Back to Top of Document
This article is based on research completed for my Ph.D thesis (as yet unfinished). The thesis is a study of the change of Element Order in early English and the causes of such changes. Early English is a valuable source of evidence for the ways in which language change can occur, for (as is well known) it underwent a change from a highly inflected language with relatively free element order to one in which grammatical function is expressed to a great degree by element order - which is comparitively fixed. It is hoped that by studying data from the period of greatest flux in the language - late OE to early ME - that some light will be thrown on the details of and reasons for such language drift. This period saw several important changes, both in Element Order and in syntax generally as well as in other features such as lexis and pronunciation, and many of these changes are related.
The article is organised as follows. First there is a background to Element Order study, next the text is discussed as to its suitability for this research and then there is a discussion of the computer analysis and tagging of the text. Finally there is an extract from the research, discussing differences within the text, and whether these point to genuine diachronic differences, or are merely examples of stylistic variation. At the end of the article is a brief description of terminology used.
The following is a very simplified account of this topic, ignoring some controversial matters.
This description of language typology is based on the well-known statistical study of Greenberg 1 which showed that languages with particular element order type patterns at sentence/clause level, also tended to have similar features in other areas of syntax . This important study has been used as the basis of theories of language development which use the most common (or as described in some theories the underlying or deep-structure) element order as a method of linguistic classification. That is, languages can be classified according to related features at various levels of their structure, and element order (being at the top level, and being able to connect other related features) is a natural candidate to be used as the key component in this classification. For instance, Subject-Object-Verb (SOV) element order usually is accompanied by a fairly complex morphology and can offer - as a result of this morphology which marks the main elements of clauses - a very free choice in the ordering of elements. Latin is a very obvious example of this kind of language.
| SOV | SVO |
| Inflected/Free Element Order | Uninflected/Fixed Element Order |
| Head-Modifier | Modifier-Head |
| Verb-aux | aux-Verb |
| Head-prep | prep-Head |
It is a mistake to think that languages can be wholly described by means of their element order, as there is never complete agreement in every level of syntax between languages supposedly of the same element order type. However, the notion of element order as a way of identifying languages is still very useful. Languages change their element order type over time, and this is very often accompanied by changes in their other syntactic features. For instance a change in a language from SOV to SVO order is often accompanied by a loss of morphology — e.g. case endings; just as Latin lost many of its case endings when it developed into modern Spanish and Italian. Since element order is thus linked to other syntactic features, it becomes a useful method of comparing different languages and different stages of development within the same language.
Venneman argued 2 that certain features of language types came about as a result of a mixture of influences such as ‘weight’ (i.e. number of syllables in word/words in a phrase), ‘given/new’ information in elements, and thematic ordering: for instance, the fact that the vast majority of languages have S before O in common, unemphatic clauses is due mainly to S being the most common thematic component 3. Also, once ‘type’ changes — or rather once a language has moved a good distance from being clearly one element order type towards becoming another — certain sub-structures become awkward for a language (e.g. a language generally does not easily accept too many elements between V and O, so a change from OV ‘type’ to VO will also involve a strong tendency for V-aux order to move to aux-V and MH to HM ). English itself — like the other Germanic languages — once had mainly MH order in its earliest known form, and V-aux order for verb phrases (e.g. ‘taken had’ rather than ‘had taken’). Such influences never bring change about quickly, however, and languages very often show ‘mixed’ types. For example Present Day English (PDE), where we have 2 possessive modifier forms:
One shows MH order, the other HM; other prepositional phrases also act as post-modifiers in PDE, in parallel with the still-existing but simpler pre-modifier:
In fact there are few languages, if any which could be described as ‘pure’ examples of their ‘types’. Although languages may begin such changes due to other, unrelated, reasons such as sound-changes (which can weaken morphology), once underway, influences such as those mentioned above extend the changes to match the features of another type. I believe it is more useful to think of language ‘types’ as surface outcomes of the effect that various influences (which affect structure) have in particular languages. Thus, ‘type’ is no more than a shorthand way of showing that some languages have structures in common, but the similarities may be close or relatively loose, in part or all of the structures that make up the type: it is only one form of description. Other relationships could give us different typologies based on other things apart from element order. This particular typology (of element order) is of particular use in the study of English as English is an excellent example of the kind of language which contained many of the features of one language type, under this kind of description, and developed into one containing many of the features of another - i.e. SOV to SVO.
Venneman theorises language change as a somewhat circular phenomenon. For example, SOV languages can lose their morphology due to sound changes and develop into SVO types; similarly, SVO languages can develop into SOV, after developing a new morphology, for instance by re-analysing pronouns, prepositions, etc. as morphological affixes. Of course, languages may change back - since (for example) an SOV language may develop a new morphology before developing fully into an SVO language and thus may reverse its development, becoming once more a fully-fledged SOV language. The term ‘drift’ is probably best used to describe these kinds of changes, since it implies the important idea that language change will not necessarily always be in one direction. Other factors are also important, such as influence of neighbouring languages, and there are many problems to be resolved in this area.
This study is based on a set of homilies compiled in the 12thC. The first, forming the bulk of the material and the core of the study, is a collection known as the Lambeth Homilies (LH) (Lambeth Palace 487). The LH are of a mixed nature. First, they were compiled towards the end of the 12thC, but contain material of earlier origin, some going back to Ælfric. This means we have a selection of material from a period when English was changing from Old English (OE) to early Middle English (eME), an important period for study because many OE features - such as a much fuller morphology, and a freer element order (but with specific features still noticable, such as inversion of S-V order in Independent Clauses and SOV order frequent in Dependent Clauses) - were apparent while the language was beginning to show these developments which would lead eventually to Present Day English (PDE). Although English had shown signs of weakening morphology for a some time before this, it is in this period we are able to see the language finally developing from one linguistic type into a new one. As homilies (sermons) were meant to be delivered to an illeterate audience, the language used is less likely to show extremes of stylistic usage than many other forms of written material of the period. However we cannot altogether rule out the possiblity of a traditional oral style which may have been familiar to the sermon’s audience.
Sisam (1951) has shown that the homilies were copied c.1200 from two earlier 12th Century texts which were themselves copied as collections of religious material (with the exception of two verse pieces all prose sermons). She calls these texts ‘X’ and ‘Y’, which relate to divisions of the material in the homilies which she names ‘A’ and ‘B’.
It is fairly clear from her arguments, based mainly on orthography, that, overall, the ‘A’ section represents an earlier production than the ‘B’, and she believes some of it may represent the language at a stage many decades earlier than the ‘B’ section. I should emphasise here that I am mainly concerned with element order evidence: there is of course in the ‘B’ section a surface coating of later language - e.g. in the use of Norman-French lexis. However, element order will change at a slower rate since it is a syntactic element, part of the basic structure of a language, and will not change as easily as lexis, which changes easily and continually. Changes could include replacing unfamiliar lexis with familiar (e.g. Lage - ‘law’ - for ae), unfamiliar morphology with familiar (perhaps even expanding for clarity by means of prepositions, diexis) . It would surely not have been changed to any great degree unless the intention was to produce what was virtually a new sermon, with a completely different structure. In fact I tested this idea by analysing the element order of several pages of one of the two sermons which are directly from Ælfric. I matched this with an earlier (Old English) version and found that although some lexis had been replaced and some phrases expanded 4, the actual element order was almost the same.
Sisam suggests that ‘A’ represents a text less developed from OE than ‘B’: this is due -apart from its being an earlier production- to its containing traceable insertions from OE, including 2 complete sermons from Ælfric which are very close to the originals. What Sisam suggests is that - excluding Ælfric’s sermons IX and X (which I did not use) - LH could be treated as two sections - one representing early 12thc language (with possibly even earlier text either inserted directly or adapted), the other the language of the end of the century. This suggested the possibility of using the data extracted to do an internal comparison between these sections as well as an external one, against data from other studies. If the differences suggested were genuine, then the LH would provide useful information about the change in the language through the 12th century. If not, then the reasons for the apparent differences could provide valuable insight into the problems of such a study, as could a comparison with other texts of the period.
Computer analysis of text first requires that the text be reproduced in machine readable form. The scanner available at Glasgow University’s Computer Services Department when I began my research required good quality text: the only copy of LH available was that edited by Morris in the 19th Century, and the quality of the print, although perfectly adequate for reading, contained too many flaws to be used by the scanner. So my first task, after deciding which homilies were to be analysed (namely, all but VI, IX and X), was to type this onto file, proof-read it and correct it. Only once a good computer-file copy was obtained could the real work begin.
This leads one to mention the greatly increased facility with which such a study could be performed now, only 4 years later. Throughout the world corpora have been and are being produced in machine-readable form (disk and CD) and one can envisage a situation in the future when all suitable corpora will be instantly available for analysis of varying kinds.
Today we are in the fortunate position of being able to use one of several existing programs which have been developed for the analysis of text. My own choice was that of the OCP (Oxford Concordancing Program) which is widely used. This was available at the STELLA teaching labs, as were other possibly useful programs such as TACT, another concordancing program. I was able to obtain access to both computer and program facilities here, as well as very welcome help and advice from the staff when it was needed.
OCP was chosen because it allowed users to insert their own ‘tagging’ into a text, so that it can be ‘marked up’ in a manner best suited for the particular analysis to be made. These tags are necessary as the analysis programs available cannot do grammatical analysis of raw text (their main use is for lexical indexing and concordancing), particularly with early English texts. Each ‘tag’ contains some of the grammatical information needed in abbreviated form. It also has a very basic set of commands which can be organised into miniature ‘programs’ - allowing users to adapt the program to suit the task.
These commands are basically to allow the user to select tagged areas of text (which can be chapters, pages, paragraphs, etc.) to be examined. Other commands allow one to specify words or phrases or word ‘frames’ (such as ‘m*ss’ which can pick out examples such as ‘miss, mass, moss’ from a text.) This could also be used to pick out certain grammatical forms, for instance ‘*ed’ to pick out past tenses. It is also possible to collocate - that is pick out particular words that occur in close proximity to each other. Combining these commands allows the user to select text where a specified set of conditions prevail.
Textual tagging. (NB ‘P’ replaces ‘V’ in the Element Order tagging, to distinguish from V as simple verb tag.)
The system of course had the advantages that one expects from a computer-program system. It was quick — that is, quick at analysing the text and producing output (either on screen, file or as ‘hard copy’ print-out). Like all computer systems, OCP is always very accurate: but this accuracy depends very much on the logic of the analytical system behind the tagging used on the text and the care taken in ensuring both the accuracy of the tagging and the text itself.
I found it was best to use the PICK command to extract element order types rather than use the standard tagging system directly. The clause types would continue to be selected with the standard tagging system, but the element order types would be inserted as normal text with the addition of an extra, rarely used character to distinguish it from the actual homily text (see diagram). Thus,
would display all examples of SVO order including examples such as SVOA, ASVO, etc. Also, with the standard tagging system being used only to select the particular clause-types to be checked for element order, the tasks that this system had to handle was much reduced.
would display all SVO order in Independent Clauses.
Note that in the discussion that follows, the Lambeth Homilies (LH) were divided into groups A and B, for reasons explained above.
The data did not give the hoped-for indications of clear diachronic difference within the text (which would have been due partly to incorporation/adaption of earlier text or the result of a continuing tradition of language style in such works originating in the OE period, or perhaps both). However, there were enough differences to suggest some inclusion of material from earlier texts, or due to a ‘traditional’ style which was weakening over time.
Firstly, there is 3% to 4% more SOV order in ‘A’ than in ‘B’ Dep. Cls., which taken with the ICl data (+5% in A) might suggest a more ‘archaic’ style for A. However, more importantly, there is completely contradictory evidence in the CjCl columns, which show a very strong (+13%) slant towards SOV order in B. The number of clauses this column’s data is based on is, admittedly, low enough (32) to feel that this difference is being artificially exaggerated, however it still seems suggestive.
To try to discover what was going on here, I collocated element order and Object type in selections (c. 40%) of text from A and B. This was not a true random sample, but it was not possible to use OCP’s random sampling feature and retain data for individual (whole) clauses. The data obtained was organised in various tables to aid comparisons: an example of one such table follows.
Table showing ratios of pronoun (Pn) object to noun (N) object in the selection (ignoring small number of NCls).
| A | B | |
| ICls | Pn/N = 11/54 | Pn/N = 3/42 |
| CjCls | Pn/N = 9/15 | Pn/N = 4/17 |
| SCls | Pn/N = 26/40 | Pn/N = 8/27 |
| Total | Pn/N = 46/109 | Pn/N = 15/86 |
It was noticeable that A had 46/109 = 42% Pn (vs N), and B had 15/86 = 17% Pn. These figures exclude Relative Clause data which is unsuitable because of the position of relative pronouns is often due solely to the chance of the semantics of a particular sentence. The percentages just quoted suggest that A will have more SOV order than B since in early English SOV was especially common when O = Pn and the over-all totals seem to verify this - except in the case of CjCls. On the other hand the B pattern is important since the CjCls have the highest percentage of SOV order in the Homilies. In early OE, the highest by far percentage of SOV order is in SCls, a lower percentage in CjCls and the lowest in ICls. The figure here is probably somewhat exaggerated due to the low figures it is based on (9/32 clauses) but it is unlikely it is completely accidental, especially when one considers the fact that B has proportionately more O = N than A in general, and in SOV order clauses in particular.
Now it has been noted in many studies of the development of English that although SOV order survived into late ME, it became more and more commonly seen where O = Pn and less where O = N. The figures above seem to be evidence enough to consider that B (with its much greater use of full Ns as O), and not A has been most influenced by OE, and here may be an explanation of the contradictory CjCl data for B. This is because in early OE CjCls seemed to have been regarded as forms of SCls, with more SOV order; later when CjCls began to be used more as in modern English, that is, as semi-independent clauses, they began to show more features in common with ICls.
This impression is strengthened by the data related to other element orders. Again this is inconsistent, and is partly a result of the variations between other elements, which is difficult to even out. However if we consider it along with its O = Pn/N data some points are clear.
SVO order has the following associated O-type data.
| A | B | |
| ICl | 2/36 Pn/N | 1/20 Pn/N |
| CjCl | 4/11 Pn/N | 1/14 Pn/N |
| SCl | 8/37 Pn/N | 1/19 Pn/N |
| Total | 14/84 Pn/N | 3/53 Pn/N |
Many argue that the development of SVO order in OE was strongly influenced by weight, and Kohonen’s data 6 shows Sawles Warde having far more Pn O in SVO than the Catholic Homilies(CH). This seems to be strong evidence that the language began to use SVO order as a grammatical signal during this period rather than as one of various ways of ordering due to influence of theme, weight, etc. The above (LH) data shows that A has more Pn O in SVO than B: another piece of evidence that B is, if not ‘older’, then the work of a scribe not bothering to adapt his OE Homily style to the same extent as the A scribe. Data from Shores’ analysis of the Peterborough Chronicle, shows a figure of 27/77 for Pn/N (all clauses, including RCls) - i.e. a ratio of 35% Pn which is closer to A than B. Since this is a text of which we have a pretty clear (mid-12thC) dating, this tends to confirm the conclusion.
Further confirmation might be found by looking at auxiliary-verb (aux-V) data, that is at verb phrases which consist of more than one element (e.g. ‘was going, had arrived’). The main things to look for are V-aux order, which is a strong sign of archaism, and clause-final aux-V order which is less archaic. Signs of more recent development would be extremely low (or non-existent) levels of V-aux order, and aux-V which was not clause-final. Kohonen’s study showed that in the Catholic Homilies (CH) there was still a great deal of use of V-aux order, although aux-V order was clearly the majority usage, even then. Now in his general figures of element order, Vices & Virtues (V&V) has the appearance of a much older text than one would expect from its date, but if we look at the aux-V data we find it fits in more with what are believed to be more contemporary (ME) texts. CH had 15% V-aux order (mainly in SCls) whereas V&V had 4%, similar to SW’s 3%.
LH shows A = 5% and B = 4% and so, in this feature, there is little difference between the two textual divisions; and, importantly, there is little difference between both parts of LH in this feature and the 2 other eME texts examined by Kohonen. In general the evidence suggests that there was a difference between the two text sections, but that, contrary to initial expectations -and contrary to the expectations resulting from Sisam’s arguments 7- it is the B section which is the older, or is at least a copy of an older text. The differences are however not so great to suggest a wide difference and it may only represent either language change over a relatively short period (perhaps less than a century) or perhaps individual differences between the scribes of the original texts from which the sections were copied. This may have been to do with the strength of the OE scribal tradition in some areas, local dialect (which may have found some ‘archaic’ features more acceptable than others) and even individual scribes’ own feelings about the language in the (OE) texts they were using as a basis for later work.
My own opinion regarding the dating of the LH sections is that B represents a collection of older material that had previously been transmitted (by various copies) over a long period in a fairly intact form, close to the original. Either some more contemporary lexis (particularly Norman French) gradually crept in, or was added in a final, late 12th Century version , while retaining the basic syntactic structure. This, combined with the later orthography, led to Sisam’s conclusion that it was the later text linguistically. In the A section however, there is more of a mixture of material - whereas B is more consistent. We find, in A, as well as some genuinely OE material (which shows signs of expansion of the the elements - a sign that the case system was no longer adequate to express the original ideas 8), material which shows distinctly different styles, and also some use of more contemporary language. When the known OE material is excluded from the analysis, we see what the ‘surface’ hides: that the remainder of the A section is based on a text or texts which are structurally (i.e. in syntax/element order) more recent than the B section.
Some worthwhile points about this kind of study have been highlighted by this article. The first is that in examining the diachronic development of a language it is not enough just to note that one particular feature changes. In the above example, differences in element order alone - despite some contradictory evidence - suggested the A section to be earlier. But an examination of other features showed that this could be explained, and gave strong evidence that it was the B section which was in fact the earlier. This shows that it is necessary to take into account a variety of evidence and determine whether or not it all (or mostly) agrees with a particular theory. It is especially important in the field of element order study where at some stages of linguistic development the word order seen may vary greatly for reasons of style, tradition and other reasons. Two contemporary OE texts might vary quite extensively in their element order, yet be very close in their time of composition. Only by examining as wide a range of other evidence (and using known certainties about the development as a guide) can we be sure that any particular piece of evidence is a sign of real linguistic change and not merely a unusual but contemporary stylistic variation.
Typology. This refers to various ways of describing and classifying languages. Languages could be described by their morphology. The usual division in such an analysis would be into isolating (no morphology), agglutinating and fusional 9. With agglutinating languages the boundary between word and the morpheme is clear so that if there is (e.g.) a case for dative and a signifier for plural, the dative plural will be word + dative + plural. With a fusional language, the dative (or any other case) may show a different plural from other plural cases or a different form of dative in the singular than in the plural. That is, with fusional languages the boundaries between morphemes are not always clear. There are clearly many differences between languages which could be classified under these headings. English and Russian are both fusional languages by this classification, but the differences between them are great: English morphology is weak, whereas Russian is highly inflected; English has fairly fixed word-order while Russian is freer, although basically SOV in element order to English SVO. Greenberg’s work opened the door for the inclusion of other features besides morphology into language description. Also, it allowed a fuller description of the way in which languages change - since the relationship between element order and various morphological and other features became obvious.
Word/Phrase description.
The abbreviated forms used here are quite commonplace, and conform closely to those found in most language textbooks.
Noun(N); Verb(V); Pronoun(Pn); Adverb(Av); auxiliary verb(aux); preposition(prep); Phrases + P (e.g. Noun Phrase = NP, etc and Prepositional Phrase [Adverbial] = AP).
Adposition: a term when referring to both prepositions and postpositions. A postposition is the same as a preposition, except that it follows the NP rather than preceding it.
When discussing diachronic change some other terms come up fairly often.
Case: Description of morphology which indicates fetaures of grammar often expressed by word order in PDE: e.g. Nominitive and Accusative case = Subject and Object.
Clitic: A form which often has low stress and becomes seen to be attached to another stressed element in a clause, for instance the French indirect object as in
Dative: Term describing case (see above),now expressed in PDE by either word order (‘ I gave the man the book’) or by use of prepositions such as ‘to, from’.
Diexis: Literally, ‘pointing’. Use of words which refer to other words (or ideas) of fuller meaning. e.g. ‘that, they, him’.
Diachronic: Across time. Diachronic development refers to the changes which a language undergoes over a period of time.
Drift: Languages change over time. Sometimes a language may seem to develop towards another structure and then its old structure will re-assert itself. To describe constant linguistic development, which is not always uni-directional, the term ‘drift’ is used.
Morpheme: a meaningful part of a word. This may be semantic meaning or grammatical meaning or both. A word may consist of a single morpheme, although sometimes there may be more than one semantic morpheme and some morphemes serve a mainly grammatical function.
Morphology: use of morpheme/s, at the beginning or end of a word for specific grammatical purposes - such as to indicate case, number, tense, etc. PDE has a very limited morphology, whereas OE had a much more complete morphology.
Element order descriptions.
Commonly used abbreviations are:
These can be combined according to how much detail is necessary to any particular analysis or argument.
The most basic divisions of language by this analysis are into OV and VO - that is into languages that show basic element order with Object before Verb or Verb before Object. More commonly, the position of other elements, particularly S, will be important and the most common divisions here are SVO, SOV, VSO - which account for a large majority of the world’s languages.
Here is shown the importance of one’s choice of description of type, since both SVO and VSO are both VO languages yet are distinct types in a more detailed description. Another description is by S, V and any other element, represented by X. This allows other problems to be discussed, such as the distinction of ‘true’ SVO from Verb-second[or V-2] languages. True SVO languages are known also as Verb-third or V-3, that is another element such as O or A may appear initially without affecting the basic element order, thus SVO languages permit SVX, XSV(X) orders whereas in V-2 languages, such fronted elements cause the subject to be shifted to a post-V position: i.e., XVS(X) and XauxS(X)V as in German or Dutch.
At phrase level, as opposed to clause level, the main element is described as H (for head) and other elements are:
in NPs (Noun Phrases): determiners - such as ‘the’ - by det; modifiers such as adjectives by M.
in VPs (Verb Phrases): the main verb is V, auxiliary verbs are aux.
in APs (Adverbial Phrases - in which for convenience I combine adverbs and prepositional phrases: the H is either an adverb or an NP in a prepositional phrase.
Bean, M.C Development of Word Order Patterns in OE. Crown Helm, 1983.
Comrie, B. Language universals and linguistic typology. Blackwell, 1981.
Denison, D. ‘On word order in OE’. in Dutch Quarterly Review, 16:1986.
Funke, O. ‘Some remarks on lOE word-order....’ English Studies, 37:1956.
Kohonen, V. and Enkvist, N.F. (eds) Reports on Text linguistics: approaches to word order. Abo Akademi, 1976.
Kohonen, V. On the development of English word order in religious prose c.1000-1200.... Abo Akademi, 1978.
Li, C.N. (ed.) Word order and word order change. Texas U.P. 1975.
Marchand, H. ‘Syntactical change from inflection to word order system...’ in Anglia 70:1951.
Parker, F. ‘Typology and word order change’. in Linguistics 18:1980.
Resziewicz, A. Order of elements in lOE prose in terms of size and structural complexity.
Rybarkiewicz, W. ‘Word order in OE prose and the functional perspective’. in Studia Anglica Ponsnanienskia 9:1977.
Smith, N.V. ‘Consistency, markedness and language change.’ in Journal of Linguistics 17:1981.
Venneman, T. ‘Topic, Subject and word order: from SXV to SVX via TVX’. in Anderson and Jones(eds)
Allen, C.L. ‘Dummy subjects and the Verb-second target in OE’. in English Studies 67:1986
Anderson. and Jones(eds) Historical Linguistics: 2 vols. Amsterdam, 1974.
Brown, W. Descriptive Syntax of King Alfred’s Pastoral Care. Hague, 1970.
Faarlund, J.T. Syntactic Change: towards a theory of historical syntax. Mouton de gruyten, 1990.
Shores, D. ‘Morpho-syntactic relations’ in English Studies 52:1971.
Morris, R. (ed.) Old English Homilies. 1868. EETS.
Sisam, C. ‘The scribal tradition of the Lambeth Homilies’ RES: 1951; pp 105-14.
1 - Results of this can be seen in: Greenberg, J. Universals of Language. 2nd ed. 1966. MIT.
Back to Text
2 - Venneman, T. "Topic, Subject and word order: from SXV to SVX via TVX". in Anderson and Jones,(1974), Vol.1.
Back to Text
3 - i.e. the theme introduces the main topic of a clause or sentence:and thus is often identified with the subject. But non-subjects can introduce clauses, especially for emphasis, e.g. ÔHim I wouldn't trustÕ. The first element here can be described as theme to distinguish it from subject which has grammatical function.
Back to Text
4 - Expansion would be of the type ( ex. not from the LH): ÔscipumÕ - pl. dative of ship - to 'fram Ôaem scipum'Ô- from the ships.
Back to Text
5 - This was to do with ÔweightÕ. As OE became less of a SOV language, heavy elements like long noun phrases occurred more often after the verb. Light elements, like pronouns could still occur before the verb due to the fact of their light wieght and longer retention of some morphological distinction, and continued to occur until late in the ME period.
Back to Text
6 - Kohonen, V., On the development of English word order in religious prose c.1000-1200. Abo Akademi, 1978 which compares 1 OE text (Aelfric's Catholic Homilies) and two ME texts: Sawles Warde and Vices and Virtues.
Back to Text
7 - I do not argue here that Sisam was mistaken: it is very likely that the A section is older in composition; however both sections are, as was said copies of earlier texts, and this suggests that the B section, though more recent, is a copy of older material which due to adaptations of lexis and orthography appears less archaic on the surface.
Back to Text
8 - The scribe may have been familiar with OE from reading old manuscripts, but the audience for the Homilies - the 12thC church-goers - would not be, so this adaption of older sermons would take place. The use of such older material was a commonplace of the medieval period when such material was regarded much more highly than anything contemporary.
Back to Text
9 - For a fuller explanation, see Comrie, B, Language Universals and Linguistic Typology: pp 30 on.
Back to Text