Objects and adverbials and the loss of OV order in English
One of the most interesting phenomena in the documented history of the English language is the systematic change that transformed it from an inecting V2 language with OV as its basic word order into the strictly SVO, isolating language of the present. The relevant syntactic and morphological changes (i.e., the changes in word order and the weakening of the in ection system) are clearly not independent of each other, though the problem of their exact relationship raises numerous questions. The focus of the present paper is the change from OV to VO order, which took place simultaneously | to some extent | with the transition from the Old English to the Middle English period. Still, it did not happen overnight, in fact, it seems to have been a gradual change that took several hundred years before it was completed. This means that for quite a long period, a signi cant number of OV and VO sentences occurred simultaneously, until at some point in the Middle English period, OV became restricted to a few, narrowly de ned linguistic contexts. Linguists have proposed various theories to explain what happened. These theories can be divided into three groups. The rst group argues for an underlying OV order in Old English, which allowed certain movements resulting in surface VO order, until VO clauses outnumbered OV ones, so a reanalysis took place into underlying VO. The second group of theories assumes two independent phrase structures di ering in base word order to coexist for a while until the VO one triumphed over the OV one. The third type of theory says Old English had underlying VO grammar with a range of elective movements that were able to produce both the OV and the VO clauses with the relevant restrictions in those cases when only one of them can be attested in textual evidence. Then, at some point in language history, the crucial movements became blocked, i.e., they were no longer possible and the underlying VO order had to surface. Pintzuk & Kroch (1989) argue in favour of an underlying OV order based on the analysis of Beowulf, where the rightward movements of objects and adjuncts are only possible as heavy object movement and extraposition, respectively. The two extremes of object structure | consistently postverbal sentential objects on the one hand and pronominal objects on the T H E E V E N Y E A R B O O K 5 (2002) 49–56
ISSN 1218–8808
49
50
Judit G or asz
other, which kept their pre-verbal position the longest|support this view. Medium heavy NPs are more problematic, at least in sources other than Beowulf. The authors emphasise that Beowulf re ects the most archaic state of the language documented in writing. This certainly explains the di erence between the linguistic situations found here and in other sources, where non-heavy objects occur post-verbally. What remains to be explained is how non-heavy objects started to move rightward|it is dicult to imagine that object clauses and really heavy NPs outnumbered other objects to such an extent that a change in the underlying word order became inevitable. The opposite view | underlying VO structure in Old English | is expressed by Fischer et al. (2000). Here, the authors give the explanation that underlyingly post-verbal objects had to undergo one of several elective movements for the purpose of case-checking and|depending on their landing site | they surfaced either in a pre- or a post-verbal position. Later, the movement taking the object left of the verb was blocked (supposedly because a pre-requisite verb movement was blocked rst) and surface VO order became obligatory. What exactly blocked the movements is left open in the theory except for a tentative suggestion regarding verbal morphology, which is considered to be problematic in several ways by the authors themselves. The question what may have triggered the change|whichever way it happened | is the question this paper will try to explore. But before this happens, some attention must be paid to the facts presented by the sources. A common feature in the theories described above is that all of them assume the existence of alternative options over the period when OV and VO clauses cooccurred in large quantities. Individual speakers had to choose between movement and non-movement or between movements or between grammars. This, of course, only re ects what we can nd in the texts, in which OV and VO orders are apparently in free variation. The Anglo-Saxon Chronicle is a valuable source for several reasons. First of all, it is a prose text, not bound by metrical considerations, secondly, it was written originally in English and last but not least, it is the product of several centuries’ scribal work, so it is ideal for diachronic study. The chronicle started at the end of the ninth century, when the annals were written in retrospective, beginning with the Romans’ rst appearance in Britain. After the end of the ninth century, subsequent events were recorded more or less contemporaneously until after the Norman conquest. That means the chronicle extends over about 200 years. The annals were continued in the twelfth century in the Peterborough Chronicle, which is an early Middle English document. With regard to word order, the text suggests that the
Objects and adverbials and the loss of OV order in English
51
two types | OV and VO | coexisted already when the Chronicle began, but later the number of OV clauses steadily declined and the type became quite rare in the eleventh century annals. At the same time, of course, VO clauses gradually gained ground. (1) Þs ymb iiii niht þered cyning 7 lfred his broþur þr micle erd to Readingum gelddon. `Four nights later King þered and lfred, his brother, led there a great army to Reading.’ (2) Her Romane gesomnodon al þa goldhord þe on Bretene wron. `Here the Romans gathered all the treasure of gold that was in Britain.’ This change, as it has been mentioned above, is sometimes linked to the loss (or reduction) of a rich verbal morphology. But this explanation does not seem to be a satisfactory one for several reasons: i. The same loss of verbal morphology (i.e., the \weakening“ of agreement) is used to explain the loss of the verb-second constraint. (See Roberts 1993.) The two changes are divided by a considerable time gap, so it is not very probable that they can be traced back to a common reason. ii. As the Anglo-Saxon Chronicle shows, the beginnings of the OV to VO change can be observed already in the Old English period, when verbal morphology was supposedly rich. The Peterborough Chronicle displays a verbal morphology as rich as that of the Anglo-Saxon Chronicle, at the same time, OV clauses are extremely rare in it. (On the other hand, V2 ourishes in the Peterborough Chronicle, which is not surprising considering (i).) iii. External evidence indicates that there need not be a direct link between a strict SVO order and poor verbal morphology. There are languages like Finnish and Estonian, where a rich verbal morphology (with no syncretism) coincides with SVO order. (Finno-Ugric languages are supposed to be originally SOV, so an SOV to SVO change can be assumed to have taken place in these languages as well.) Here, however, the necessity of SVO can be linked to nominal morphology, more precisely case-endings. Finnish, for example, has an extremely rich nominal in ection system (comparable to Hungarian) but in place of the usual nominative vs. accusative distinction, there are three cases associated with subject and object position: nominative, accusative
52
Judit G or asz
and partitive. All three endings can signal (di erent kinds of) objects; nominative and partitive may equally stand in subject position, while the singular accusative ending in full NPs is identical with the genitive ending, which again may mark NPs in subject position in certain constructions. In the plural the nominative and accusative endings are identical. Under these circumstances, the separation of subject and object NPs by the verb seems to be desirable, and a strict SVO order achieves just that. The situation is even more complicated in Estonian, where nominal case markers have undergone phonological reduction and the accusative/genitive ending has disappeared, which has led to a signi cant amount of syncretism in the crucial positions. This may suggest that Old English OV may have turned into VO for similar reasons, due to problems with overt case marking. Old English distinguished a nominative, an accusative, a genitive and a dative case but in many paradigms the nominative and accusative forms were identical. An example is shown below: singular
plural
nom.
nom.
sunu acc. sunu gen. suna dat. suna
suna suna gen. suna dat. sunum acc.
In spite of the syncretism, many subject and object NPs could still be distinguished by means of demonstratives and adjectives but clearly there must have been a signi cant number of instances where situational knowledge and/or phonological stress were needed to disambiguate the utterance. This is most probable in a linguistic situation displayed by the Chronicle, in which word order is considerably free. A little later, Early Middle English texts already show a signi cant simpli cation in nominal in ections, including the forms of articles and demonstratives. This situation must have favoured xed positions for NPs until nally SVO came to be regarded as the basic order. As a xed order, SVO is more favourable than SOV, partly because of the clear separation of nominal elements, partly because it does not lead to ambiguity when objects are topicalized: (i) NPS + NPO + V ! object fronting ! NPO + NPS + V (ii) NPS + V + NPO ! object fronting ! NPO + NPS + V In the absence of overt case-marking, object fronting in (i) may be obscured, while in (ii) the clause remains unambiguous.
Objects and adverbials and the loss of OV order in English
53
It is also worth noting that pronouns, which often do not follow the same rules as full NPs (they tend to be pre-verbal when full NPs are often post-verbal, they behave as clitics and remain preverbal even in V2 clauses), display a greater morphological richness than nouns, and they retain it over the Middle English period. The process then can be summarised as follows: syncretism in the nominal in ection encouraged the choice of SVO clauses and this led to a situation in which the identi cation of subject and object NPs depended less and less on overt case marking but more and more on the positioning of constituents. This situation may have played a role in the loss of overt case endings in general, which again reinforced and stabilised SVO order. This explanation can be linked to either of the above-described theories. If we accept that Old English objects were underlyingly pre-verbal, we may suggest that the existence of some post-verbal constituents (e.g., heavy objects) provided a model for solving the syncretism problem by moving other objects right of the verb. On the other hand, the case-checking theory has itself connected the problem to NPs and the accusative case. The present explanation permits to leave out the assumed blocking of verb movement as the reason for object movement blocking and regard the weakening of overt case marking as a sucient reason to block the leftward movement of verbs. A tentative question may be how much the loss of overt case marking contributed (by analogy) to the subsequent loss of verbal in ection, which in turn led to the loss of the verb-second constraint. The word order of Old English is often said to resemble that of German. As a result of the changes being discussed here, the language became similar to Scandinavian. The question of possible Scandinavian in uence naturally arises here. Of course, more study is needed to give a de nite answer. But it is a possibility which deserves some consideration. In any case, in theory at least, such an in uence may have taken place either by directly changing the word order or by contributing to the loss of nominal in ection as a result of direct language contact, which then could lead to word order changes in the above-described manner. That would suggest the creolization of Old English and explain the weakening of case endings to a critical extent, when it inevitably led to major syntactic changes. While the arguments presented so far can explain why objects lost the possibility of a pre-verbal position, they do not say much about adverbials, which have undergone a similar change. Old English adverbials appear to have been quite free in their positioning. It must be noted, however, that many of them had a \loose” semantic connection to the rest of the clause, in the sense that they were not \essential“ or absolutely necessary to make the
54
Judit G or asz
clause stand | grammatically or otherwise. The examination of adverbials in clearly OV clauses in the Anglo-Saxon Chronicle reveals that adverbials often take a post-verbal position there. What the data suggests is that this post-verbal position is closely linked to \weight” or structural complexity in the sense that only prepositional phrases and clausal adverbials are positioned on this side of the verb. While clausal adverbials are rare in the corpus (and in any case, they constitute a special class of adverbials which do not like to mix with the phrasal constituents of the main clause), prepositional phrases can often be found in a pre-verbal position as well. The di erence between pre- and post-verbal prepositional phrases seems to depend on the obligatory or optional nature of the adverbial: if the verb phrase is grammatical and semantically complete without the prepositional phrase, then the prepositional phrase is truly peripheral and occurs post-verbally, while an obligatory prepositional phrase normally precedes the verb. (On peripheral and non-peripheral adverbials cf. Matthews 1981 : 121{141.) (3) 7 Cantware him feoh geheton wiþ þam fri þe `and the people of Kent promised them money in return for peace’ (4) 7 hiene him to biscepsuna nam `and adopted him as his godson’ (5) 7 he hine to cyninge gehalgode `and he consecrated him king’ (6) hiene þa Cynewulf on Andred adrfde `then Cynewulf drove him to (the forest of) Andred’ The PP is optional in (3), but it is obligatory or in (4), (5) and (6). Looking at this distribution from a broader aspect, it can be said that pre-verbal prepositional phrases share the feature [+obligatory] with other pre-verbal constituents, i.e., subjects and objects, while post-verbal prepositional phrases are [ obligatory]. The other adverbial constructions | unmodi ed adverbs as well as noun phrases and pronouns in the dative case | invariably occur pre-verbally. It can be assumed that these phrases are structurally light enough to stay in pre-verbal position. Why are grammatically optional heavy constituents delayed until the end of the clause? This arrangement may help to balance the information expressed in the clause and to avoid the unreasonable delay of important information by
Objects and adverbials and the loss of OV order in English
55
long and complicated phrases transmitting perhaps supplementary information | after all the number of optional adverbials in a clause is limited only by practical considerations. Structurally light phrases are not likely to disturb the ow of information, so they can keep their pre-verbal position. Post-verbal position, on the other hand, seems to be at the same time a post-sentential position in verb-late clauses and that may explain why complements, which must keep their close connection with the verb, are not free to move there | yet |, regardless of their weight. With the loss of OV, post-verbal ceased to be identical with post-sentential and most adverbials moved into this position as well. In any case, the question of weight does seem to have some relevance in the history of word order in English. We have seen that certain heavy constituents (both objects and adverbials) were post-verbal at a very early stage of English, while pronouns, the lightest possible material, were the last to give up their pre-verbal position. Medium weight objects, however, were exibly treated during the transition period: the optional nature of the movement can be observed with regard to them in the rst place. Objects of phrasal verbs in Present-day English are in a situation very similar to this: pronominal objects must precede the particle, heavy objects must follow it, while simple NPs are optionally positioned before or after the particle. (7) Put it on. (8) Put on the coat that you like best. (9) Put your coat on. (10) Put on your coat. While the presence of a considerable amount of post-verbal material may not have been enough in itself to trigger a major change in the word order of English, it may have encouraged the choice in favour of this position where such an option existed. This, on the other hand, may have made the shift from OV to VO easier, while syncretism may have represented the necessity which made the shift inevitable, as discussed above. It may be suggested that the role of weight in the history of English deserves some deeper research, since it may turn out to be an important force or perhaps a constraint behind various linguistic phenomena. The principle that heavy elements tend towards the end of the clause and light elements
56
Judit G or asz
towards the beginning explains the use of discontinuous NPs frequent in Old English. (For instance, a part of a heavy subject is left behind to keep up the weight asymmetry between the beginning and the end of the clause.) To sum up, the paper attempts to show a connection between the loss of OV and the syncretism found in the nominal in ection of Old English, especially with regard to the nominative vs. accusative distinction. The issue seems to have been coloured by the features [ heavy] and [obligatory]. references
Fischer, Olga, Ans van Kemenade, Willem Koopman and Wim van der Wurff. 2000. The Syntax of Early English. Cambridge: Cambridge University Press. Matthews, Peter H. 1981. Syntax. Cambridge: Cambridge University Press. Pintzuk, Susan and Anthony Kroch. 1989. The rightward movement of complements and adjuncts in the Old English of Beowulf. Language Variation and Change 1 : 115–143. Roberts, Ian. 1993. Verbs and Diachronic Syntax: A Comparative History of English and French. Dordrecht, Boston & London: Kluwer.
