UDK 802.0-022-25-54-559.1
THE SYNTAX OF THE OLD ENGLISH PREVERBAL GE-
IN THE LIGHT OF THE THEORY OF LANGUAGE CHANGES AS STRENGTHENINGS OR WEAKENINGS
Frančiška Trobevšek-Drobnak
The following paper reports on the results of 'the investigation into some aspects of the usage of the Old English preverbal ge- from the point of view of the theory of language changes as strengthenings or weakenings. The investigation served as the basis for the author's Ph. D. dissertation (1990).
1. The meaning and the function of the Old English preverbal ge-: some current doctrines and what they leave to be desired
1.1.	The Old English preverbal ge- is one of the most discussed morphemes in the history of the English language. During the last century some thirty-five dissertations, monographs and articles purporting to explain its meaning and function have been published. Some of them admit openly that »... ueber ae. ge- herrscht noch immer weitgehende Unklarheit« (Pilch 1953), while others appear to be philosophical hypotheses rather than strict linguistic observations, although at times they »... reveal a very seductive cogency, but only in the light of data that have been especially selected to substantiate them« (Lindemann 1970, p. 1).
1.2.	The Old English preverbal ge- has traditionally been related to the Gothic g a-, which in turn is explained as corresponding to the Latin verbal prefix cum-jcom- (from IE. *kom-). Streitberg (1891) insisted that the original meaning of the Gothic ga- (,mit, zusammen') had faded out and that its function was to »perfectivate« the verb. The Old English ge- was described along the same lines as
a)	either completely devoid of any lexical or grammatical meaning: »... Ge- apud Saxones semper fere superfluum ...« (Benson 1701),
b)	stressing/intensifying the action of the verb (Bernhardt 1870),
c)	converting an intransitive verb into a resultative verb that is transitive (Lenz 1886, Lorz 1908),
d)	indicating completion, pluperfect or the future perfect (Grimm 1878, Lenz 1886, Mosse 1938, Samuels 1949),
e)	expressing perfective aspect (Martens 1863).
1.3.	The doctrine sub e) is similar to the doctrine cited sub d), except that it is thrust into a verbal system analogous to that in Slavic languages. Its advocates believed that the aspectual situation in Old English was like
13
in Slavic dialects and that the prefixation of the Old English verbs should be considered as the encodement of aspectual information. From that point of view ge-verbs would always denote the perfective aspect and simplex verbs would always denote the imperfective aspect. The alleged dichotomy was sometimes pushed to the extreme. When Friedrich Weick (1911), a disciple of Streitberg's, examined the Lindisfarne Gospels and discovered that the forms of the verbs were different from what he believed they ought to be (i. e. ge-verbs were found to express the imperfective aspect and simplex verbs were found to express the perfective aspect), he suggested they simply be changed: »...Das Simplex sollte stehen... Das Kompos. sollte stehen«. He apparently believed more in his doctrine than in the glossator's mastery of his own language.
1.4.	In his extensive study of the meaning and function of the Old English preverbal ge- (1970), J. W. Richard Lindemann re-examined the existing doctrines and accused their authors of having carefully selected data to substantiate them, of »romantic preoccupation with Slavic dialects«, and of »transcendental linguistics«. According to Lindemann, some scholars substituted pure speculation for the laborious and objective examination of the prefix per se (Lindemann 1970, pp. 1, 19).
Lindemann cites an impressive number of instances from the Old English Gospels and from Orosius that speak against the so-called »syntactic function« of the preverbal ge- (i. e. against its perfectivating role). He agrees with Streitberg's critics in that -the grammatical category of verbal aspect in Slavic languages differs from the aspectual situation in Germanic languages. While the former define aspect (Aspekt) as a grammatical (mandatory, syntactic) category, Germanic languages operate with a lexical category »manner of action« (Aktionsart)1. On the other hand, Lindemann had also become weary of excessive »... dwelling upon (...) Aktionsart, for an Aktionsart is merely a secondary characteristic, a by-product of the action expressed by a verb« (Lindemann 1970, p. 21). He suggests that the real origin of ge- is to be found in the Indo-European demonstrative pronominal stem *gho-, a highly deictic morpheme, stressing the progression of an action from one point towards another, not unlike the German morphemes von, zu, zu-, ver-.
Afraid of being accused of »transcendental linguistics« himself, Lindemann examined 45,000 simplex and compound verbs and concluded that the lexical meaning of the Old English ge- was roughly equivalent to the modern English morphemes to, out, at, on, toward, forth and away. Lindemann referred to this meaning of ge- as its abstract meaning and its semantic nucleus. In Lindemann's study the prefix ge- seems to »do something« to the information conveyed by the verb, to add direction, concretion, and — even Aktionsart. The nature of this »additional information« seems to depend on the context and the meaning of the verb.
1.5.	The ample — albeit selected — evidence substantiating each and all of the above doctrines indicates that the Old English preverbal ge- had no single lexical meaning nor grammatical function. Furthermore, there seems to be no consensus whatsoever among the various scribes (speakers?) as to when exaotly the prefix ge- was called for. Even in different manuscripts
1 For more information on the distinctions between the aspectual situations in Slavic and Germanic languages cf. C. R. Goedsche, »Aspect versus Aktionsart«, JEGP, xxxix, 1940, pp. 189—97.
14
of texts like the translations of the Holy Gospels, which were probably treated with little licencia poetica, we find ge-verbs alternating with simplex forms or with verbs prefixed with some other particles translating ¡the same verbal phrase in the Latin original, e. g.:
Readings in:	CORPUS MS	HATTON MS	LINDISFARNE MS	RUSHWORTH MS
Mark 3.10:	ge-haelde	haelde	gehaelde	gihaelde
Mark 5.12:	gegan	gan	ingeonga/gae	ingonge/ingae
Mark 5.29:	gefredde	fredde	gefoelde	gifoelde
Mark 11. 2:	gelaedad	ge-laedect	to-laedes	to-gi-laedas
John 3.18:	gelyfd	lyfd	gelefes	gilefes
1.6. If, on the one hand, Lindemann offers a putative explanation of the inconsistency of the use of the prefix ge- in Old English (— the writer/ speaker was free to add it to the simplex form thereby enriching its lexical meaning with some vague sense of direction), and if, on the other hand, there seems to be little to add to the information about the origin of the ge-prefix and to the ultimate effects of its disappearance from the English language (Van Draat 1902), not enough notice has ever been taken of the diachronic aspect of its role and function in Old English. While the gradual withdrawal of the ge-prefix (and of other prefixes for that matter) from the English language was described by Mosse (1938) as the collapse of the former Germanic aspectual system, which was eventually replaced by expanded verbal forms in Modern English, the process itself was never given any particular attention.
In the study described in the present paper the focus of attention was placed on the loss of the ge-prefix as a linguistic change. The point of departure was as follows:
a)	(At least) throughout the late Old English period, the preverbal ge- was gradually withdrawing from the (written) language; after 1200 it existed virtually only in past participles.
b)	(At least) in the late Old English period the preverbal ge- was weakened to the point that its usage was not mandatory any more and that verbs with the prefix ge- functioned as syntactic variants of corresponding simplex verbs (and possibly also of verbs with other prefixes)2,:
c)	The actual usage of either syntactic variant can be empirically evaluated in terms of its (grammatical) environment. The statistical significance of the differences of the values of individual parameters can be computed.
d)	If statistical evidence for ge-verbs shows significantly different values of individual parameters than for simplex verbs, these differences can be assessed in the light of the theory of language changes as strengthenings and weakenings.
2. The theory of language changes as strengthenings or weakenings
2.1. The theory of language changes as strengthenings and weakenings was first introduced in the framework of the natural phonology (Stampe 1979,
2 Other prefixes were not the subject of the study.
15
Danegan 1985) and natural morphology (Dressier 1985). In syntax it was partially assumed by Ryden (1979). Syntactic changes as strengthenings or weakenings have also been the subject of research work carried out under the guidance of Prof. Janez Oresnik at the Department of Germanic Languages and Literatures at the Faculty of Arts in Ljubljana, Slovenia.3
2.2. On a synchronic level, the theory of strengthenings and weakenings assumes that of two roughly equivalent language variants one is »stronger« and the other one »weaker«. Formally, the stronger variant is more elaborate, it demands more effort from the speaker but it is easier to decode by the hearer. The weaker variant is less elaborate, more economical for the speaker and more difficult to decode for the hearer. Typical examples of strong variants are periphrastic constructions (as compared to simple forms), whereas typical examples of weak variants are assimilations, contractions, generalizations etc.
In each language a speaker/writer can encode his/her message in different ways: he chooses from the possible synonyms (to express the lexical meaning) or from the possible grammatical instruments (to express some grammatical meaning/relation). By analogy, in each language there is evidence of a hearer/reader being able to decode a message in more than one way (homonyms or constructions covering more than one grammatical meaning).
E. g.: When referring to a person who teaches English, the speaker of the English language can use at least two constructions:
a)	»English teacher«
b)	»teacher of English«.
The hearer/reader4 can decode the construction »English teacher« in at least two different ways:
a)	»teacher of the English language«
b)	»teacher who is of English nationality«.
The probability of (in)correot decoding depends on several factors: the similarity of the cultural background of the two participants in the communication process, the medium of communication (written language excludes instruments like emphasis, intonation, gesticulation etc.), context and extra-linguistic circumstances. The more the situation referred to, hence the message, is ooncrete and simple, the greater the probability of correct interpretation and vice versa: the more complex the information (conveying the meanings that the addressee cannot detect from context or from the extra-linguistic circumstances), the greater the probability of misunderstanding. From that point of view the speaker may be expected to choose the more elaborate of the possible variants, i.e. the »stronger« variant, in more complex, less concrete circumstances. In simple, more concrete circumstances, when the message is clear or the extra-linguistic elements self-evident, the speaker may be expected to economize and choose the less elaborate, i. e. the »weaker« of the possible variants.
3	The theoretical premises of the research work were published in Lingüistica XXX, Ljubljana 1990 (Oresnik, Snedec, Terzan, Trobevsek-Drobnak). See also Oresnik and Trobevsek-Drobnak, »Expanded Tenses in the Old English Orosius: A Strehgthened Construction«, in: Language and Civilization I, Frankfurt 1992, pp. 146—161.
4	We shall here disregard the intonation as the means of expressing different meanings.
16
2.3.	Language changes can consequently also be examined as the assertion of new stronger or weaker linguistic variants. The assertion of a stronger variant is called strengthening, the assertion of a weaker variant is called weakening.
2.4.	On a diachronic level, the theory of strengthenings and weakenings assumes that after the nascent state, i. e. before the grammaticalization, stronger variants assert themselves under relatively complex (grammatical) conditions, and weaker variants assert themselves under relatively simple (grammatical) conditions. The criteria of grammatical complexity/simplicity were first discussed by the Prague School (see Jakobson 1932). The Praguians operated with the terms marked and unmarked, which would roughly correspond to the terms simple and complex in this paper. According to Mayer-thaler (1980), individual parameters of grammatical categories have the following degrees of markedness:
1.	An independent clause is less marked than a dependent clause.
2.	The affirmative prepositional modality is less marked than the nonaffirmative propositional modalities.
3.	The present tense is less marked than the non-present tenses.
4.	The indicative mood is less marked that the non-indicative moods.
5.	Verb + direct object (in the accusative) is less marked than verb + prepositional object or verb + object clause.
6.	The singular is less marked than the non-singular.
7.	The active voice is less marked than the non-active voices.
3. The application of the theory of strengthenings or weakenings in the attempt to clarify the inconsistent use of the preverbal ge-in Old English
3.1. Points of departure
In the research presented in this paper the theory of strengthenings and weakenings was applied from the following points of view:
3.1.1.	The Old English ge-verbs and the Old English simplex verbs were syntactic variants, since ample empirical evidence shows that they performed the same syntactic functions. The former can be defined as the stronger and the latter as the weaker syntactic variants, which agrees with the basic definition of stronger or weaker constructions — the former are more elaborate, less economical and easier to decode; the latter are simpler, more economical and more difficult to decode.
3.1.2.	The process of the withering away of the prefix ge- was the assertion of the weaker syntactic variant. The disappearance of the preverbal ge-meant that the tendency to economize prevailed over the tendency to make the message more elaborate and hence clearer.
3.1.3.	In accordance with the theory of strengthenings and weakenings the ge- prefix was expected to weaken faster under relatively simple/unmarked (grammatical) conditions and to persist longer under relatively complex/ marked grammatical conditions. On the basis of the criteria of markedness stated sub 2.4, the following predictions were made:
1.	Ge-verbs will be more frequent in dependent clauses than in independent clauses.
2.	Ge-verbs will be more frequent in clauses of non-affirmative propo-
2 Acta
17
sitional modalities than in clauses of the affirmative propositional modality.
3.	Ge-verbs will be more frequently in non-present tenses than in the present tense.
4.	Ge-verbs will be more frequently in non-indicative moods than in the indicative mood.
6.	Ge-verbs will be more frequently in the non-singular than in the singular.
7.	Ge-verbs will be more frequently in non-active voices than in the active voice.
3.2. Grammatical analysis
Two types of samples were formed: basic samples containing ge-verbs and control samples containing simplex verbs.
3.2.1.	The first set of samples was taken from King Alfred's Old English translation of Orosius' Historiarum Adversum Paganos (Sweet, ed., 1883), on the history of the World from the Creation to A. D 416. Orosius was the third Alfred's translation from Latin and it was completed probably circa 890. The translation was rather free. Not only did Alfred omit what he considered unimportant, he also added his own remarks and observations and inserted original essays (Bosworth 1859).
The basic sample (BO) consisted of all 820 occurrences of ge-verbs in the edition. The sample did not include the auxiliaries beon/wesan and weorpan, nor the verb habban when used with a past participle.
The control sample (CO) consisted of 1000 instances of simplex verbs in the edition. The sample was made up of every other simplex verb in the edition, with the exception of modal verbs, the auxiliaries beon/wesan and weorpan and the verb habban in constructions with past participles. The sample covered most of the text in the edition (294 of 298 pages).
3.2.2.	The second set of samples was taken from the Old English translations of the Gospels according to St. Mark and St. John.
The Old English translations of the Holy Gospels date from the 9lh century (Skeat, ed., 1871, 1878). Several manuscripts are preserved, none of which, however, retains the original version. All the scribes adjusted their language to that of their time and dialect. Nevertheless, some manuscripts are earlier and probably closer to the original translation. The last preserved MSS were written quite some time after the Norman invasion, the oldest were probably written some hundred years earlier.
The samples were taken from the Corpus MS5, but other manuscripts6 were consulted, as well as the Latin original of the Lindisfarne MS and the Modern English translation of the Bible7.
5	CORPUS MS. — MS. No. CXL (before S. 4), in Corpus Christi College in Cambridge, described by Wanley (1705) on page 116.
6	HATTON MS. (before Hatton 65, now "Hatton 38), in Bodleyan library in Oxford, decribed by Wanley (1705) on page 76.
LINDISFARNE MS., also known as the DURHAM BOOK, now part of the Cotton MSS in the British Museum (Nero D. 4), with the Latin original and a Northumbrian gloss.
RUSHWORTH MS., in the Bodleyan library in Oxford (Auct. D. ii 19).
7	The Holy Bible containing the Old and New Testaments, set forth in 1611 and commonly known as the King James Version, American Bible Society, New York 1816.
18
The basic sample taken from the Gospel according to Saint Mark (BM) was made up of all 227 occurrences of g <2-verbs in the Corpus MS edited in Skeat (1871). The same types of verbs were excluded from the sample as in the case of the basic sample taken from Orosius.
The control sample (CM) consisted of 405 occurrences of simplex verbs. It was made up by taking every third simplex verb in the Corpus MS of the edition. The same types of verbs were excluded from the sample as in the case of the control sample taken from Orosius.
To extrapolate the differences between the basic and the control samples, two sub-samples were formed. The basic sub-sample (BM1) consisted of 179 instances of ge-verbs which had the prefix ge- in all four consulted MSS, even if the verb itself was different. The control sub-sample (CM1) was made up of those 250 instances of the control sample that took no prefix in any of the MSS cited in the edition.
3.2.3.	The third set of samples was taken from St. John's Gospel. The basic sample (BJ) consisted of 150 occurrences of ge-verbs taken from the first seven chapters of the Corpus MS in the edition. The control sample (CJ) was made up of 160 occurrences of simplex verbs (the first five verbs on each page of the first seven chapters of the Gospel were taken).
On account of a relatively small basic sample, only one sub-sample was formed, the control sub-sample (CJ1), on the basis of the same criteria as in the case of the sub-samples taken from St. Mark's Gospel.
3.2.4.	The choice of parameters was partly diotated by Mayerthaler's criteria of markedness (see 2.4), and partly by what the author considered to be potentially significant in terms of the complexity of the grammatical environment.
3.2.5.	Each instance in each sample was analyzed as follows:
a)	The clause in which the verbal phrase contained a finite ge-verb (or a simplex verb respectively) was defined as having affirmative or negative prepositional modality. The interrogative prepositional modality was not taken into consideration on account of its low absolute frequency.
b)	The clause in which the verbal phrase contained a finite ge-verb or a simplex verb respectively was defined as a simple sentence, as a main clause in a complex sentence or as a subordinate clause.
c)	The tense of the verb was defined as preterite or as present. In the samples taken from the two Gospels, future reference of verbs in the present tense was noted. The criterion was the verbal tense in the Latin original of the Lindisfarne MS and in King James' Version of the Bible.
d)	The mood of the verb was defined as the indicative, the subjunctive or the imperative. In ambiguous cases the Modern English verb in King James' Bible was consulted. Verbs which took modals in the modern English translation of the Bible were defined as subjunctive. In other dubious cases, verbs were defined as indicative.
e)	Verbs were defined as intransitive, monotransitive or ditransitive. The type of object(s) was defined with transitive and ditransitive verbs as simple accusative, complex accusative8, genitive, dative, prepositional object, object clause or non-finite object clause.
8 Here the complex accusative denotes an accusative expanded by another accusative (object complement), prepositional phrase (adjuct) or relative clause.
2*
19
f)	The person of the verb was defined as the 1st, the 2nd or the 3rd.
g)	The grammatical number of the verb was defined as plural or singular. The voice of the verb was not noted on account of the low frequency of
non-active voices in all three Old English texts. The aspect was omitted on account of the possible aspectual function of the .preverbal ge-.
3.2.6. To compute the probability that a given parameter assumed its marked value in a given sample, the following formula was used:
The probability rates of individual values of parameters in basic samples were compared with the corresponding probability rates an the control samples.
The statistical significance of the differences between probability rates of (marked) values of parameters was computed with the help of the index here named Id:
Two probability rates were considered to be significantly different ¡if the value of the index Id exceeded 2. (Pavlic 1985).
3.2.7. To better illustrate the relation between different values of individual parameters in terms of the presence or absence of the ge-prefix, the frequency of ge-verbs at individual values of a given parameter was determined. One drawback of this method is that it does not reflect the absolute frequency of a given value of a parameter and can therefore be misleading in terms of the statistical significance of the results, unless taken in combination with the method described sub 3.2.6.
4. Statistical results of the grammatical analysis in the light of the theory of language changes as strengthenings or weakenings
4.1. Parameter: Type of clause
In Orosius independent sentences were significantly more frequent in the basic than in the control sample. The frequency of main clauses with subordinate clauses was almost the same in both samples, whereas the frequency of subordinate clauses was higher in the control sample. Ge-verbs were therefore most likely to appear in independent sentences (52.3 %), in main clauses expanded with more than one subordinate clause (51.6 %), in main clauses with one subordinate clause, and least likely to appear in subordinate clauses.
9 a = a favourable event, i. e. a marked value of the parameter; Pa = probability of a favourable event; na = number of favourable events; n = number of all possible events.
_ 10 Pj= mean probability of a favourable event in both samples (Pi + P2)/2; q = 1 — P; ni = total number of events in sample 1; n2 = total number of events in sample 2.
P(a) == iu/n'
Id =
Pi — P:
10
20
frequencies of parameter values (in %)
sample independent main clause subordinate
	sentence	with subor.	clause
BO	48.0	19.0	33.0
CO	35.7	19.2	45.1
BM1	41.9	16.2	41.9
BM	42.3	17.2	40.5
CM	62.7	13.8	23.5
CM1	67.6	12.8	19.6
BJ	35.3	18.7	46.0
CJ	52.5	15.6	31.9
CJ1	57.4	15.7	26.9
P(B)"	41.8	17.8	40.4
P(C)	55.2	11.6	29.4
frequencies of ge-verbs (%)
Orosius Mark John Total
independent sentence	52.5 27.4 38.7 39.5
main cl. with subordinates	44.8 41.1 52.8 46.2
subordinate clause	37.5 49.2 51.9 46.2
value of parameter
In St. Mark's Gospel the frequency of independent clauses was found to be increasing consistently from the basic sub-sample to the control sub-sample. The frequency of main clauses and of subordinate clauses consistently declined in the same direction. That means that ge-verbs were most likely to be found in subordinate clauses (49.2 %), in main clauses (41.1 %) and least likely in independent sentences (27.4 %). In St. John's Gospel the results were more or less the same for subordinate clauses and for main clauses (i. e. for complex sentences), while independent sentences seemed to reduce the probability of ge-verbs.
The results of the analysis of the two Gospels were different from the results of the analysis of Orosius, which made them inconclusive, although
they confirmed the prediction stated sub 3.1.3.
4.2. Parameter: Propositional modality
In all sets of basic samples the negative propositional modality was more frequent than in the control samples. In all three sets of samples ge-verbs were more frequent in the negative sentences than in the affirmative sentences. The reliability of such results, which completely supported the prediction cited sub 3.1.3., was diminished only by the fact that the absolute presence of the negative propositional modality was low (less than 10 %, except in the basic sample of St. John's Gospel).
11 P(B) = mean probability rate in all basic samples; P(C) = mean probability rate in all control samples.
21
frequencies of parameter values (in %)
sample -;---:
negative affirmative modality modality
BO	96.8	3.2
CO	98.1	1.9
BM1	92.2	7.8
BM	92.6	7.4
CM	94.8	5.2
CM1	96.8	3.2
BJ	87.3	12.7
CJ	91.2	8.8
CJ1	92.6	7.4
~P(B)	92.2	7.8
P(C)	94.7	5.3
value of	frequencies of ge-verbs (%)		
parameter	Orosius Mark	John	Total
affirmative modality	44.8 36.4	47.3	42.8
negative modality	57.8 44.7	57.6	53.4
4.3. Parameter: Tense
sample	frequencies of parameter values (in %)		
	preterite	present total	present with future ref.
BO	97.0	3.0	_
CO	83.7	16.3	—
BM1	59.2	40.8	26.0
BM	59.5	40.6	22.8
CM	66.9	33.1	16.4
CM1	62.8	37.2	16.1
BJ	54.7	45.3	20.6
CJ	64.3	35.7	17.5
CJ1	59.3	40.7	13.6
P(B) P(C)	67.6 67.4	32.4 32.6	23.1 15.9
value of	frequencies of		ge-verbs (%)	
parameter	Orosius	Mark	John	Total
preterite	48.7	33.2	44.3	42.1
present tense total	13.3	38.8	53.5	35.5
present tense — future ref.	—	48.8	58.3	53.6
22
The prediction cited sub 3.1.3., that preterite would be more frequent in the basic samples than in the control samples, was corroborated by the results of the analysis of Orosius. The probability rate for the parameter tense to assume its marked value (preterite) was significantly higher in the basic sample than in the control sample. But the results of two other sets of samples were contrary to the prediction stated above. Preterite was 'foun'd to be more' frequent in the control than in the basic samples. Such a marked difference between the results obtained from the analysis of Orosius and the results obtained from the analysis of the two Gospels made it imperative to reexamine the nature of tenses in both sets of samples. A closer look at the verbs in the present tense (in all samples) revealed that the nature of most of such verbs in Orosius was the universal time reference:
10/19-22: Of paere ie Indus, pe be western eallum paem lande liged, betux paere ie Indus / paere pe be western hiere is, Tigris hatte, pa flowad buta sup on pone Raedan Sae;v-
However, the real time reference of the present-tense verbs in the two Gospels was much more complex:
Mark 1/2, Corpus MS: Se ge-gearwad pinna weg be-foran de. (KJV: ... which shall prepare thy way before thee.);
Mark 4/16 Corpus MS: Sona paenne hi paet word ge-hyrad and paet mil blisse onfod. (KJV: ... who, whan they have heard the word, innediately receive it with gladness)-,
John 5/45 Corpus MS: Ne wene ge paet ic cow wrege to faeder. (KJV: Do not think that I will accuse you to the father)
To account for the real time reference and its relation to the presence or absence of the prefix ge-, the verbs referring to future time were selected from all the verbs in the present tense in the samples taken from the two Gospels. This time reference was chosen because it was possible to test it against the Modern English translation of the verb in the present tense and/or the Latin original in the Lindisfarne MS. It was assumed that of all possible real time references of verbs in the present tense the future reference was one of the more complex (marked) ones (as compared to the universal time reference for instance). The results confirmed the assumption: not only was the frequency of future time reference expressed by the present tense higher in the basic than in the control samples, but the trend of its presence was found to be consistently rising from the control sub-sample, via the control and the basic samples to its highest level in the basic sub-sample. Ge-verbs were found in 35.5 % of all analyzed present tenses, in 42.1 % of all preterites and in 53.6 % of all present tenses with future reference.
4.4. Parameter: Mood
The prediction cited sub 3.1.3., namely, that non-indicative moods were to be expected more frequently in the basic than in the control samples,
13 The verbs underlined are the verbs in the present tense. " King James Version of the Holy Bible (1611).
23
sample	frequencies of parameter values (in %)		
	indicative	subjunctive	imperative
BO	88.1	11.6	0.3
CO	90.9	9.0	0.1
BM1	78.3	12.8	8.9
BM	77.1	13.2	9.7
CM	86.0	8.6	5.4
CM1	85.5	5.6	8.8
BJ	85.0	11.3	4.7
CJ	91.9	5.0	3.1
CJ1	93.3	1.9	4.6
P(B) P(Q	82.1 89.5	12.2 6.0	5.9 4.4
frequencies of ge-verbs (%)
parameter	Orosius	Mark	John	Total
indicative subjunctive imperative	44.3 51.4 75.0	33.5 46.2 50.0	46.2 68.0 58.3	41.3 55.2 61.1
proved to be correct for all three sets of samples. The low frequency of the non-indicative moods (especially of the imperative) in Orosius lessened to some extent the reliability of the results, but their consistency still spoke in favour of the prediction cited above.
4.5. Parameter: Transitivity
sample	frequencies of parameter values (in °/o)		
	intransitive	monotransitive ditransitive	
BO	14.4	63.9	21.7
CO	32.3	56.3	11.4
BM1	18.4	76.6	5.0
BM	18.1	77.5	4.4
CM	37.5	50.0	12.3
CM1	41.2	46.8	12.0
BJ	14.7	80.0	4.7
CJ	45.6	45.0	9.4
CJ1	45.4	47.2	7.4
P(B) P(C)	16.4 40.4	74.5 49.1	9.1 10.5
24
type of object:
frequencies of parameter values (in °/o)
sample	acc. obj.c.		non-fin.	com. acc.	gen.	dat.	prep. ob.
BO	58.1	11.4	2.1	11.0	2.9	4.8	9.7
CO	43.0	15.8	7.6	17.8	1.2	2.5	12.1
BM	52.9	8.0	4.5	17.6	1.7	10.2	5.1
CM	60.1	6.8	1.5	4.9	1.5	14.3	10.8
BJ	39.6	11.6	3.3	18.2	_	9.1	18.2
CJ	55.6	9.7	1.4	6.9	—	4.2	22.2
P(B)	50.2	10.3	3.3	15.6	1.5	8.0	11.0
P(C)	52.8	10.8	3.5	10.0	0.9	7.0	14.8
							
	value of		frequencies of ge-verbs (%)				
	parameter		Orosius	Mark	John	Total	
	intransitive		26.8	21.2	23.2	23.7	
	monotransitive		48.3	46.3	62.7	52.4	
	ditransitive		60.9	16.7	31.8	36.5	
	— accusative		55.8	43.2	54.5	51.2	
	— obj. clause		40.3	50.0	66.7	52.3	
	— non-finite c.		20.4	72.7	80.0	57.7	
	— complex acc.		36.7	75.6	81.5	64.6	
	— genitive		68.2	38.2	—	53.2	
	— dative		64.1	50.0	78.6	62.2	
	— prepos.obj.		42.9	29.0	57.9	43.3	
In all sets of samples there was a marked (significant) difference between the probability rate (frequency) of intransitive verbs in the basic or in the control samples respectively. Intransitive verbs were found more frequently in the control samples and transitive verbs were more frequent in the basic samples. The situation with the ditransitive verbs was more controversial. In Orosius ditransitive verbs were more frequent in the basic than in the control sample, whereas in the two Gospels they were more frequent in the control than in the basic samples. In Orosius the frequency of the prefiix ge- was highest with ditransitive verbs. In St. John's Gospel the frequency of ge-verbs was highest with mono'transitive verbs, lower with ditransitive verbs and lowest with intransitive verbs. In St. Mark's Gospel the frequency of ge-verbs was highest with monotransitive verbs, lower with intransitive verbs and lowest with ditransitive verbs.
The presence of the prefix ge- significantly correlated with the transitivity of the verbs in that it was higher with transitive than intransitive verbs, but the number of objects showed no correlation with the presence or absence of the prefix.
As far as the form of the object is concerned, the frequency of the construction verb + accusative was found to be higher in the control samples than in the basic samples of St. Mark and St. John, which would agree with the prediction, cited sub 3.1.3., but it was lower in the control sample than in the basic sample of Orosius. The form of the object did not correlate with
25
the presence or absence of the ge- prefix. Quite contrary to the prediction stated above, the prefix ge- was found less frequently in construction of verb + prepositional object than, for example, in constructions of verb + dative, which would roughly express the same deep case relation.
4.6. Parameter: Number (3rd person only)
	frequencies of parameter values (in %)	
sample		
	singular	plural
BO	66.2	33.8
CO	65.8	34.2
BM1	53.7	46.3
BM	56.9	43.1
CM	64.0	36.0
CMl	66.4	33.6
BJ	51.3	48.7
CJ	74.3	25.7
CJ1	78.3	21.3
P(B)	57.0	43.0
P(C)	69.8	30.3
value of	frequencies of ge-verbs (%)			
parameter	Orosius	Mark	John	Total
singular	45.5	36.4	39.3	40.4
plural	45.1	40.2	64.0	49.8
The prediction cited sub 3.1.3., that plural would be moore frequent in the basic than in the control samples, proved to be correct an the case of St. Mark's and St. John's Gospels. In Orosius the results of the analysis did not corroborate the prediction, however, the difference between the frequencies of plural in the basic and in the control samples was not statistically significant. The consistency of the results in the basic and control sub-samples nevertheless substantiated the conclusion that there existed a correlation between the presence of the prefix ge- and the grammatical number of the verb.
4.7. Parameter: Person
The reason for abstaining from any assumption about the relation between the person and the prefixatio-n of the verb was that the structure of person(s) in the verb is too complex to be squeezed into the dichotomy of simple-complex or marked-unmarked. Nevertheless, the relation of the third person versus the non-third persons was noted in the sense that the third person was considered to be the »non-person«, neither the speaker nor the hearer of the message, and from that point of view less complex (less marked) than the first and the second persons.
26
frequencies of parameter values (in %)
sample	1st sg	2nd sg	3rd sg	1st pi	2nd pi	3rd pi
BO	1.1	0.1	65.4	_	_	33.5
CO	0.8	0.1	64.2	1.1	0.5	33.3
BM1	3.4	3.4	46.9	3.9	11.7	30.7
BM	4.0	4.4	48.5	3.5	11.0	28.6
CM	5.4	7.2	51.4	1.0	6.4	28.6
CM1	7.6	9.2	49.6	1.2	6.0	26.4
BJ	5.3	5.3	40.7	6.7	16.7	25.3
CJ	8.8	4.4	61.3	1.9	5.0	18.8
CJ1	11.1	5.5	62.0	1.9	4.6	14.4
P(B)	3.5	3.3	50.4	3.5	9.9	29.5
P(C)	6.6	5.3	57.7	1.4	4.5	24.3
value of	frequencies of ge-verbs (%)			
parameter	Orosius	Mark	John	Total
third persons	45.4	38.9	43.6	42.6
non-third persons	28.6	39.1	61.4	43.0
The results of the analysis showed slightly higher frequency values for the non-third persons in the basic than in the control samples. However, no conclusion could be reached on account of the fact
a)	that it is difficult to separate the notion of the person from the notion of the number (all persons in plural tend to be more frequent in the basic than in the control samples);
b)	that individual persons cover more than one notion, e. g. the 1st person plural can be inclusive (I + you) or exclusive (I + they); the 3rd person can design human beings or inanimate subjects, abstract or concrete subjects etc.
CONCLUSION
5.1.	The results of the analysis of the grammatical environment of Old English ge-verbs and their comparison with the results of the analysis of the grammatical environment of Old English simplex verbs in the same Old English texts confirm the initial hypothesis that the presence of the Old English preverbal ge- correlated with the complexity (markedness) of the grammatical environment of the respective verbs.
5.2.	Of seven observed parameters five (propositional modality, tense, mood, transitivity, number) showed higher probability rates of assuming marked values in the basic than in the control samples. That means that at the marked values of these parameters verbs were more likely to have the prefix ge- than at the unmarked values of the same parameters.
According to our statistical results, the type of clause and the person of the verb do not seem to affect the frequency of the preverbal ge-.
27
5.3.	A closer look at the results, those confirming the initial hypothesis and those contradicting it, shows that the criteria of the complexity (marked-ness) of grammatical environnment as formulated by Mayerthaler (1981) need further elaboration if they are to be applied in the examination of the correlation between the complexity of grammatical environment and the assertion of stronger or weaker language variants. The impact of grammatical environment seems stronger if deep, notional grammatical relations are considered instead of the formal expressions of individual parameter values. Some facts established in our paper substantiate the preceding conclusion.
The impact of the real time reference (in our case of future time) prevailed over the impact of the fact that this (marked, complex) time reference was expressed with a (simple, unmarked) verbal form in the present tense. Moreover, in all observed texts the frequency of the preverbal ge- was higher with verbs in the present tense referring to future time than with verbs in the preterite referring to past time. A marked value of a given parameter seems to aggravate the complexity of the grammatical environment even more if it is not transparent, i. e. not formally expressed with a definite marker. This conclusion gains weight with the results presented in the dissertation on the diachronic and synchronic aspects of the »werden« future tense as a syntactic variant of the present tense referring to future time in German, written and defended by Karmen Terzan at the Faculty of Arts in Ljubljana in 1991. In her extensive analysis of German sentences referring to future time Terzan found sentences with verbs in the present tense to contain more complex (marked) elements than sentences containing verbs in the explicit future tense. It has to be noted, however, that in neither study the presence/ absence of an adjunct or of an adverbial clause of time was taken into consideration. The relation between the distance of the futurity marker from the root syllable of the verb and the impact that it has on the choice of potential syntactic variants within the sentence would certainly deserve further investigation.
It is perhaps worth recalling at this point that the results show for prepositional objects lower correlation with the frequency of ge-verbs than for the bare datives, which express roughly the same deep case relation. The fact that ditransitive verbs did not feature as more complex than the mono'transi-tive ones could perhaps be ascribed to the fact that only verbs with explicit two objects were marked as ditransitive.
The need to consider verbal persons from the paint of view of their deep complexity was mentioned in chapter 4.7.
5.4.	The conclusion that the occurence of the Old English preverbal ge-correlated with the complexity of the grammatical environment in the clause in which the verbal phrase containing a ge-verb occurred is defensible. However, the complexity of the grammatical environment must be primarily considered in its deep and also cumulative sense: the greater the number of data (meanings) to be expressed and the more complex the temporal and argumentative information, the greater the pressure to expand the formal expression of the communication unit. In other words: if complex, marked relations are not (cannot be) explicitly encoded with specific language units, they will affect the choice of language variants in the sense that the weight of (grammatical) information is balanced with the length of the utterance.
28
REFERENCES
Benson, T., Vocabidarium Anglo-Saxonicum, Léxico Gui. Somneri Magna Parte auctius, Oxford 1701.
Benveniste, E., »Structure des relations de personne dans le verbe«, Bulletin de la Société de Linguistique, XLIII, 1946.
Bernhardt, E., »Die Partikel Ga als Hilfmittel bei der gotischen Conjugation«, ZfdPh, II, 1870.
Bosworth, J., ed., King Alfred's Anglo-Saxon Version of the Compendious History of the World, London 1859.
Brugmann, K., Die Demonstrativpronomina der indogermanischen Sprachen, Leipzig 1904.
Chomsky, N., Knowledge of Language: Its Nature, Origin and Use, Praeger Publishers, New York 1986.
Goedsche, C. R., »Aspect versus Aktionsart«, JEGP, xxxix, 1340.
Goropius, J., Hermathena, Antwerp 1580.
Grimm, J., Deutsche Grammatik, 2nd ed., rev. Scherer, Berlin 1878.
The Holy Bible containing the Old and New Testaments, set forth in 1611 and commonly known as the King James Version, American Bible Society, New York 1816.
Jakobson, R., Zur Struktur des russischen Verbums, Prague 1932.
Lenz, P., »Der syntactische Gebrauch der Partikel 'ge' in den Werken Alfreds des Grossen«, dissert., Darmstadt 1886.
Lindemann, J. W. R., Old English Preverbal Ge-: Its Meaning, The University Press of Virginia, Charlottesville 1970.
Anton Lorz, »Aktionsarten des Verbums in Beowulf«, dissert., Wuerzburg 1908.
Martens, H., »Die Verba perfecta in der Nibelungendichtung«. KZ, xii, 1863.
Mayerthaler, W., Morphologische Natuerlichkeit, Wiesbaden 1981.
Mosse, F., Histoire de la forme périphrastique être + participe présent en Anglais de 1200 à nos jours, thèse pour le doctorat ès lettres, Paris 1938.
Orešnik, J., »Introduction to the Subsequent Three Papers in the Present Volume«, Lingüistica XXX, Ljubljana 1990.
Orešnik, J., Trobevšek-Drobnak, F., »Expanded Tenses in the Old English Orosius: A Strengthened Construction«, Language and Civilization I, pp. 146—161, Peter Lang Publishers, Frankfurt 1992.
Pavlic, I., Statistička teorija i primjena, Tehnička knjiga, Zagreb 1985.
Pilch, H., »Das AE. Praeverb ge-«, Anglia. LXXXI, 1953.
Pott, A. F., Etymologische Forschungen auf dem Gebiete der indogermanischen Sprachen, Lemgo/Detmold 1859.
Ryden, M., An Introduction to the Historical Study of English Syntax, Stockholm 1979.
Samuels, M. L., »The Ge- Prefix in the Old English Gloss to the Lindisfarne Gospels«, Transactions of the Philological Society, London 1949.
Skeat, W., ed., The Gospel according to Saint Mark. In Anglo-Saxon and Northumbrian versions synoptically arranged, with collations exhibiting all the readings of all the MSS, Cambridge University Press 1871.
Skeat, W., ed., The Gospel according to Saint John. In Anglo-Saxon and Northumbrian versions synoptically arranged, with collations exhibiting all the readings of all the MSS, Cambridge University Press 1878.
Snedec, A., »The Progress of Syntactic Change: English Do-Support«, Lingüistica XXX, Ljubljana 1990.
Stampe, D., A Dissertation on Natural Phonology, New York, London 1979.
Streitberg, W., »Perfective und imperfective Aktionsart im Germanischen«, PBB, XV, 1891.
Sweet, H., ed., King Alfred's Orosius, EETS Nr 79, London 1883, repr. 1959.
Teržan, K., »Das Wörden-Passiv als syntaktische Verstaerkung im Deutschen«, Lingüistica XXX, Ljubljana 1990.
Trobevšek-Drobnak, F., »Expanded Tenses in the Old English Orosius: a Syntactic Strengthening«, Lingüistica XXX, Ljubljana 1990.
29
Trobevšek-Drobnak, F., »Skladnja st aroangleških glagolov s predpono ge- v primerjavi s skladnoj staroangleških gagolov brez prepopne«, disertacija, Filozofska merjavi s skladnjo staroangleških glagolov brez predpone«, disertacija, Filozofska fakulteta, Ljubljana 1990.
Wanley, H., Antiquae Literaturae Septentrionalis Uber alter, seu Humphredi Wanleii Librorum Vett. Septentrionalium Catalogus, Oxoniae 1705.
Weick, F., »Das Aussterben des Praefixes ge- im Englischen«, dissert., Darmstadt 1911.
Wise, F., ed., Annales rerum gestarum AElfredi Magni auctore Asserio Mene-vensi, recensuit Franciscus Wise, Oxon. 1722.
30