (PDE), ISL shows ‘no clear evidence of a length effect’ (pp. 179, 180).
The second type of non-canonical prepositional relative is ‘preposition intrusion’, resulting from the insertion of an extra preposition in the clause or a mismatch between the two prepositions used. Though theoretically questionable as a category of prepositional relative, since the intrusive preposition serves no apparent syntactic function, Radford's analysis of the unconventional relativiser in which (also spelled as inwhich) is very interesting (as is the proposed reanalysis of where and whereby from relative adverbial pronouns to complementisers in chapter 2). Radford's survey of different examples of in which/inwhich, together with his review of the literature on the subject, makes a convincing case for the treatment of the form ‘as a single word relativiser, and not a PP which comes about through a preposition insertion operation in the syntax’ (p. 162). His reanalysis of the combination as a single word relativiser is also supported by its occurrence in written discourse, including examples collected from student assignments, and could have been further reinforced with examples from GloWbE and COCA.
One hypothesis to account for the low frequency of non-canonical prepositional relatives (Radford does not report their exact frequency with respect to canonical ones) is that they are the result of hypercorrection (p. 168), which leads to the insertion of a fronted preposition in an attempt to satisfy a prescriptive rule and emulate prestige speech styles. This hypothesis holds if the examples presented are taken from spontaneous speech; however, its validity is weakened by the availability of examples in written texts, which are usually subjected to processes of output monitoring and editing.
The last category of non-canonical relative clauses in Radford's survey are ‘gapless relatives’. Unlike canonical relative clauses, gapless relatives do not seem to show any filler–gap dependency (p. 191). Chapter 4 deals mainly with the more commonly occurring gapless prepositional relatives, and pays less attention to ‘non-prepositional gapless relatives’, of which 113 examples are reported. The author makes little attempt to offer a clear explanation of these examples (e.g. ghosting or assumption of an additional string of words works without violating different conditions) and the reader is left no wiser as to how to analyse them. On the subject of ‘prepositional gapless relatives’, Radford presents a review of the literature, in which experts differ over whether to analyse the preposition as stranded or fronted. Irrespective of the analysis chosen, a silent spellout of a stranded/fronted preposition must be assumed. The interpretation of the preposition as stranded seems more accurate since, as Radford explains, it violates fewer grammatical constraints and the only alternative analysis is to adopt different ad hoc operations and ‘an extremely powerful (and in certain respects ad hoc) descriptive apparatus which could be argued to be incompatible with the goal of developing maximally simple grammars’ (p. 204). The stranded analysis does not, however, account for relative clauses in which only pied-piping is possible. One question that remains to be answered is why speakers favour these structures over their grammatical alternatives (e.g. preposition stranded or pied-piped structures). For that, however, a variationist approach which quantifies and analyses the specific contexts (intra and extra-linguistic) of all the variants present would be required. Since both of the silent preposition spellout analyses have flaws, a third analysis is mentioned, whereby ‘the relation between the relative pronoun and the rest of the relative clause is determined by pragmatic inferencing’ (p. 216). Pragmatics is mentioned here as a potential factor to account for these non-canonical structures. Pragmatics could also have been used to account for other non-canonical clauses dealt with in the volume, since they belong to spontaneous speech and are part of conversations.
Kortmann & Lunkenheimer's Electronic World Atlas of Varieties of English (Reference Kortmann and Lunkenheimer2013) is referenced for the first time in chapter 4 in order to illustrate the wide use of ‘preposition chopping’ (feature 198, according to Kortmann & Lunkenheimer's classification) in World Englishes. Radford's underuse of this interactive database of morphosyntactic variation in spontaneous spoken English is surprising, considering it would also have been a useful source of evidence for his analysis of other non-canonical structures, such as ‘resumptive/shadow pronouns’ (feature 194) and ‘linking relative clauses’ (feature 197), present in numerous varieties of English and both featured in the monograph.
Overall, Radford's latest work is an excellent and significant addition to the study of relativisation in English. The aims and research questions established at the beginning are met and answered amply by the rich and wonderfully comprehensive collection of data, analysis and theoretical apparatus, imbued throughout by the author's deep-rooted interest in the origin and status of non-canonical structures. The few shortcomings identified in this review (e.g. creation of ad hoc rules to account for some examples, analysis of ISL, questionable examples) are only minor criticism when compared with the remarkable advance in research into non-canonical grammar in general and relativisation in particular which Radford's study represents.
Usage examples from spontaneous and colloquial speech offer an insightful perspective on formal approaches to grammar. This monograph clearly demonstrates how usage can inform grammar, showing that the choices made by speakers are not dictated by an invariant grammar, but by modelling patterns of usage in which pragmatics, discourse, prescriptivism and psychological factors interact with one another. It is an excellent example of Bybee's (Reference Bybee2005) maxim that ‘grammar is usage and usage is grammar’.
