Roman and Runic in the Anglo-Saxon Inscriptions at Monte Sant’Angelo: A Sociolinguistic Approach

This paper addresses the Anglo-Saxon personal name inscriptions at Monte Sant’ Angelo in Southern Italy from a sociolinguistic angle. The main interest lies in the mix between Roman and runic writing and its inter pretation in the light of indi vidual literacy and the cultural context of medieval pilgrimage. Four from a total of five inscriptions were written in runes; two of these show sig nifi cant in fluence from Anglo-Saxon scribal practices and Roman epi graphic writ ing. The fifth Anglo-Saxon name is written entirely in Roman letters. Draw ing on theo retical approaches from modern sociolinguistic studies of multi lin gualism in writ ing, this study suggests that the use of mixed Roman-runic prac tices reflects the biscriptal background of the respective carvers and was applied in situ to indi vidualize the inscriptions. However, not all the in scrip tions show such a mix; hence either skill or personal preference varied among the pilgrims. The prac tice of mixing evident in the runic inscriptions does not fully correspond to previ ously described features of multilingual and multi scrip tal writing, which is why a new term, “heterographia”, has been coined in this study to include mix ing not only in a language and a writing system, but also on a graphetic and ortho graphic level. Finally, the use of runes or Roman script for one’s personal name is inter preted as an expression of social identity dependent on the person’s social embedding.

abroad. There are within this heterogeneous group of inscriptions ten instances in Italy which form a subgroup from one particular historical context: between A.D. c. 650 and 850, Anglo-Saxon Christians, probably pil grims, traveled to Rome and Jerusalem and along the way they inscribed their names on the walls of the Roman catacombs of Commodilla, St Marcel linus, St Peter and St Pamphilus and at the Sanctuary of San Michele in Monte Sant'Angelo (Gargano).
These graffiti were first discovered by the Italian epigraphist Carlo Carletti (Carletti 1980(Carletti , 1984 and thereafter presented largely separately in different publications (Mastrelli 1980;Arcamone 1981Arcamone , 1992Derolez andSchwab 1983, 1994). Ray Page remarks on some of the inscriptions in his Intro duction to English Runes (1999), and Maria Giovanna Arcamone (2007) subse quently presents an overview of all the interpretable inscriptions, includ ing pictures and a description of the epigraphic, runographic and etymo logical characteristics of the inscribed personal names. The most recent find in the catacombs of Pamphilus in Rome is presented by Luisa Izzi (2014, 147 and n. 28) and "possibly read as '[+ +] CYNRIC'". The period to which the inscriptions are historically dated begins with the conversion of the Anglo-Saxons in the beginning of the 600s, and from 650 there are accounts of Anglo-Saxon pilgrims to mainland Europe. The terminus ante quem, on the other hand, is determined by the alleged Saracen attack of the sanctuary after which the place was abandoned (cf. Foxhall Forbes 2019, 178 f.).
In her article, Arcamone (2007, 128, 137) points out that the pilgrims not only knew how to carve runes but, as Christians, must also have mastered the Roman alphabet. On the walls of the Sanctuary of San Michele and in the Roman catacombs there is further evidence for this assumption since we find Anglo-Saxon names written not only in runes but also in Roman letters (Arcamone 2007, 128;Carletti 1984Carletti -85, 2002. Page also (1994, 182;1999, 224) mentioned the inscriptions' connection to Roman alphabetic literacy and a possible learned background of the Anglo-Saxon pilgrims in Italy but did not delve deeper into the matter. Thus, previous runological research had no systematic focus on the effect Roman writing and the social and situational context of pilgrimage might have had on the runic inscrip tions far away from the "runic homelands".
In this paper, I address such Roman-runic "contact effects" in the inscrip tions of the Anglo-Saxon pilgrims but limit my study here to the inscriptions at the Sanctuary of San Michele. 1 My main aim is to con-trib ute to the hitherto neglected social contextualization of the graffiti, examin ing the literate skills of the writing pilgrims and possible social mean ings of the inscriptions by taking a sociolinguistic approach. In a recent paper, Helen Foxhall Forbes (2019, 170) also points out that the main focus of scholarly research has so far been on the philological and ono mastic features of the inscriptions while spiritual and social aspects have not been sufficiently addressed. Forbes focuses her study on the situa tional and contextual embedding of the graffiti on the walls at Monte Sant' Angelo and seeks to explore their significance for issues of devotion and identity with a historical and theological approach (see below).
In my analysis, I follow the method applied in modern sociolinguistic studies of writing systems in contact and include visual, material and linguistic characteristics of writing (cf. Sebba 2012, 102-06), the first con cerning such matters as spatial arrangement, layout and script style; the second technical execution. Given that the Anglo-Saxon inscriptions in Italy ap pear only as single, syntactically isolated names, the linguistic analysis will only touch on orthography. Mark Sebba's studies have shown not only that visual aspects in particular are often neglected in research on multi lingual texts (Sebba 2012, 102) but also that such features can serve a socio linguistic function, such as underlining equality, for example, or expressing a certain identity or ideology (Sebba 2009, 39 f.;2012, 109).
In the following, I will first present the historical and local context of the Anglo-Saxon inscriptions at Monte Sant'Angelo in Gargano. The effects of contact will then be separately addressed for each inscription in the section "Roman and runic writing in the Anglo-Saxon names". In the sub sequent sections, sociolinguistic explanations for the use and choice of script will be discussed in the light of modern theories of multilingual lan guage use and with respect to the historical and local context.

Anglo-Saxons in Italy
Pilgrimages were extremely popular among medieval Christians, and both clerics and laypeople undertook journeys to the Holy Lands or other places of great religious significance (cf. e.g. Webb 1999). The Anglo-Saxon pil grims who visited Monte Sant'Angelo most likely continued their journey to Jerusalem by boat from ports close to Gargano (Sinisi 2014, 47, 59 f.). Their pilgrimage was probably religionis causa, i.e. undertaken in order to find "a closer link with the divinity in order to reach a state of perfection of the soul". The function of the pilgrims' graffiti should cer tainly be seen in this religious context as having spiritual, devotional and commemorative purpose. The names may represent either a kind of ratification of "a pact with God that they had come all that way to sign" (p. 60) or, more likely, may be compared with medieval memorial practices known from Libri vitae, where people left their name to represent their souls, to be prayed for and commemorated during their lifetime and in the after life (cf. also Foxhall Forbes 2019, 188-92).
There is no historical data on the Anglo-Saxon pilgrims in Gargano in particular (cf. Handley 2013, 759-61) and hence we do not know who the per sons behind the names investigated in this article were, whether they were clerics or laypeople, what exactly their motivations for the pil grimage were or whether they were traveling in a group or on their own. It is more over unclear whether the names indicate the pilgrims them selves or those close to them (p. 747). According to Carletti (2002, 354), the large number of Anglo-Saxon name graffiti in uncial writing in the cata combs of Rome points to a learned and clerical background for the writers. It might, however, be reasonably assumed that the group of people traveling to Italy was a heterogeneous one, as their level of literacy may also have been.

Literacy in Anglo-Saxon England
When Christian Latin literacy was introduced in Anglo-Saxon England in the 500s, it "did not spring up in a scriptless desert" (Derolez 1990, 400). Runic writing was found at that time, although evidence is scarce (Barnes 2012, 42 f.;Page 1999, 25), and the Anglo-Saxons presumably already had knowl edge of the Roman script from previous contacts with the Continent (Page 1999, 213) and from Latin inscriptions from Roman colonial times still present in the Anglo-Saxon linguistic landscape.
The epigraphic and manuscript evidence of Anglo-Saxon England strongly suggests that runes and Roman script existed contemporaneously from early on and for a long period. Much Anglo-Saxon runic writing is mixed or inter woven with Roman script and Latin (cf. most recently Okasha 2018). Objects variously occur with inscriptions in both scripts, runic and Roman, and both languages, vernacular and Latin, side by side (see below); alter natively, occasional single runes occur in Roman inscrip tions and manu script texts (cf. Page 1999, 213-24, andOkasha 2018 for examples).
As concerns the functional distribution of the runic and Roman writing systems in Anglo-Saxon England, the Roman alphabet was used more frequently than the runic one in both epigraphic and manuscript writing. In manuscripts, runes are - apart from the characters ê and W which were borrowed into insular script - clearly less common than Roman letters. The two writing systems seem otherwise to have enjoyed a comparable status since they occur in similar contexts, although, in general, Latin texts appear in Roman script rather than runic whereas both scripts were used for Old English (Orton 2014, 231 f.;Okasha 2018, 35). For personal names, a socially coded distribution can be noted on East Anglian coins from the 700s where names of royals were written in the Roman alphabet whereas names of officials or moneyers appear in runes (Page 2003, 551). Regard ing the epigraphic use of the two writing systems, Parsons (1994, 110 f.) shows that their spatial distribution is very similar. The evidence clearly indicates that active knowledge of both Roman and runic writing systems existed contemporaneously (Okasha 2018, 35).
Of the total of around 350 inscriptions from Anglo-Saxon England, seven teen inscriptions dating from the 700s to the 800s contain both runic and Roman writing according to Elisabeth Okasha (2018, 31 f., 34, 41 f.), who lists them and classifies them into four groups. The first group comprises Roman-script texts with occasional single runes (e.g. the Chester-le-Street stone with the personal name EADmVnD); the second only comprises the Franks casket, the texts are runic but for one in Roman script; the third shows parallel inscriptions in both scripts (e.g. the Falstone stone, see below); the fourth contains different texts in each script (e.g. the Hack ness stone with a prayer in Latin written in Roman script and a now indecipherable runic inscription and a text in runic ciphers).
On a syntactic level, the two scripts are kept apart and not mixed. Par al lel inscriptions where the same text appears in Roman and runic on the same object provide particular evidence for this practice. An example is the Falstone stone with an Old English memorial formula * aeftae… becun * f… gebidaedþe * saule in runes and … ITA…AEFT * R HROETHBERHT * BECUNAEF… EOMAEƷEBIDAEDDERSAULE in insular majuscules, the translation being '(NN set up) a monument after Hroeth berht after his uncle. Pray for his soul' (cf. Page 1999, 142, trans-liter ation according to RuneS-DB). Okasha explains this parallel use of scripts as "artistic, to exhibit knowledge" (Okasha 1971, 72;cf. also 2018, 40). Chris tine Fell (1994, 129) suggests two further expla nations, of which the first is "to reach as wide an audience as possible and present your text for all who are literate in either script", which corresponds to sociolinguistic con sid er ations on modern parallel script use (cf. Sebba 2012, 109, 112). Okasha (2018, 39) considers this explanation unlikely, however, since the Anglo-Saxon context makes a reader capable of reading only one script improbable. Fell's second explanation, that the multi scriptal practice for epitaphs might be a reference to Christ's inscription on the cross, which according to the Gospel of St Luke 23:38 was written in letters of Greek, Latin and Hebrew is more far-fetched and fairly unlikely. Rather, I consider this use of parallel script might be linked to aspects of local identity (see section "Script choice and identity in personal names" below). As noted above, recent socio linguistic research has shown that differ ent scripts can express such features and thus, as regards choice and use of writing systems, it is not only readability which is significant (Sebba 2009, 39 f.;2012, 109).

Local context of the inscriptions at Monte Sant'Angelo
The runic inscriptions at Monte Sant'Angelo in Gargano in Southern Italy are situated in the cave-sanctuary of San Michele, a sacred Christian place where the Michaelian Cult has been practiced since the sixth century (Sinisi 2014, 51). It is a place of religious importance and spiritual power to which pilgrims have traveled since the beginnings of Christianity in Europe.
In 1949, when the sanctuary was renovated after an earthquake, a late antique chapel with a long porticus to the sanctuary was rediscovered (cf. Arca mone 2007, 129). 2 On its walls were found around two hundred medieval graffiti carved by pilgrims. These walls comprise several historical layers of inscriptions accidentally brought together. The graffiti are mainly personal name inscriptions, but symbols such as the signum crucis and short prayers can also be found (cf. Carletti 1980;Foxhall Forbes 2019, 184). Among these inscriptions are, as mentioned above, five interpretable Anglo-Saxon name graffiti, one in the Roman alphabet and four in the Anglo-Saxon futhorc. Four of these carvings were discovered in 1976 when the inscriptions in the late antique porticus leading to the sanctuary were systematically investigated (Carletti and Otranto 1980), and the fifth, leofwini, was added to the corpus after a subsequent examination by Maria Giovanna Arcamone (1992). According to Derolez and Schwab (1983, 114 f.), a further runic inscription can be detected, but due to erosion and overwriting it can barely be deciphered, much less interpreted. Three runic inscrip tions are located on the western façade of the wall at the entrance to the long porticus, hereberehct, herraed and wigfus, in addition to the uninter pretable runic scratches. The fourth inscription, leofwini, is located on a column inside the porticus, as is the Anglo-Saxon name EADRHID in Roman script. 3 Since four of the inscriptions are grouped closely together, three of them on the same ashlar, it may be assumed that they were inscribed on the same occasion (cf. Derolez and Schwab 1983, 113) and they may even be interpreted as belonging to pilgrims who had traveled in a group. This conclusion is, however, only tentative. As already men tioned, we lack historical data about these Anglo-Saxon pilgrims, and it was more over not uncommon to inscribe names of relatives, i.e. people who were not present.

Bind-runes and hc-spelling in Hereberehct
The inscription of the name Hereberehct is clearly visible and can be found at the main entrance to the gallery on the wall on the right side (see fig.  1). The individual runes are 4.2 to 5.2 cm high (cf. Derolez and Schwab 1983, 117) and the inscription can be read as (-í,ᚳᛏ h︠ e︦ r︡ e͡ be͡ re͡ hct Hereberehct. This renders a common Anglo-Saxon male name with here-< Germanic *harja-as the first element and -berehct < *bertha-(with an epenthetic vowel) as the second. The technical execution with deep and neat carvings bears witness to an experienced carver (cf. also Arcamone 2007, 131).
The inscription shows a sophisticated and rather unusual use of bind-runes. There are three consecutive instances of ᛖ as the first element: -, í, ,, and word-initially we even find the tri-partite bindrune (. According to Mindy MacLeod (2002,81,94), the bind-runes found in runic inscriptions in Anglo-Saxon England nearly all appear in a context connected to an ecclesiastical environment. This fact is particularly highlighted in the only other example of a triple bind-rune in an intelligible text: The inscription on the Whitby comb, an inscription dated to the late 600s or early 700s is read as d͡ aẹ ụsm͡ aeus:godaluwalu d͡ ah︠ e︦ l︡ ipaecy… Daeus maeus God aluwaluda helipae Cy… (cf. Waxenberger 2011, 74 f.) and interpreted by Bammesberger (2010) as 'deus meus, may God the Allruler help Cy-[' or 'deus meus, O God Allruler may you help Cy-' (cf. also Page 1999, 164 f.) where Cy-is assumed to be the first part of a personal name. The inscription is partly in Latin and contains a formulaic prayer. Hence, the use of bind-runes in Anglo-Saxon England suggests that the writer of the hereberehct inscription belonged to a Latin-speaking milieu. Two new finds with multipartite bind-runes in the inscrip tions on the Shropham lead tablet (not yet dated) and the Ipswich belt buckle (dated to the 700s; cf. Waxenberger 2016) further indicate that the use of multipartite bind-runes might have been specific to the regions of Northumbria and Mercia. 4 The hereberehct inscription, with its extensive use of bind-runes, suggests their use as a stylistic feature perhaps linked to the functional context in which the inscrip tion was embedded. Together with its prominent place ment on the western façade of the hall leading to the porticus and the tidy execution of relatively large runes, bind-runes could have served to high light visually and emphasize the inscription at this particular sacred place.
One feature that points to the influence of scribal practices is the spelling ᚻᚳ hc in the second name element -berehct. The spelling of the Germanic velar fricative [χ] in both initial and medial position in Old English runic inscrip tions is exclusively ᚻ (cf. Derolez and Schwab 1983, 121). The digraphic <ch>-spelling, however, occurs in Old English texts written in the Roman alphabet, but even here the single-graph spellings <c> and <h> are common. 5 Less frequently, <hc> (and <gh>) also occur (cf. Seiler 2014, 163). The spelling hc should not in any case be deemed a mistake as Arca mone (2007, 131) suggests but a spelling variant. Interestingly, in Seiler's (2014, 150) study of early Old English spelling, the variant <hc> was found only in Kentish and Mercian texts dated between 750 and 775. Due to the lack of comparative material, this data is insufficient to date the hereberehct inscription or indicate the writer's origin. Never theless, it is remark able that both spelling practice and use of bind-runes can be linked to Roman script practices in Anglo-Saxon records dating to the 700s and located in the south of Northumbria, Mercia and Kent. These two features surely indicate that the carver was familiar with vernac ular scribal practices in Roman script (cf. also Derolez and Schwab 1983, 122;Page 1994, 182).

Serifs and double spelling in Herraed
The herraed inscription is also placed on the west façade of the entrance hall leading to the long porticus, more specifically on the right capital supporting the entrance arch. The characters are on average c. 1.7 cm high (cf. Derolez and Schwab 1983, 116). The carvings are technically tidy and were probably executed by an experienced scribe (see fig. 2). A frame around the entire inscription was used as a graphic element to outline the runes. It marks the writing space, possibly adding visual emphasis to the inscription after the runic characters were carved. The reading of the inscrip tion is ᚻᛖᚱᚱᚨᛞ herraed Herraed, and thus it also renders a male name. The first name element is the same as in Hereberehct, her-< *harja-, but with no linking vowel -e-, and the second element is -raed < *rādi-.
Even though this inscription makes no use of bind-runes, which has been shown to be related to familiarity with Roman writing, it does exhibit two other features prominent in Roman writing: serifs and double spelling. Serifs are not limited exclusively to Roman alphabet writing in Anglo-Saxon England; they do occur in runic inscriptions. According to Page (1999, 103), however, the practice was transferred from monu mental Roman writing to monumental runic writing. He therefore assumes that the rune-carvers who employed serifs also produced Roman inscrip tions, and that they were responsible for the transfer of this graphic feature to runic epigraphy. Three Anglo-Saxon examples of runic inscrip tions with serifs further suggest the connection of this practice to a biscriptal milieu since they occur in combination with serifed Roman script inscriptions on the same object (Hartlepool I, Lindisfarne I, Monkwearmouth II). It cannot be determined whether the serifs were applied to the herraed inscription ad hoc or whether they were an established trait of runic epigraphy; with this stylized graphic feature, however, the herraed carver in Italy visually highlighted the inscription, in addition to employing a frame, as mentioned above.
The second feature, double spelling, is clearly influenced by Roman writing practices although not unusual in Old English runic inscriptions. In his investigation of Anglo-Saxon runic inscriptions from 1962, which to my knowledge is still the most recent on the subject, Ray Page considered whether the "runic practice" of avoiding double runes could in fact be regarded as a general runic spelling rule or instead only applied to certain local runic corpora. In his thorough account of Anglo-Saxon runic inscriptions featuring either double runes or single runes where vernacular Roman writing would expect double spelling, he demonstrated that a signifi cantly high number of inscriptions with double runes derived from mixed Roman-runic writing contexts and might have been influenced by non-runic spelling, such as unneg on the Franks casket or fearran on the Ruth well cross, both objects with runic and Roman texts (cf. Page 1962, 900-02;additionally, <g> for [χ] in unneg can be interpreted as a Roman spelling occurring in manuscripts, cf. Parsons 1994, 98 f.). Other sequences with double runes render Latin words or names (e.g. … ohann…s Iohannis on St Cuthbert's coffin) and can likewise clearly be linked to Latin and Old English scribal spelling practices. The instances show ing single spelling where in contrast double spelling in manuscript texts would be expected are, according to Page (1962, 902), not proof of a runic spelling rule since even these cases find parallels in non-runic texts. Lastly, even the unconventional spelling of long vowels with double runes in riicnae (Ruthwell) and liin (Brunswick) can be paralleled in manu script texts, for example in the Lindisfarne Gospels and the Durham Ritual (cf. Page 1962, 903). Although some forms remain unexplained (e.g. hilddigyþ, Hartlepool II), it can generally be concluded that the use of double runes was probably influenced by Roman script and associated scribal practices. This context might indicate that the herraed carver was familiar with scribal spelling practices, as was the hereberehct carver. The double spelling in herraed marking the end of the first name element and the beginning of the second also occurs in Old English manu script sources (Searle 1897, 294 f.).
The runic spelling ᚨ ae for the Old English sound resulting from fronting and i-mutation (< Germanic *ē 1 ) in -raed indicates that the inscription is not a transliteration of the name in Roman script, but rather that the carver (also) followed runic spelling conventions.

Wigfus and Leofwini
The remaining two interpretable runic inscriptions show no remarkable characteristics with respect to script mixing. The wigfus inscription is located on the same ashlar as herraed. The single characters are clearly visible, 2.5 cm high and neatly executed (cf. Arcamone 2007, 134; see fig. 3). The reading is ᚹŬᚷᚠᚢᛋ wigfus which can be interpreted as a dithematic male name composed of the elements wig-< *wīga-and -fus < *funsa-. A signum crucis with serifs is carved after the final s-rune, but it is not possible to identify whether the cross and the runic inscription correspond visually in terms of carving technique. An inscription in uncials follows the signum and it could just as easily belong to this inscription. It is, however, not unlikely that the signum was executed independently of both inscrip tions since this symbol appears all over the walls.
The leofwini inscription, finally, is notable among the runic inscriptions because it is placed on a column inside the long porticus, has very small characters (1 cm high) and is less tidily executed. The single carvings have frayed edges and the characters are not aligned horizontally but curve down wards towards the end of the inscription. The round surface of the column was obviously more difficult to inscribe than the plain wall of the entrance hall. Nevertheless, the inscription gives the impression of having been executed by a scribe less experienced than those already discussed. The inscription's reading, however, is not impeded. It is ᛚᛖᚪᚠᚹŬᚾŬ leofwini and renders the dithematic male name Leofwini with the common name elements leof-< *leuƀa-and -wini < *wini-(cf. Arcamone 2007, 133).

Uninterpretable inscription in runes
On the capital on the western façade of the entrance hall, directly below the wigfus inscription, is one more carving featuring runic characters. They are difficult to decipher, partly due to scratches and erosion, and partly due to overlap with overlaid inscriptions. Additionally, the runes have not been as tidily executed as the other inscriptions on the same wall. Derolez and Schwab (1983, 115, 124;cf. fig. 4) tentatively read s maegu or s maegy but offered no interpretation. This reading, however, requires reexamination since what was taken by Derolez and Schwab to be the lower stave of the ae-rune in fact seems to be a longer, shallower line belonging to the circled cross inscription carved over the m-rune. Further more, a d-rune following u - which Derolez and Schwab also noted but did not include in their final reading - can be deciphered and positively identified. The reading should additionally include a clearer indi cation of the characters of which traces can be detected, even if they cannot be deciphered. This pertains to remnants of at least two likely characters following d.
Due to the current condition of the graffiti, it is hard to judge whether this carving originally comprised an intelligible runic text or was instead a rune-like imitation by a non-literate scribe. In any case, however, it certainly shows no mixed-writing features.

EADRHID: an Anglo-Saxon name in Roman script
Further Germanic names can be found on the walls at Monte Sant'Angelo, but only one can clearly be assigned Anglo-Saxon origin (cf. Sinisi 2014, 43). This last legible inscription of an Anglo-Saxon name is entirely in Roman script. It is placed inside the gallery on the so-called rulers' pillar on the left-hand side of the scala tortuosa, one of the staircases lead ing from the gallery to the sanctuary. The inscription begins with a cross and reads: ). 6 Eadrhid has been interpreted as a male name comprised of the elements ead-< *aud-and -rhid < *-raed < *rādi-(cf. Arcamone 1980, 287). The obscure spelling of the second element is extensively discussed in Derolez and Schwab (1983, 127-30). They propose that the name is a "clumsy Langobardian rendering of OE Eadred" following the common <h>-spelling of Langobardian names in -hari such as <Bertarith> or <Perctharith> for <Perchthari>. How ever, this suggestion fails to explain the <rhid>-spelling since the Lango bardian spel lings referred to in Derolez and Schwab concern <th>-spellings as a variant of <t>-spellings. More convincing, especially in regard of the fact that an autograph is less likely than a copy executed by a professional carver (see below), is their second suggestion "that the carver was pre sented with a model in insular script, and that he misinterpreted e.g. Ead frid written <eꝺꝼꞃɩꝺ> producing EADRHID". This inscription is remarkable for several reasons. According to Derolez and Schwab (1983, 105), its placement just below three commemorative Latin inscriptions naming the renowned Langobardian rulers Grimuald, Romuald and Gunperga suggests that the Anglo-Saxon Eadrhid had some relation ship with the ruling Langobardian family. Such a connection between Anglo-Saxons and Langobards is probable in light of the historical context since Cunincpert, the Langobardian king from 688 to 700, was married to Hermelinda "ex Saxonum Anglorum genere" (cf. Derolez and Schwab 1983, 109). 6 Arcamone (2007, 130, 149) gives the original but wrong reading, EADRIHD, and in addition the incorrect drawing from Carletti (1980, 72, no. 56), although she had corrected the reading of the second name element to RHID in one of her earlier publications, referring to Carletti's plate 13 (Arcamone 1981, 167 and n. 34). Derolez and Schwab (1983, 125) con firmed the reading EADRHID after reexamination of the inscription. See also Carletti (2004, 526) for the correct reading.  (Derolez and Schwab 1983, 125, fig. 12) Additionally, the technically professional execution of the inscription makes it stand out among other carvings in the same area (e.g. Carletti 1980's numbers 50, 51, 53, 54). Derolez and Schwab (1983, 107) therefore consider that it may have been the work of a professional carver rather than an autograph, something that would indicate the higher social status of Eadrhid.
Carved in addition to Eadrhid's name are the epithets SAXSO 'Saxon' and the abbreviation V H, v(ir) h(onestus) - an honorific title "used by lay men of middling to high status" in Langobard Italy (Handley 2013, 747). The latter attribute together with the spatial and technical context of Eadrhid's inscription serve to explain why Roman script was used for this inscrip tion rather than runes. Eadrhid might have been an Anglo-Saxon living abroad and thus acquainted with Latin customs. It is on the other hand remarkable that the appellation SAXSO was added to the name as a lexical marker of origin and ethnicity.
This inscription shows no spatial correspondence to the runic inscriptions in the entrance hall and also stands out among the inscriptions with Anglo-Saxon names in its use of Roman script. Nevertheless, a graphic correspondence can be found in the letters showing the same type of triangular-shaped serifs as in the runic herraed inscription. The assumption that Eadrhid's inscription was made by a professional carver strengthens the argument that this graphic feature might have been applied to the runic inscription in order to stylize and highlight it and to make it appear "elegant".
Roman and runic practices intertwined: "heterographia" and individual style The analyses in the preceding section have uncovered various forms of parallels and entanglement of runic and Roman writing in the Anglo-Saxon pilgrims' inscriptions at Monte Sant'Angelo. First, both writing systems were used to render Anglo-Saxon personal names at the same sacred Christian place. Second, graphic features (serifs in herraed) and spelling practices (hc-spelling in hereberehct and double-spelling in herraed) were transferred from epigraphic or scribal practices in Roman letters to runic writing. Additionally, the extensive use of bind-runes in the hereberehct inscription has been interpreted as the influence of a Latin-speaking milieu even though it is not directly linked to Roman alpha bet writing practice. There are, however, also runic inscriptions show ing no Roman script influence. In view of the kind of mixed-writing inscrip tions found in the Anglo-Saxon homelands (see the subsection "Literacy in Anglo-Saxon England" above), it is furthermore remarkable that no parallel inscriptions with the same name recorded in both writing systems occur at Monte Sant'Angelo. Moreover, the letter inventories of the two writing systems were strictly separated, i.e. changes of script or hybrid letter forms within one inscription do not occur. Seen in the light of the sociolinguistic theoretical frameworks established in studies on modern writing systems in contact, these results cause certain problems. The writing conventions of the historical material discussed here cannot be seamlessly embedded into previous models. Mark Sebba's (2012) model of multi lingualism in written discourse concerns multilingual texts, i.e. texts using more than one language, and his analytical focus relates to the visual and grammatical mixing of languages in a text. The former refers to the space occupied by texts written in the respective languages while the latter focuses on the content. A "mixing type" according to Sebba's under standing implies that a visual unit, e.g. a paragraph, contains text in two languages or that two languages are used within a grammatical unit, i.e. a sentence. Consequently, the Roman-runic practices of Monte Sant' Angelo do not correspond to Sebba's (pp. 107 f.) category of "mixing type" in multi lingual writing, since his analytical starting point are texts or parts thereof in different languages. An analytical framework which is more applicable to the data is provided by Philipp Sebastian Angermeyer's (2012) study of Russian-American writing which discusses script choice in "digraphic usage", i.e. "the use of more than one writing system for the same language and by the same speech community" (p. 255). 7 For such a context of writing systems in contact he delineates the categories "script alternation" (alternating code-switching of languages and script in one and the same text), "transliteration" and "script hybridity". These terms still do not fully describe the situation relating to the inscriptions con sidered here, however. Script alternation and transliteration occurred in England and constitute a characteristic of Anglo-Saxon literacy. But neither practice is found in the Italian inscriptions. Lastly, Angermeyer's 7 The term "digraphic" is, however, problematic since it is used in graphemics as a designation of a two-letter spelling, such as <sh> for /ʃ/ in English or <ch> for /ç/ in German. A more appropriate term for the literacy described by Angermeyer is "biscriptal". In Bunčič et al. (2016, 51-54), "biscriptal" is used as a cover term for "the simultaneous use of two (or more) writing systems (including different orthographies) for (varieties of) the same lan guage". The term "digraphia" is also included, however, but restricted to specific types of biscriptality (pp. 56-59).
(2012, 263) definition of "script hybridity" cannot be applied since it covers only instances of letters from two writing systems being mixed at the lexical level. As discussed here, however, the mixing in the Anglo-Saxon inscriptions in Italy is neither on the level of language nor alpha bet, but rather on the level of different resources specific to written lan guage: the graphic (serifs, perhaps also bind-runes) and orthographic (spel ling practices) level. They integrate parts of originally Roman script traditions - both epigraphic and scribal - into runic writing. This type of mix is not mentioned in Daniel Bunčič, Sandra L. Libbert and Achim Rabus's (2016) comprehensive discussion of biscriptal writing either. Although they describe and define a wide range of sociolinguistically determined usages in biscriptal communities and contribute much to the field, they do not address the mixing of scripts on a textual, syntactical (called "graphic code-switching") or lexical level (pp. 69-71).
A parallel modern case for integrated spelling practices affected by foreign orthographies can, however, be found in modern German. In German magazines and texts deriving from computer-mediated sub-cultures, English graphemes are sometimes used, such as <sh> for German <sch> in e.g. <shatz> and <x> for German <chs> in e.g. <Abwexlung> (cf. Androutsopoulos 2000;2006). Just as with the case of digraphic ᚻᚳinstead of single ᚻ-spelling in hereberehct, this linguistic integration is of course mediated by the same sound value of the respective variants in the two written languages. This feature of written use in German is based on an active knowledge of English and can more broadly be linked to the "hetero glossia in practice"-approach to multilingualism established in the context of late modern mobile and superdiverse societies (Blommaert and Rampton 2011). Within this approach, "meaning making is not confined to the use of languages as discrete, enumerable, bounded sets of linguistic resources … Rather, signs are available for meaning making in communicative repertoires that extend across languages and varieties …" (Black ledge, Creese and Kaur Takhi 2013, 192). Thus far, this sense of "hetero glossia" has mainly referred to spoken language but has not been con fined to a particular medium. 8 Written language use, however, also requires the consideration of visual semiotic resources and spelling, and the term "heteroglossia" can be modified to "heterographia" to clarify its specificity to writing. Even though established to describe practices in late modern society, the definition just provided also applies to the use of written resources in the Anglo-Saxon inscriptions in Italy: graphic and ortho graphic resources of different languages and writing systems were utilized in the inscriptions according to the respective skill inventories, or personal preferences, of the different carvers.
Whereas the term "heterographia in practice" refers to the sociolinguistic situation and the way users of written languages apply linguistic resources to make meaning, the written product of such a practice, i.e. a word or a sentence showing characteristics of different linguistic repertoires, still remains undefined. I suggest broadening the terms coined by Angermeyer and Sebba to a more general "mixed writing" in order to include the mixing not only of languages and writing systems, but also of graphic features, orthography and other semiotic writing resources.
Such use of resources from two different writing traditions indubitably implies biscriptal competence among the carvers of runes in Italy. The question is, however, whether the heterographic use is an intended, individ ual act of the carvers of hereberehct and herraed, or whether the two systems were "momentarily blurred in their mind", as Angermeyer (2012, 263) suggests for the use of hybrid forms in his studies of biscriptal Russian-English speakers and writers in New York. The latter expla nation cannot however be tested and the possibility must therefore remain open. It is of course not improbable that biscriptal writers used to writing with both scripts unconsciously mix resources from the respective systems. Yet a closer look at the visual presentation of both inscriptions rather suggests that the inscriptions were intentionally and creatively designed. The hereberehct inscription is comparatively large, clearly and visibly placed on the western façade of the entrance hall and carved deeply; herraed is framed and additionally visually emphasized by serifs. The two runic in scrip tions lacking features linked to Roman script practices, on the other hand, show no visual highlighting. We can therefore assume that the carvers of hereberehct and herraed made intentional use of these resources for ostentatious stylistic purposes, possibly consciously with respect to the sacred location intended to eternally house their inscriptions.

Script choice and identity in personal names
The evidence of active knowledge of scribal practices among the Anglo-Saxon pilgrims in Italy raises one further question: why did they use runes at all? In the Sanctuary of San Michele they were surrounded by Roman inscriptions and as pilgrims in a foreign Christian context might be expected to use the script of the Roman church. Instead, they chose to use runes. In addressing this fact, Ray Page (1999, 224) tentatively assumed that "runes may have been a common script for giving one's name in an informal setting" and, alternatively, considered whether there "could … be a distinctive usage here, Englishmen writing their signatures in what they regarded as a characteristically English script". In an earlier publication, he suggested another alternative, namely a possible local vari ation within Anglo-Saxon England with "Northumbrians then as now advertising their distinctive culture" (Page 1994, 182).
In sociolinguistic research, choice-making in general (Coulmas 2013, 9) and script choice in particular are socially interpreted. Sebba (2009, 39-41) gives several examples of socially motivated script choices, e.g. as an identifier of religious culture (Urdu script as a marker of Muslim and Devanagari of Hindu culture respectively, see King 1998, 84, referred to in Sebba 2009. Both Sebba and Angermeyer (2012, 269) state that writ ing systems or even different styles and graphic features within the same script can potentially function as an iconic and powerful symbol of the identity of certain language communities. As applied to the use of runes by Anglo-Saxon pilgrims, the results from studies on modern script choice support Page's second and consequently also third assumption. Runes might indeed have served as a symbolic marker of Anglo-Saxon or even Mercian or Northumbrian ethnicity, and they might even have been intended to stand out visually from the Roman inscriptions. Foxhall Forbes (2019, 201-03) arrives at a similar conclusion and stresses that the choice of runes should be seen as a deliberate act of expressing or per form ing ethnic identity. She makes the further interesting point that runes express social belonging and possibly also the pilgrims' shared acceptance of what English ethnicity might have meant; at the same time, as the only inscriptions on the walls of the porticus not written in Roman script, they consciously or otherwise mark the difference between their carvers and other pilgrims.
It is moreover not surprising that symbolic markers of identity occur in connection with personal names. Names serve as identifiers and indi vidu al izers of human beings as well as indicating social identity (cf. Aldrin 2016). This demarcation might have assumed particular importance in light of the memorial function of the names in the religious setting of the sanc tuary (cf. the section "Historical context" above). On the other hand, I con sider Page's initial suggestion that runes might have commonly served to render personal names in informal settings less probable for two main reasons: firstly, I doubt that the setting at Monte Sant'Angelo can be regarded as informal; the stylistic adaptions discussed in the previous section instead indicate that the setting contributed to a more formal exe cution of the names as skills allowed. Secondly, Eadrhid's inscription provides a counterexample to the use of runes for Anglo-Saxon names. Eadr hid's choice of Roman script, as noted above, is unsurprising against the background of his possible relationship to a Langobardian family. Interestingly, however, two symbolic markers of social identity were added to his name, v(ir) h(onestus), giving his social status presumably within the Continental Langobardian community and the other, SAXSO, indicating his ethnic identity. Hence, this lexical marker fulfilled the same indexical social function as the runic writing system in the other inscriptions. This signifies that the marking of ethnic identity was considered important, even though it is unclear whether it was Eadrhid himself or the Lango bardian community who decided to add this attribute. As regards the runic inscriptions, on the other hand, it is reasonable to assume that it was the pilgrims themselves who made the carvings.

Conclusion
The Anglo-Saxon runic inscriptions in Italy have fascinated scholars and attracted public interest ever since their discovery and their value for both historical and linguistic research certainly cannot be overestimated. While vari ous aspects of their reading and structural linguistic interpretation as well as their broader historical context have been addressed in previous studies, the present paper has concentrated on sociolinguistic aspects of the use of runes for personal names in this specific context of Christian pil grim age. The main focus concerned the mixing of writing practices from Roman script and runic traditions and its interpretation against the background of the pilgrims' literacy and social identity. The findings suggest that the choice of either runes or Roman script for Anglo-Saxon per sonal names in the Sanctuary of San Michele at Monte Sant'Angelo was based on the social context in which the pilgrim was embedded. Roman script was used only once, for a person with probable connections to the Langobardians. Runes seem otherwise to have been the primary writing system for Anglo-Saxons to have their names eternalized on the walls of the sacred place. The writing system may have functioned as a semiotic marker of ethnicity, parallel to the lexical indicator SAXSO added to the Anglo-Saxon name Eadrhid in Roman script. Nevertheless, two of four runic inscriptions show graphic and orthographic influences from epigraphic or scribal Roman script practices found in Anglo-Saxon England which on the one hand implies a biscriptal Roman-runic background for the respective carvers. On the other hand, they might have functioned as stylistic features dis played by the individual writers in order to highlight their inscriptions that more over were visually emphasized through their technical execution and spatial placing. The other two runic inscriptions show none of these features of mixed writing or visual stylistics, however. Hence, these practices varied and were implemented on an individual basis dependent either on the respective carver's literacy resources or per sonal preference.
The term "heterographia" has been coined to describe the phenomenon of mixed writing practices in biscriptal use shown in two of the Anglo-Saxon runic inscriptions following principles delineated for heteroglossic lan guage use in modern multilingual speech communities. In contrast to previous theoretical accounts of multilingualism in writing, the "heterographic" approach adds the important feature that contact in written lan guage concerns not only parallel and mixed use of more than one lan guage and writing system in a text, but also that semiotic resources from other levels of written language, such as graphetic (or typographic) features and spelling practices, can be "selected" by biscriptal users to produce meaning in mixed writing.