Clause structure in Old English.

Clause structure in Old English. By Masayuki Ohkado, 2005. Pp. viii, 285.

The volume under review is a collection of seven papers devoted to a number of issues concerning the structure of Old English (henceforth OE) clauses, with special emphasis on word order. The papers were written by Masayuki Ohkado from 1994 to 2004 and all but one have already been published in linguistic journals or collections of essays. Each paper constitutes a separate chapter and may be read without reference to the rest of the volume. Alternatively, it may be treated as part of a larger whole united by its common subject matter. The thought underlying the volume is to present some well-known empirical facts of OE syntax from a novel perspective, i.e. through the prism of formal syntactic theories. With this intention in mind, the author addresses the selected syntactic issues and analyses them by employing a theoretical apparatus worked out by the generative scholarship. The papers incorporate the basic tenets of Chomsky's Theory of Government and Binding as well as some major contributions of the GB-based research to the study of early English. By departing from the traditional (i.e. purely descriptive) approach the author hopes that his work will throw new light on old data and give a better insight into the selected aspects of OE clause structure.

The first paper, entitled "On the underlying structure of Old English with special reference to Verb (Projection) Raising", sets out to defend the claim that OE clauses do not have a single underlying structure. The claim draws on the seminal Double Base Hypothesis (Pintzuk 1991) and takes OE clauses to be head-final or head-initial at the VP level.

Ohkado's argumentation in favour of the double base in OE is based on the analysis of constructions with three verbal elements: a modal (M), an auxiliary (A) and a lexical verb (L). Theoretically, the verbal elements involved may enter six different combinations, corresponding to six different surface orders of verbal strings: (I) MAL, (II) MLA, (III) LAM, (IV) LMA, (V) AML and (VI) ALM. Under the scenario that the underlying order is OV (Allen 1975; van Kemenade 1987; Koopman 1990; Lightfoot 1979), the patterns (I-VI) may be derived by means of operations such as Inversion, Reanalysis, Verb Raising and Verb Projection Raising. Ohkado points out that, despite its initial appeal, the analysis of OE as OV at the base does not stand up to scrutiny on empirical grounds. On the one hand, it is too permissive because it fails to block the derivation of clause patterns which are not attested in OE (e.g., pattern (VI)). On the other, it fails to derive certain patterns which do occur in OE. Moreover, Ohkado expresses reservations about the theoretical validity of Inversion and Reanalysis, both of which are regarded by him as ad hoe.

Without examining the validity of the VO scenario, Ohkado proceeds to demonstrate the superiority of the double base analysis, in which the relevant orders may be derived by employing two movement operations, i.e. Verb Raising and Extraposition. As for the non-existent pattern (VI), Ohkado claims that its derivation can be readily blocked by the OV/VO system. In fact, however, he does not devote much space to the precise mechanics by which the blocking is assured. The author concludes that the evidence afforded by OE constructions involving three verbal elements tips the scales in favour of the OV/VO base. Throughout the discussion Ohkado implicitly follows the assumption that the double base analysis holds only at the VP level. Thus, he does not consider the possibility that functional categories (or other lexical categories) could also be analysed in a double base fashion, nor that projections of the same category (whether functional or lexical) do not necessarily have to be uniformly branching at all levels. (1)

The second paper, "NEG 1 constructions in Old English", focuses on negative constructions of the type illustrated in (1) below (his la):

1) Ne forseah Crist his geongan cempan

'Christ did not despise his young champions'

One of the telltale signs of such constructions is that the initial position is occupied by the negating particle (ne), followed by the finite verb. The reason why such constructions are of interest is that they cannot be easily reconciled with the customary analysis of OE as a V2 language. The V2 analysis appears to run aground in view of the fact that the particle ne is usually treated as an Xo-clitic, which is attached to the left of the finite verb, as in (2) (his 13a):

2) [[sub.CP] [[sub.C'] [ne-forseah.sub.i] [[sub.IP] Crist his geongan cempan [t.sub.i]]]]

As it stands, the NEG 1 construction in (2) exemplifies a Verb First phenomenon, which calls for some explanation in the light of the general V2 character of OE. To maintain the V2 analysis, Ohkado makes an assumption that [Spec CP] in NEG 1 constructions is filled by an empty NEG operator. On the theoretical plane, the presence of an element in [Spec CP] may help explain the motivation for the V-to-C movement if it is assumed that V-to-C is triggered when [Spec CP] is filled by some material. (2) On the empirical plane, Ohkado observes that constructions with a genuine V1 order are restricted to interrogatives, sentences with extraposed subjects or sentences without nominative subjects. Taking into account the rarity of such contexts it seems that the analysis of fairly common NEG 1 constructions in the V 1 spirit is not likely to be correct.

In the third paper, "On nominative case assignment in Old English", the author discusses OE passive constructions with extraposition and attempts to provide a uniform account of them. The account is based on the postulate that in the constructions under discussion extraposed NP subjects are assigned nominative Case in their base position, i.e. as a complement of the verbal head. What this proposal entails is that since the NPs in question receive Case in their base position they are required, by the minimalist considerations of economy, to stay in situ. Such a state of affairs does not apply to similar constructions in Present Day English (henceforth PDE), which differ from their OE counterparts in a number of respects. Firstly, passive constructions with extraposition require the presence of a locative phrase in PDE but not in OE. Secondly, the extraposed subject may only be followed by non-arguments in PDE whereas in OE the appearance of arguments after the extraposed subject is not infelicitous. Thirdly, passive constructions with extraposed subjects are confined to main clauses in PDE whereas they occur in both main and subordinate clauses in OE. In view of these differences passive constructions with extraposition in PDE and OE cannot be considered identical, despite their superficial similarity. Given this, Ohkado suggests two different analyses for each of them. Thus, for PDE he proposes that prior to being extraposed the subject undergoes leftward movement to [Spec IP], where it receives nominative Case. In OE the subject receives nominative Case in its base generation site, i.e. case assignment is prior to extraposition. Significantly, the latter is not preceded by NP-movement to [Spec IP].

The fourth paper, "On MV/VM order in Old English", focuses on the so-called "modal constructions", consisting of a modal verb (M) and its infinitival complement (V). The author is concerned with the relative ordering between the two types of verbs, i.e. MV and VM. As the facts from OE demonstrate, the majority of main clauses exhibit the MV order whereas in subordinate clauses some vacillation between the two orders can be observed. It is specifically the question about what determines the relative ordering between M and V in subordinate clauses that is at the core of this paper. In search of an answer, Ohkado puts forward two working hypotheses, formulated in (3a-b) (his 13a-b):

3a) The longer the syllable length (3) of non-finite verbs, the more likely it becomes that they follow modal verbs.

3b) If non-finite verbs are heavier than modal verbs, the former tend to follow the latter. If non-finite verbs are lighter than modal verbs, the former tend to precede the latter.

The author shows that both claims are untenable. As an alternative to (3a) and (3b), a different solution is offered, i.e. that the relative order between M and V is conditioned by the presence or absence of some extra lexical material in a given VP. In the presence of such lexical material the order MV is preferred, whereas in the absence thereof the order VM prevails.

The fifth paper, "Subordinate clauses in Old English with special reference to paet clauses", draws attention to the issue of how to distinguish properly between main and subordinate clauses. Ohkado claims it is necessary to recognise two distinct clause types: genuine subordinate clauses, which are verb final, and embedded main clauses, which are verb second. The motivation for such a distinction comes from certain word order patterns observed in PDE clauses which are often regarded as subordinate and which appear in the so-called "assertive contexts" and in so that-clauses expressing result (both exhibit a typically main clause phenomenon, i.e. Subject Auxiliary Inversion (SAI)). As for their structural analysis, Ohkado adopts Authier's (1992) idea of CP recursion, where the embedded main clause corresponds to the lower CP. Assuming this for PDE, he subsequently aims to determine whether the same observations pertain to OE. The data from AElfric's Catholic Homilies demonstrate that there is indeed a high degree of similarity between embedded main clauses in OE and in PDE. Firstly, complement clauses of assertive predicates show a preference for SVX, which is indicative of their main clause character. By contrast, complement clauses of non-assertive predicates tend to exhibit SXV, which is suggestive of their subordinate character. Secondly, (swa) paet-clauses expressing result show a preference for SVX, which is also indicative of their main clause character. Conversely, (swa) paet-clauses expressing purpose usually exhibit SXV, which is in line with their subordinate character.

The sixth paper, "The position of subject pronouns and finite verbs in Old English", is concerned with the relative placement of pronominal subjects and finite verbs. More specifically, its focus is on the syntactic status of pronominal subjects, which have been widely recognised to behave differently from full NP subjects in terms of their distribution. In many classical analyses (e.g., van Kemenade 1987; Pintzuk 1991) pronominal subjects in OE are given the status of clitics. The analysis in terms of cliticisation, however, begs a further question as to how OE pronominal clitics should be represented structurally. Two specific explanations have been proposed: by van Kemenade (1987) and by Pintzuk (1991). Each of them makes different predictions concerning the possibility of separating the pronominal subject from the finite verb (in main clauses with inversion) and from the complementiser (in subordinate clauses). What Ohkado wishes to demonstrate is that the evidence afforded by his analysis of AElfric's Catholic Homilies (first series) supports van Kemenade's analysis. As for main clauses, the data are straightforward insofar as they include not a single instance where the sequence of the finite verb and the pronominal subject is discontinued. As for subordinate clauses, the situation is not equally straightforward because there is a handful of examples which involve a split between the complementiser and the pronominal subject. Although the existence of this pattern cannot at first sight be reconciled with what van Kemenade predicts, Ohkado argues that the clauses in which it appears are not unambiguously subordinate but may be treated as embedded main clauses. There is thus no reason to abandon the uniform treatment of pronominal subjects as head clitics adjoined to Co.

The seventh paper, "Coordinate clauses in AElfric's Catholic Homilies", is devoted to two types of coordinate clauses: coordinate main clauses and coordinate subordinate clauses. The distinction between the two types is significant insofar as the word order patterns observed in each of them are not identical. Rather, as Ohkado notices, they seem to exhibit some tendencies which are linked to the use of specific connectors, either coordinators (in coordinate main clauses) or subordinators (in coordinate subordinate clauses).

The author observes that in both PDE and in OE there is a potential difficulty in differentiating coordinate subordinate clauses without overt subordinators from coordinate main clauses because they are identical on the surface. In order to disambiguate their status he analyses the relevant clause types appearing in AElfric's Catholic Homilies (first series). The first stage of his examination leads to the conclusion that in main clauses the verb typically precedes its complementation and in subordinate clauses the verb is usually placed after its complement. In the next stage, 4 types of coordinate clauses are examined: (i) coordinate main clauses with a subject (MC), (ii) coordinate main clauses without a subject (MVC), (iii) coordinate subordinate clauses with a subject (SC), (iv) coordinate subordinate clauses without a subject (SVC). On the basis of the analysis of (i-iv) two major tendencies can be observed. The first tendency concerns MC and MVC, which show a higher frequency of verb-final order in comparison with ordinary main clauses. This is taken to confirm the validity of distinguishing coordinate main clauses from ordinary main clauses, as in Mitchell (1985). The second tendency has to do with the fact that within the whole group of coordinate clauses, i.e. (i)-(iv), those without subjects tend to be head-final whereas those with subjects are more often head-initial. Both tendencies are subsequently discussed at length.

The book under review is characterised by an exemplary organisation of its contents as well as by the clarity of argumentation. Each paper contains a clear statement of objectives, description of method, and conclusions. The claims made throughout the discussion are meticulously examined and a precise line of reasoning is accordingly spelled out. In these and other respects, the volume constitutes a solid contribution to the research on OE syntax in the framework of generative syntactic theory. The book can be recommended both to students and scholars.

In the introductory chapter the author regrets that too many linguistic studies confine themselves to presenting crude statistical data, supplemented by a handful of representative examples. Such a method does not provide a full array of data used for analysis, which makes it impossible for the reader to verify the conclusions reached by a scholar. In this light, the development of electronic (syntactically) tagged corpora is promising insofar as it creates a potential for making the data more readily accessible. However, Ohkado emphasises that it is not enough for corpus-based studies merely to refer the reader to a given corpus. Rather, he contends that it should become customary for corpus-based studies that they provide a step-by-step description of the methodological procedure used for a study or, better still, that they are accompanied by a machine-readable database (attached in electronic form or publicised on the Internet). To give an example of how this can be put into practice, the author makes available on the web the complete database which he employed in paper seven.

Unfortunately, the work is not free from errors and inaccuracies. Although some of them are enumerated in the errata sheet, their presence makes reading difficult. A random selection of errors is given below:

a) spelling errors, e.g., Gernerative instead of "Generative" (p. 167), constuctions instead of "constructions" (p. 225), convienience instead of "convenience" (p. 234), Ghrist for "Christ" in (69b) (p. 242), etc.

b) grammatical errors, e.g., the second analysis ... treat them as verb second (p. 52), the language of the text belong to Late Old English (p. 128), van Kemenade's new approach make the same prediction (p. 177), etc.

c) wrong numbering of examples, e.g., (33) instead of (32) (p. 150), (39) instead of (38), (40) instead of (39) (p. 152), etc.

d) wrong calculations, e.g., wrong percentages in Table (42c) (p. 125), wrong figures in Table (35) (p. 220), which are further repeated in (61) (p. 234) and in (84) (p. 248), etc.

Summing up, the collection of papers under review constitutes an interesting account of various word order phenomena observed in OE clauses. It is hoped that the book will find a broad readership and serve as a stimulus to more work on the syntax of OE, integrating the tools of formal syntactic theory and corpus methodology.

REFERENCES

Allen, Cynthia 1975 "Old English modals", in: Jane B Grimshaw (ed.), 89-100.

Authier, Jean-Marc 1992 "Iterated CPs and embedded topicalisation", Linguistic Inquiry 23: 329-336.

Fuss, Eric--Carola Trips 2002 "Variation and change in Old and Middle English: On the validity of the Double Base Hypothesis", The Journal of Comparative Germanic Linguistics 4:171-224.

Grimshaw, Jane B. (ed.) 1975 Papers in the history and structure of English. Amherst: University of Massachusetts.

Kemenade, Ans van 1987 Syntactic case and morphological case in the history of English. Dordrecht: Foris.

Koopman, Willem 1990 Word order in Old English, with special reference to the verb phrase. [Ph.D. dissertation, University of Amsterdam.]

Lightfoot, David W. 1979 Principles of diachronic syntax. Cambridge: Cambridge University Press.

Mitchell, Bruce 1985 Old English syntax. Oxford: Clarendon Press.

Pintzuk, Susan 1991 Phrase structures in competition: Variation and change in Old English word order. [Ph.D. dissertation, University of Pennsylvania.]

Reviewed by Agnieszka Pysz, Adam Mickiewicz University, Poznan

(1) For a proposal that functional categories are uniformly head-initial see Fuss--Trips (2002).

(2) Yet, Ohkado is well aware that the generative literature has offered a multitude of other explanations for V-to-C movement.

(3) The meaning in which the author employs the term "syllable length" refers to the number of syllables of which a given verb consists. In this sense, the length of syllables is apparently not affected by the length of the vowels which they contain.