Across the literature on historical linguistics, the English language can be noted for the dramatic changes that it has undergone over the past 1,300 years or so. The loss of case marking, shifts in basic word order and numerous phonetic changes are but a few examples. Changes in argument structure, i.e. how the participants expressed with verbs and other predicate words are morphosyntactically realised, are also abundantly attested. A case in point is the loss of impersonal constructions, as exemplified with the Middle English verb longen ‘long’ in (1) below. This is the chief focus of Castro-Chao's investigation.
(1) To Þe me longeð swuðe.
to you me longs greatly
‘I feel a great desire for you’
(Middle English, a1250, source: OED; cited on p. 11)
In Old English and Middle English, one of the arguments of many verbs (typically some kind of Experiencer) could be realised as a case-marked NP or objective pronoun, with no overt nominative subject in the clause, as in (1), instead of a (nominative) subject in grammatical agreement with the verb, as shown by example (2). While similar constructions subsist in many modern Germanic languages, such as German in example (4) below, English impersonal constructions gradually fell into disuse over the Late Middle English period (1400–1500), and the corresponding predicates in Present-day English (PDE) can now only realise their Experiencer argument as subject, as shown in example (3).
(2) Ich langy so swiÞe after Gorloys his wife.
I long so greatly after Gorloys his wife
‘I have such a great desire for Gorloy's wife’
(Middle English, c1300, source: MED; cited on p. 11)
(3) Dana longs for a sunny day. (Present-day English; cited on p. 11)
(4) Mir ist kalt. (Present-day German)
me.dat is cold
‘I'm cold’
This shift typifies one of the leitmotivs of the historical changes to English: its gradual shift from a synthetic, heavily case-marked language with relatively flexible word order, akin to some of the Present-day Germanic languages, to an isolating language with rigid word order, obligatory subjects and mere traces of case marking that are mostly limited to pronouns. It is thus not surprising that these impersonal constructions have received much attention in the literature. However, much of the extant scholarship focuses on Old English and Middle English, and mostly aims to investigate the factors that may have brought about the demise of impersonal constructions. The Early Modern English period (EModE, 1500–1700) has, however, been comparatively neglected by previous research, and although these constructions were already at best archaic by that time, the question remains whether and when they ceased to be recorded after the Middle English period, and what patterns have replaced them in EModE. Castro-Chao proposes to fill this gap in the literature by focusing on verbs of Desire in particular, such as long in the examples above. These questions are examined in a Construction Grammar, usage-based, cognitive-functional linguistics framework, which lends much importance to semantics and function as drivers of speakers’ choices and ultimately of language change. Methodologically, the study broadly uses quantitative corpus linguistics methods, and is based on data from the Oxford English Dictionary (OED), the Middle English Dictionary (MED) and, perhaps most importantly, a bespoke corpus of EModE texts drawn from the Early English Books Online (EEBO) database.
After a short introduction in chapter 1 outlining the aims of the investigation (pp. 11–15), chapter 2 provides the main background and literature review for the study (‘The functions and development of English impersonal constructions’, pp. 17–41). Much of the chapter is concerned with defining the terms and notions used in the study, in particular ‘impersonal verbs’ and ‘impersonal constructions’. There seems to be no consistency in the literature as to what these terms are meant to refer to, as they are given different meanings by various scholars. The author settles on a definition of impersonal verbs and impersonal constructions that combines syntactic and semantic criteria. Impersonal verbs are defined as verbs that occur in impersonal constructions at any point in time and take an Experiencer argument. Impersonal constructions are defined as patterns displaying the two syntactic features typically cited by work on the topic, i.e. the invariable third-person inflection of the verb and the lack of agreement with any other clause constituent, excluding constructions with an overt non-referential subject like (h)it. The author then proceeds to document earlier accounts of the history of the construction. In particular, Jespersen (Reference Jespersen1927) explains the demise of impersonal constructions and their replacement by nominative clause subject constructions as a case of reanalysis caused by case syncretism: the disappearance of case marking is argued to have led to structural ambiguity in the assignment of subject versus object grammatical roles, and as word order increasingly rigidified, the experiencer argument was recruited as subject and moved to pre-verbal position, notably due to its animacy. This account deserves to be repeated here as it is re-examined later on in the book using corpus data, and it is actually found to be inconsistent with empirical evidence (see below).
Chapter 3 describes the theoretical framework of the study, in particular with respect to the description of verb meanings and the representation of grammar (‘The nature of verb meaning and constructional meaning’, pp. 43–60). Particular properties of verb meanings are identified that are claimed to be especially relevant for the study. Dowty's (Reference Dowty1991) concept of proto-roles is introduced as a useful approach to explain the selection of subject and object arguments for a given verb in terms of a set of semantic features that characterise prototypical agents (or proto-agents) and prototypical patients (or proto-patients). Dowty's (Reference Dowty1991) account predicts that the argument with the greatest number of proto-agent properties will be selected as subject, and the one with the most proto-patient properties as object. Verbs of Desire represent an interesting case for Dowty's proto-roles approach, because the participants to these verbs do not quite fit the proto-agent and proto-patient categories, which explains the relevance of impersonal patterns for these verbs, and more generally the variation in their syntactic patterning both synchronically and diachronically. As for grammar, the study is couched in a Construction Grammar framework, more particularly as defined by Goldberg's (Reference Goldberg1995) work on argument structure constructions.
Chapter 4 (pp. 61–78) defines the focus and scope of the study by describing how the verbs to be investigated were selected. All verbs listed in Levin's (Reference Levin1993) class of verbs of Desire were searched for in the OED and the MED, and only those with attested impersonal uses (according to the definition mentioned above) in the relevant ‘desire’ sense in the entries’ citations were kept. This led to four verbs: hunger, long, lust and thirst. Of these, hunger was eliminated from further study, on the grounds that it is semantically quite similar to thirst and shares with it the same diachronic development (extension of the physical sensation sense) and similar complementation patterns. The rest of the chapter outlines a semantic characterisation of verbs of Desire. In particular, verbs of Desire and their two participant roles, Desirer and Desired, are analysed with respect to Dowty's proto-roles. It is shown that verbs of Desire differ from typically transitive verbs in terms of properties of the process and especially in that the two participants are not clearly differentiated in terms of proto-agent and proto-patient roles, in the same way that typical agents and patients would be. This explains why these verbs were used in impersonal constructions in Old English.
Chapter 5 (pp. 79–100) introduces the corpus used in the study and discusses how the data were collected and analysed. The study uses corpus data from EEBO, a massive collection of digitised historical English texts from the fifteenth to the seventeenth century. It is more specifically based on EEBOCorp 1.0 (Petré Reference Petré2013), a 525-million- word corpus based on the EEBO Phase I texts, pre-processed for use in linguistic research. EEBO has become an increasingly popular resource for EModE studies, as it is extremely comprehensive in its scope. The undesirable implication for corpus linguistics is that it is also unbalanced, both over time and in terms of the type of texts it contains. To address these issues, the author built a balanced subset of the corpus consisting of four 5-million-word subcorpora assembled from randomly selected texts (excluding verse) from four successive fifty-year periods: 1500–49, 1550–99, 1600–49 and 1650–1700. Instances of long, thirst and lust, in all spelling variants attested in the OED, were extracted from this corpus and manually filtered, resulting in 273 relevant instances of lust, 304 instances of thirst and 341 instances of long.
Chapters 6, 7 and 8 report the findings of the corpus study on the diachronic development of the complementation patterns of the verbs lust, long and thirst respectively. For lust (chapter 6, pp. 101–34), impersonal uses are found to be rather rare (8 tokens, 2.93 per cent), and are restricted to the beginning of the period. In all cases, the Desired arguments always falls into the NO PROP category, i.e. cases in which the Desired is not overtly realised, and the verb is used in a construction with as or when, e.g. It is not lawful for me to do as me lusteth. This suggests a very low degree of productivity of impersonal constructions for lust in EModE, as they are essentially frozen in relatively fixed expressions, the last remnant of a type of complementation that has almost completely vanished. With personal patterns, the only type of realisation of the Desired argument that maintains its frequency over time is the zero complement, i.e. cases in which the Desired is not overtly expressed but is recoverable from the context or receives a generic interpretation, which appears to be largely accounted for by the (semi-)fixed phrase the flesh lusts against/contrary to the spirit, typical of religious discourse. Clausal realisations of the Desired decrease in frequency, which is hypothesised to be due to the replacement of lust by more semantically general verbs like wish or please in their clausal uses, an explanation which dovetails with the semantic specialisation of lust in its narrower ‘carnal desire’ meaning. Interestingly, about 70 per cent of the remaining clausal complements are NO PROP constructions (e.g. as he lusteth), which parallels the remnant impersonal construction mentioned above, and might have helped the impersonal pattern to be maintained for a while. NP realisations of the Desired (viz. as object) were attested only for a brief time and were always very infrequent, which is explained by the fact that the Desired does not show any of the proto-patient properties, hence this construction was quickly dismissed by speakers in favour of prepositional realisations (in particular with after).
The results for the verb thirst (chapter 7, pp. 135–75) differ quite a bit from those for lust. The initial sense of thirst is the physical sensation (i.e. the need to drink). The second and more relevant sense for this study is that of desire, presumably derived from the physical sensation sense, although both senses have been attested for at least as long as the written English records can take us back. Unlike lust, thirst is not attested at all with impersonal patterns in the corpus of EModE. In personal patterns, the realisation of the Desired as a PP increases over time at the expense of NP, clausal or zero realisations. Traces of the source domain of thirst as a verb of Desire are still found in the data, for instance in its collocation with nouns referring to drinks. This metaphorical link leads the author to describe the prepositional pattern as a metaphorical extension of the intransitive motion construction, called the ‘move-attention’ construction, in which directing one's attention is construed as motion, following the meaning of the intransitive motion construction. Like for lust, NP complements are initially rare and quickly disappear, which is similarly explained by the fact that the Desired does not show any of the proto-patient properties.
Chapter 8 reports the findings for the verb long (pp. 177–206). Like for thirst, no impersonal uses are attested in the EModE corpus for long. In the personal patterns, the Desired can be realised as a prepositional complement, a clause, a zero complement or a directional adverb; patterns with NPs are unattested. Prepositional complements start the period of interest as the most frequent pattern and continue to gain in prominence over time at the expense of the other patterns, in contrast to the other two verbs, for which the prepositional pattern only becomes prominent later in time. In the early sixteenth-century section of the corpus, long is attested with the Desired realised as the adverb therefore, which is used as a directional adverbial in other contexts and is related to the locative preposition for. These uses are argued to point to a directional construal of desire for the verb long, similar to the case of thirst, and captured with the same metaphorical ‘move-attention’ construction. When the Desired is realised as a clause, subordinate clauses with till/until are attested, which are normally adjuncts but here are equivalent to a complement clause with that. This illustrates a functional shift from adverbial subordination to verb complementation. Finally, zero-complements with long are similar to the ones with lust, with which it is sometimes coordinated. In these uses, long with an unexpressed Desired argument is equivalent to ‘long for sin’ in many contexts.
Chapter 9 (pp. 207–23) summarises the main findings for the three verbs, and reflects on the implications of the study for the history of impersonal constructions more generally. In sum, impersonal uses are only attested with lust in the EModE corpus, albeit with low frequency, and they are only residual and seem to correspond to archaisms or idiomatic, frozen expressions. All three verbs are predominantly found in religious contexts, which are typically considered to be more conservative and to preserve archaisms. This confirms that the transition from impersonal to personal patterns for verbs of Desire was all but completed by the EModE period, in which the three verbs are attested with a range of personal patterns, yet with a distribution still largely in flux. At the same time, the fact that the verbs differ as to when their last impersonal uses are recorded shows that this transition was largely lexically specific, and that it was a gradual change affecting verbs individually, rather than a brutal shift impacting the entire lexicon at once.
With all three verbs, no clear support is found for Jespersen's (Reference Jespersen1927) reanalysis hypothesis mentioned above, as it is actually contradicted by empirical data. The Desirer argument is found to be predominantly realised as a pronoun in most cases, across all complementation patterns. This finding is not surprising in itself and can be largely explained by language-external factors. However, this makes Jespersen's (Reference Jespersen1927) hypothesis a rather unlikely explanation for the shift in these three verbs: since most uses do not contain two lexical NPs but at least one case-marked pronoun, there is no rampant ambiguity in case marking, hence it cannot be the reason for the reinterpretation of Desirers as subjects, and the demise of impersonal constructions. As an alternative, usage-based explanation, pronominal Desirers can be seen as a driving force for changes in the argument structure of these verbs, and especially their entrenchment as experiencer-subject verbs: following the principle of end-weight (longer constituents come last) and the idea that given information tends to be placed before new information, pronominal Desirers are more likely to occur at the beginning of a clause, and thus to be reanalysed as subjects, especially when pitted against the Desired argument.
As a whole, the study makes a valuable contribution to the documentation of historical English and to our understanding of one of the major changes in argument structure in the history of English. The book is well written and nicely presented overall, with the caveat that some figures are rather hard to read in the black-and-white hard copy. Also, maybe some chapters could have been made shorter, as it is not clear whether some passages are really needed, which might be due to the fact that the monograph is based on a revised PhD thesis; in particular, some of the earlier chapters include discussions of a number of theoretical concepts and analyses whose relevance to the study is not always immediately clear, and which are not always drawn on or returned to.
The book will be of interest to scholars working on the history of English and other Germanic languages, as well as to diachronic linguists at large, especially those working in a Construction Grammar and/or usage-based framework. The results are admittedly rather modest when it comes to the main topic, i.e. impersonal constructions: since only one verb was found to be attested with impersonal patterns, and only marginally so, most of the content discussed in the study is actually not concerned with impersonal constructions, making the subtitle of the book, ‘The development of impersonal constructions in Middle and Early Modern English […]’, a bit of a misnomer. That said, the analysis of the corpus data is very detailed and provides a wealth of information on other patterns of these verbs. Such a rich dataset would perhaps have deserved slightly more refined corpus methods, as the study is largely restricted to reporting raw frequency. However, these data are aptly analysed and cogently discussed. At the very least, the study establishes, more precisely than previous ones, the time of death of impersonal constructions for verbs of Desire. It also demonstrates the importance of documenting diachronic studies with abundant and representative data, using corpus-based methods. In that connection, the careful and insightful way in which the EEBO data were handled and turned into a balanced and representative corpus of EModE is to be praised, and should certainly be used as a point of reference for scholars interested in using this resource for their own research. In conclusion, this book is a nice and valuable addition to scholarship on impersonal constructions and on the history of English in general.
