The poetics of attention in Old English verse: A cognitive stylistic approach to the depiction of Grendel in Beowulf

In this article, I apply Stockwell’s framework of attentional phenomena in language to Old English literature. I outline the special attentional features that we find in Old English poetry, encompassing alliteration, poetic vocabulary and apposition. I then move on to provide an analysis of the figure of Grendel through an attentional lens, arguing that the Beowulf-poet consistently pulls our attention towards his demonic or monstrous aspects rather than his residual humanity. In my discussion, I argue that alliterative patterning and its intersection with lexical content is central to how the poet filters our perception of Grendel. I then move on to examine how the poet uses stock poetic lexis to define him, drawing on the Cognitive Grammar concept of reference points, suggesting that the various synonyms for ‘man’ or ‘warrior’ applied to Grendel are consistently decentred from his overall description. Finally, drawing on Nuttall, I conclude with some observations on how the narrator controls our perception of Grendel’s mind.


Introduction
Attention is a fundamental aspect of language processing, being central to the way in which we organise perceptual information and select what is most salient to our immediate context (Tomlin and Myachykov, 2015). In cognitive stylistics, the study of attentional phenomena in written discourse has been pervasive, a fact that is unsurprising given that the manipulation of readerly attention essentially underpins old formalist notions of stylistic deviance and foregrounding (see Sanford and Emmott, 2012). While attention often forms part of cognitive poetic approaches to literary texts, Stockwell's (2009a) monograph Texture centred attention as one of theprimary factors in his theory of literary resonance, or the potential of a text to 'linger' in the mind of a reader. Drawing on psychological studies of attention, he developed a list of phenomena to which readers are naturally drawn, labelling such features attractors, and arguing that they interact at the discourse level to produce fully figured, attention-grabbing textual objects (2009a: 24-26). Attractors combine over the course of a text to direct a reader's attention to structural and thematic patterns, though they may be variously neglected, in which case attention is redirected to newly salient meanings. In addition to helping account for resonant effects, a focus on textual attractors is also useful for 'testing' conventional (and competing) interpretations of literary works. As with much cognitive poetic scholarship, however, there has been relatively little work on attention in Old English literature. Harbus's research (2012Harbus's research ( , 2015 has explored the application of cognitive approaches to Old English, though her book-length treatment of the subject was primarily interested in the application of conceptual metaphor theory, text world theory, and blending to poetic texts, with attention being alluded to only briefly. The scarcity of cognitive poetic scholarship for the early medieval period is not without good reason, however: the study of Old English requires extensive training in philological practice and the cultural contexts of the early medieval period, while cognitive poetics is already a dauntingly wide-ranging field despite its relative youth, encompassing an array of theoretical frameworks. However, from the earliest days of formalist treatments of Old English literature, it has been acknowledged that attention was central to the art. In a highly influential treatment of the epic poem Beowulf, for example, Renoir (1962:156) stated that 'it always goes without saying that if the poet must hold the attention of his audience while recounting a tale of action, he must make them visualise that action' (italics are the original author's). In this article, I will use Beowulf as a test case for examining how stylistic aspects of Old English poetry modulated audience attention, using Stockwell's framework as my foundation, supplemented with the observations of Talmy (2010). I will focus particularly on the figure of Grendel, the first adversary of the hero Beowulf, and a figure who has garnered plenty of critical comment, particularly for the terror he causes and the ambiguity of his depiction, sitting somewhere between man and monster. While the monstrous aspects of Grendel are not in doubt (Mearns, 2015: 219), scholarship has also emphasised the extent to which the Beowulf-poet humanised Grendel, and the "human or monster" question has become stale, even trite. Literary interpretation is nothing if not an evolving, iterative process, however, and in this article I use the depiction of Grendel as a test case for extended cognitive stylistic analysis of Old English verse, and especially the way attentional effects influence our interpretation. This is to say that the following paper does not argue that Grendel is either more monster than man or vice-versa, but rather that the Beowulf-poet stylistically nudges readers toward the former interpretation. To this end, I will focus on three different aspects of Old English poetic style: alliteration and its patterning, the use of stock poetic language, and, drawing on Nuttall's recent work (2015,2018), the way the poet manipulates our perception of Grendel's mind. First, however, I will give a brief overview of cognitive poetic theories of attention and how they might be applied to an Old English context.

Attention phenomena
The most extensive treatment of how attention operates within literary texts remains Stockwell's book Texture (2009a) and a companion article from the same year (2009b; see also 2020: 75-80). Using the idea of figure and ground, he outlines several cognitive principles about what makes good textual objects, including characteristics like closure (where an object has a well-defined 'outer circumference') and symmetry (where an object has a 'regular, stable or predictable' shape), among several others (2009a: 23). Characterising literary works as cognitively 'cluttered arrays' consisting of a range of 'colours, edges, forms and textures that are resolved into gestalt objects ' (2009b: 29), Stockwell suggests that our interpretation of textual information is configured by the relative attractiveness of different objects from moment- to-moment (2009a:20). Perceptual elements that demand our attention are called attractors and in a literary context they are figured by a range of conceptual and stylistic features. Attractors need regular maintenance to remain within our conscious awareness, otherwise they may become backgrounded or neglected in Stockwell's nomenclature (2009a: 21). Overall, he outlines twelve different features that tend to make for good textual attractors, encompassing characteristics like newness, agency, brightness, aesthetic difference from the norm (2009a: 25). Attractors, then, are the basis of a psychologically grounded form of close reading, and one that is more expansive than traditional conceptions of stylistic 'deviance'. In the decade since his book was published, further cognitive linguistic research has fleshed out and reinforced Stockwell's framework, perhaps most importantly by Talmy (2010). His own investigation into attentional phenomena emphasises that, while attention does work at the lexical, morphemic and phonological level (a bound morpheme is less attractive than a free morpheme, for example), listeners and readers tend to process these elements at a more global discourse level, taking into account the pragmatic context also (2010: 266). Cognitive stylistic work has been carried out by Sanford and Emmott, who have published a monograph (2012) and several articles with other authors (Emmott et al., 2006(Emmott et al., , 2007(Emmott et al., , 2013, which examine readerly processing of literary texts, and in particular the psychological underpinning of formalist ideas of foregrounding. Together, each of these scholars emphasise the holistic nature of how attention is directed linguistically, though Stockwell is arguably more expansive in that he accounts for perceptually attractive phenomena (e.g. lightness, largeness, etc.) in addition to linguistic phenomena (e.g. concrete nouns are more attractive than abstract nouns) (see also Harbus's comments, 2012: 79).
Among other things, a focus on attention has the potential to give credence to-or challenge-established interpretations of texts, something that Stockwell demonstrates to good effect in a reading of Shakespeare's Sonnet 116 that justifies the most 'popular' interpretation of the poem as the most psychologically plausible (2009a: 35-39). The value of attention, then, is to help recentre careful textual engagement in critical practice and make us more attuned to the concepts, images and themes that emerge during reading. I suggest that, for pre-modern literature, especially Old English poetry, such a framework is potentially invaluable, especially given how an ever-expanding tradition of contemporary translations and literary criticism inevitably filters our judgement of the original language. As generations of anxious undergraduates can attest, Old English must be learned as a foreign tongue and therefore any approach that puts emphasis on the metacognitive awareness of our reading process is beneficial, something that Harbus (2012: 77) has been keen to emphasise. It is not just the language that is obscure, however, but also the poetic form, and Old English verse has several unique features that make it distinct from modern poetry.

Attention and the Old English poetic tradition
While the perceptual attractors that Stockwell outlines hold as much for Old English as they do for Contemporary Englishes, the former has several special formal and linguistic features that I argue contribute to attentional manipulation. In this section, I provide a brief overview of three that are closely interconnected: alliteration, apposition, and poetic vocabulary.

Alliteration
Old English poetry, along with other early Germanic poetic traditions, is stress-based, with each line being formed of two half-lines (the a-verse and b-verse respectively) separated by a caesura. Each half-line contains two stressed syllables and a varying number of unstressed or secondary stressed syllables depending on metrical context (Pons-Sans, 2016: 569-70). Stress is thus the primary structuring principal of Old English verse, but its most noticeable feature to modern ears is the decorative alliteration that binds the two half-lines together; see, for example, the following two lines from the lyric poem The Wanderer: Ðonne onwaecneð eft wineleas guma, gesihð him biforan fealwe wegas (45-46;Klinck, 1992: 76) (When the friendless man awakes again, he sees before him grey waves) 1 Here, line 45 alliterates on/w/(-waec-and win-), while 46 alliterates on/f/(-for-and feal-). It is the first stressed syllable of the b-verse (known as the head stave) which dictates the aural pattern of a line, alliterating with either one or both stressed syllables in the a-verse. Lines with only two alliterating stressed syllables can be termed simple alliteration (Le Page, 1959: 435). When the stressed syllables in the a-verse both alliterate-for a total of three alliterating stressed syllables across the whole line-this is known as double alliteration (Le Page, 1959: 435): Ðonne sorg ond slǣp somod aetgaedre earmne anhogan oft gebindað... 2 (39-40;Klinck, 1992: 76) (When sorrow and sleep together often bind the wretched lone thinker) Double alliteration is common, to the extent that some have suggested that the proportion of a-verses containing the feature are stylistically insignificant (Hoover, 1985: 54). This is based on the fact that double alliteration appears to be nearly compulsory in certain metrical contexts (Griffith, 2018: 76-77), though just because an alliterative feature might be conditioned by a formal demand, it does not necessarily follow that it is stylistically meaningless. In our above example, the poet highlights the interconnected, overwhelming nature of sorrow (sorg) and sleep (slǣp) by alliterating the couplet with somod ('together') in the next half-line, which is given extra emphasis by the tautological aetgaedre (also meaning 'together'). The alliteration, in combination with semantic repetition, thus makes the audience attend to the anguish felt by the earmne anhogan ('wretched lone thinker'). In addition to these two commonplace line-types, we also find crossed alliteration (Griffith, 2018: 72): Þǣr aet h yðe st od hringedstefna (Fulk et al., 2009: 432 Here, the different half-lines feature more elaborate patterning, with two alliterating sounds alternating between stressed syllables. The alteration of alliteration between h yðe ('harbour') and hringed-('ringed') and st od ('stood') and stefna ('prow') perhaps serves to emphasise the sturdiness of these vessels, but Griffith (2018: 133-34) has shown how crossed alliteration seems to have served narratological role signalling a new section of the story or the beginning of a speech.
In addition to these intralinear arrangements, alliteration could also function interlineally in several ways. Griffith (2018: 70-72) has given a thorough outline of these types, and I have provided an overview of some common patterns below with illustrative examples and some basic interpretations of how we might read their effects in attentional terms.
Continued alliteration is where the alliterating pattern of one line continues into one or more immediately following lines: Werod eall aras, modigra maegen, swa him Moyses bebead, maere mago-raeswa, metodes folce... Exodus 100b-102 (Anlezark, 2011: 212) (The whole army rose up, the creator's [i.e. God's] people, strong in their pride, just as Moses commanded them, the famous leader of men…) In this example from the biblical poem Exodus, the poet's continued alliteration of m binds together several important images: Moses (Moyses), the military and spiritual leader of the Hebrews, is tied to his army modigra maegen ('strong in their pride'), who are also described as metodes folce ('the creator's people'), while Moses himself is afforded extra praise as maere mago-raeswa ('famous leader'). Through this sound patterning, the poet strings together strong attractors-the nouns maegen, Moyses, mago-raeswa and metod-that serve to stylistically reinforce the unity between Moses, his people and their covenant with God.
Next, we have enjambed or strong-linked alliteration, where the final non-alliterative stressed syllable of a b-verse alliterates with the pattern of the immediately following line.
Waeron hleahtor-smiðum handa belocene, alyfed lað-sið leode gretan... Exodus 43-44 (Anlezark, 2011: 208) 4 (The hands of the laughter-smiths were restrained, the people were allowed to make the hateful journey…) These couple of lines describe the Hebrews (leode 'people, nation') fleeing from Egypt, with the 'laughter-smiths' (hleahtor-smiðum) representing the Egyptians who have just suffered the killing of their firstborn children. Here alyfed ('allowed') continues the alliteration on l from belocene ('restrained') in the previous line. In doing so, the poet draws our attention to the contrast between the constrained and grief-stricken Egyptians and the moment of (albeit toilsome) freedom for the Hebrews, the aural patterning underscoring the dichotomy between stasis and release.
End-linked alliteration is where the final non-alliterating stressed syllable of a b-verse alliterates with the non-alliterating stressed syllable of the following b-verse:

Rand-byrig waeron rofene, rodor swipode
Mere-deaða maest, modige swulton... Exodus 464-65 (Anlezark, 2011: 236) (The shield-walls were riven, the greatest sea-deaths scourged the sky, the brave ones died) This is an excerpt from later in Exodus describing the destruction of the Pharoah's army as it tries to cross the Red Sea, with swipode ('scourged') and swulton ('died') providing another alliterative dimension to the already strongly alliterating couplet. The alliteration combines with simple semantic repetition to keep our attention fixed on destruction.
Finally, we have back-linked alliteration, where the final non-alliterating stressed syllable of a b-verse alliteratives with the primary pattern of the previous line: Ahleop þa for haeleðum hilde-calla, bald beadu-hata, bord up ahof… Exodus 252-54 (Anlezark, 2011: 222 (The battle-herald, the brave messenger, leapt before the heroes, raised his shield) Here our battle-herald (hilde-calla, beadu-hata) steps before the army of the Hebrews to hush them so Moses can address his men. The alliteration draws attention to the battleherald's urgent movement (ahleop 'leapt', ahof 'raised'), linking up with poetic diction of war in the form of haeleðum ('heroes') and hilde-('battle'). The combination of backlinked alliteration and poetic vocabulary thus serves to render this couplet particularly attention-grabbing, a textual figuration of the battle-herald's own call to attention in the next few lines.
Intralinear and interlinear alliteration could be combined in quite elaborate ways, spilling across entire episodes or images to give a richly aural texture, one that must have been particularly resonant for a listening audience. Indeed, the complex intertwining of alliteration with stylistic features like assonance, stock lexical material and regular metrical conditions, would have been important for the mnemonic transmission of oral poetry (DuBois, 2011: 214-19). Other work has borne this out in areas like language learning, where alliteration is recognised as a valuable tool for memorisation (Boers and Lindstromberg, 2004). On an artistic level, Quirk (1963: 152-53) long ago recognised that alliteration was one of the primary ways in which poets could make semantic and thematic connections between collocates, often of a 'complementary or contrastive' type. Crucially, he also argued that a combination of alliteration and lexical variation was 'close to our starting point in estimating the original audience's pleasurable experience, as it is close to our starting point in criticism of the poetry today' (154). More recently, Harbus (2017: 209) has discussed how 'acoustic patterning' in Old English forged conceptual links across larger stretches of poetic discourse, while Griffith (2018: 152-53) emphasised that critics should be more sensitive to the way in which alliteration contributes to 'the meaning of individual passages', as well as how more elaborate alliteration relates to wider thematic and structural patterns.
Psycholinguistic research has in turn supported the work of stylisticians, with Egan et al. (2019) suggesting that alliteration supports attentional engagement and aids readers in linking concepts 'beyond the level of literal semantics' (111). While their study focuses explicitly on reading rather than listening, it is striking that they find that 'stylistic manipulation' such as alliteration seems to dynamically link 'semantic processing and attentional orienting' in the minds of readers (2019: 116). Where alliterating words are semantically 'congruent', and therefore straightforward to process (such as in a phrase like 'dazzling diamond'), their experiments found that 'semantic processing' increased, leading to continuous cognitive arousal (2019: 116). Interestingly, where the words were not semantically congruent (e.g. 'dangerous diamond'), readers would still perceive meaning relatedness, seemingly as a result of the effects of alliteration. As Stockwell notes (2009a: 58), the importance of phonological effects is something that modern readers tend to overlook as oral engagement with poetry has declined. The results of Egan et al.'s study, however, lends scientific support to the notion that alliteration was not just a decorative aspect of Old English poetry, but rather an integral part of the way poets made meaning and drew associations between different concepts (Pons-Sans, 2016: 571).

Apposition
Old English poetry makes extensive use of apposition: at the most basic level, this is the syntactic juxtaposition of linguistic elements via syntactic parallelism (Robinson, 1985: 3). We saw a good example of this in Exodus above in the line: 'swa him Moyses bebead,/maere mago-raeswa' (just as Moses commanded them, the famous leader of men). Here, maere mago-raeswa (famous leader of men) serves to laud Moses and maintain attention on an important dimension of his character. It is also not unusual to find a word or phrase modified by more than one appositional element, giving the poet the ability to vary different aspects of a character or other textual object. This is known as variation, and it is an important stylistic aspect of Old English poetry (Pons-Sanz, 2016: 270). In his important monograph on apposition, Robinson (1985: 60-61) suggested that the technique 'forces us to read reflectively', not to mention having the effect of slowing down the narrative so that 'a state or situation seems to be dwelled upon.' A good example from early on in Beowulf comes when Hrothgar considers the damage done by Grendel to his people (elements in variation with one another are in bold or underlined): Sw a ð a mǣlċeare maga Healfdenes sing ala s eað; ne mihte snotor haeleð, w ean onwenden; waes þaetġewin t o sw yð, l aþ ond longsum, þ e on ð a l eode bec om, n ydwracu n ıþgrim nihtbealwa mǣst.
Beowulf 189-93 (Fulk et al.: 9) (So the son of Halfdan brooded constantly upon that sorrow; nor could the wise hero turn from that woe; the hardship was too great, loathsome and long-lasting, that came upon the people, cruelly savage torment, the greatest of night-attacks.) This section nicely figures Halfdan's brooding via the variation of emotive words and phrases that each add a slightly different description to the sorrow (mǣlċeare) that kicks off the chain. Almost every appositional element is also in alliteration: w ean ('woe') witḣ gewin ('hardship'), l aþ ond longsum ('loathsome and long-lasting'), and nydwracu n ıþgrim ('cruelly savage torment') with nihtbealwa ('night-attacks'). Such extravagant repetition is in line with the sort of synonym variation that Stockwell (2009a: 24) identifies as being a key way in which textual attractors can maintain their salience. This is further reinforced in this example by the fact that the verbal repetition intersects with the 'empathetic recognisability' of a leader who is troubled by his inability to protect his people and see off the threat of Grendel (Stockwell, 2009a: 25). This impression is only heightened by the fact that Halfdan is called snotor haeleð ('wise hero') in the same breath that his wisdom is found wanting.

Poetic vocabulary
Old English poetry is known for its verbal richness and is replete with compound words that are richly evocative and frequently inventive; kennings-compounds with metaphorical meaning-are arguably the apotheosis of this lexical art, though in the present discussion they are less relevant (Godden, 1992: 499-500). According to Talmy's account of attention phenomena (2010: 268-69), open-class morphemes are more salient than bound, and can be more pronounced in compounds that require more cognitive work to unpack. As Davis-Secord (2016: 41-43) has noted, it is now agreed that compounds do not seem to undergo full lexicalisation, and are instead processed at least in part via the analysis of the semantic and syntactic relationship between their constituent elements; this seems to be true in both literate and oral contexts (45). Indeed, they require a fair degree of 'extralinguistic, contextual information' in general (25), and they were an integral part of the imaginative and intellectual life of literate writers. It is little wonder, then, that compounds are often reserved for important objects and characters, especially those that intersect with martial culture like warriors or weapons (Brady, 1979(Brady, , 1983. Compounds also always form part of the alliterative scheme of a line (Cronan, 1986: 146), and often occupy an entire half-line, which gives them extra textual salience. Compounds are thus attentionally important on a number of levels: they are mostly formed from content words (Kastovsky, 1992: 362-76), they are semantically and metaphorically evocative, and they can potentially occupy entire half-lines.
Joining compounds are a set of stock poetic simplexes: these were integral for filling out the metrical demands of an Old English poetic line, as performers and writers needed multiple synonyms for expressing frequent concepts like warrior or sword (Godden, 1992: 498-99, 509-11;Fulk, 2017: 243-44). Such poetic words were the bread and butter of a poet's craft, and most of them seem to have an ancient pedigree, with cognates being shared across different early Germanic languages (see for example Townend, 2015). They alliterate to a very high degree (Cronan, 1986), to the extent that their use in nonalliterating positions might be seen as stylistically marked in some way. Given their commonplace nature and presence within quite narrow semantic fields, it is probable that such words had low semantic salience in and of themselves (Boroff, 1962: 60;Cronan, 1986: 145), especially when compared to compounds. This could vary depending on their contextual use, however, and skilled poets might be able to exploit them for stylistic effect.

Beowulf and the depiction of grendel
Beowulf is found in the second codex of Cotton MS Vitellius A XV, which is traditionally dated to around the year 1000, though the poem itself was almost certainly composed earlier, perhaps in the first half of the 8th century (Neidorf, 2014(Neidorf, , 2017; for alternative views see Chase, 1981). The poem is centred around the career of the Geatish hero Beowulf in 6th-century Scandinavia, though this narrative is intertwined with the knotted dynastic politics of the period (Leneghan, 2020). While the structure of the poem is complex and filled with digression, it is largely centred around the eponymous hero's three battles with monstrous foes: Grendel, Grendel's mother, and, at the end of his life, a fire-breathing dragon. Until J.R.R. Tolkien's famous 1936 lecture 'Beowulf: the Monsters and the Critics', these individual fights tended to be perceived as unserious (if dramatically realised) distractions from wider historical and thematic interest, but as a result of his argument, it is now accepted that the struggles against the monsters are integral aspects of the poem's wider reflections on (among other things) the nature of good kingship and the value of worldly reputation (Fulk et al., 2009: cxxiii-cxxiv). After Tolkien's lecture, critics began to take Beowulf's adversaries more seriously, as well as taking a much greater interest in the art of the poem on its own terms, which largely resulted in an enormous amount of broadly historicist criticism, but also much greater consideration of style; it is perhaps no coincidence that his argument coincided with the emergence of formalism as a major strand of literary study in Anglophone countries (Fulk et al., 2009: cxxiv).
In early stylistic studies of Beowulf, no part of the poem garnered more attention than Beowulf's fight with Grendel. Tolkien himself (1963: 89) drew attention to the emphasis on Grendel's emotions during his battle with Beowulf, as well as his labelling as rinc, guma and haeleþ (stock poetic terms for 'man' or 'warrior'). The studies of Brodeur (1959) and Renoir (1962) were foundational in this regard, focusing on the poet's rendering of Grendel's approach to the hall of Heorot where Beowulf lay in wait; they emphasised the careful structuring of this section, the skilful modulation of point of view, and the general aura of terror suffusing the scene. Greenfield (1967Greenfield ( , 1972) offered a more rigorously stylistic analysis of the syntax of the entire passage, essentially giving linguistic substance to the observations of Brodeur andRenoir. In the 1980s, O'Brien O'Keeffe (1981) and Dieterich (1983) both argued that the poet sought to collapse the distinction between Beowulf and his monstrous adversary, with the former notingamong other things-the fact that Grendel appears to become more corporeal and human as he approaches Heorot, and that, during the fight itself, the syntax renders the actions and descriptions of the two antagonists ambiguous at certain points. Later in the same decade, Müller (1988) pushed back against various claims made by O'Brien O'Keeffe, arguing that the grammatical structure of the passage in fact seems to reify the division between man and monster. Lapidge (1993), noting that scholarly focus on the fight with Grendel neglected other parts of the poem, surveyed the various words used to refer to him elsewhere, and suggested that the sheer variety of terms-human and monstrous, corporeal and incorporeal-was designed to make Grendel seem indefinable and 'frighteningly unfamiliar' (393). In the following analysis, I focus on attentional phenomena and how these contribute to past arguments on the nature of Grendel, shifting the focus away from the fight itself. In section 4.1, I look at Grendel's initial introduction into the poem, focusing especially on alliterative patterning, before moving on to look at the use of lexis in section 4.2. Section 4.3 then considers the applicability of recent cognitive stylistic work on the textual representation of Grendel's psychology.

Grendel's introduction
Grendel is introduced to the audience in an early passage (ll. 86-114) that garnered special attention in Lapidge's study of the Beowulf-poet's handling of terror. As mentioned above, one of the central aspects of his argument is that the poet consistently resists giving fully figured descriptions of Beowulf's adversary, the unknown nature of his aspect and nature making him more intimidating since it lets the audience's imagination take over: Ð a se ellengǣst earfoðl ıce þr age geþolode, se þe in þ ystrum b ad þaet h e d ogora gehw am dr eam geh yrde hl udne in healle. Þǣr waes hearpan sw eġ, swutol sang scopes (86-90a).
(Then the powerful gǣst, he who remained in darkness, suffered in hardship for a long while, because each day he heard merriment loud in the hall-there was the harp's sound, the sweet song of the court poet).
The lines are effectively disorientating: the definite article in se ellengǣst ('the powerful gǣst') should help the audience establish an intended anaphoric referent, but we have yet to encounter Grendel. Lapidge (1993: 377-78) noted that the compound ellengǣst was 'not a precise, defining term', suggesting also that the second elementspelt < gaest> in the manuscript-could conceivably stand for the word gǣst ('spirit') or gaest ('guest'), the poet playing on the idea that Grendel is 'both ghost and guest'. Even accepting that the word is gǣst as most modern editions do, there is still a problem of conceptual selection: the DOE offers fourteen separate senses, of which eleven can be discounted for line 86, leaving senses 10 ('incorporeal spirit'), 11 ('living soul') and 12 ('an incorporeal being, spirit', including 12.b 'demon' and 12.b.ii 'living creature of malignant nature') as possible options. There is, then, some ambiguity at the outset in terms of knowing how gǣst should be interpreted, with significant differences depending on whether it profiles a spirit, a living human, or a demonic entity (incorporeal or otherwise). While this is difficult for modern-day readers, I would suggest that the same would have been true for medieval audiences given the multivalent nature of the word: humans are reliant on retrieving cognitive schema -'prototypical' bundles of readerly knowledge about the world (Emmott, 1998: 176-77)that help inform the understanding of discourse, but no obvious one presents itself here.
There is, however, an indication across lines 88-90 that we are dealing with an alien being: although Grendel's perception of the noise in the hall is construed objectively ('because he heard loud joys in the hall every day'), this is embedded within a global narrative construal that is subjective. Grendel's experience is thus filtered through a human conception of the world, and the audience is left with the uneasy question of why this unnamed being should suffer at the sound of human joys. Along with the uncertainty surrounding ellengǣst, the rest of the sentence over lines 86-90a also gives a general impression of inertia: the main clause is stative, describing how Grendel þr age geþolode ('suffered for a long time'), as is the following relative clause discussing how he b ad in þ ystrum ('remained in darkness'), while the immediately following main clause is cognitive but passive ('each day he heard merriment loud in the hall'). In contrast to the terrifying, constantly moving figure that we encounter later, Grendel's introduction is thus characterised by ambiguity and a notable stillness-not ideal qualities for a new textual attractor.
This shadowy introduction contrasts with an immediately following display of alliterative excess: though line 86 is conventional, with alliteration on two strongly stressed syllables per half line (ellen-and ear-), the immediately following lines consist of double alliteration with three alliterating syllables (þr age geþolode… þ ystrum) followed by a line with crossed alliteration (with d ogora and dr eam intersected with gehw am andġeh yrde, with additional secondary alliteration on h e). Line 89 is strongly linked with line 88, continuing the alliteration on h, with this line in turn linking up with line 90, and alliteration between sw eg and swutol. This initial alliterative pattern animates two strongly attractive images, one psychological and another noisy: Grendel's disturbed condition and the source of that disturbance in the form of the noise of hall life. Between lines 90 and 98, where most editions tend to end the immediately following sentence, we find six more lines including double alliteration rounded off with a final line of cross alliteration: (Then the powerful gaest, he who remained in darkness, suffered in hardship for a long while, because each day he heard merriment loud in the hall-there was the harp's sound, the sweet song of the court poet. He who knew how to recite the ancient tale of the origins of men said that the almighty made the earth, that bright plain, encircled by waters, and set the sun and moon in triumph to light up the land-dwellers with their beams, and he adorned the corners of the land with branches and leaves, he shaped life as well, all the creatures, those living things that moved about).
The aural texture of this passage can be read as deliberately coinciding with the socalled 'Song of Creation' wrought by Heorot's scop, the extravagant alliteration linking the creative energies of the poet with the Christian God's more cosmic powers (for the scriptural background to this passage see Helder, 1987). The Beowulf-poet also varies his alliterative pattern internally to direct audience attention: as mentioned above, the first line introducing Grendel only has a simple alliterating pattern between ellengaest and earfoðl ıce ('in hardship'), but the next standard line coming at 92 describes God's making of earth with similar assonance between AElmihtiga ('almighty') and eorðan ('earth'). The echoic nature of the alliterating syllables here, being formed respectively of vowel + l and vowel + r, draws attention to the ironic parallel between God and Grendel as the 'powerful gǣst', playing on the meaning of gǣst as the Christian 'holy ghost' (see the senses under DOE gǣst, 4. And 4.a). It also contrasts Grendel's passive suffering (earfoðl ıce) with the more dynamically attractive image of God's all-powerful agency (eorðan worhte). Across this entire passage our attention is modulated in such a way that the antagonist is immediately backgrounded even as he is introduced: not only is the compound used to describe him vague and possibly laced with irony, there are also other elements that make him a rather paradoxical textual attractor. While he has newness, topicality and definiteness, he lacks activeness, brightness (he dwells in þystrum, 'in darkness') and a general sense of fullness. Heorot, by contrast, is filled with noise (hl udne 'loud', sw eg 'sound', sang 'song', saegde 'said'), and God himself is active, his creation scintillatingly bright (wlitebeorhtne 'bright', sunnan 'sun', monan 'moon', l eohte 'light', gefraetwode 'decorated') and generative (landbuendum 'land-dwellers', leomum 'branches', l eafum 'leaves', l ıf 'life', cynna 'life', cwice 'living things'). The 'Song of Creation' is rounded off with a line of crossed alliteration that underlines this abundance with a flourish, cynna alliterating in variation with near-synonym cwice that roam around the earth. The ellorgǣst is thus occluded (Stockwell, 2009: 21-22) from our attention in almost the same moment that he is introduced, with readerly focus being shifted toward God and creation, the deity's ineluctable might reified through a skilful consonance between content and form, and underpinned by a range of attractors.
Having had our attention shifted to the heavenly, the immediately following section running from lines 99-114 functions as the introduction to Grendel 'proper', revivifying Grendel as an attractor: (Thus the men lived in joy blessedly, until something began to commit crimes, a fiend from hell. That fearsome demon was called Grendel, powerful walker on the borderland, who held the moors, fen and fastness. That cursed man dwelt for along time in the land of monsters, after the creator had condemned him among Cain's descendentswhen he killed Abel, the eternal Lord avenged that murder).
The poet makes extended use of double alliteration across lines 99-104, connecting a series of important words: we are told explicitly that he is a f eond ('enemy; demon') who carries out fyrene ('sins, crimes; violence') in line 101, with the poet reusing the term gǣst in the following line in alliteration with the antagonist's name. In revitalising Grendel, the emphasis is placed on his otherworldliness: while the respective meanings of f eond and gǣst are potentially ambiguous in isolation, the fact that both are used in variational proximity encourages the selection of a 'demon' or 'monster' schema in imagining Grendel, particularly since the former word is profiled explicitly against hell (helle). Following the findings of Egan et al. (2019), the words that are in alliteration with one another serve to reinforce the semantic association between the Grendel and the hellishly supernatural. This is given further grammatical reinforcement in line 102, where Grendel is grounded both intrinsically (Grendel) and overtly (s e grimma gǣst) within an imperfective, stative construction. We need to be wary of modern editions and punctuation influencing our interpretation of the original Old English, and I think there is a good argument to be made for the half-line f eond on helle being linked with both the preceding and following clauses, and thus further underscoring his otherness: (until something began to commit crimes, a fiend from hell, that fearsome spirit was called Grendel, powerful walker on the borderland).
Following this repunctuation, across lines 101-104 we would then have three-part variation, with f eond and gǣst mutually reinforcing Grendel's monstrous nature, and mǣre mearcstapa ('powerful walker on the borderland') grounding him against the tangible world of creation. The interconnection between these lines is given further emphasis via end-linked alliteration across helle, h aten and h eold. Despite being revivified as an attractor, then, Grendel largely remains static, a direct contrast with the agency ascribed to the God who created (worhte), placed (gesette), adorned (gefraetwade) and shaped (gesc eop) the earth. His inertia is underlined further immediately following this when he is called a 'cursed man' (wonsaeli wer) who has dwelt in the land of monsters (fifelcynnes eard), a description that serves to associate Grendel with a key stock figure from Old English literature: the exiled individual, forced to live on the fringes of society (Baird, 1966;Hill, 1975).
As mentioned above, there has been anxiety surrounding the complex net of (seemingly contradictory) words used to describe Grendel throughout the poem, though critics have tended to treat each lexical item in an atomised fashion, shorn of wider context. Some like Irving (1989: 10-12) have gone so far as to suggest that the poet was not particularly concerned with consistency of Grendel's image, using descriptors that were only intended to have a fleeting effectiveness in the context of ephemeral oral performance. What my above analysis suggests, however, is that the demonic, marginal aspect of Grendel is emphasised via a careful combination of alliterative patterning and lexical variation, and this is bolstered in lines 111-14a where various creatures are said to descend from Cain, placing him among monsters (following exegetical tradition, see Reinhard, 2013). The description wonsaeli wer nods to his ultimate connection with his ancestor Cain, but his humanity is weakened: the phrase appears sporadically in Old English poetry in reference to particularly evil men, notably in the wisdom poem Maxims I, where we are told that 'won-saelig mon genimeð him wulfas to geferen' (Bjork, 2014: 76, 'a cursed man takes wolves as companions'). The word wer itself is furthermore highly generic, being found across both poetry and prose, and contrasts some of the more specialised poetic vocabulary used of Grendel later in the poem (a whole-word search for wer in the Dictionary of Old English Corpus yields 1459 hits) (Cameron et al, 2018). As Mearns (2015: 215, 222) has argued, what constituted 'the supernatural' in the early medieval English imagination was closely aligned with notions a broader notion of the other associated with alienness and the natural world, so wonsaeli wer is entirely congruent with Grendel's monstrosity. The 'demon' or 'monster' schema that has been invoked is thus modified rather than completely discarded at this point in the narrative, establishing that the wonseli wer is not just outside the proper bounds of human community, but outside humanity itself.
We should also consider the appositive style of Old English poetry and how this intersects with cognitive linguistic ideas of scanning-that is, the way language users process events. I argue that this passage encourages the reader towards summary scanning, or in other words, the compression and non-linear interpretation of textual information. As Nuttall (2015: 29) notes, summary scanning essentially 'involves examining multiple targets in a cumulative manner, building up to a gestalt conception which is apprehended as a whole'. Old English poetry is particularly geared towards encouraging summary scanning: among other things, nominalising compounds, structurally binding alliteration, the frequent variation of diction and phrasing, not to mention use of apposition, all contribute to the somewhat 'static' nature of the form (Robinson, 1985: 60-61). The multiplicity of different descriptors applied to Grendel, then, are not simply processed one-by-one, but are instead taken together as different dimensions of his being, even as certain aspects (f eond, gǣst) are afforded greater salience. While the poet maybe gestures to his residual humanity with wonsaeli wer, the monstrous aspect of his lineage is quickly revivified and his separation from humankind underlined in God's punishment of Cain: Ac h e hine feor forwrac Metod for þ y m ane mancynne fram.
(But the creator chased him far from mankind for that crime. From then arose all foul creatures, giants and elves and evil monsters, and the giants who fought with God for a long time).
Here, the poet extends the alliteration on /f/ in line 109, with secondary alliteration on intensifying prefix for-(the final stressed alliteration being -wrac) emphasising how far Cain was expelled, and is further backlinked with the alliteration of line 110, tying in fram ('away from'). Line 112 in turn shows continued vowel alliteration from line 111, again with assonantal secondary alliteration on on-, emphasising the alienness (untydas) and relationship (ealle) of these beings. Throughout the passage introducing Grendel, then, we see that the poet stylistically foregrounds Grendel's terrifying alterity at the expense of whatever vestigial humanity he might retain.

'
In the shape of a man': Grendel and stock poetic vocabulary In the section above we saw that the Beowulf-poet varies the terms used to refer to Grendel, from simplexes such as f eond or gǣst, to more evocative compounds like ellengǣst and mearcstapa. Later in the poem, Grendel is called (among many other things), sceaduganga, m anscaða and synscaða. Each of these compounds profile a particular aspect of Grendel that reinforces his monstrous nature, whether directly as in the evil harm of m anscaða, or more obliquely as in mearcstapa and the potential inhuman implications that accompany the invocation of an 'exile' schema (Mearns, 2015: 222). This variation in the words used of Grendel can be thought of as the relationship between reference points and targets in Cognitive Grammar terms (Stockwell, 2009: 180, Harrison, 2017: 73-76, Nuttall, 2018. In effect, Grendel functions as the global, discoursal reference point around which variational lexis (or targets) coalesce and to give him meaning and a sense of literary texture (Harrison, 2017: 74). While targets are often much broader than simply lexical synonyms (see especially Stockwell's analysis of Bleak House, 2009: 178-83), for the case of Old English poetic form it is useful to think about the stylistic variation mentioned in section 3.2 in terms of reference point and target. Collectively, the set of targets and associations elicited by a reference point are its dominion (Harrison, 2017: 33), that is the 'network of all the possibilities that [a] reference evokes from experiential memory' (Stockwell, 2009: 180;Nuttall, 2018: 94). Dominions can evoke anything from 'conventional meanings for an expression' to 'personal, experiential associations, intertextual connections, and register-specific connotations' (Nuttall, 2018: 94). Whether such associations are active is dependent on their 'centrality'-that is, whether they are more likely to be activated in the minds of an audience or not (Nuttall, 2018: 94).
Throughout the poem, the dominion surrounding Grendel integrates associations with monsters, demons, death, destruction, and exile into a relatively coherent whole. Methodologically, it is worth distinguishing between local and global reference points (Harrison, 2017: 42). The former might be important across a passage, as in the example of sceaduganga acting as our springboard for the analysis in section 4.1. Globally, a reference point might be a character, or a setting, or any other major textual element that spans the entirety of a text (Harrison, 2017: 42;73-101). While reference points, targets and their encompassing dominions have not been discussed in detail in terms of attention, I suggest in this section that the coherence-or not-of the targets relating to a reference point are important in how readerly attention is manipulated. Targets that distract from the overall dominion forming around a reference point have the potential to draw or attention, or alternatively incongruent targets might end up neglected depending on their contextual surroundings.
While most of the targets in relation to Grendel cohere with the idea of him as a monstrous being, at various points the poet interlaces simplexes that are normally used of humans into his description of Grendel-as we saw in the previous section with the example of wonsaeli wer. At lines 1349b-53, for example, after Beowulf has defeated Grendel, Hrothgar provides his own revealing comment on the Danes' perceptions of the monsters that have plagued Heorot: Ðǣra oðer waes, þaes þe h ıe gewislicost gewitan meahton, idese onl ıcnaes; oðer earmsceapen on weres waestmum wraecl astas traed, naefne h e waes m ara þonne ǣnig man oðer.
(The second of them [i.e. Grendel's mother], as far as I could most clearly tell, had a woman's likeness; the other was wretchedly shaped in the form of a man, except that he was bigger than any other).
The king is clear that both Grendel and his mother are roughly akin to humans in shape, if not size. In a highly influential close reading of Grendel's approach to Heorot, O'Brien O'Keeffe (1981: 487-88) suggested that what we see is an increasing solidification of Grendel as he moves from the dark and monstrous sceadugenga through to a rinc ('warrior; man') at the point that he reaches the doors of the hall: 'He journeys to the hall as an exile, a man deprived of the joys of society'. She goes on to suggest that Grendel 'assumes a human identity' in the vicinity of the hall and that the significance of this is underlined by a further four uses of rinc-either as a simplex or a compound-during the following combat (though not in reference to Grendel himself).
In addition to this, Watanabe (2009: 192) argues that when Grendel first sees Heorot, the compounds put emphasis the hall as the centre of heroic community: W od under wolcnum t o þaes þe h e w ınreced, goldsele gumena gearwost wisse, fǣttum f ahne (714-16a).
(He walked under the clouds until he could most clearly see the wine-hall, the gold-hall of warriors, richly adorned).
The poet profiles different social aspects of the hall as a place of festivity (winreced), of material wealth (goldsele), and martial culture (gumena), all emphasised with back-linked alliteration of wisse. These are the elements of aristocratic human life that Grendel and his ancestors have been denied, something that is underscored again at the point that he is described as rinc: (He came then to the hall, the warrior travelling, deprived of joys. The doors immediately split apart, fast bands forged in fire, after he touched them with his hands).
Alongside earlier wer, the use of rinc adds to the matrix of possible targets for Grendel, and thus potentially also expanding the dominion to encompass images of humanity. I would suggest, however, that rinc is only weakly attention grabbing and therefore only marginally integrated into the overall dominion connected to Grendel. In the above passage, for example, our attention is directed in such a way as to render any hint of Grendel's humanity rather hollow. Watanabe (2009: 192) notes that, after being presented with a multifaceted description of the hall, in the following section the poet reverts to much less evocative words, including reced, and this is the word that is in simple alliteration with rinc at line 720. These are stock poetic words that are shorn of any particularly rich imagery, especially in comparison with other descriptions of Grendel (sceadugenga, synscaða, ellengaest etc.) (Cronan, 1986: 148). Indeed, after this simple alliterating line with its generic vocabulary, Grendel's separation from human joys is emphasised (dreamum bedaeled), with the double alliterating line leading straight into violently active attractor of the strong, fire-forged doors of the hall bursting open at his mere touch (Stockwell 2009: 25). The combination of double alliteration and backalliteration serve to emphasise his intrusion in contrast to the comparatively weak alliterating scheme of line 720, resting as it does on commonplace vocabulary. The wider stylistic choices of the passage thus distract from rinc and the strength of the connotative value it might have in relation to Grendel as a global reference point.
Grendel is, however, referred to with other generic terms for 'man' or 'warrior' later in the poem: twice as guma and one each for haeleð, r enweard, healþegn and hilderinc (O'Brien O'Keeffe, 1981: 486). Both healþegn ('hall-thegn') and renweard ('mighty guard') appear to be used ironically in a quite similar fashion: the first is applied to Grendel after he has terrorised Heorot, while the latter refers collectively (pl. renweardas) to both Grendel and Beowulf during their chaotic battle in the hall itself. I agree with Irving (1966: 164-65) in seeing these words as ironically casting Grendel as a 'mockwarrior': in neither instance is Grendel contributing to the protection of the hall, and is instead being actively destructive, the words' effectiveness as attractors being only to throw Grendel's inhumanity into even sharper relief. Within the structure of the remainder of the poem, the other poetic words are used in quite a distinctive way: while Beowulf calls Grendel þyrs ('giant') at line 426 before he has encountered his adversary, the poet later has the hero draw on stock poetic synonyms for 'man/warrior' instead. Indeed, outside the narrator, he is the only character to use such terms of Grendel. Guma is used at line 973 when he is recounting his fight with Grendel: Beowulf first describes the dismemberment of Grendel's arm, linking f eond with folme (which is in turn in variation with earm ond eaxle), with the recognisable bodily elements of his foe profiled-but crucially leaving the rest of Grendel's appearance to our imagination. The alliteration in line 973 then places emphasis on his wretchedness, the line linking f easceaft and fr ofre, rather than the label guma itself, which is striking given that stock poetic words tend to alliterate at very high rates (Cronan, 1986: 146-47). Given the role alliteration plays in attentional orienting (Egan et al., 2019: 216), the fact that guma is left out of the alliterating scheme for the line serves to de-emphasise his humanity at the expense of his emotional and social condition, and picks up again on the f-alliteration a couple of lines prior and his status as f eond. Similarly, at line 2072, when Beowulf tells Hygelac of his grappling with Grendel, he recounts how they engaged in hondraes haeleða ('wrestling of heroes'). The use of both guma and haeleð as potential targets, then, should be taken in the context of Beowulf boastfully talking himself up in the presence of his lordly benefactor. It would make sense that he might use stock poetic terms to this effect, invoking a schema that calls to mind heroic narrative, even as he controls the image of Grendel with the instantiating local reference points f eond and gǣst in the same breath.
The other instance of guma and the stock compound hilderinc are embedded within the narrator's descriptions: the latter occurs at line 986 shortly after Beowulf ends his speech in which he calls Grendel guma, and the poet is describing the monstrosity of the injured opponent's talon: [aeþelingas]  As Cavell (2014: 163-67) has argued, this section others Grendel's body by blurring the lines between his natural, if lethal, appendage and a warrior's weapon, turning his arm into a token of victory. It would make sense, then, that the narrator would opt for a word that attempts to absorb Grendel into a recognisable heroic world even as he distances him: the narrator is careful to call him a f eond first-much like Beowulf does in his speech, and the narrator does earlier at Grendel's introduction-before containing the target hilderinc within a description that emphasises alterity. Much like the use of rinc before Grendel enters Heorot is quickly undermined by his bursting through the doors, here his description as a hilderinc is connected alliteratively with target words profiling different aspects of his otherness (haeþen, handsporu), with secondary alliteration and rhyme at 987a with the adjective unh eoru. The narrator's use of guma comes as Hrothgar receives the sword hilt retrieved from the mere where Grendel and his mother dwelt: Once again, the reference to Grendel as a warrior is preceded by a description of him (and his mother) in demonic terms through the reference point d eofla, with the double alliteration of line 1682 giving emphasis to his hostility and opposition. In contrast to those words that invoke the image of Grendel as a demonic, monstrous entity, the stock poetic words for 'man/warrior' afford very little granularity in Cognitive Grammar terms (Langacker, 2008: 55-57). While they are not as low salience as function words, I would suggest that they are lower when compared to richer compound terms and the repeated appellations that fall within the semantic field of f eond (including gǣst and d eofol). Grendel's integration into the world of men is as weak in stylistic terms as it is in narrative reality, the stock poetic language contrasting with the much richer vocabulary used of his monstrous aspect.
On a global level, then, Grendel's effectiveness as a textual attractor rests heavily on the fact that there is consistent synonym variation that places the focus on his 'aesthetic difference from the norm' (Stockwell, 2009: 24-25). When there is a departure from terms that imply alienness to stock poetic terms for man or warrior, these are not necessarily a nudge for us to reconsider Grendel's humanity, but instead contextually serve to emphasise his alien nature through stylistic contrast. Relevant to this discussion is the notion of 'shallow processing' Sanford and Emmott, 2012: 95-96), or the idea that, for the most part, readerly processing of discourse is relatively surface level. I argue that our stock poetic simplexes relating to the man/warrior domains, unless used in complex coreference with both other congruent variant synonyms and imaginative poetic coinages (especially compounds), do not elicit the same degree of 'semantic analysis' in a reader (Sanford and Emmott, 2012: 95-96), nor would they encourage the sort of backtracking and reassessment of previous textual information that more sustained deviance from the established monster schema might (Nuttall, 2018: 108). Ultimately, readerly attention is never distracted to the extent that Grendel is fully humanised.

Monstrous intent
One final aspect of how our attention is manipulated throughout the poem is in regard to the mental representation of Grendel. Various scholars have noted that we are treated to some of his thoughts and feelings during his fight at Heorot: O'Brien O' Keeffe (1981: 484), for example, points to the poet's emphasis of his mental statesbolgen, yrremodwhile Larrington (2001: 254) notes that Beowulf and Grendel share similar mental states through their descriptions as bolgenmod and gebolgen respectively. However, Cronan (2017: 647-48) has recently argued that Grendel's 'human-like quality only serves to emphasise the spiritual and moral vacuity of his existence'. In particular, he notes that Beowulf displays a much calmer sense of purpose in the lead up to his fight with Grendel, first by showing a more controlled rage, and second by continuing to observe his enemy even after he's entered the hall. This focus on Grendel's psychological representation is an interesting one, but has received relatively little attention from stylisticians. Drawing on Nuttall's recent work (2015, 2018) on how readers assign motivation and thought to nonhuman entities, this final section will briefly suggest a few ways in which the poet construes Grendel as being motivated by instinct rather than intellect.
First, it is worth fleshing out Cronan's suggestion in stylistic terms. In highlighting Beowulf's controlled nature in comparison to Grendel's more instinctive aspect, we can compare the abovementioned lines that show lexical parallelism with the element bolgen: Ac h e waeccende wr aþum on andan b ad bolgenm od beadwa geþinges (708-9) (And he [i.e. Beowulf], waking [i.e. awake] for the enemy in anger, awaited greatly enraged the outcome of battle). onbrǣd þ a bealoh ydig, ð a (h e ge)bolgen waes (723) (He [i.e. Grendel] stalked then, intent on evil, he was swollen with rage).
While the alliterative echo of this word brings both combatants' psychological states into alignment, there are some crucial differences. Cronan is right that Beowulf shows more control: his mind is set on the outcome of the battle, his anger subordinated to a series of appositional words and phrases. He may be keyed up for the fight ahead, but his aggression is emphasised as being secondary to his sense of purpose. Line 723 echoes the structure of line 709 alliteratively and structurally, the first half-line consisting of a verb then adjective, but the second half-line is used to reemphasise Grendel's rage, linking together bealohydig and gebolgen. Crucially, this is expressed in an intensive relational process that emphasises the fact that his anger is the defining aspect of his motivation. A few lines later at 726a, the idea that Grendel's anger controls his action is further underlined when we are told he eode yrremod ('went angry in mind') across the floor of Heorot. Grendel's effectiveness as an attractor in the build-up to combat thus rests on the fact that his mental state is figured physically through his actions, while Beowulf's internal anger serves to background him. This interpretation of Grendel also feeds subtly into the overall dominion surrounding him as a global reference point, connecting the more demonic or monstrous vocabulary associated with him to a mental state that appears more instinctive.
In her analysis of Richard Matheson's I Am Legend, Nuttall argues (2015: 29) that one of the key ways in which the vampiric antagonists are dehumanised is by presenting their behaviour as vague and undirected. I argue that something similar appears to be happening with the presentation of Grendel. As his attacks on Heorot begin from line 115 onward, he is presented as a constantly moving attractor with some degree of intent: he seeks out the hall, he snatches thirty Danes, and we are told that he fought against all that was right. However, we are never really given a clear idea of his motivation other than that he is inherently evil and, in some way, disturbed by Heorot's noisiness (though see the possible implication of this below). Indeed, once Grendel has driven the Danes away from Heorot, he does not seem to do much with the gutted hall: (But the powerful one, the dark death-shadow, was hunting warriors young and old, remained and lied in wait… thus that enemy of mankind, the terrifying stalker, performed many crimes, hard humiliations; he occupied Heorot).
Grendel's actions are presented as continuous, with the initial past progressive construction incorporating the verb ehtan: this can generally mean 'to attack', but also has a more specific subsense that encompasses animals attacking humans (DOE s.v. ehtan 2.b.). The subsequent attacks on the young and old are thus rendered indiscriminate, their intent unknowable. Later, during Grendel and Beowulf's fight, we are given a limited view of his interiority, but even then, the poet's language renders this ambiguous: at three points, Grendel mynte besyrwan (intended to ensnare, ll. 712-13), mynte gedaelde (intended to separate, l. 731) and mynte gewindan (intended to flee, ll. 762-63). While besyrwan and gedaelan might imply some sort of reasoning, it is important to be mindful of the fact that the description of Grendel is subjectively construed via the narrator, and each of these actions conveys impulse: catching prey, killing prey, fleeing from danger. Overall, the Beowulf-poet stresses these parts of Grendel's emotional state at the expense of points at which his inner life seems to show complexity, such as in his anguished reaction to the sounds of hall-life at Heorot we saw in 4.1. The overall result is that Grendel's humanity is consisitently backgrounded and decentered psychologically (Nuttall, 2018: 96), with his effectiveness as an attractor resting on his activeness and aesthetic difference from the norm, especially through his 'violence' and 'wilfulness' (Stockwell, 2009a: 25).

Conclusion
Throughout this article I have sought to show how paying attention to attentional phenomena in Old English poetry can shed new light on old arguments surrounding the stylistic portrayal of Grendel. While I have not sought to provide a radical new interpretation, my analysis has shown how the poet draws careful attention to certain aspects of Grendel's character over others, and that he is ultimately only very weakly humanised. His monstrous, animal nature remains the most central part of him as a textual attractor throughout. This is not, however, to argue that Grendel is simply a monster, only that the Beowulf-poet leads us in that direction throughout the text through careful stylistic choices, even when glimpses of an alternative interpretation are provided. My outline of attentional features in Old English remains only provisional and there are plenty of further avenues for exploration. Any treatment of how attention and poetic metre intersect, for example, would likely require a dedicated study. I hope that my analysis will at least emphasise the importance of considering the stylistic aspects of the Old English poetic tradition in attentional terms and potentially encourage more engagement with cognitive stylistic approaches to early medieval poetry.