4.1.1 Topic switching

The texts under scrutiny exhibit a varying propensity towards topic switching, and the type of pronoun used to topic switch. Topic switching makes for a more complex patterning of anaphoric relations. Se-pronouns unambiguously mark topic switch when they pick up non-nominative discourse-new referents, and usually only co-refer with nominative discourse-new referents in contexts involving no competing referents. Contrastively, personal pronouns can pick up both nominative and non-nominative discourse-new referents even when competing referents are available, which makes for potentially ambiguous reference tracking, particularly in contexts involving two competing referents that coincide in number and/or gender, as in (33), where hi could potentially refer to Iudeiscan ‘Jews’ or boca ‘books’, and in (34) where both Gallicanus the man of God and the discourse topic Julian are possible referents.

1. (33) We hedað þæra crumena ðæs hlafes. and ða Iudeiscan gnagað þa rinde. for ðan ðe we understandað þæt gastlice andgit þæra boca. and hi rædað þa stæflican gereccednysse buton andgite; (ÆCHom II,8,0060.70.108)

   ‘We heed the crumbs of the bread and the Jews gnaw the crust because we understand the spiritual meaning of those books and they read the literal explanation without meaning.’

1. (34) Adræfde þa Gallicanum þone godes man aweg. And he ferde to Alexandrian þære ægiptiscan byrig. (ÆLS (Agnes), 0144.398–0145.399)

   ‘Then (Julian) drove away Gallicanus the man of God. And he went to Alexandria the Egyptian city.’

Based on the data sample discussed in section 3.1 (table 2), table 3 summarises the relative frequency of clause-initial pronominal anaphora across texts according to the information-structural properties of their referents. The findings indicate that certain texts exhibit greater topic continuity; for example, 91 per cent of clause-initial pronouns continue a discourse-old topic in the scientific writing of LchI, and only 9 per cent topic switch. WHom also involves greater topic maintenance than ÆLS and ÆCHom.

Table 3.

Frequency of clause-initial pronominal anaphora depending on the information-structural properties of their referents

A breakdown of the figures in table 3 reveals that texts differ with regard to the type of pronoun employed to pick up discourse-new material. Table 4 summarises the rate at which personal pronouns and se-pronouns are used to highlight discourse new referents across texts.

Table 4.

Frequency of personal pronouns and se-pronouns used to highlight discourse-new referents (including nominative and non-nominative antecedents)

At 58 per cent (N = 51/88) and 69 per cent (N = 60/87), respectively, ÆLS and ÆCHom rely heavily on personal pronouns to pick up discourse-new referents regardless of whether these referents constitute nominative or non-nominative antecedents. The use of topic-switching personal pronouns is relatively common, and quantitatively remarkably similar in both ÆLS and ÆCHom. In ÆLS there are 52 contexts in which a pronoun picks up a non-nominative discourse-new referent; personal pronouns are employed at a rate of 63 per cent (N = 33/52) and se-pronouns occur 36.5 per cent of the times (N = 19/52). In ÆCHom personal pronouns occur at 66 per cent (N = 29/44) and se-pronouns at 34 per cent (N = 15/44). ÆLS and ÆCHom also show a propensity towards using personal pronouns to mark topic switch even in contexts involving competing referents that coincide in gender and number; 30 per cent (N = 10/33) of the personal pronouns in ÆLS that pick up non-nominative discourse-new referents fall into this category, and 48 per cent (N = 14/29) in ÆCHom. LchI employs personal pronouns to highlight discourse-new referents on eleven occasions (N = 8/11: 72%), but none of these instances involves competing referents, or two co-referents that coincide in gender and number, and topic switching is in any case rare in LchI.

One notable feature of Wulfstan's homiletic writing is his tendency in WHom to mark discourse-new referents unambiguously using se-pronouns; at 81 per cent (N = 9/11), se-pronouns are the preferred anaphoric type for highlighting discourse-new referents. WHom also uses se-pronouns more so than any other text to pick up discourse-old referents (N = 3/12: 25%), as table 2 indicates. In fact, WHom employs the highest overall rate of se-pronouns out of all the texts. Based on the figures in table 2, the relative frequency of se-pronouns across texts (as opposed to personal pronoun usage) is as follows: WHom 25% (N = 12/48) > ÆLS 15% (N = 40/226) > ÆCHom 11% (N = 31/267) > LchI 4% (N = 5/119). Se-pronouns are thus notably rare in scientific writing, but relatively abundant in Wulfstan's homiletic writing, with Ælfric's homiletic writing and the saints’ lives falling somewhere between.

4.1.2 The texts

Medical recipes of the kind found in LchI are a well-defined genre with a clear purpose as a handbook intended for consultation. Their communicative context differs from the other genres treated here in that recipe collections were probably intended for individual consultation (Taavitsainen Reference Taavitsainen2001: 107). The recipes provide instructions on how to prepare medicine and follow a highly regular structure. The entries in LchI are organised according to plant type. A description of the plant is followed by indications of preparation, application and efficacy, generally using the formula he bið gehæleð ‘he will be healed’. The recipes possess many of the text-type features that Taavitsainen (Reference Taavitsainen2001: 98–106) has identified for Late Middle and Early Modern English recipes. These features explicate the text's very clear instructive communicative strategy. They include the chronological sequencing of ingredients and actions in order to reflect the steps in preparing the prescription, the redaction of instructions in short paratactic sentences, and the use of the imperative in accordance with the instructive text style (Werlich Reference Werlich1983). Taavitsainen (Reference Taavitsainen2001: 100) does not find strong evidence in her Middle English material of the type of object pronoun deletion that has been identified for later medical writing (see Görlach Reference Görlach, Rissanen, Ihalainen, Nevalainen and Taavitsainen1992; Carroll Reference Carroll1999). At 1.4 per cent, the overall rate of oblique pronoun forms in LchI is in line with other texts in the present study, and the use of oblique pronouns is often plentiful, as in (35), but object pronoun deletion of the type illustrated in (36) is not uncommon.

1. (35) Wið eagena sare genim þas wyrte þe man argimoniam & oðrum naman garclife nemneð, cnuca hy swa grene þurh hy selfe. Gyf ðu hy þonne grene næbbe genim hy drige & dype on wearmum wætere swa þu eaþelicost hy brytan mæge. (LchI (Herb), 0385.32.1)

   ‘For sore of eyes take this wort which is called agrimony and by another name garclive, pound it so green by itself. If then you have it not green take it dry and dip (it) in warm water so you most easily can use it.’

1. (36) Gif man gewundud sy genim wegbrædan sæd, gnid to duste & scead on þa wunde. (LchI (Herb), 0075.2.6)

   ‘If a man be wounded take seed of waybread rub (it) to dust and shed (it) on the wound.’

With regard to the reference-tracking system of the recipes, as previously mentioned, 91 per cent of clause-initial anaphors pick up discourse-old referents, which indicates that topic continuity is characteristic of the linguistic discourse of medical recipes. The excerpt in (37) exemplifies the type of unambiguous reference tracking that characterises LchI.

1. (37) Wið næddran slite genim ðas sylfan uiperinam, cnuca hy, mengc mid wine, syle drincan. Heo hælð wundorlice þone slyte & þæt attor todrifð. (LchI (Herb), 0181–0182.6.1)

   ‘For adder bite take the same viperina, pound it, mix with wine, give to drink. It heals wonderfully the perforation and drives away the poison.’

In medical writing, narration is not the core text function; in other words, the central aim of ‘telling a story’ does not shape the composition of such text types. The main textual concern in scientific writing is to convey facts succinctly and unambiguously, a textual concern that is explicated by a system of pronominal referencing that minimises ambiguity.

The same textual concern of maximal transparency applies to the law texts. The redaction of Old English law codes conforms to the underlying logical structure that has been identified for modern-day legal documents: ‘Reduced to a minimal formula the great majority of legal sentences have an underlying logical structure which says something like “if X, then Z shall be Y” or alternatively “if X, then Z shall do Y”. . ..’ (Crystal & Davy Reference Crystal and Davy1969: 203). In the excerpt taken from Æthelred's laws (LawIAtr) in (38), disambiguation between the various participant lines, comprising freoman ‘freeman’, borh ‘guarantor’ and hlaford ‘lord’, is achieved by the repetition of the lexical items borh and hlaford. The text's reliance on lexical items rather than an excessive use of pronominal anaphora no doubt correlates with the relatively low rate of nominative pronoun use in LawIAtr (N = 192/9,194: 2%), and reflects a concern for exactness of reference expected from legal writing.

1. (38) Þæt is, þæt ælc freoman getreowne borh hæbbe, þæt se borh hine to ælcon rihte gehealde, gif he betyhtlad wurðe. Gyf he ðonne tyhtbysig sy, gange to þam þryfealdan ordale. Gif se hlaford sæcge, þæt him naðer ne burste ne að ne ordal, syððan þæt gemot wæs æt Bromdune, nime se hlaford him twegen getreowe þegenas innan þam hundrede . . . (LawIAtr 0003.1–0005.1.2)

   ‘That is that every freeman have a true guarantor that the guarantor might present him to every justice if he should be accused. If he then be of bad repute let (him) go to the threefold ordeal. If his lord should say that he had failed neither in oath nor ordeal since the assembly was at Bromdus, let the lord take with him two true thanes within the hundred . . .’

We might compare the concern for exactness of reference witnessed in LawIAtr with Ælfric's apparent occasional disregard for precise reference. Often in sections involving two female or male participants, Ælfric's heavy reliance on pronominal anaphora, rather than lexical items, makes the grammatical ties supplied by the anaphors ambiguous, as in (39), where it is not immediately obvious that heo refers to Melantia and not Eugenia. Similarly, in (40), the use of pronominal anaphors instead of lexical items seriously complicates tracking the three different participant lines Chromatius, Tranquillinus and preoste ‘priest’.

1. (39) Eugenia þa hi [= Melantia] gesmyrode mid gehalgodum ele and eac gemearcode mid rodetacne. And heo [= Melantia] þæt reðe attor eall ut aspaw, þe hyre [= Melantia] dærede, and wearþ gehæled þurh þæt halige mæden. (ÆLS (Eugenia), 0041.136–0042.138)

   ‘Eugenia smeared her [= Melantia] with hallowed oil and also marked (her) with the sign of the cross. And she [= Melantia] that evil poison all vomited out that harmed her and was healed through the holy maiden.’

1. (40) Chromatius þa bæd þæt he [= Tranquillinus] him [= Chromatius] gebrohte ðone mann þe hine [= Tranquillinus] gefullode and fram þære coðe gehælde. Tranquillinus ða eode to ðam arwurðan preoste, sæde hu hi spræcon, and hine [= priest] sona gelædde to þam heahgerefan. And he [= Chromatius] cwæð him [= priest] to . . . (ÆLS (Sebastian), 0055.190–0056.192)

   ‘Chromatius then asked that he [= Tranquillinus] brought him [= Chromatius] the man who had baptised him [= Tranquillinus] and healed from the illness. Tranquillinus then went to the venerable priest, said how they had spoken and brought him [= priest] immediately to the prefect and he [= Chromatius] said to him [= priest] . . .’

With regard to the saints’ lives and the homilies, a strict correlation between anaphoric strategy and genre is not immediately apparent. Despite the statistically significant difference in the overall rate of pronominal anaphora between ÆLS and ÆCHom, noted in table 1, the reference-tracking systems of Ælfric's saints’ lives and homilies are as strikingly similar, in terms of the distribution of anaphoric types, as those of ÆCHom and WHom are dissimilar. Differences in usage across different text types could be idiosyncratic to particular authors or the product of a text's subject matter. If the parameter of author remains the same across text types, this helps to control for differences in pronoun usage being idiosyncratic to a particular author rather than the result of content, but in this case, it is the similarities in usage across seemingly different texts (ÆCHom and ÆLS), and divergences in usage in apparently similar types of writing (ÆCHom and WHom), that require explanation.

The parallels in linguistic usage across Ælfric's homilies and saints’ lives could reflect an idiosyncrasy of Ælfric's writing style, but overlaps in subject matter are more likely to explain the similarities. ÆCHom and ÆLS both draw from similar content involving the accounts of saints’ lives and scriptural narratives. The parallels in subject matter explain the strong narrative elements found in both texts, such as the predominance of preterite forms (Ogawa Reference Ogawa1989: 104) and the high overall rate of third-person pronouns. ÆCHom and ÆLS belong to different genres but to the same narrative text type. The discourse of narrative text types has a strong ‘story telling’ element, which often involves an intricate set of referents that appear and reappear in changing configurations as the story develops. Hagiographical narratives are particularly complex in that the communicative situation in saints’ lives functions at two levels; in mediating the narrative, the narrator communicates with the reader/audience while at the same time the characters communicate with each other (von Contzen Reference Contzen, Fludernik and Jacob2014: 188).

Text tradition is also an important consideration. The saints’ lives under scrutiny here derive from Christian Latin sources, and the Anglo-Saxon hagiographical material was part of that tradition in its details and arrangement. This raises the question of whether linguistic choices in the Latin original may have influenced Ælfric's system of pronominal reference. The Latin text would have exerted no influence on Ælfric's employment of personal pronouns as the unmarked praxis in Latin is not to use pronominal subjects, thus providing no model for the referential pronouns that had to be added to the English text. While Ælfric's stylistic devices do not owe a heavy debt to Latin rhetorical tradition (see Randall Lipp Reference Randall Lipp1969 for discussion), his indebtedness to the content of his Latin source texts is often substantial. A line-by-line comparison of Ælfric's system of pronominal reference with that of his Latin sources is beyond the scope of the present study, but a brief comparison suggests that the highly conventionalised narration associated with saints’ lives is drawn from Latin hagiographical tradition, and may have influenced Ælfric's choice of pronominal anaphora. A case in point is the formulaic introduction of the saint that begins with an appropriate epithet, e.g. ‘noble maiden’, followed by the saint's name and biographical details that reiterate the saint's devout Christian nature. A comparison of Ælfric's initial description of St Agnes in (41) with that of his Latin source in (42) indicates that he followed his source closely in this case. The formulaic introduction of the saint is clearly modelled on the Latin and could explain the use of the OE se-form to render Latin quæ.

1. (41) On þære tide wæs sum æðelboren mæden Agnes gehaten, on ðone Hælend gelyfed, binnan Rome byrig, bilewit and snotor, cildlic on gearum and ealdlic on mode. Seo wan þurh geleafan wið þa feondlican ealdras, and on ðam þritteoðan geara þone deað forleas and þæt ece lif gemette, forðan þe heo lufode Crist. Heo wæs wlitig on ansyne, and wlitigre on geleafan. (ÆLS (Agnes), 0004.6–0006.13)

   ‘At that time (there) was a certain noble maiden called Agnes, (she) believed in the Saviour, in the city of Rome, gentle and wise, a child in years and old in mind. She contended through faith against the fiendish rulers and in her thirteenth year lost mortality and found eternal life because she loved Christ. She was fair in countenance and fairer in faith.’

1. (42) Tempore illo erat quædam virgo Romæ nobilibus orta natalibus, nomine Agnes, quæ certans contra principes et potestates tenebrarum, tertiodecimo ætatis suæ anno mortem perdidit, et vitam invenit . . . Infantia quidem computabatur in annis, sed erat senectus mentis in moribus: corpore quidem juvencula, sed animo cana. Pulchra facie, sed pulchrior fide, et elegantior castitate. (Ambrose, ed. Migne Reference Migne1845: col. 735, §1)

   ‘At that time there was a girl in Rome, born from noble parents, called Agnes, who, fighting against rulers and powers of the dark, lost death in her thirteenth year and found life . . . Though her childhood was computed in years, there was old age/maturity of mind in her behaviour: though of juvenile body, she was grey/old in mind. She had a pretty face, but was prettier in her faith, and more elegant in her chastity.’

Bethurum's (Reference Bethurum1932) comparative study of Ælfric's Old English renderings of the Latin originals reveals that Ælfric varied his method of translating and not all the lives are rendered with the same degree of literalness. The lives of St Lucy, St Martin and particularly St Agnes present ‘a fairly literal rendering of the Latin text, adding little of his own and omitting little of the Latin’ (Bethurum Reference Bethurum1932: 519). In other lives, such as those of St Sebastian and St Eugenia, summary rather than translation was his ‘usual practice’ (Bethurum Reference Bethurum1932: 519), which might explain some of the divergences in approach to reference tracking between the Latin and Old English versions. For instance, in (43) taken from the Life of St Eugenia, the Latin original relies heavily on lexical items, unlike Ælfric's reference tracking, which, as we have seen, refrains from employing lexical items, regardless of the resulting ambiguity.

1. (43) Hec et his similia prosequente Eugenia, Melantia surda aure transiebat. Interea dum Eugenia conaretur Melantiam de laqueo mortis eripere, et ageret Melantia quatenus Eugeniam inretiret. (Fábrega Grau Reference Fábrega Grau1955: 89, §18)

   ‘While Eugenia continued this and similar such things, Melantia kept going on with a deaf ear. Meanwhile, while Eugenia tried to pull Melantia out of the snare of death, [and] Melantia was acting so as to ensnarl Eugenia.’

With regard to the differences between Ælfric's and Wulfstan's homiletic writing, the authors’ divergent pronominal systems add to the list of linguistic differences between the homilists that have been noted in the literature, such as Wulfstan's more widespread use of present-tense finite verbs and modal auxiliaries (Ogawa Reference Ogawa1989: 101, 103).Footnote ⁹ Once again the differences between the two homilists in matters of linguistic choice can partly be explained on the basis of differing subject matter and text type. ÆHom and WHom represent different types of homilies (Ogawa Reference Ogawa1989: 142). Bethurum (Reference Bethurum1957: 96) suggests that ‘if we accept the distinction that a homily is an elaboration of a text, a sermon, with or without a text, an elaboration of some subject, then Wulfstan wrote more sermons than homilies’. Often ‘homilies’ and ‘sermons’ are treated synonymously in the literature because they share the central activity of preaching (see Kohnen Reference Kohnen, Fitzmaurice and Taavitsainen2007), but important differences between the genres have been identified with respect to their source material and the context of the speech act. Whereas a homilia was ‘an exegetical commentary on scripture for a specific liturgical celebration’, a sermo was ‘an edifying discourse meant to bring the faithful closer to Christian truth with no prescriptive theme – scriptural or otherwise – and no obligatory connection to the liturgical event’ (Heffernan Reference Heffernan1984: 179, cited in Kohnen Reference Kohnen, Fitzmaurice and Taavitsainen2007: 142). Wulfstan's homilies are mainly sermo-type homilies that treat dogmatic, moral or hortatory themes, whereas ÆHom largely comprise exegetical homilies, or vita-type homilies that draw from accounts of saints’ lives and scriptural narratives (Ogawa Reference Ogawa1989: 103). All of Ælfric's homilies included in the present data set are of the homilia or vita type and all of Wulfstan's represent the sermo type, except the eschatological homily ‘Matthew on the Last Days’ (Bethurum's II), which mostly relies on adapted Gospel passages. The religious instruction provided by Wulfstan is more directive in nature, a textual concern that is reinforced by the frequent use of imperatives in direct instruction in action-demanding sentences, as exemplified in (44). Ælfric's homilies, on the other hand, offer instruction that involves exposition, exegesis and narration.

1. (44) & scyldað eow wið galscypas & swyðe georne wið æwbrecas, & wið oferfylle beorgað eow georne. (WHom 8c,0057.163)

   ‘And protect yourselves against luxury and very eagerly against lawbreakers and against excess protect yourselves eagerly.’

Divergent subject matters and text types alone do not account for the differences between the two homilists in matters of linguistic usage. Each writer adopts a style that is best suited to his intended audience. Wulfstan's homilies were composed for public delivery to a largely ad populum audience, whereas Ælfric's intended audience comprised a clerical congregation. Ogawa (Reference Ogawa1989: 120, 144) interprets Wulfstan's preference for present-tense finite verbs and modal auxiliaries as indicative of the fact that he seems to be writing in a form of Old English prose that is closer to colloquial language, and thus better suited to his lay audience. There are frequent traces of orality in Wulfstan's homilies which belie their conceptually oral nature and the fact that they were composed with public delivery in mind: for instance, there are moments of direct address to his lay congregation, as in swa hit þincan mæg ‘as it can seem’, and hortatory phrases, like understande se þe wille ‘may he understand who wants to’ and gecnawe se þe cunne ‘may he understand who knows how to’ (Ogawa Reference Ogawa1989: 120).

Wulfstan's style in WHom also demonstrates a very clear communicative aim or text strategy. He is noted for mastering an impassioned, heightened style that was intended to move his audiences and warn them in no uncertain terms that impending judgment loomed nigh. To the same purpose, however, he also masterfully employed a plain style suitable for exposition and teaching that adhered to the Augustinian tenet of ‘sacrificing everything to clarity’ (Bethurum Reference Bethurum1957: 89). Ælfric presupposes that his audience shares the same learned Christian background and worldview; Wulfstan does not. Indeed, Ælfric's apparent lack of concern for maintaining entirely unambiguous participant lines may reflect the fact that his saints’ lives and homilies were written for a clerical audience already familiar with the narratives. The need for clarity when addressing large audiences ‘whose capacity for abstruse thought he [Wulfstan] did not overestimate’ necessitated a clear preaching style (Bethurum Reference Bethurum1957: 92). Wulfstan adds explanatory rephrasings, introduced by þæt is, to potentially confusing comments (Bethurum Reference Bethurum1957: 89–90).Footnote ¹⁰ His use of pleonastic pronouns and pronominal adverbs serves the same elucidatory purpose, as does his rejection of potentially confusing poetic imagery, such as metaphor and simile (Bethurum Reference Bethurum1957: 90, 91).

Similarly, Wulfstan's system of pronominal referencing, which involves (a) an overall low rate of anaphoric material, (b) a tendency towards topic maintenance and c) a high overall use of se-pronouns rather than personal pronouns to highlight not only discourse-new referents but also discourse-old referents, minimises the chances of ambiguous reference tracking and is part of a communicative strategy that is motivated by textual concerns including the instructive, and above all, a need for clarity. In minimising opacity, the system of reference tracking employed is perfectly suited to discourse that was intended for a large lay congregation. The colloquial, conceptually oral nature of Wulfstan's homilies and the orator's desire for force, which motivated a general proclivity in Wulfstan for intensifying words (Bethurum Reference Bethurum1957: 90), might also explain his comparatively heavy reliance on se-pronouns. Commonalities can also be drawn here with the interaction between style and information structure that has been identified for anaphoric strategies in Modern German whereby spoken language exhibits a striking preference for demonstrative usage (Bosch et al. Reference Bosch, Katz, Umbach, Schwarz-Friesel, Consten and Knees2007).

More in-depth analysis in future research is naturally still required, but the present analysis would suggest that text type, text strategy and communicative intent, together with intended audience, play important roles in conditioning a text's pronominal referencing system. Such factors affect both the relative frequency of the different anaphoric pronoun types and their distribution.