Connective or “disconnective” discourse marker? Old English þa, multifunctionality and narrative structuring

Brita Wårvik, Department of English Language and Literature, Åbo Akademi University

1. Introduction

Narratives are probably the most thoroughly studied type of text and their structuring is well investigated in a great number of languages. Signals of narrative structuring have been the focus of much early work on text and discourse linguistics (e.g., Grimes 1978) and they continue to interest linguists of various orientations. Some structuring signals present challenges for the analysis, because their uses in the texts do not seem to warrant a coherent description. One such signal is the Old English adverb þa, which has been attributed both connective and ‘disconnective’ functions. The aim of this study is to investigate the contradictory hypotheses about the discourse functions of þa in an attempt to find a common denominator that can be taken as the core meaning of þa.

1.1 Þa and þonne: homonymy and polysemy

The Old English adverb þa has been included in the list of pragmatic particles (Östman 1982, Östman 1995), discourse markers (Enkvist & Wårvik 1987), pragmatic markers (Brinton 1996: 9–12), and connectors (Lenker 2010: 64ff). Þa is perhaps no longer one of the ‘linguistic Cinderellas: familiar, drab, hard-worked, and lacking in morphological, phonological and etymological glamour’ (Enkvist 1972: 95), though it is still occasionally considered meaningless enough to be sent to the ‘lunacy ward [of the] grammar or lexicon [of language] where mindless morphs stare vacantly with no other purpose than to be where they are’ (Bolinger 1977: ix). For þa the ‘mindless’ job involves being in initial position in the clause and keeping the verb in the second position.

According to A Microfiche Concordance to Old English, þa is the third most frequent word in Old English: there are 65,208 occurrences of þa including alternative spellings in the extant Old English texts (Venezky & Butler 1985). [1] However, this huge number of cases in the concordance list subsumes all homonymous, polysemous and multifunctional cases of the item, which are illustrated by example (1). [2]

(1)

Þa (1) hie þa (2) hamweard wendon mid þære herehyþe, þa (3) metton hie micelne sciphere wicenga, & þa (4) wiþ þa (5) gefuhton þy ilcan dæge, & þa (6) Deniscan ahton sige; (ChronA 885.10)

‘When (1) they then (2) returned homeward with that booty, then (3) they met a large fleet of pirates, and then (4) fought against them (5) the same day, and the (6) Danes had the victory.’

In this example, we find þa appearing as a conjunction (1), an adverb (3, 4), pronoun (5), and demonstrative determiner (6). The second instance of þa (2) is either an adverb in the subclause introduced by the conjunction þa (1) or the second member of a doubled conjunction þa þa, which can be separated from the first member by an inserted element, though the two members mostly appear together. The focus of this study is on the adverb þa, excluding the demonstrative and pronoun cases totally and limiting the comments on the conjunction to noting the somewhat problematic distinction between the adverb and the conjunction. Mitchell (1985: §§ 2536, 3010) characterizes þa as an ambiguous adverb/conjunction, acknowledging the possibility of an intermediate stage between the categories (cf. Lenker 2010: 64–66). The distinction is traditionally determined by word order, as outlined in Bosworth & Toller, s.v. þa (= BT). BT do not mention þa in non-initial positions, in which it is always an adverb.

adv. conj. Then, when. When the word stands at the beginning of a clause and may be translated by then, the verb generally precedes its subject; if it is to be translated by when the subject generally precedes the verb.

They list the different senses of þa:

1. then, at that time …
2. marking sequence, then, after that, thereupon …
3. as adverbial connective, 

   (1) of time, when

   (2) of cause or reason, when, since, as

As a temporal adverb, þa can be deictic, referring to a point or period of time, removed from the present moment, typically in the past (sense I in BT), a sense that it shares with Present-Day English then (Fretheim 2006, Schiffrin 1987, Schiffrin 1990, Schiffrin 1992, among others). This sense can be derived from the demonstrative origin of the word, as described in OED (s.v tho; abbreviations expanded, BW):

originally a case form of the demonstrative stem þa- of the, that; either the actual accusative singular feminine, Old English and Old Norse þá, or (as some think) a stressed form of the original accusative masculine; meaning ‘that time’, the noun being omitted

The deictic sense can be nicely illustrated by þa in a subordinate clause. In example (2), þa refers to the time in the biblical story when Adam gives names to all creatures.

(2)

and adam him eallum naman gesceop. and swa swa he hi þa genamode. swa hi sindon gyt gehatene (ÆCHom I, 1.14.13)

‘and Adam created names (for) them all and as he them then named so they are still called’

The other temporal sense shared with Present-Day English then is anaphoric, referring back to a previous point in time in the text world, taking the reference time forward: ‘then, next, after that’ (sense II in BT; Fretheim 2006, Schiffrin 1987, 1990, 1992 among others). In example (3), this sense is reinforced by æt nihstan ‘next’.

(3)

Swa he Philippus þa miclan ricu geniþerade, þeh þe ær anra gehwelc wende þæt hit ofer monig oþru anwald habban mehte, þæt hie þa æt nihstan hie selfe to nohte bemætan. (Or3 III.7.63.25)

‘So much, he, Philippus, humiliated those great kingdoms, though earlier each of them thought that it might have power over many others, that they then next compared themselves to nothing.’

The deictic and anaphoric senses blur together. In example (4), this point in the story is preceded by the fall of Lucifer and we are told what happens next, each þa making reference to the preceding one and moving the reference time forward and at the same time referring to the time of the story.

(4)

And god þa geworhte ænne mannan of lame. and him on ableow gast. and hine geliffæste. and he wearð þa man gesceapen on saule and on lichaman. and god him sette naman adam. and he wæs þa sume hwile anstandende. god þa hine gebrohte on neorxna wange. and hine þær gelogode. and him to cwæð; (ÆCHom II, 1.12.28)

‘And god then created a man of clay and blew spirit on him and made him alive and he was then created man in soul and in body and god gave him the name adam and he was then some while alone. God then brought him to paradise and lodged him there and said to him’

As if homonymy and polysemy were not enough, Old English also presents us with another complication. Today’s then is not the successor only of þa, but there is another word þonne, which is also a high-frequency item: the Microfiche Concordance (Venezky & Butler 1985) lists 17,263 occurrences. [3] The difference between þa and þonne is outlined by BT (s.v. þanne) as follows:

þanne and þá differ in force; the former is used where the time of an action is indefinite, and is found with the future, the indefinite present and the indefinite past; the latter is used where a definite action has taken place.

The reference to an indefinite time is most obvious in the causal sense ‘in that case’, which þonne has in if – then pairs, as in example (5).

(5)

Gif we willað areccan ealle ða gewitnyssa þe be criste awritene sind. þonne gæð þær swiðe micel hwil to. (ÆCHom I, 9.214)

‘If we want to relate all the witnesses that are written about Christ, then it takes such long while.’

Þonne is also used as a temporal deictic and anaphora in cases where the reference time is in the future or the situation is repeated or hypothetical, that is, where we do not have a single definite event or action in the past. Example (6), which refers to future events, also illustrates the adverbial and conjuctive uses of þonne. In example (7), þa accompanies a single action, while þonne marks a situation which is repeated an indefinite number of times.

(6)

Þonne bið sib on eorðan þonne ure drihten cymð to urum lande. and ðonne he gæð into urum husum (ÆCHom I, 7.138)

‘Then (there) will be peace on earth when our Lord comes to our land and when he goes to our houses’

(7)

Þa foron hie siþþan æfter [4] þæm wealda hloþum & flocradum. bi swa hwaþerre efes swa hit þonne fierdleas wæs. (ChronA 894.11)

‘Then they went afterwards through the woods in bands and troops, on whichever side it then was without an army.’

In example (8), we find þonne ‘then’ contrasting with nu ‘now’; nu in the first clause refers to the time of preaching or reading the text and þonne to a time after nu, indicated by the imperative. When the text turns back to past events and the story continues, we find þa.

(8)

Nu þencð mænig man and smeað hwanon deoful come; Þonne wite he þæt god gesceop to mæran engle þone þe nu is deoful. ac god ne sceop hine na to deofle. ac þa ða he wæs mid ealle fordon and forscyldgod þurh ða miclan upahefednysse and wiðerweardnysse þa wearð he to deofle awend. se ðe ær wæs mære engel geworht. ða wolde god gefyllan and geinnian þone lyre þe forloren wæs of ðam heofonlicum weorode. and cwæð þæt he wolde wyrcan mannan of eorðan. þæt se eorðlica man sceolde geþeon and geearnian mid eaðmodnysse þa wununga on heofonan rice. þe se deoful forwyrhte mid modignysse; And god þa geworhte ænne mannan of lame. (ÆCHom II, 1.12.18–28)

‘Now many a man thinks and ponders whence the devil came; Then let him know that god created as a great angel the one that now is devil, but god did not create him as devil, but when he was altogether destroyed and condemned through the great pride and hostility then he was changed to devil, who earlier was created a great angel. Then god wanted to fill and supply the loss that lost was of the heavenly host and said that he wanted to create a man of earth, that the earthly man should thrive and earn with humility the dwelling in heaven’s kingdom that the devil forfeited by pride; And god then created a man of clay.’

Mitchell (1985) discusses the differences between these two items (adverbs §§ 1116–1117, conjunctions §§ 2562ff) and notes the complexities involved in a great number of cases. A thorough investigation of the division of labour between þa and þonne is outside the scope of the present study, as here we concentrate only on their connective vs. “disconnective” functions. More studies are also needed of the relationship between these two ambiguous adverb/conjunctions in the later stages of the history of English, when the adverbial and conjunction functions are differentiated and divided between then and when, and þa gives way to þonne (see Wårvik 1995 for a preliminary study of the discourse contexts of this change).

Before we turn to the hypotheses about the discourse functions of þa, let me introduce some basic terms used in the present study.

1.2 Basic terms: discourse marker, text structure and continuity

“Discourse marker” is a term which is used in somewhat varying ways for a varying group of items. To list some of the other terms that have been used to refer to roughly the same group of items, we find particles (Leibniz 1839/1940: 323–325), pragmatic particles (Östman 1981), modal particles (Weydt 1969), mystery particles (Longacre 1976), discourse particles (Kroon 1998), pragmatic markers (Brinton 1996), discourse markers (Schiffrin 1987), conjunctions (Halliday & Hasan 1976), and connectors (Lenker 2010). The items in this group have a variety of functions on different levels of text/discourse. The different repertoires of levels that have been used in modelling the functions of these items can be illustrated by the ideational, rhetorical, sequential, and topical levels (Fraser 1999), the levels of ideational, rhetorical, and sequential structure and inferential component (González 2004), the propositional, interpersonal, and textual metafunctions (Halliday 1978), the representational, presentational, and interactional levels (Kroon 1998), the parameters of coherence, politeness, and involvement (Östman 1981), the coherence components of ideational, rhetorical, and sequential structure (Redeker 1991), and the planes of ideational, action, and exchange structure, information state, and participation framework (Schiffrin 1987). The focus of this study will be on structuring functions that can be found in the different models as part of the textual or presentational level, of coherence, of rhetorical and sequential structures, and of planes of ideational and action structure.

To describe text structure for the purposes of the present study, we can start from a simplified view of narrative structure, as consisting of units filling functional slots in the narrative schema. For the stories in the sample, a well fitting schema can be found in the Labovian model: (abstract) introduction + complicating action + resolution (coda) + evaluation (Labov & Waletzky 1967); the only modifications that are needed are omissions of some parts in very short stories like the Ohthere interpolation in (9) and in the Anglo-Saxon Chronicle. In addition to the narrative schema, we need to take into account cohesion, coherence, connectivity, continuity, conjunction – a variety of terms covering the ways of keeping a text/discourse together, creating textuality at different levels (cf. e.g., Enkvist 1987a, Enkvist 1987b, Givón 1983, Halliday & Hasan 1976).

In whatever way we see the forces keeping the narrative text together, we cannot model them as just one single unified line of progress through the story: the text consists of units, for which we can adopt the frequently used term “episode”, which form a hierarchical structure of larger and smaller units. The hierarchical episode structure is of course only one dimension of the organization of text/discourse. In addition to chunking – marking the hierarchical organization of the units, a full multi-dimensional model of text/discourse structure would have to take into account at least cohesion – linking units together, grounding – keeping track of the main line of development, and view-point – indicating the perspective from which the content is presented. The focus of this study is only on cohesion and chunking as they relate to connectivity and discontinuity.

The identification of episodes gives rise to various problems depending on the type of materials; even in the most straightforward case of standard written text, the relationship between the orthographic paragraphs and episodes is not a simple match, and spoken discourse or historical data in which scribes have filled the whole space of writing materials available involve a more challenging task. In studying historical materials with no or very few consistent graphic signals of unit boundaries, we have to rely on text-internal criteria to identify episode boundaries. For narrative texts, we can make use of the continuities of time, place, and participants, which are prominent structuring features of narrative (cf. the unities of time, place, and action, attributed to Aristotle; cf. e.g., 1954). So, breaks in temporal continuity can be signalled by temporal expressions and tense-aspect forms moving the story ahead or jumping backwards or forwards in time; the continuity of participants can be broken by introductions of new participants into the story or reintroductions of participants already present; and breaks in the continuity of place (and action) can be marked by locative expressions and reorienting verbs indicating shifts of place. Such breaks in continuities, or discontinuities are good signs of episode boundaries (Enkvist 1987a, Enkvist 1987b, Givón 1983, Virtanen 1992; Chafe 1979, Ji 2002).

Even smaller units can be problematic in historical data, in which punctuation can differ drastically from modern conventions. As it is not always easy, or even possible in all cases, to determine unambiguously what constitutes a “sentence” in Old English texts, I have opted for the “clause” as the basic unit of analysis, focusing on main clauses in the study presented here. These basic units are identical with sentences in the case of simple sentences, consisting of one main clause only, and complex sentences, consisting of one main clause and the subordinate clause(s) embedded in it, but they do not match in the case of compound sentences, consisting of main clauses co-ordinated with each other. Defining the basic unit as a clause avoids making arbitrary decisions about sentence boundaries for strings of clauses co-ordinated with and, which are very common, and long, in many Old English narratives. [5]

The repertoire of potential signals of text and discourse structure is huge: for a textualist, everything in language is, more or less, at the service of text and discourse, from case marking (e.g., intraclausal cohesion, grounding as salience) to word order (information structure as cohesion, grounding and chunking), and to items listed as connectives, conjunctions, discourse markers, pragmatic markers, or particles (with functions at different levels or planes). This study concentrates on signals of continuities: participant continuity measured by topicality/givenness and participant status, and continuity of place assessed in terms of verb types.

1.3 Discourse functions of þa: multifunctionality

Let us begin our investigation of the discourse functions of þa with a text which is often one of the first texts met by students approaching Old English: the Ohthere interpolation in the Old English version of Orosius’ Historiae adversus paganos. Example (9) is divided into main clauses and clauses with þa , the latter being marked by capital letters and darker background.

(9a)

Ohthere sæde his hlaforde Ælfrede cyninge þæt he ealra Norðmonna norþmest bude.[...]

‘Ohthere said to his lord, King Alfred that he lived northernmost of all the Northmen.’

Click to close.

(9a)

Ohthere sæde his hlaforde Ælfrede cyninge þæt he ealra Norðmonna norþmest bude.[...]

‘Ohthere said to his lord, King Alfred that he lived northernmost of all the Northmen.’

(9b)

cwæð þæt he bude on þæm lande norþweardum wiþ þa Westsæ.

‘He said that he lived in the land northward along the Western Sea.’

Click to close.

(9b)

cwæð þæt he bude on þæm lande norþweardum wiþ þa Westsæ.

‘He said that he lived in the land northward along the Western Sea.’

(9c)

He sæde þeah þæt þæt land sie swiþe lang norþ þonan; Ac hit is eal weste, buton on feawum stowum styccemælum wiciaþ Finnas, on huntoþe on wintra, and on sumera on fiscaþe be þære sæ.

‘He said, however, that the land is very long to the north from there; But it is all waste, except that Sámi people camp out in a few places here and there – by hunting in winter and fishing in summer along the sea.’

Click to close.

(9c)

He sæde þeah þæt þæt land sie swiþe lang norþ þonan; Ac hit is eal weste, buton on feawum stowum styccemælum wiciaþ Finnas, on huntoþe on wintra, and on sumera on fiscaþe be þære sæ.

‘He said, however, that the land is very long to the north from there; But it is all waste, except that Sámi people camp out in a few places here and there – by hunting in winter and fishing in summer along the sea.’

(9d)

He sæde þeah þæt he æt sumum cirre wolde fandian hu longe þæt land norþryhte læge, oþþe hwæðer ænig mon be norðan þæm westenne bude.

‘He said that on a certain occasion he wanted to find out how long the land extended to the north, or whether any man lived north of the wasteland.’

Click to close.

(9d)

He sæde þeah þæt he æt sumum cirre wolde fandian hu longe þæt land norþryhte læge, oþþe hwæðer ænig mon be norðan þæm westenne bude.

‘He said that on a certain occasion he wanted to find out how long the land extended to the north, or whether any man lived north of the wasteland.’

(9E)

Þa for he norþryhte be þæm lande;

‘Then he travelled northwards along the land.’

Click to close.

(9E)

Þa for he norþryhte be þæm lande;

‘Then he travelled northwards along the land.’

(9f)

let him ealne weg þæt weste land on ðæt steorbord & þa widsæ on ðæt bæcbord þrie dagas.

‘All the way he kept the wasteland on his starboard, and the open sea on his port side for three days.’

Click to close.

(9f)

let him ealne weg þæt weste land on ðæt steorbord & þa widsæ on ðæt bæcbord þrie dagas.

‘All the way he kept the wasteland on his starboard, and the open sea on his port side for three days.’

(9G)

Þa wæs he swa feor norþ swa þa hwælhuntan firrest faraþ.

‘Then he was as far north as the whalehunters farthest travel.’

Click to close.

(9G)

Þa wæs he swa feor norþ swa þa hwælhuntan firrest faraþ.

‘Then he was as far north as the whalehunters farthest travel.’

(9H)

Þa for he þa giet norþryhte swa feor swa he meahte on þæm oþrum þrim dagum gesiglan.

‘Then he travelled north still as far as he could sail in the next three days.’

Click to close.

(9H)

Þa for he þa giet norþryhte swa feor swa he meahte on þæm oþrum þrim dagum gesiglan.

‘Then he travelled north still as far as he could sail in the next three days.’

(9I)

Þa beag þæt land þær eastryhte, oþþe seo sæ in on ðæt lond,

‘Then the land turned eastward there, or the sea into the land,’

Click to close.

(9I)

Þa beag þæt land þær eastryhte, oþþe seo sæ in on ðæt lond,

‘Then the land turned eastward there, or the sea into the land,’

(9j)

he nysse hwæðer

‘he did not know which,’

Click to close.

(9j)

he nysse hwæðer

‘he did not know which,’

(9k)

buton he wisse ðæt he ðær bad westanwindes & hwon norþan

‘but he knew that he waited there for a wind from the west and a little from the north,’

Click to close.

(9k)

buton he wisse ðæt he ðær bad westanwindes & hwon norþan

‘but he knew that he waited there for a wind from the west and a little from the north,’

(9L)

& siglde ða east be lande swa swa he meahte on feower dagum gesiglan.

‘and sailed then east along the land as far as he could sail in four days.’

Click to close.

(9L)

& siglde ða east be lande swa swa he meahte on feower dagum gesiglan.

‘and sailed then east along the land as far as he could sail in four days.’

(9M)

Þa sceolde he ðær bidan ryhtnorþanwindes, for ðæm þæt land beag þær suþryhte, oþþe seo sæ in on ðæt land,

‘Then he had to wait for a wind directly from the north, because the land turned southward there, or the sea into the land,’

Click to close.

(9M)

Þa sceolde he ðær bidan ryhtnorþanwindes, for ðæm þæt land beag þær suþryhte, oþþe seo sæ in on ðæt land,

‘Then he had to wait for a wind directly from the north, because the land turned southward there, or the sea into the land,’

(9n)

he nysse hwæþer.

‘he did not know which.’

Click to close.

(9n)

he nysse hwæþer.

‘he did not know which.’

(9O)

Þa siglde he þonan suðryhte be lande swa swa he mehte on fif dagum gesiglan.

‘Then he sailed from there due south along the land as much as he could sail in five days.’

Click to close.

(9O)

Þa siglde he þonan suðryhte be lande swa swa he mehte on fif dagum gesiglan.

‘Then he sailed from there due south along the land as much as he could sail in five days.’

(9P)

Đa læg þær an micel ea up in on þæt land.

‘Then a great river extended there up into that land.’

Click to close.

(9P)

Đa læg þær an micel ea up in on þæt land.

‘Then a great river extended there up into that land.’

(9Q)

Þa cirdon hie up in on ða ea, for þæm hie ne dorston forþ bi þære ea siglan for unfriþe, for þæm ðæt land wæs eall gebun on oþre healfe þære eas. (Or 1, 1.13.29–14.18)

‘Then they turned up into the river, because they dared not sail forward across the river because of hostilities, because the land was all settled on the other side of the river.’

Click to close.

(9Q)

Þa cirdon hie up in on ða ea, for þæm hie ne dorston forþ bi þære ea siglan for unfriþe, for þæm ðæt land wæs eall gebun on oþre healfe þære eas. (Or 1, 1.13.29–14.18)

‘Then they turned up into the river, because they dared not sail forward across the river because of hostilities, because the land was all settled on the other side of the river.’

This text has many instances of þa and it is not exceptional in this respect. Þa is sometimes so frequent that its frequency is taken to mean that it cannot have any, or at least not much content. Many early scholars of Old English have said so and such views are occasionally found even now. Þa is characterized as a ‘cover’ for verb-initial order (German Deckwort), a prompt for getting the verb in a position before the subject.

an introductory word [..] this is merely a formal sign, and instead of being the cause of inversion, is perhaps the result (McKnight 1897:187–188)

Une phrase comme þa ridon hie þider ne serait donc qu’une phrase à verbe initial: ridon hie þider, à laquelle on a adjoint þa, comme on peut mettre her devant n’importe quelle phrase, sans que sa construction soit modifiée. (Fourquet 1938: 68)
‘A clause like þa ridon hie þider would thus be only a verb-initial clause ridon hie þider to which one has appended þa, like one can put her in front of any clause without modifying its structure.’

nothing more than a syntactic slot filler (Breivik 2002: 49)

Such empty words, occurring so frequently in a text have been seen as indicators of a simple, primitive style. Probably the best example of this view is found in Andrew’s study of Old English syntax and style (Andrew 1940/1966). He discusses a great number of instances where þa introduces two or more consecutive clauses and argues that they must be interpreted as complex sentences with subclauses headed by the conjunction þa and a main clause introduced by adverbial þa (ibid.: 3–18). In conclusion he presents a model of word order in which no instances of clause-initial þa are unambiguously adverbial (ibid.: 18).

Instead of accepting that the high frequency of þa is a sign of undeveloped style and proof of the emptiness of the word, historical syntacticians and text and discourse linguists have identified many different functions that þa has both at the syntactic level and at the level of discourse in Old English. In this study I focus on hypotheses about the discourse level functions of þa as a signal of continuity vs. discontinuity, and thus the syntactic hypotheses will not get the attention that they deserve, but will only be referred to when they directly affect the hypotheses about the discourse functions.

In some early studies of Old English syntax, þa was characterized as a signal of ‘expressive type’ (Fourquet 1938), demonstrative order (Andrew 1940/1966), and marked order (Bacquet 1962), referring to the frequent occurrence of initial þa followed by verb-subject word order. More recently, historical syntacticians have given þa, together with certain other items, an important role in language change: the behaviour of clause-initial þa is taken as evidence for the verb-second stage of English, and it is argued that þa functioned as a catalyst in the change from verb-final to verb-medial order and in the differentiation of main clauses and subclauses (Stockwell 1984, Stockwell & Minkova 1991). In the same vein, other historical syntacticians have seen the role of clause-initial þa in the development of hypotaxis from parataxis as related to clause-typing particles in other early Germanic languages (van Kemenade & Los 2006 and references therein). Though most research has dealt with clause-initial þa, other positions, too, have turned out to be relevant: studying the position of þa and the types of subjects, van Kemenade found that clause-internal þa functions as a discourse partitioner, marking the boundary between given or discourse-linked information and new information in certain types of clauses (van Kemenade 2009). [6]

Even as a practically empty word, þa nevertheless seems to contribute something to the effects and functions of its context of occurrence. As just a way of inducing subject-verb inversion, þa is associated with the stylistic and rhetorical force of that word order. [7] Verb-initial order has been attributed similar senses, and when clause-initial þa is seen as a mere ‘cover’ for verb-initial order, it shares these senses, as it does with inversion. Thus the inverted verb-subject order has been characterized as the expressive type, the demonstrative and marked order, and associated with vividness, emphasis, and relative stress (Bacquet 1962, McKnight 1897: 44, Dorgeloh 1997, Petrova & Solf 2008). In another line of research, the verb-subject order has been ascribed an introductory function: it is the word order found in episode-initial clauses (Hopper 1992) and in presentative constructions, introducing new participants (Breivik 2002) (cf. also Traugott 1992: 277–279). However, not all researchers attribute these effects to word order alone, but instead prefer to give some of the credit to þa. Enkvist (1972) describes þa as an action marker and Enkvist and Wårvik characterize its functions as foregrounding in narrative texts (Enkvist 1986, Enkvist 1994, Enkvist & Wårvik 1987, Wårvik 1990; on grounding see Wårvik 1996/2006). Likewise, in Hopper’s (1992) model for Old English narrative structure, þa is the element that calls for inversion at the beginning of an episode, and Breivik (2002) cites cases where þa and ðær ‘there’ appear interchangeable in presentative constructions.

The short example text in (9) is enough to illustrate the association of þa with inversion: eight of the nine clauses with þa have inverted word order: (9E), (9G), (9H), (9I), (9M), (9O), (9P) and (9Q); the only exception, (9L), is a co-ordinated clause sharing its subject with the preceding clause, for which subject-verb inversion does not apply.

Evaluating vividness, expressivity, emphasis, and other such effects is problematic and it is difficult to identify the source of the effect and separate the part played by þa from the role of word order in producing the effect. One possible method is to study the verbs in clauses with þa, clauses with inversion and clauses with both þa and inversion, and look for co-occurrence patterns of verb types with þa and word order. When we take the main clauses that have both subject and verb in the example text and divide them according to their word order into subject-verb (SV) and verb-subject (VS) clauses, we get the following distribution of verbs (capital letters refer to clauses with þa).

The example text in (9) first shows some negative support for the vividness hypothesis, as the clauses with subject-verb order (SV-clauses) can hardly be taken as vivid: the verbs we find in them are sæde ‘said’ in (9a), (9c) and (9d), cwæð ‘said’ in (9b), nysse ‘knew not’ in (9j) and (9n), and wisse ‘knew’ in (9k). The clauses with inversion (VS-clauses), which all begin with þa, present a more varied list of verbs. One group, for ‘travelled’ in (9E) and (9H), siglde ‘sailed’ in (9O), cirdon ‘turned’ in (9Q), and beag ‘turned’ in (9I), can be taken as dynamic actions on the journey, though beag ‘turned’ in (9I) is not a real action, as it is a description of the land turning. The rest of the verbs, wæs ‘was’ in (9G), sceolde bidan ‘should wait’ in (9M), and læg ‘lay’ in (9P) are definitely not vivid actions. Still, dynamic verbs represent a small majority of the verbs in the þa-clauses.

The vividness associated with inversion and the action signalled by þa are not directly involved in the investigation of connectivity, but they need to be mentioned, because they are the starting-points of the studies of the discourse functions of þa and its role in contributing to connectivity in text.

Þa has been characterized as a signal of unit boundaries, which puts it in the group of markers of disconnectivity or discontinuity. One hypothesis is that þa marks sentence boundaries, insofar as we can talk about sentences in Old English texts. In a study of Ælfric’s Lives of Saints, Waterhouse (1984) found that 45 per cent of units that can be characterized as sentences have þa at or near the beginning; however, the strength of this hypothesis is seriously weakened by the fact that the proportion of the sentence units introduced by þa varies between 25 per cent in some Lives and 63 per cent in others. By looking at the examples presented, we can find two further arguments against the hypothesis of þa as a sentence marker. First, as shown by the beginning of the passage in (9), þa is firmly bound to narrative contexts and thus it cannot mark discourse-level units in other types of textual material. Secondly, þa is also occasionally found in non-independent main clauses, like (9L) in (9), which is a co-ordinated main clause without a subject of its own, and in subordinate clauses, like those in (2) and (3).

When þa is associated with discontinuities at a higher structural level, it has been taken to express topic discontinuity (Kim 1992) and as already noted, to mark episodes (Foster 1975, Hopper 1992) and function in presentative constructions (Breivik 2002). In another view, the text-structuring functions of þa have been linked to its foregrounding function, so that þa participates in marking narrative structure by absence from episode boundaries at higher levels, where backgrounded material (setting, new participants) can signal beginnings of episodes (Enkvist & Wårvik 1987). In this view, þa is associated with continuities and its association with discontinuities arises from its cooperation with other markers of text structure.

As noted earlier, Old English texts are not graphically divided into paragraph units, and we are obliged to rely on text-internal criteria for capturing the hierarchical structure of the text. The criteria employed in the present study are participant continuity and continuity of place. There is a risk of circularity in using text-internal criteria for determining text structure, because the same expressions that form the basis of our identification of the structure may turn up as signals of that structure. Thus, temporal continuity cannot be used as a criterion for dividing the stories into episodes when the study has the purpose of investigating the role of þa in text-structuring, because þa is a temporal adverb.

When we look at the participants in the example story, we find some cases of new participants being presented, while the story mostly focuses on one main participant. The main participant, Ohthere, is introduced in a SV-clause in (9a). New participants, or rather props, appear in þaVS-clauses twice, in (9I): Þa beag þæt land þær eastryhte ‘Then the land turned eastward there’ and (9P): Đa læg þær an micel ea up in on þæt land ‘Then a great river extended there up into that land’. These two clauses support the hypothesis of þaVS as a presentative construction, though neither of the participants introduced in these clauses plays any active role in the narrative. But most of the þa-clauses have old subjects: they refer to the main participant of the story, Ohthere, who is referred to as he in (9E), (9G), (9H), (9M), and (9O), and by zero anaphora in (9L), which shares the subject he with (9k), and who is included in hie ‘they’ in (9Q). Thus the hypothesis of þa as a presentative form does not get support from this small sample, nor does the hypothesis of þa as a marker of participant discontinuity.

Yet, the hypothesis of þa as a signal of episode beginnings deserves to be investigated more closely and with special attention to participant continuity. The majority of þa-clauses in this short narrative have as their subject the main participant of the story. There are only two exceptions: clauses (9I) and (9P), which have different subjects and which present new participants as subjects. If we wish to find episodes in this narrative, þa alone is not a very good signal of episode beginning, unless we are prepared to accept very short episodes, but when we combine the occurrence of þa with participant continuity, we get units that could be taken as episodes – at least in this short narrative. Ohthere’s journey begins in (9E), and (9I) and (9P) indicate the stages of the journey by referring to landmarks. As we shall see later, the system is more elaborate than suggested by this example, but this illustrates the basic principles.

Another argument for the view of þa as a discontinuity marker comes from co-occurrence patterns of þa with certain types of verbs and word orders. Hopper, who has investigated word order and text structure in Old English and other languages, studied the Anglo-Saxon Chronicle (the annals up to 891 in MS A) and identified the following functions of word orders in this text (Hopper 1992: 218):

Episode-initial: (and) þa Verb Subject
Episode-internal: Verb-final (Object Verb)
Background: Subject Verb

Hopper argues that þa together with the verb-subject order marks the beginning of a new episode. What is essential in this pattern is the inverted verb-subject order, while þa at the beginning of the episode may be ‘amplified’ or even replaced ‘by a full adverbial phrase’, such as her ‘here’ (Hopper 1992: 219). To test his hypothesis, Hopper looks at verbs occurring in different functional slots in the episodes and observes that in his data þa typically occurs together with verbs indicating some kind of shift of orientation: motion in a direction (come, go, send), static location (stay, remain), identified subject (come to throne, die) or newly introduced patient (find, take). According to Hopper, “[p]erhaps even more striking are the verbs which are virtually excluded from occurring after þa. These are ordinary transitive verbs in which an Agent actively affects a Patient” (Hopper 1992: 221). Such verbs occur in the episode-internal clauses.

The Ohthere interpolation in (9) is both stylistically and content-wise close to the Chronicle, and thus it is not surprising to find the same kinds of reorienting verbs in the þa-clauses: verbs of motion: for ‘travelled’ in (9E) and (9H), siglde ‘sailed’ in (9L) and (9O), beag ‘turned’ in (9I) and cirdon ‘turned’ in (9Q), and verbs of static location and position: bidan ‘wait’ in (9M), wæs ‘was’ in (9G) and læg ‘extended’ in (9P). As noted earlier (Table 1), only (9L) does not have the verb-subject order. The only transitive verb in the þa-clauses is bidan ‘wait’, which is not ‘ordinary’ in the sense of ‘an Agent actively affecting a Patient’. However, this particular story deals with a journey and thus verbs expressing motion and location on the journey are only to be expected. There are no clauses in (9) illustrating the episode-internal verb-final order, but we find the subject-verb order for background in (9a), (9b), (9c), (9d), (9j), (9k), and (9n).

In addition to contributing to signalling episode boundaries, þa has been attributed a more connective kind of structuring function: its task is to mark foreground or sequential story-line in the narrative text (Enkvist 1986, Enkvist & Wårvik 1987, Wårvik 1990) or, in a different terminology, it works as an additive and transitional connective (Lenker 2010). When we consider the clauses with þa in our example text, we get the progress of Ohthere’s journey:

Such a string of clauses is not a very exciting account of the journey, but just the main line of narrative, from which all background material, i.e. setting, descriptions and explanations, have been left out. For short stories like this one, the function of marking the main line of the narrative does not appear as obviously essential, but for longer narratives, keeping track of the progress of the story is a structurally central function (on grounding and story-line sequentiality see e.g. Hopper & Thompson 1980, Wårvik 1996/2006). [8]

The short narrative of Ohthere’s journey can illustrate the hypotheses about the functions of þa and give us a starting-point for evaluating the conflicting views. Before we turn to more data for judging whether þa should be seen as a connective or a “disconnective” discourse marker, I present the data used in the study.

1.4 The data

The studies reported in the statistics are based on a sample of 4,116 main clauses of Old English narrative prose excerpted from four works. The sample is presented in Table 4.

All the four works from which the samples are taken have narrative as their dominant text type and represent the historical and the religious registers. The Anglo-Saxon Chronicle is the most purely narrative of the works: as a report of the events relevant to England during the years concerned, it contains only very short pieces of other kinds of textual material. The longer sample, the annals up to 891, consists of the earliest part of the Chronicle, all written in one hand in the A-manuscript (Bately 1986). The second extract from the Chronicle, is a convenience sample: it consists of the annals that are as contemporary as possible with the rest of the later materials in the sample, i.e. the period 950–1050, CO3 in the Helsinki Corpus. [10] The other historical text, the Old English Orosius, is a history of the world, which combines geographical description and argumentative commentary with historical narrative (Bately 1980). The Ohthere and Wulfstan interpolations are two travel reports by contemporary travellers inserted in the geographical introduction (Bately 1980). The main sample from the Orosius is taken from the beginning of the historical part of the work. The bulk of the later materials is taken from Ælfric’s Catholic Homilies, composed around 1000 (Godden 1979, Godden 2000). The sermons in the sample are selected so that they represent Ælfric’s narrative and instructive writing in the two collections as equally as possible; obviously, as sermons, all of the texts are instructive, but all of them also contain narrative parts, varying in extent and number (cf. Wårvik 2009).

It could be argued that most of these texts are invalid for studying the structure of Old English narrative prose, because they can be considered translations. The Anglo-Saxon Chronicle is the only purely native English text. Ælfric relied on Latin sources in composing the Catholic Homilies. The Orosius is based on a Latin original; even the interpolations were most likely first presented in another language than English. However, these texts would hardly be called translations by modern standards: they are mostly rather unfaithful translations, and more accurately characterized as paraphrases or adaptations. Because of the problems, or impossibility, of finding the Latin originals of Old English texts, the nature of the translations cannot be fully studied, but several scholars investigating specific constructions or stylistic features have commented on the relative independence of the works from Latin, among them, Clemoes 1966, Jones 1999 and Waterhouse 1992 for Ælfric’s works, and Liggins 1986 and Traugott 2007 for the Orosius. Furthermore, since there is nothing in Latin that resembles Old English adverbial þa in distribution or function, I have some confidence in using these texts to study discourse-pragmatic features of Old English narrative prose.

2. Discourse markers, continuity and discontinuity in Old English narrative prose

The different hypotheses that have emerged in the research on the discourse functions of Old English adverbial þa can be grouped into two conflicting sets of views: the connective view, which sees þa as a marker of continuity in discourse and the “disconnective” view, according to which þa signals discontinuity. In this section, I will discuss these hypotheses in relation to other signals of text structure, with the aim of finding a common core in these conflicting views. To show how the discourse functions of þa make it different from other time adverbs in the same semantic field, it will be compared to two closely related items, þonne ‘then’ and siþþan ‘afterwards’ (BT s.v. þanne, s.v. siþþan); both of these words have the potential of functioning as discourse markers in narrative text.

2.1 Þa as marker of participant continuity vs. discontinuity

One of the bases for the arguments about the functions of þa concerns the topicality of subjects in clauses in which þa occurs. The term “topicality” is here used both for the assumed familiarity of the subject (Prince 1981) and its continuity, or more exactly its forward continuity or persistence in the text (Givón 1983). Both kinds of topicality are encoded in the form of the noun phrase, according to an iconically motivated scale on which the degree of topicality correlates with the length or weight of the noun phrase, so that the highest degree of topicality is encoded by zero anaphora and the lower degrees with an increasing amount of material (cf. the topic accessibility scale by Givón 1983 and the givenness hierarchy by Gundel, Hedberg & Zacharski 1993). The more topical, in the assumed familiarity sense, an entity is, the easier it is to identify and the shorter the form needed to refer to it, the extreme case being zero anaphora, which can be used when the entity is so unambiguously present in the immediate textual context that it does not need to be identified by any referring expression. At the other end of the scale, entities lowest in topicality need to be introduced as “brand-new” in the discourse, typically by an indefinite noun phrase (the terms are from Prince 1981). The forms in the middle of the topicality scale, definite noun phrases, proper names, and pronouns, can be used for entities that are recoverable from a wider textual or situational context. Entities that are introduced earlier in the discourse can be identified by pronouns when they are “current-evoked”, that is, active in the interlocutors’ consciousness so that their referents can be unambiguously identified. When they are “displaced-evoked”, that is given but no longer active, they need to be identified by longer and more informative forms, and thus definite noun phrases and proper names are used. A simplified scale for the purposes of this study consists of the categories in Table 5 (examples are from the translation of text (9)):

Topicality and the choice of appropriate referring expressions are of course an integral part of cohesion, keeping the text together, but topicality also plays a role in grounding, as one parameter differentiating foregrounded, main-line material from backgrounded, subsidiary material. High topicality of the subject is one of the features associated with foregrounding and story-line sequentiality: narratives by definition consist of actions performed by and events happening to certain main characters, which typically appear as given or old information in the text. As a foreground marker, þa is expected to co-occur with subjects that are high on the topicality scale (Hopper & Thompson 1980, Wårvik 1996/2006). However, some studies have found that þa is not associated with highly topical subjects, but with those lower down on the topicality scale. Thus, in her study of the West-Saxon Gospels, Kim (1992) argues that þa marks different kinds of discontinuities and co-occurs with topics low on the topicality scale: full definite noun phrases and stressed pronouns. [11] This finding tallies with the view of þa as an episode marker, having a disconnective function (Breivik 2002, Foster 1975, Hopper 1992). In my study of topicality and participant status in a sample of Ælfric’s Catholic Homilies, it turned out that þa was not associated with just any definite noun phrases, but with those that were used to refer to the main participants when they were reintroduced into the story: thus it could be seen as marking a specific kind of continuity of the participants (Wårvik 1994).

The examples cited above show that þa can occur with subjects at different points on the topicality scale, but as illustrated by Ohthere’s story in example (9), þa-clauses prefer to have the main participant as their subject. Dividing the subjects in the whole data sample into five rough categories on the basis of their form, we find distributions that are at first sight contradictory: þa prefers definite noun phrases and pronouns and disprefers zero anaphora; þa is slightly less common with clausal and impersonal subjects and it is indifferent to indefinite noun phrase subjects (Table 6).

There are significant positive correlations between þa and definite noun phrases and pronouns, whereas zero anaphora has a significant negative correlation with þa (Table A1 in the Appendix). Þa thus prefers forms associated with old or familiar participants as subjects, but this preference concerns only explicit subjects, while zero anaphora, which is the most given type of subject, is an exception. These patterns tally with the findings suggesting that þa reintroduces main participants (Wårvik 1994).

Before we look for more explanations for this anomaly, let us compare þa to two time adverbs that are semantically very close to it: þonne ‘then’ and siþþan ‘afterwards’ (BT s.v. þanne, s.v. siþþan). Both are high-frequency items, though they are much less frequent than þa (Venetzky & Butler 1985). [12]

For þonne, we find the same kinds of preferences of subject type as for þa: definite noun phrases and pronouns are more frequent than the other categories (Table A2). There is a positive correlation between the occurrences of þonne and pronoun subjects and a strong negative correlation between þonne and zero anaphora as subject (Table A1). Þonne thus appears to have a more limited association with given types of subjects, but the same kind of ‘repulsion’ relationship with zero subjects as þa has. [13] Siþþan differs from þa and þonne in occurring equally often with different explicit subjects, but what distinguishes siþþan from the other two time adverbs is its significant positive correlation with zero anaphora (Table A1 and A3). By these measurements, siþþan is the most continuous of the three adverbs, while þa and þonne are associated with subjects in the middle of the topicality scale and thus a slightly lower degree of continuity.

When we look at the contexts of the three adverbs, another parameter emerges as relevant: co-ordination. Siþþan is significantly more common in co-ordinated main clauses, while both þa and þonne are significantly less common in them (Table A4). As zero anaphora is the only type of subject that has a positive correlation with co-ordinated clauses (Table A5), it is not unexpected to find an association between siþþan and zero anaphora. However, co-ordination is also relevant for continuity.

The measurements so far have only taken into account the form of the subject and its assumed familiarity, rather than the continuity of participants in the text. Continuity as a more dynamic notion can be studied as the lookback and the persistence of a participant: lookback refers to the distance to the previous mention of the participant and persistence to how long the participant remains active. Both kinds of continuity can be counted as the number of clauses between the mentions of a referent. In the study reported in Tables 7 and 8, distance is counted as the length of the chain of co-ordinated clauses following a specific clause and sharing the same subject – the chain is broken when an explicit subject is used in a clause. [14] Table 7 presents the figures for continuity in clauses with different types of subjects.

Though the numbers are fairly small because most clauses are not followed by co-ordinated same-subject clauses (two or more in 14.7 per cent and three or more in 3.7 per cent of all clauses), there are still significant differences between the categories. Definite noun phrase subjects and pronoun subjects are more continuous and more forward-looking, as shown by significant positive correlations between these categories, while indefinite noun phrase subjects are slightly less continuous, though their negative correlation with length is not significant; zero anaphora and empty subjects fairly obviously correlate negatively with participant continuity measured as length of chains (Table 7 and A6).

Turning to þa, we find that participant continuity is higher in clauses with þa than in other clauses. Again, even if the numbers reported in Table 8 are small, the differences are significant. There is a significant positive correlation between the occurrence of þa and the continuity of the subject measured as persistence (Table A7). Þonne and siþþan, which are much less frequent than þa, do not show any significant preferences as to partipant continuity of subjects, which suggests that they do not participate in this aspect of discourse structuring as actively as þa (Tables A7, A8, and A9).

One more parameter needs to be addressed in this section: word order. The position of þa in the clause and its effect on the order of elements have been thoroughly investigated and discussed in the literature on Old English syntax (cf. e.g., van Kemenade 2009, van Kemenade & Los 2006, Kohonen 1978, Mitchell 1985, Petrova & Solf 2008, Pintzuk & Haeberli 2008, Stockwell 1984, Traugott 1971, Traugott 1992, and Trips & Fuss 2009). In the present study word order is considered only as one parameter in evaluating the connective function of þa, and thus word order is taken into account only as the relative order of the subject and the finite verb, focusing on the discourse functions of inversion, as a presentative order and as a signal of vividness (cf. above in 1.3 and note 7).

The general tendency of the subject to follow the finite verb in clauses with þa holds for this material: of the þa-clauses that have both a subject and a finite verb, 65 per cent have the inverted verb-subject order and the majority of them have þa in initial position in the clause (Table A10). But we should also note that the majority of the clauses with inversion do not have þa and 39 per cent of the cases of þa occur in clauses with other word orders. Of the other two adverbs, þonne displays a similar pattern to þa, with an even higher proportion, 89 per cent, of þonne-clauses having the finite verb before the subject (Table A11 and A12). Siþþan shows no significant correlations with either word order type, though most siþþan-clauses are of the Subject-Verb type (Table A11 and A13). Thus word order preferences do not reveal any differences between þa and þonne at this level of discourse structure, but again siþþan turns out to be different (for more local level discourse functional similarity between þa and þonne, see discourse partitioners in van Kemenade 2009).

As noted earlier, one of the functions attributed to inversion is related to participant continuity, namely the presentative function of introducing new participants. The figures in Table 9 show a higher proportion of indefinite subjects and subjects of the ‘other’ type with Verb-Subject-order, while definite noun phrases and pronouns are more common as subjects in clauses with Subject-Verb-order. However, when we look at correlations between subject type and word order in the whole material (Table A14), we find that definite noun phrases correlate positively and ‘other’ subjects negatively with both word order types, while for indefinite noun phrases the correlations show a strong preference for Verb-Subject-order and for pronouns we find a positive correlation with Subject-Verb order.

Considering only subject type and word order, we find these results perfectly in line with studies treating pronouns as light elements and even clitics, placed before the verb, and with research pointing to the discourse dependence of word order in Old English, especially the principle of end-weight adhering to the given-new order of information structure (Kohonen 1978, van Kemenade & Los 2006). We can now remind ourselves of the preferences of þa as to subject type and word order. In the data for the present study, þa was found to correlate positively with definite noun phrases, pronouns and Verb-Subject-order, but not to have any particular affinity to indefinite noun phrase subjects, which for their part correlate positively with Verb-Subject-order (cf. Table 6 and A1). This discrepancy can be explained by arguing that Verb-Subject-order is presentative on its own, rather than that þa functions as a presentative (Breivik 2002). The subject types that correlate positively with þa, definite noun phrases and pronouns, are typically used for familiar or given participants and thus these findings support the view of þa as a connective.

To sum up, the results so far support the hypothesis that þa has functions which are more tightly associated with continuity than discontinuity in text. Þa seems to prefer subject participants which are higher on the topicality scale to those lower down. Same subject participants encoded as zero anaphora are an exception, as þa rarely co-occurs with them. However, this pattern could be seen as a natural consequence of the role of þa as a signal of the continuity of the main story-line: signalling continuity with þa would be redundant in such highly continuous contexts; when þa is used with zero anaphora, it tends to emphasize the progress of the story-line (like þa (4) in example (1)). Another measurement of continuity, by the length of same subject chains, suggests that þa tends to occur with more persistent subject participants, thus signalling forward continuity. Taken together, these parameters show that þa is a signal of connectivity as it is more closely associated with high topicality (given subject participants) and participant continuity (persistence).

2.2 Þa as marker of action, episodes and story-line

The second disagreement about the functions of þa has its origins in views concerning the status of þa as an action marker. Though Enkvist’s original hypothesis of þa as an action marker (1972) has been developed into the hypothesis of þa as a foreground or story-line marker (Enkvist 1986, Enkvist & Wårvik 1987, Wårvik 1990), “action” still appears as a parameter in studies of the discourse roles of þa, evaluated as dynamicity and transitivity. Thus we find a conflict between the hypothesis of þa as a story-line signal, which associates it with dynamic actions and transitive verbs and with the continuity of the main story-line, and the hypotheses of þa as a presentative particle and as an episode marker, which associate it with intransitive and reorienting verbs (Breivik 2002, Hopper 1992).

In order to study the co-occurrence patterns of verbs of certain types with þa and the two other time adverbs, þonne and siþþan, the main verbs in all clauses in the data were divided into semantic categories. [15] Table 10 lists the numbers of occurrences for verb types that occur in more than one hundred clauses; the full list of categories is presented in the Appendix (Table A15).

These distributions show that verbs of motion are indeed the most frequent type with þa, but they are also generally common, and they are the most common type also with siþþan and the second most common with þonne. These preferences are supported and specified by correlations between the three adverbs and verb types (Table A16). We find significant positive correlations between þa and verbs of motion, between þonne and verbs of being, and between siþþan and verbs of motion and of staying. Significant negative correlations show the tendencies for ‘repulsion’ between these adverbs and certain verb types: þa avoids verbs of being, dying, and killing, þonne the company of ‘military’ verbs of fighting and ruling, and siþþan verbs of saying. When we look at significant correlations between individual verbs and these three adverbs, the top ten verbs, which occur more than fifty times in the data, show very similar patterns (Table A17). There are significant positive correlations between þa and cweðan ‘say’, fon to rice ‘accede to the throne’, and cuman ‘come’, but significant negative correlations with beon ‘be’, forþfaran ‘die’, ofslean ‘kill’, and healdan ‘hold’. A difference from þa is revealed by þonne’s positive correlation with beon ‘be’. The preference of siþþan for motion verbs is shown by a positive correlation with faran ‘go’. On the basis of these tendencies it is difficult to draw the conclusion that þa is associated with reorienting verbs to such a degree that this association would support the hypothesis of þa as a marker of discontinuity. Rather, the three verbs that have the strongest associations with þa form a group of ‘coming, saying, and acceding to the throne’. These verbs refer to activities and actions that can play a central role in the stories found in the data, though the specific patterns vary according to the genre of the text and, for instance, the high frequency of fon to rice ‘accede to the throne’ is due to its use in the Chronicle. [16]

From the analysis of verb types we can move to another criterion dependent on the type of the verb, namely the number of participants. In their study of co-occurrence patterns of morpho-syntactic features in several different languages, Hopper and Thomson (1980) found that transitivity, which is traditionally seen as a factor of the presence of object(s) in the clause, is rather a cluster of different parameters, one of which is the number of participants. [17] In his study of text-building strategies in Old English, Hopper (1992) found that þa tended to appear in episode-initial clauses with intransitive verbs or verbs low in Transitivity, i.e., verbs which do not express action, in the sense of dynamic actions by an agent affecting an object, while the episode-internal clauses depicted actions and thus had more transitive verbs, both in the traditional sense and in Hopper and Thompson’s sense of dynamic action affecting an object. As we saw earlier (Table 1), the Ohthere interpolation (example 9) supports Hopper’s hypothesis about the intransitivity and lack of action of the verbs in þa-clauses. However, more data reveals the opposite: þa has a preference for structures with two or more participants (Table 11; Chi square (df=2) 12.437, p<.002). Þonne, in contrast, prefers one-participant structures and dislikes those with more participants (Table A18), while siþþan appears indifferent to this parameter (Table A19). There are significant positive correlations between þa and two-participant structures and between þonne and one-participant structures, and conversely, significant negative correlations between þa and one-participant structures and between þonne and two-participant structures (Table A20).

When we compare the number of participants in clauses including þa and in clauses co-ordinated with þa-clauses, we find a difference that explains Hopper’s view of þa-clauses as intransitive. Changing the scope of þa, so that we include only clauses co-ordinated with þa-clauses, we find the pattern that Hopper found in his data: structures with two or more participants are significantly more typical of clauses co-ordinated with þa than of clauses including þa (Table 12; Chi square (df=1) 29.148, p<.000).

The data studied by Hopper may explain the conflict between his results and mine: the Anglo-Saxon Chronicle represents a specific genre, and the earlier annals that he focuses on often report only chains of events starting with movement or meeting and resulting in victory or defeat, such as in (10).

(10)

Þa foron hie to & gefliemdon þone here, & þæt geweorc abræcon, & genamon eal þæt þær binnan wæs, ge on feo. ge on wifum, ge eac on bearnum, & brohton eall in to Lundenbyrig, & þa scipu eall oðþe tobræcon, oþþe forbærndon, oþþe to Lundenbyrig brohton oþþe to Hrofesceastre. (ChronA 894.48)

‘Then they proceeded there and put the army to flight, and stormed the fortification, and took all there there was inside, both money, and women, and also children, and brought everything to London, and all the ships either broke to pieces, or burned up, or brought to London or Rochester.’

Such repeated patterns are not typical of the other texts in my sample, where we find a more varied set of verbs in the þa-clauses and the co-ordinated clauses following them, even in the short exemplary stories in Ælfric’s Catholic Homilies, such as (11).

(11)

Þes foresæda halga wer wæs gewunod þæt he wolde gan on niht to sæ. and standan on ðam sealtan brymme oð his swyran. syngende his gebedu; Þa on sumere nihte hlosnode sum oðer munuc his færeldes. and mid sleacrre stalcunge his fotswaðum filigde. oð þæt hi begen to sæ becomon; Ða dyde cuþberhtus swa his gewuna wæs. sang his gebedu on sælicere yðe. standende oð þone swyran. and syððan his cneowa on ðam ceosle gebigde. astrehtum handbredum to heofenlicum rodore; Efne ða comon twegen seolas of sælicum grunde. and hi mid heora flyse his fet drygdon. and mid heora blæde his leoma beðedon. and siððan mid gebeacne his bletsunge bædon. licgende æt his foton on fealwun ceosle; Þa cuðberhtus ða sælican nytenu on sund asende. mid soðre bletsunge. and on merigenlicere tide mynster gesohte; Wearð þa se munuc micclum afyrht. and adlig on ærnemerigen. hine geeadmette to ðæs halgan cneowum. biddende þæt he his adl eallunge afligde. and his fyrwitnysse fæderlice miltsode; Se halga ða sona andwyrde. Ic ðinum gedwylde dearnunge miltsige. gif ðu ða gesihðe mid swigan bediglast. oð þæt min sawul heonon siðige. of andwerdum life gelaðod to heofonan; Cuðberhtus ða mid gebede his sceaweres seocnysse gehælde. and his fyrwites ganges gylt forgeaf; (ÆCHom II, 10.83.74–93)

‘The aforementioned holy man had the custom that he would go at night to the sea, and stand in the salty water up to his neck, singing his prayers; Then one night another monk was on the watch for his going, and with stealthy tread followed his footsteps, until they both came to the sea; Then Cuthbert did as was his custom, sang his prayers on the waves of the sea, standing up to the neck (in the water), and afterwards bent his knees on the sand, with the palms of his hands stretched out to the heavenly firmament; Just then two seals came from the bottom of the sea, and with their fleece they dried his feet, and with their breath bathed his limbs, and after that with signs asked for his blessing, lying at his feet on the pale sand; Then Cuthbert sent the sea animals out to swim away, with a true blessing, and returned to the monastery in time for matins. The monk was then greatly frightened, and ill in the early morning submitted himself to the holy man’s knees, begging that he drive away his illness, and show mercy on his curiosity in a fatherly way. The holy man then at once replied: I show mercy on your error privately, if you keep that sight secret by silence, until my soul goes from the present life called to heaven; Cuthbert then healed his spy’s sickness with prayers and forgave the sin of his curious steps;’

To sum up, the co-occurrence patterns of þa with various verb types provide further support for the hypothesis of þa as a signal of connectivity and continuity in the text. The verbs that have the strongest associations with þa are all dynamic verbs, which we can expect to find in story-line clauses. The preponderance of verbs of movement is hardly surprising, because the main narratives in the historical texts and to some extent even in the sermons involve a lot of movement. As verbs of movement typically involve intransitive, one-participant structures, the preference of þa for transitive clauses with two or more participants may appear contradictory, but in fact it brings out more clearly the association of þa with dynamic verbs: though movement verbs are the largest groups, they are not the majority. The pattern that emerges from these two parameters put together is the preference of þa for verbs that are typical of the story-line, whether the story proceeds by movement, acceding to the throne or saying something.

3. Conclusion

Old English þa, the temporal adverb, is primarily a connective element, creating cohesion in the narrative text by signalling the progress of the story-line and by keeping track of the main participants of the story. Þa enters into the field of disconnectivity or discontinuity when it cooperates with other devices to signal narrative structure. But even then it can be seen as primarily connective, because its main task is to keep the narrative together and the story going. Thus the hypotheses of þa as an action marker (Enkvist 1972), as an episode introducer (Foster 1975, Hopper 1992), and as a signal of the main story-line (Enkvist 1986, Enkvist & Wårvik 1987) do not contradict each other, though they may at first sight seem to be mutually exclusive. They can all be included under the core function of marking the main line of the narrative. Even the most conflicting views of þa as an episode marker, expressing discontinuity, and of þa as a story-line marker, signalling continuity, can be consolidated when we see þa as marking those episode-initial clauses that are part of the main story-line. The patterns of participant reference support this view: the frequent co-occurrence of þa with definite noun phrases combines the discontinuity of episode marking with the continuity arising from the reintroduction of main participants. This interpretation of the functions of þa tallies with its classification as an additive and transitional connective (Lenker 2010). The co-occurrence patterns with verb types reveal a preference of þa for dynamic verbs, which are typical of the main-line of the narrative, though the repertoire of verbs obviously varies according to the needs of the story.

Discussing discourse marker meaning, Schourup (1999) comments on the interaction between the discourse marker and contextual factors:

Meanings conveyed by the entire utterance and assumptions accessed in assigning it an interpretation ‘leak’ as it were, into the proposed [discourse marker] core. (Schourup 1999: 250).

In the core of the discourse marker þa is the function of marking the main line of the narrative, and the meanings that ‘leak’ into that core are marking action and text structuring. Action and vividness are meanings that arise out of the cooperation between þa and verb types and word order, while the text structuring function and the presentative interpretation grow out of the interaction of þa with participant continuity and other temporal expressions. Figure 1 presents these functions and meanings in a graphical format.

Old English þa signals a specific kind of continuity in narrative text, combining story-line sequentiality and text structuring; it marks the continuation of (or return to) the main-line at different levels: the level of the temporal sequentiality of the story line (next, after that), the level of participant continuity (return to main participant), and the level of text structuring (next episode or next substory).

In the Middle English period, þa was affected by a number of changes, which finally led to its merger with the longer form þonne and eventually its almost total disappearance from standard language and formal contexts. However, the importance of the function of structuring narrative can be seen in the survival of inversion after then longer than in other contexts (Bækken 2000, Warner 2007) and in the use of then, in much the same way as þa, in simple stories in oral story-telling still today (cf. e.g., Gonzalez 2004, Schiffrin 1990). Though the reasons for the disappearance of the foregrounding form must be left for further research to look for, we can note some potentially relevant points. For one, þa was much more frequent than þonne, but it was also more frequently ambiguous both as an adverbial and as a conjunction – and it was þa as a conjunction that was first replaced by when. Secondly, þa was fairly highly specialized in its functions, while þonne had a wider sphere of uses. Thirdly, þa may also have been at a disadvantage because it was so short and thus easily eroded by phonological reduction.

The purpose of this study has been to try to reconcile the conflicting views of the discourse functions of þa as either a signal of discontinuity or a connective expressing continuity in the text. A detailed overview of the uses and functions of Old English adverbial þa can give insights into the pragmatics of narrative structuring in Old English, the division of labour between þa and þonne, two high-frequency words for ‘then’, and even the gradual disappearance of þa in favour of the longer form after this period and the variety of functions carried out by their Present Day English successor then. This study has been limited to examining contradictory characteristics in the co-occurrence patterns of þa with textually relevant parameters. The investigation has shown that it is possible to find a common denominator among the different hypotheses about the functions of this item: a core function of signalling the main line of the narrative and a number of peripheral functions that arise through cooperation with other parameters.

Acknowledgments

I am grateful for feedback on earlier versions of this article to the participants of the workshop “Connectives in synchrony and diachrony in European languages”, to professor Tuija Virtanen, to the editors of this volume and to the anonymous referee for the e-journal. I wish to thank Dr Martin Gill for language checking. I am of course responsible for all the remaining short-comings, many of which might have been avoided, if I had been able to take all their comments into account.

Notes

[1] This is the number of cases as given in the concordance itself; the figure in the rank-order list is 64,402, which still keeps þa in the third place. The most frequent word in the list occurs 113,087 times. It is the symbol for ‘and’, which appears in the list as &. The word and is sixth in the rank order list, with 53,150 instances. The second most frequent word is on, which occurs 83,905 times.

[2] The examples are cited from the editions listed under Sources. The abbreviated text titles follow the short titles in the Dictionary of Old English (DOE) and the references to the texts follow their system for each text. The list of texts cited in the dictionary is available at http://www.doe.utoronto.ca/st/index.html

The translations are meant to reflect the Old English text as closely as possible and they are thus, obviously, not idiomatic Present-Day English.

[3] Þonne appears in different spelling forms and thus its rank in the frequency lists cannot be determined as easily as the rank of þa. This number of occurrences includes the two most common spellings: þonne, 16,846 cases, and þanne, 417 cases (Venetzky & Butler 1985).

[4] Bately’s edition (1986) seems to skip one line as the text there does not contain the words hie ænigne feld secan wolden; Þa foron hie siþþan æfter. Thus the beginning of this example is cited from the electronic version of the Dictionary of Old English Corpus (DOEC; di Paolo Healey et al. (eds) 2000).

[5] The “sentence” issue is discussed for instance by Mitchell (1980, 1985: § 2546); cf. also Greenbaum 1979 for Present-Day English.

[6] I wish to thank the anonymous referee who evaluated my article for this publication for emphasizing the connection between the behaviour of þa in the information structure of the clause at the syntactic level and its function as a marker of narrative continuity. The position of þa is not considered in the study reported in this article, because the co-operation of the position of þa and the textually relevant parameters in this study is worth a more thorough investigation than could be included here.

[7] In the present study the term “inversion” is used to refer to all word orders in which the finite verb precedes the subject of the clause. Though the term is inaccurate for Old English word order from a syntactic perspective, it is appropriate for my purposes of studying the textual functions of þa. For one thing, as all the data are main clauses, the issue of different word order rules in main and subordinate clauses does not arise. Secondly, the aim is to investigate how þa functions on the discourse level and what gives rise to the conflicting hypotheses; thus occurrences of þa in all positions in the clause are included and the presence of þa is taken into account as one factor separate from word order and type of subject. Thirdly, word order and inversion are investigated only in main clauses which have both a subject and a finite verb; consequently, co-ordinated same-subject clauses with zero anaphora are excluded from these investigations. It should also be pointed out that co-ordination is not considered a separate factor, but is included only in the study of the lengths of co-referential chains in section 2.1.

[8] Keeping track of the progress of the story by marking the story-line is of course not a central concern in narratives where the story-line or plot is intentionally hidden or distorted for literary and artistic purposes (cf. e.g., Björklund & Virtanen 1991). Old English þa does not belong to the same story-telling tradition as such narratives.

[9] The periods given here follow those of the Helsinki Corpus (Kytö 1996). The date of the composition of the original text can rarely be determined exactly and the date of the extant version can usually be given only approximately. The matter is further complicated by the variation in the degree to which later copies reproduce or modernize their earlier originals. In this study, the periods are not used for diachronic comparisons, but only for describing the data.

[10] These annals are written in the extant copy by various scribes from the middle of the tenth century onwards (Bately 1986).

[11] Kim studies the ‘primary topic, i.e., the nominal which is most focused in a clause’ (1992: 120). This is typically the subject.

[12] Including the two most common spellings, siþþan and syþþan, we get a figure of 2,850 for this high-frequency item (Venetzky & Butler 1985). 

[13] Renouf & Banerjee define lexical repulsion as ‘the intuitively-observed tendency in conventional language use for certain pairs of words not to occur together’ (2007: abstract). I stretch the original meaning of the term slightly to cover even the collocational ‘dislikes’ of discourse markers with different types of subjects.

[14] These explicit subjects that break the same-subject chain include co-referential subjects, like the pronouns referring to Ohthere in the example text (9). Though according to this method the continuous chains are shorter than the chains including all co-referential items, evaluating all cases of potential co-reference would have involved a higher degree of subjectivity in the analysis and my decision was to include only the unambiguous cases of co-reference indicated by same-subject zero anaphora.

[15] The categories include verbs of similar semantic fields and situation types (cf. Vendler 1967). This list includes all the categories of verbs and examples of the most common verbs with approximate translations in each category; the category label is underlined. The full list of clauses divided into verb categories is found in Table A15.

Affecting an object without concrete change: geceosan ‘choose’, beswican ‘deceive’
Being: beon ‘be’
Beginning: onginnan ‘begin’
Changing an object or changing itself: awendan ‘turn, change’, gehælan ‘heal, save’, gehalgian ‘consecrate, ordain’
Continuing: þurhwunian ‘continue’
Creating: acennan ‘bring forth’, aræran ‘raise up’, don ‘do, make’, gesceppan ‘create’, wyrcan ‘work, make’
Destroying: forbærnan ‘burn up’, forhergian ‘plunder’
Dying: forþfaran ‘die’
Displaying: æteowian ‘appear, show’, getacnian ‘signify, witness’
Feeling: lufian ‘love’, ondrædan ‘fear’, þrowian ‘suffer’
Finishing: geendian ‘end, accomplish’, gefyllan ‘fulfil, complete’
Giving/taking: forlætan ‘permit’, niman, geniman ‘take, obtain’, sellan, gesellan ‘give, sell’, onfon ‘take, receive’, underfon ‘receive’
Happening: geweorðan ‘happen’, weorþan ‘become’
Having: gedafenian ‘behove’, habban ‘have’
Holding: healdan ‘hold’
Killing: ofslean ‘kill, strike’, slean ‘slay’
Military verbs of fighting and ruling; dynamic verbs whose end-points are specified in the context: feohtan, gefeohtan ‘fight’, fon to rice ‘accede to the throne’, hergian ‘ravage, make war’, ricsian ‘rule’, winnan ‘fight’
Moving: arisan, ‘rise’, cuman ‘come’, faran ‘go’, feran ‘go’, gan ‘go’, gecyrran ‘turn, return’, gewitan ‘go, depart’, metan, gemetan ‘meet’
Saying: biddan ‘ask, pray, command’, cweðan ‘say’, hatan ‘call’, secgan ‘say’
Seeing/hearing: geseon ‘see’, gehyran ‘hear’
Staying: licgan ‘lie’, sittan ‘sit’, standan ‘stand’, wunian ‘remain, dwell’
Thinking: nytan ‘not know’, smeagan ‘consider’, wenan ‘suppose, think’, witan ‘know’
Transport: asendan ‘send out’, geflieman ‘drive away’, gegaderian ‘gather’, gelædan ‘lead’, gesettan ‘set, appoint’, sendan ‘send’

[16] The role of genre conventions in determining the uses of þa and other discourse markers in Old English is a topic that cannot be discussed in this article. 

[17] Other parameters in the cluster of Transitivity include the dynamicity and telicity of the action, the agentivity of the subject and the affectedness of the object (Hopper & Thompson 1980).

Sources

Bately, Janet M., ed. 1980. The Old English Orosius (=Early English Text Society. Supplementary series, 6.) London: Oxford University Press.

Bately, Janet M., ed. 1986. The Anglo-Saxon Chronicle: A Collaborative Edition. Volume 3: MS A. Cambridge: Brewer.

BT = Bosworth, Joseph & T. Northcote Toller. 1954. An Anglo-Saxon Dictionary. Oxford etc: Oxford University Press. [1st printing 1898] 30 Nov. 2010. http://bosworth.ff.cuni.cz/

DOEC = di Paolo Healey, Antoinette, Joan Holland, Ian McDougall & Peter Mielke. eds. 2000. Dictionary of Old English Corpus in Electronic Form. Toronto: Dictionary of Old English Project. Available from the Oxford Text Archive. http://ota.ahds.ac.uk/headers/2488.xml.

Godden, Malcolm, ed. 1979. Ælfric's Catholic Homilies. The Second Series Text (=Early English Text Society, Supplementary Series, 5.) London etc.: Oxford University Press.

OED = Oxford English Dictionary Online. 2010. 30 Nov. 2010. http://www.oed.com/.

Venezky, Richard & Sharon Butler. 1985. A Microfiche Concordance to Old English: The High-Frequency Words (=Publications of the Dictionary of Old English, 2.) Toronto: The Pontifical Institute of Mediaeval Studies.

References

Andrew, S.O. 1940. Syntax and Style in Old English. Cambridge: Cambridge University Press. [Reissued 1966. New York: Russell & Russell]

Aristotle. [1954]. Rhetoric and Poetics (The Modern Library). Translated by W. Rhys Roberts and Ingram Bywater. New York: Random House.

Bacquet, Paul. 1962. La structure de la phrase verbale à l'époque Alfrédienne (=Publications de la Faculté des Lettres de l'Université de Strasbourg, Fascicule 145.) Paris: Les Belles Lettres. 

Bækken, Bjørg. 2000. “Inversion in Early Modern English”. English Studies 5: 393–421.

Björklund, Martina & Tuija Virtanen. 1991. “Variation in narrative structure: A simple text vs. an innovative work of art”. Poetics 20: 391–403.

Bolinger, Dwight. 1977. Meaning and Form (=English Language Series, 11.) London: Longman. 

Brinton, Laurel J. 1996. Pragmatic Markers in English: Grammaticalization and Discourse Functions. (=Topics in English Linguistics, 19.) Berlin & New York: Mouton de Gruyter.

Breivik, Leiv Egil. 2002. “Old English þa revisited”. Language and Function: To the Memory of Jan Firbas (=Studies in Functional and Structural Linguistics, 49), ed. by Josef Hladký, 39–55. Amsterdam & Philadelphia: Benjamins.

Chafe, Wallace L. 1979. “The flow of thought and the flow of language”. Discourse and Syntax (=Syntax and Semantics, 12), ed. by Talmy Givón, 159–81. New York etc.: Academic Press. 

Clemoes, Peter. 1966. “Ælfric”. Continuations and Beginnings. Studies in Old English Literature, ed. by Eric Stanley, 176–209. London: Nelson.

DOE = The Dictionary of Old English Website. 18 May 2011. http://tir.doe.utoronto.ca/index.html.

Dorgeloh, Heidrun. 1997. Inversion in Modern English: Form and Function (=Studies in Discourse and Grammar, 6.) Amsterdam & Philadelphia: Benjamins.

Enkvist, Nils Erik. 1972. “Old English adverbial þa – an action marker?”. Neuphilologische Mitteilungen 73: 90–96.

Enkvist, Nils Erik. 1986. “More about the textual functions of Old English adverbial þa”. Linguistics across Historical and Geographical Boundaries: In Honour of Jacek Fisiak on the Occasion of his Fiftieth Birthday. Vol. 1: Linguistic Theory and Historical Linguistics (=Trends in Linguistics: Studies and Monographs, 32), ed. by Dieter Kastovsky & Aleksander Szwedek, 301–309. Berlin etc.: Mouton de Gruyter. 

Enkvist, Nils Erik. 1987a. “A note towards the definition of text strategy”. Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung 40(1): 19–27.

Enkvist, Nils Erik. 1987b. “More about text strategies”. Perspectives on Language Performance: Studies in Linguistics, Literary Criticism and Foreign Language Teaching and Learning. To honour Werner Hüllen on the Occasion of his Sixtieth Birthday, ed. by Wolfgang Lörscher & Rainer Schulze, 337–350. Tübingen: Gunter Narr.

Enkvist, Nils Erik. 1994. “Problems raised by Old English þa”. Writing vs Speaking: Language, Text, Discourse, Communication (=Tübinger Beiträge zur Linguistik, 392), ed. by Světla Čmejrková, Frantisek Daneš & Eva Havlová, 55–62. Tübingen: Gunter Narr. 

Enkvist, Nils Erik & Brita Wårvik. 1987. “Old English þa, temporal chains, and narrative structure”. Papers from the 7th International Conference on Historical Linguistics (=Current Issues in Linguistic Theory, 48), ed. by Anna Giacalone Ramat, Onofrio Carruba & Giuliano Bernini, 221–37. Amsterdam & Philadelphia: Benjamins.

Foster, Robert. 1975. “The use of þa in Old and Middle English narratives”. Neuphilologische Mitteilungen 76: 404–414.

Fourquet, Jean. 1938. L’ordre des éléments de la phrase en germanique ancien: Études de syntaxe de position. Paris: Belles Lettres.

Fraser, Bruce. 1999. “What are discourse markers?”. Journal of Pragmatics 31: 931–952.

Fretheim, Thorstein. 2006. “English then and Norwegian da/så compared: A relevance-theoretic account”. Nordic Journal of Linguistics 29: 45–93.

Givón, Talmy. 1983. “Topic continuity in discourse: An introduction”. Topic Continuity in Discourse: A Quantitative Cross-language Study (=Typological Studies in Language, 3), ed. by Talmy Givón, 1–41. Amsterdam & Philadelphia: Benjamins.

Godden, Malcolm. 2000. Ælfric’s Catholic Homilies: Introduction, Commentary and Glossary (=Early English Text Society. Supplementary series, 18.) London etc: Oxford University Press.

González, Montserrat. 2004. Pragmatic Markers in Oral Narrative. The Case of English and Catalan (=Pragmatics & Beyond New Series, 122.) Amsterdam & Philadelphia: Benjamins.

Greenbaum, Sidney. 1979. “The treatment of clause and sentence in A Grammar of Contemporary English”. Studies in English Linguistics for Randolph Quirk, ed. by Sidney Greenbaum, Geoffrey Leech & Jan Svartvik, 17–29. London: Longman.

Grimes, Joseph E., ed. 1978. Papers on Discourse. Arlington, Texas: Summer Institute of Linguistics.

Gundel, Jeanett K., Nancy Hedberg & Ron Zacharski. 1993. “Cognitive status and the form of referring expressions in discourse”. Language 69: 274–307.

Halliday, Michael A.K. 1978. Language as Social Semiotic. The Social Interpretation of Language and Meaning. London: Arnold.

Halliday, Michael A.K. & Ruqaiya Hasan. 1976. Cohesion in English. Longman: London.

Hopper, Paul J. 1992. “A discourse perspective on syntactic change: Text-building strategies in Early Germanic”. Reconstructing Languages and Cultures (=Trends in Linguistics, Studies and Monographs, 58), ed. by Edgar C. Polomé & Werner Winter, 217–238. Berlin & New York: Mouton de Gruyter.

Hopper, Paul J. & Sandra A.Thompson. 1980. “Transitivity in grammar and discourse”. Language 56: 251–299.

Ji, Shaojun. 2002. “Identifying episode transitions”. Journal of Pragmatics 34: 1257–1271.

Jones, Christopher A. 1999. Ælfric’s Letter to the Monks of Eynsham. Cambridge: Cambridge University Press.

Kemenade, Ans van. 2009. “Discourse relations and word order change”. Information Structure and Language Change: New Approaches to Word Order Variation in Germanic (= Trends in Linguistics. Studies and Monographs, 203), ed. by Roland Hinterhölzl & Svetlana Petrova, 91–118. Berlin & New York: Mouton de Gruyter.

Kemenade, Ans van & Bettelou Lou. 2006. “Discourse adverbs and clausal syntax in Old and Middle English”. The Handbook of the History of English, ed. by Ans van Kemenade & Bettelou Los, 224–248. Malden, Mass.: Blackwell.

Kim, Taejin. 1992. The Particle þa in the West-Saxon Gospels: A Discourse Level Analysis (=European University Studies, Series XIV: Anglo-Saxon Language and Literature, Vol. 249.) Berne, etc: Lang.

Kohonen, Viljo. 1978. On the Development of English Word Order in Religious Prose around 1000 and 1200 A.D. – A Quantitative Study of Word Order in Context (=Publications of the Research Institute of the Åbo Akademi Foundation, 38.) Turku: Research Institute of the Åbo Akademi Foundation.

Kroon, Caroline. 1998. “A framework for the description of Latin discourse markers”. Journal of Pragmatics 30: 205–233.

Kytö, Merja, comp. 1996 (1991). Manual to the Diachronic Part of the Helsinki Corpus of English Texts. Coding Conventions and Lists of Source Texts. (3rd ed.). Helsinki: Department of English, University of Helsinki. 30 Nov. 2010 http://clu.uni.no/icame/manuals/HC/INDEX.HTM.

Labov, William & Joshua Waletzky. 1967. “Narrative Analysis: Oral versions of personal experiences”. Essays on Verbal and Visual Arts: Proceedings of the 1966 Annual Spring Meeting of the American Ethnological Society, ed. by June Helm, 12–44. Seattle: University of Washington Press.

Leibniz, Gottfried Wilhelm von. 1839/1940. Opera Philosophica quae Extant Latina Gallica Germanica Omnia. Ed. by Johann Eduard Erdmann. Berolini: Eichler.

Lenker, Ursula. 2010. Argument and Rhetoric. Adverbial Connectors in the History of English (=Topics in English Linguistics, 64.) Berlin & New York: de Gruyter.

Liggins, Elizabeth M. 1986. “ Syntax and style in the Old English Orosius”. Studies in Earlier Old English Prose, ed. by Paul E. Szarmach, 245–273. New York: State University of New York Press.

Longacre, Robert E. 1976. “‘Mystery’ particles and affixes”. Papers from the Twelfth Regional Meeting Chicago Linguistic Society, April 23–25, 1976, ed. by Salikoko S. Mufwene, Carol A. Walker & Sanford B. Steever, 468–77. Chicago: Chicago Linguistic Society.

McKnight, George H. 1897. “The primitive Teutonic order of words”. Journal of English and Germanic Philology 1: 136–219.

Mitchell, Bruce. 1980. “The dangers of disguise: Old English texts in modern punctuation”. Review of English Studies 31: 385–413. [Reprinted in Mitchell, Bruce. 1988. On Old English: Selected Papers. Oxford: Blackwell, 172–201.]

Mitchell, Bruce. 1985. Old English Syntax. 2 Vols. Oxford: Clarendon Press.

Östman, Jan-Ola. 1981. You Know. A discourse-functional approach (=Pragmatics & Beyond, II:7.) Amsterdam: Benjamins.

Östman, Jan-Ola. 1982. “Pragmatic particles in an applied perspective”. Neuphilologische Mitteilungen 83: 135–153.

Petrova, Svetlana & Michael Solf. 2008. “Rhetorical relations and verb placement in the early Germanic languages: A cross-linguistic study”. ‘Subordination’ versus ‘Coordination’ in Sentence and Text (=Studies in Language Companion Series, 98), ed. by Cathrine Fabricius-Hansen & Wiebke Ramm, 329–351. Amsterdam & Philadelphia: Benjamins.

Pintzuk, Susan & Eric Haeberli. 2008. “Structural variation in Old English root clauses”. Language Variation and Change 20: 367–407.

Prince, Ellen F. 1981. “Toward a taxonomy of given-new information”. Radical Pragmatics, ed. by Peter Cole, 223–55. New York etc.: Academic Press.

Redeker, Gisela. 1991. “Linguistic markers of discourse structure”. Linguistics 29: 1139–1172.

Renouf, Antoinette & Jayeeta Banerjee. 2007. “The search for repulsion: A new corpus analytical approach”. Towards Multimedia in Corpus Studies (=Studies in Variation, Contacts and Change, 2), ed. by Päivi Pahta, Irma Taavitsainen, Terttu Nevalainen & Jukka Tyrkkö. http://www.helsinki.fi/varieng/series/volumes/02/renouf_banerjee/.

Schiffrin, Deborah. 1987. Discourse Markers (=Studies in Interactional Sociolinguistics, 5.) Cambridge etc.: Cambridge University Press.

Schiffrin, Deborah. 1990. “Between text and context: Deixis, anaphora, and the meaning of then”. Text 10: 245–270.

Schiffrin, Deborah. 1992. “Anaphoric then: aspectual, textual, and epistemic meaning”. Linguistics 30: 753–792.

Schourup, Lawrence. 1999. “Discourse markers”. Lingua 107: 227–265.

Stockwell, Robert P. 1984. “On the history of the verb-second rule in English”. Historical Syntax (=Trends in Linguistics, Studies and Monographs, 23), ed. by Jacek Fisiak, 575–592. Berlin, New York & Amsterdam: Mouton.

Stockwell, Robert P. & Donka Minkova. 1991. “Subordination and word order change in the history of English”. Historical English Syntax (=Topics in English Linguistics, 2), ed. by Dieter Kastovsky, 367–408. Berlin & New York: Mouton de Gruyter.

Traugott, Elizabeth Closs. 1971. The History of English Syntax. New York: Holt, Rinehart & Winston.

Traugott, Elizabeth Closs. 1992. “Syntax”. The Cambridge History of the English Language. Volume 1: The Beginnings to 1066, ed. by Richard M. Hogg, 168–286. Cambridge: Cambridge University Press. Cambridge Histories Online. DOI: 10.1017/CHOL9780521264747.005

Traugott, Elizabeth Closs. 2007. “Old English left-dislocations: Their structure and information status”. Folia Linguistica 41: 405–441.

Trips, Carola & Eric Fuss. 2009. “The syntax and semantics of the temporal anaphor ‘then’ in Old and Middle English”. Advances in Comparative Germanic Syntax, ed. by Artemis Alexiadou, Jorge Hankamer, Thomas McFadden, Justin Nuger & Florian Schäfer, 171–196. Amsterdam & Philadelphia: Benjamins.

Vendler, Zeno. 1967. Linguistics in Philosophy. Ithaca, NY: Cornell University Press.

Virtanen, Tuija. 1992. Discourse Functions of Adverbial Placement in English: Clause-initial Adverbials of Time and Place in Narratives and Procedural Place Descriptions. Turku: Åbo Akademi University Press.

Warner, Anthony. 2007. “Parameters of variation between verb-subject and subject-verb order in late Middle English”. English Language and Linguistics 11(1): 81–111.

Wårvik, Brita. 1990. “On grounding in English narratives: A diachronic perspective”. Papers from the 5th International Conference of English Historical Linguistics (=Current Issues in Linguistic Theory, 65), ed. by Sylvia Adamson, Vivien A. Law, Nigel Vincent & Susan Wright, 559–575. Amsterdam & Philadelphia: Benjamins.

Wårvik, Brita. 1994. “Participants tracking narrative structure: Old English þa and topicality”. Topics and Comments: Papers from the Discourse Project (=Anglicana Turkuensia, 13), ed. by Sanna-Kaisa Tanskanen & Brita Wårvik, 115–139. Turku: University of Turku.

Wårvik, Brita. 1995. “The ambiguous adverbial/conjunctions þa and þonne in Middle English: A discourse-pragmatic study of then and when in Early English Saints’ Lives”. Historical Pragmatics. Pragmatic Developments in the History of English (=Pragmatics & Beyond New Series, 35), ed. by Andreas Jucker, 345–357. Amsterdam & Philadelphia: Benjamins.

Wårvik, Brita. 1996/2006. “Grounding”. Handbook of Pragmatics 1996/2006, ed. by Jef Verschueren, Jan-Ola Östman, Jan Blommaert & Chris Bulcaen. Amsterdam & Philadelphia: Benjamins, 14 pages. 30 Nov. 2010. https://www.benjamins.com/online/hop/.

Wårvik, Brita. 2009.“Teaching by stories: Ælfric’s instructive narratives”. Instructional Writing in English: Studies in Honour of Risto Hiltunen (=Pragmatics & Beyond New Series, 189), ed. by Matti Peikola, Janne Skaffari & Sanna-Kaisa Tanskanen, 13–34. Amsterdam & Philadelphia: Benjamins.

Waterhouse, Ruth. 1984. “Sentence determination in Ælfric’s Lives of Saints”. Neuphilologische Mitteilungen 85: 257–272.

Waterhouse, Ruth. 1992. The Triangular Clause Relationship in Ælfric’s Lives of Saints and in Other Works (=American University Studies, Series IV: Anglo-Saxon Language and Literature, Vol. 1.) New York etc: Peter Lang.

Weydt, H. 1969. Abtönungspartikel. Die deutschen Modalwörter und ihre französischen Entsprechungen. Bad Homburg: Gehlen.