The rolling snowball: lone English-origin lexical items in Guernésiais

Long-term contact with English has led to the presence in Guernésiais of a considerable number of lone English-origin lexical items (Jones, 2015). Although the presence of such items was being noted as far back as the nineteenth century, this is the first study to analyse and document them systematically. Using extensive original data, it examines these lexical items in relation to each part of speech and discusses their use in Guernésiais in the broader context of language contact. The study also considers whether, and how, lone English-origin lexical items become assimilated phonologically and morphosyntactically and whether frequency and motivation have a bearing on their usage. Résumé Le contact de longue durée avec l ’ anglais a conduit à la présence en guernésiais d ’ un nombre considérable d ’ éléments lexicaux isolés d ’ origine anglaise (Jones, 2015). Bien que la présence de ces éléments ait été notée dès le dix-neuvième siècle, cette étude est la première à les analyser et à les documenter de manière systématique. À l ’ aide de nombreuses données originales, elle examine ces éléments lexicaux en relation avec chaque partie du discours et discute de leur utilisation en guernésiais dans le contexte plus large du contact linguistique. L'étude examine également si, et comment, les éléments lexicaux isolés d ’ origine anglaise sont assimilés sur le plan phonologique et morphosyntaxique et si la fréquence et la motivation ont une influence sur leur utilisation.


Introduction
When two languages are used by the same speech community, linguistic evidence of this contact is commonly present (Thomason and Kaufman, 1988).Matras and Sakel (2007) distinguish what they term "MAT borrowing", defined as "when morphological material and its phonological shape from one language is replicated in another language" (Sakel, 2007:15) from "PAT borrowing", defined as "where only patterns of the other language are replicated i.e. the organisation, distribution and mapping of grammatical or semantic meaning, while the form itself is not borrowed" (Sakel, 2007:15). 1The present study offers a detailed examination of MAT borrowing, specifically, lone English-origin lexical items, in contemporary Guernésiais, the Norman language spoken in Guernsey.Although the presence of such lexical items is well attested in Guernésiais (cf.Jones, 2015), this is the first study to analyse them systematically using extensive and original data.It examines them in relation to each part of speech and discusses their use in Guernésiais in the broader context of language contact.
Guernsey's Norman speech community has been in contact with English since the installation of a small garrison on the island to protect against the threat of a French attack after the Channel Islands became formally annexed to the English Crown in 1259.Though initially small, the garrison grew steadily as Guernsey's strategic significance as a military base increased when England became more involved in wars outside its shores.During the Napoleonic Wars, for example, almost 6,000 men were stationed in the island, whose local population at the time was recorded as 16,155: the troops inevitably brought tradespeople and other locals into contact with English.From the nineteenth century, trade with England, in particular the development of the horticultural industry, integrated Guernsey's economy firmly with that of the UK and the improvement of regular communication by sea allowed tourism to be set on a serious footing, bringing thousands of people from the UK to the Channel Islands each year.Language contact was accompanied by cultural contact, with English customs being adopted, many local streets being renamed (from French 2 to English) and English influence becoming increasingly visible in Guernsey's architecture.During the Second World War, the evacuation to the UK of over half of Guernsey's population prior to the island's occupation by German military forces also brought islandersvery abruptlyinto contact with English, with many of the evacuated children growing up with English, rather than Norman, as their mother tongue.Since the War, immigration from the UK, associated with the expansion of Guernsey's off-shore finance industry, now its largest employer, has resulted in UKborn individuals representing nearly one quarter of Guernsey's population. 3Today, English is spoken fluently by all 63,448 residents and dominates every domain of island life.Speaker-numbers of Guernésiais, not recorded officially since the 2001 Census, are estimated at no more than a few hundred, most of whom are elderly.
As a result of this extensive and long-term contact, lone English-origin lexical items permeate contemporary Guernésiais.Their presence was noted as far back as the nineteenth century, with one guidebook to Guernsey commenting on how amusing it was "to wait for the English words to peep out of so different a language" (Anonymous, 1847).A contemporary travel writer observed "It [Guernésiais] is a good old dialect, which, during the last century, at least, has proceeded in a steady course of gathering, like a rolling snowball, from everything it encountered, and increasing its vocabulary by various compounds of Latin, Welsh, Scotch, German, English, and Italian, added to the original stock, which was Norman French" (Lane-Clark, 1880:1).That English-origin lexical items were used frequently in the everyday language of the nineteenth century may be seen from their inclusion in a large body of Guernésiais translations of the Bible and theatre dating from this time (see (1) and (2)).
(1) Et n'allouait pouin aie guiablles de dire qui l'couniesaie 'And did not allow the devils to say that they knew him' (Mark 1.34).

Methodology
The data analysed in this study were collected from interviews with 46 native speakers of Guernésiais, most of whomin keeping with the overall demographics of this particular speech community, had close connections to agriculture and farming.All speakers were fluent in Guernésiais although it was not necessarily still their main everyday language.For logistical and ethical reasons, the data presented were collected before the Covid-19 pandemic. 4Given the advanced degree of language contact in the speech community (as stated in §1, all speakers of Guernésiais are also fluent in English: no monolinguals remain) and the cessation of intergenerational transmission (Jones 2015: §4.2), with most speakers aged over 65 at the time the data were collected, it was not possible to consider usage related to proficiency in English, intensity of contact, age or social stratification. 5All interviews were conducted by myself and in Guernésiais and took the form of free conversation.In an attempt to obtain naturalistic data and to lessen the observer's paradox (Labov, 1972:32), I was accompanied at all times by a fluent speaker of Guernésiais who was well known to the people being interviewed and who often took the lead in the conversation, a strategy which has, in other contexts, proved an effective way of enhancing the elicitation of casual speech, especially in cases where the researcher is not a native speaker of the variety under investigation (Turpin, 1998:223;Milroy and Gordon, 2003:75;Bowern, 2010:351).Involving a research assistant also made it possible to use social networks to locate speakers (cf.Milroy, 1987), a strategy whose effectiveness has been demonstrated in other studies made of Norman (see Jones, 2001Jones, , 2015)).
A language takes lexical items from another language when it lacks a word for a particular referentin other words, when there exists a referential gap (such as in the case of the words pizza, taken by French from Italian, or alligator and sushi, taken by English from, respectively, Spanish and Japanese) (cf.Hock, 1991:408;McMahon, 1994:201).However, this can also occur when a native word for the referent in question does exist but, for reasons of culture or prestige, speakers opt to use instead a form from another language (such as with French soldat, taken from Italian despite the presence of indigenous soudart) (cf.Hock, 1991:409;McMahon, 1994: 202;Jones and Singh, 2005:31-32).Lexical items with both these motivations were examined.
Following other studies, a wide definition of "lone English-origin lexical item" has been adopted.This includes well-established, dictionary-attested loans, which would not be likely to be instantly perceived by all speakers as English-origin (e.g.allouaïr 'to allow'), and nonce items that only appear once in the corpus (e.g subsidy) (cf.Poplack, Sankoff and Miller, 1988;Poplack, 2018:28).To limit its focus, the study only considers i) lone English-origin lexical items and compounds and ii) frozen expressions acting as a unit and thereby functioning as single lexical items.Both i) and ii) occur in otherwise Guernésiais discourse (cf.Turpin, 1998: 224;Poplack, 2018:41).Each different lone English-origin lexical item is classed as a "type" and all occurrences of the same lone English-origin lexical item are considered tokens of the same "type" (cf.Poplack, 2018:42;Poplack, Sankoff and Miller, 1988).Multi-word English fragments are not considered (i.e.what would be codeswitches [unmixed donor languages constituents] in Poplack (2018)'s terms and inter-sentential codeswitches for Myers-Scotton (1993:3)).
Lone English-origin lexical items in contemporary Guernésiais are likely to represent a mixture of borrowings and single-word intra-sentential codeswitches (Jones, 2015:143-154;Winford, 2003:126-167;Turpin, 1998;Poplack and Meechan, 1995:224).Drawing a distinction between such forms is not straightforward and, as Poplack and Meechan (1998:127-128) point out, many studies have been written advancing different facets of this debate.To summarise these briefly, according to studies such as Poplack and Meechan (1998) and Poplack (2018), although intrasentential codeswitches and borrowings may bear some resemblance in their surface manifestation, codeswitches conform to the grammar of the donor language (i.e. the morphosyntax of the donor language is retained) whereas borrowings (whether established or nonce) conform to the grammar of the recipient language.They therefore consider most lone English-origin lexical items to be borrowings on the grounds that they pattern in the same way as their indigenous base-language counterparts.Myers-Scotton, however, sees all nonce borrowings as intra-sentential codeswitches in (donor languagerecipient language) mixed constituents, arguing that, when mixed constituents are accessed, interaction of the two grammars necessarily occurs at an abstract level (2002:154-155).She further argues that all such forms occur as part of the same developmental continuum (Myers-Scotton, 1993:63) so that, from a synchronic point of view, there is no need, strictly speaking, to distinguish between borrowings and intra-sentential codeswitches (Myers-Scotton, 2002:153) (cf.also Thomason, 2003 andvan Coetsem, 2000).Matras (2009:110), Thomason (2001:133) and others also see borrowings and intrasentential codeswitches as related points on a continuum rather than in terms of a sharp dichotomy.Rottet (2019:199), for instance, argues that "la binarité traditionelle empruntalternance codique est simpliste et ne reflète pas la complexité réelle des communautés profondément bilingues" and Gardner-Chloros writes "[a]t a synchronic level, there is no failsafe method of distinguishing between loans and codeswitches, as only time can tell if a loanword is more generally adopted over time" (2010:186).It is intended, in future research, to explore questions of this kind in relation to Guernésiais and it is hoped that the extensive data presented for the first time in this study will be of interest to the debate.However, given the lack of published data on the lexis of Guernésiais, before research can focus on the different "statuses" of these lone English-origin lexical items, it is first necessary to establish a clear picture of their nature and usage.An attempt has been made to offer some illustration of the way in which these data could potentially inform such a debate during the discussion of noun pluralisation ( §3. 3.1.2.).
In order to make the Guernésiais data accessible to readers more familiar with French than with Norman, the lone English-origin lexical items are transcribed phonetically but the the utterance (with the lone English-origin lexical item in bold) is given an orthographic rendering based on the (largely French-based) spelling system used in the Dictiounnaire Angllais-Guernésiais (De Garis, 1982, hereafter DAG: the only contemporary dictionary of Guernésiais).

Hierarchy of lone English-origin lexical items
The Guernésiais corpus contained 652 types (2,157 tokens) of lone English-origin lexical items.As Table 1 reveals, a clearly demonstrable "hierarchy" was present, with nouns far more likely to be borrowed than any other part of speech and adverbs and discourse markers far less likely.This is directly comparable to the now almost canonical borrowability scale put forward by Haugen (1950: 224) for Norwegian and Swedish immigrant speech in the US and, within the French-speaking context, to Poplack's Ottawa-Hull corpus (2018:48), King's (2000) study of the French of Prince Edward Island, Péronnet's work on the French of New Brunswick (Péronnet 1989) and Flikeid's (1989) study of Acadian French in Nova Scotia.It also correlates broadly with that proposed more recently by Matras (2007:61) on the basis of 27 different languages. 6In this respect, therefore, despite the advanced degree of language shift, Guernésiais usage seems to conform to commonly observable borrowing patterns.154 of the 652 English-origin types were recorded in the DAG, suggesting therefore that approximately one quarter of the English-origin lexical items in the data are established or generally accepted forms.In line with other studies such forms, that are "included in the standard lexicography of the receptor language" (Zenner and Kristiansen, 2014:4), are termed "listed" (L) (cf.Muysken, 2000:71;Poplack, Sankoff and Miller, 1988;Deuchar, 2020).

Frequency
Poplack , Sankoff and Miller (1988) and Poplack (2018) usefully distinguish lone English-origin lexical items by their frequency and by whether they are used by a single speaker or by several different speakers.By adapting their framework in accordance with the different sizes of the corpora, the lone English-origin lexical items in Guernésiais were divided into the following five categories: 7 a) Widespread lone English-origin lexical items (w).
Lone English-origin lexical items uttered once or more by more than 5 speakersin other words, lexical items that seem to have achieved a certain level of recognition within the speech community (Hasselmo, 1969;Mackey, 1970;Poplack and Sankoff, 1984): (3) Nou'va par pleine /pleɪn/ 'We go by plane.'8(4) Nou'n'tait pas allouaïs /aluɑj/ d'allaïr à la maïr 'We weren't allowed to go to the sea.' Total number of types in the corpus: 53.
b) Occasional lone English-origin lexical items (o). 9 Lone English-origin lexical items used more than once but fewer than 5 times in the corpus and by more than one speaker: (5) I'peut affectaïr dauve mes tablets /taebləts/ 'It can affect my tablets.' (6) Énne cruise /kru:z/ au Caribbean 'A cruise in the Caribbean.' Total number of types in the corpus: 121.

7
Given the greater size of the Ottawa-Hull corpus compared to the Guernésiais corpus (120 speakers compared to 46 speakers -19,579 tokens compared to 2,157 tokens), the benchmark for a lone English-origin lexical item to be labelled as "widespread" was reduced from 10 tokens to 5 tokens and as "recurrent" from more than 10 times though not necessarily by as many speakers to more than 5 times though not necessarily by as many speakers.c) Idiosyncratic lone English-origin lexical items (i).
Lone English-origin lexical items uttered frequently by a single speaker but not by any other speaker: (7) I'soulait s'n allaïr yodellaïr /jəʊdlɑj/ 'He used to go yodelling.'(8) Les daeux teams /ti:m/ 'The two teams.'Total number of types in the corpus: 30.d) Recurrent lone English-origin lexical items (r).
Lone English-origin lexical items uttered more than 5 times in the corpus, though not necessarily by as many speakers: (9) Nou'n'pouvait pas faire des sentences /sɑ tɑ :s/ en Angllais 'We couldn't make sentences in English.' (10) A'maettait l'robish /rɔbɪʃ/ dans la shed 'She put the rubbish in the shed.'Total number of types in the corpus: 18. e) Nonce lone English-origin lexical items (n).

10
For the nonce lone English-origin lexical items, the number of types is, of course, identical to the number of tokens.
From the above, it can be seen that nonce lone English-origin nouns account for the largest category by far, making up more than half the lone English-origin lexical items in the corpus.
While the speech sample analysed for this study may be considered representative of the Guernésiais speech community as a whole, somewhat inevitably, it cannot match it exactly.For this reason, a category assigned to a given lone English-origin lexical item in the corpus can necessarily only serve as its statistical descriptor within this particular study rather than as a definitive descriptor within the Guernésiais language in general.A lone English-origin lexical item that is widespread in the corpus may reasonably be assumed to be widespread within the wider speech community since, in such a bilingual community, the stocks of lone English-origin lexical items of any two speakers are likely to contain at least a few words in common, especially if there exists a referential gap (cf.Poplack, Sankoff and Miller, 1988:83;Poplack, 2018).However, the same will not necessarily hold true for other categories.For example, although a lone English-origin lexical item labelled "nonce" in the corpus might also be momentary in the context of the wider speech community, it could also be that, by chance, the referent denoted by that particular lexical item was simply not talked about much during the corpus interviews, thereby leading to its labelling as "nonce" for the corpus even though it may prove more frequent within the wider speech community (cf.Poplack, Sankoff and Miller, 1988:95).As will be seen (for example in §3.3.2.1.),nonce English-origin lexical items in the Guernésiais corpus do not display any intrinsic propertiesother than frequency of usethat distinguish them from other categories of English-origin lexical items (cf.Bouchard, 2023:4).Nouns are the lexical items with most lexical content (Weinreich, 1968;Muysken, 1984;Poplack, 2018:49) and cover the most differentiated inventory of labelling concepts, objects and roles (Matras, 2009:168).As discussed in §3.1.,nouns represent the most common lone English-origin lexical items in Guernésiais (506 out of 652 types: 37 widespread, 94 occasional, 25 idiosyncratic, 15 recurrent, 335 nonce) (cf.Matras, 2009:167).They are the easiest English-origin part of speech for Guernésiais to accommodate since nouns are, structurally, relatively less well integrated into the recipient discourse.As can be seen from the above examples, some nouns are phonologically assimilated into Guernésiais ( 9), (10), whereas others (5), ( 6), ( 8), ( 11), (13) are not.Variable integration between 2 tokens of the same type (i.e.where some tokens of a type are produced with Guernésiais phonology and other tokens of the same type with English phonology) was most common with so-called "international" words.For example, télévisiaon (12 tokens) was assimilated phonologically in 9 cases (/televizijaõ/) but its "English" pronunciation (/telɪviʒən/) was maintained in 3 cases.No clear overall pattern emerged, however, as most "international" words in the corpus were nonce lexical items.As examples, recitâtiaon, processiaon and conversâtiaon were assimilated to Guernésiais phonology (/rεzitasjaõ/, /prosεsjaõ/, /kɔ ṽεrsɑsjaõ/), whereas invâsiaon and programme kept their "English" pronunciation (/ɪnveɪʒən/, /prəʊgraem/).
3.3.1.1.Gender.All Guernésiais nouns have grammatical gender (Tomlinson, 2008:4), meaning that all lone English-origin nouns in Guernésiais must be assigned as either masculine or feminine for purposes of specification (i.e.articles, demonstratives) and agreement.However, unlike languages such as Italian and Spanish, gender in Guernésiais is not suggested by a word's phonological "shape". 11In theory, therefore, most lone English-origin nouns stand an equal chance of being assigned either masculine or feminine gender in Guernésiais.
In French, most lone English-origin nouns are assigned the unmarked masculine gender (Wise, 1997:93).Some lone English-origin nouns are also assigned masculine gender in Guernésiais (where, like in French, the masculine is also unmarked): examples include lunch (w), ouayeur (o), lorry (w), pénni (o), sacepàn (o), wireless (o), carre (w) (all of which are listed) and bookcase (n), carving (i), spelling (n) (all unlisted).However, many lone English-origin nouns are also assigned the (marked) feminine gender: examples include listed lexical items such as (aero)pleine (w), choppe (w), gaume (w), picture (w), grappe (w).Feminine gender assignation was particularly common with unlisted nonce nouns: examples include football, scarf, lighthouse, kitchen, jelly, handbag, library, fruitcake and highlight.Moreover, some less widespread (and unlisted) English-origin nouns were assigned different genders by different speakers.Examples include ( 14) -( 21 In order to test whether, as found in the French of Ottawa-Hull (Poplack, Sankoff and Miller, 1988:66), gender assignment became less variable as a lone Englishorigin noun increased in frequency, the gender of the 20 most widespread lone English-origin nouns in the corpus was examined (Table 2). 12 The lone English-origin nouns visiteur, German, touriste and picture were always used in the plural and without an accompanying variable adjective, which meant that it was not possible to determine which gender these words were being assigned.Of the remaining 16, 9 showed complete gender agreement across speakers.

11
Exceptions to this occur in some "international" words with a -tiaon ending in Guernésiais, which are usually feminine.However, the fact of being suggested by a lone English-origin noun's "shape" does not guarantee that a particular gender will be assigned by all speakers: cf. for example the form aen conversâtiaon 'a conversation' (m.) produced during one of the interviews.In this study, touriste is considered to be a lone English-origin noun rather than a native word of Guernésiais.The Dictionnaire Jersiais-Français describes its cognate in the Insular Norman of Jersey (Jèrriais) as being "d'introduction récente" (Le Maistre 1966:319) and, although it is most frequently pluralised via an oikomorphological strategy (see §3.3.1.2and, specifially, Table 3), in the present corpus this noun is never integrated into Guernésiais phonologically and is always given an "English" pronunciation (/tʊərɪst/).
However, despite their frequent use, 7 lone English-origin nouns were assigned both masculine and feminine genderand, in the case of djob and carre, different genders were assigned by the same speaker at different points in the same interview (( 22) -(25)).
It seems therefore that, in Guernésiais, for some nouns, more widespread usage determines a conventionalised gender (cf.bosse) which, for listed forms, usually corresponds to that documented in the DAG (cf.greenhouse, piti, choppe, chaoux, groweur, guaine, radio, télévisiaon, traffic) but, for others, gender remains variable despite their widespread use (cf.holiday, span, byke, djob, carre).Moreover, the gender most frequently assigned to a noun in the corpus did not always correspond to its listed gender (cf.djob, carre, bosse, radio).
14 Note that in this utterance, it is clear from the accompanying adjective pllôines (/pjɔ ĩn/) (which is phonologically distinct from the masculine pllôins (/pjɔ ̃ŋ/) that carre is assigned feminine gender (it is listed as masculine in the DAG).
Using one pluralisation strategy did not necessarily trigger the use of the same strategy within a given utterance ( 38)-( 40 In order to determine whether frequency of usage had a bearing on the pluralisation strategy applied, the plurals of the 20 most frequent lone English-origin nouns were examined (Table 3).15 of the 20 most widespread lone English-origin nouns in the corpus occurred in plural form.Of these, 7 plurals were formed exclusively via the oikomorphological strategy.The plurals of 8 lone English-origin nouns were formed variably via both strategies but, unlike Turpin's findings (1998:231), none were formed exclusively via the xenomorphological strategy.It was striking that a large degree of variation in plural formation was found in such widespread Englishorigin nouns: indeed, 5 of the 8 nouns with varying plurals (byke, (aero)pleine, visiteur, carre, touriste) were established enough to be listed in the DAG.Since the Guernésiais speech community consists entirely of bilinguals, it may be that it is so easy to produce and to understand different plural forms of these nouns with reference to Englishthe other language in which both the speaker and interlocutor are fluentthat they remain beneath the notice of speakers to the extent that no pressure exists in favour of the establishment of a single "normalised" plural form.
At the other end of the frequency spectrum, unlisted nonce lone English-origin nouns in the corpus were found to occur with both oikomorphological plurals (examples include pies /paɪ/, cesspits /sεspɪt/, boats /bəʊt/) and xenomorphological plurals (examples include earphones /iəfəʊnz/, meetings /mi:tɪŋz/, peatbags /pi:tbaegz/).Interestingly, the lone English-origin plural form chips (n), a noun which speakers would presumably be more used to hearing (in English) in its plural rather than singular form, occurred with an oikomorphological plural (/sε: tʃɪp/) (see (11) above).This is further evidence that, despite the advanced degree of language shift, the morphological patterns of Guernésiais remain strong for its speakers.
In her study of Acadian French, Turpin suggests that, following Poplack, it may be possible to determine the status of lone English-origin nouns in otherwise French discourse by examining whether they are integrated both syntactically and morphologically into the recipient language (1998:221, 223).In other words, do they conform (i) to the grammar of French but not to that of English (in which case they are considered to be borrowings) or (ii) to the grammar of English but not to that of French (in which case they are considered to be intra-sentential codeswitches).In relation to plurality specifically, Turpin sees codeswitching as entailing both the presence of the English plural morpheme /s/, /z/ and the absence of the French  (Turpin, 1998:229, after Béniak, Mougeon andValois, 1985), where the lone English-origin noun is conforming to the grammar of English but not to that of French.In contrast, a lone English-origin noun is considered a borrowing if both a French article and a zero plural mark are present (Turpin, 1998:228) as, for example, in the utterance Des hatch-back ø de rouvert, c'était après de boire (1998:229).The use of the French plural article in combination with the English affix /s/, /z/ is seen as indicating an intermediate degree of morphological integration.As discussed, the Guernésiais corpus revealed a greater tendency overall to opt for oikomorphological plural marking.However, drilling down, the relative proportions of plural marking strategies were found to vary from one speaker to another (cf.Turpin, 1998:231).Although most speakers tended to favour the oikomorphological strategy (29 used this strategy for more than 60% of their English-origin noun plurals), some clearly did not (8 used the xenomorphological strategy for more than 60% of their Englishorigin noun plurals), and others showed no clear preference for either strategy (9 used both these strategies for fewer than 60% of their English-origin noun plurals).Indeed, 7 speakers used variable pluralisation strategies for the same word (cf.( 32) -(37) above). 15According to Turpin, such variation in strategy may suggest that, for some of these speakers, several of the /s/-or /z/-marked lone English origin nouns may be codeswitches at the equivalence site between the determiner and the noun.
3.3.2.2.Phonological assimilation.Listed lone English-origin verbs often showed less phonological integration of the stem than the forms recorded in the DAG.For example, although listed as tchêtaïr (suggesting the pronunciation /tʃεtɑj/) 'to cheat', the form produced during the interviews showed less evidence of assimilation (63).Similarly, évacuaïr (/evakwɑj/) 'to evacuate', lecteuraïr (/lεktørɑj/) 'to lecture', pâtchi (/pɑ:tʃi/) 'to park', and bllâmaïr (/bjɑmɑj/) 'to blame' ( 64)-( 67): (63) Nou'n' les cheaterait /tʃi:tərε/ pas 'We wouldn't cheat them' (i, L). (64) Y en avait aen amas qui évacuaitit /ɪvaekueɪti/ 'There were many who evacuated' (w, L). (65) Il s'en va lectureaïr /lεktʃərɑj/ à l'University dé Southampton 'He's going to lecture at Southampton University' (n, L). (66) Il avait parkaï /pɒ:kɑj/ l'môto 'He had parked the car' (o, L). (67) Ch'n'est pas les écoles qui saont blameaïes /bleɪmɑj/ 'It's not the schools which are blamed' (o, L).Poplack and Sankoff (1984) describe such variable integration of lone otherlanguage lexical items as occuring when those lexical items are newly incorporated in the recipient language.Mougeon and Béniak, on the other hand, report that Haugen (1953) "found that bilinguals could "touch up" the form of older nativised loanwords to bring them more in line with donor-language phonology" or even "reborrow" them, and they provide compelling evidence of the phonological "denativisation" of loanwords as a minority language community undergoes language shift over successive generations (1989:304-306).The fact that, in the Guernésiais corpus, vacillation was found in both listed and less established forms seems to corroborate Mougeon and Béniak's view that "the answer is not as simple as ::: only incipient loanwords are able to show variable phonological integration" (1989:307).As all speakers are fluent in both languages, the degree of phonological integration is presumably a matter of choice.If listed lone English-origin lexical items can be considered as forms which have been accepted by the speech community to the extent that they have to all intents and purposes "become part" of Guernésiais, then their use with non-assimilated stem-forms suggests that speakers may be sourcing lexical items such as (63)-( 67) "afresh" from their English mental lexicon, presumably in a similar way to nonce lexical items.In such a fully bilingual community, differently integrated forms are, of course, all readily understood.3.3.2.3.Bare forms.Bare forms (lone English-origin verbs with no overt Guernésiais inflectional morphology and therefore structurally unintegrated into the language) were also present in the corpus, although these were extremely rare (6/221 verb tokens): examples include ( 68)-( 69).Such forms are also attested in the French of Québec (Bouchard, 2023) and Louisiana (Picone, 1994(Picone, , 1997;;Dubois and Sankoff, 1997;Rottet, 2016Rottet, , 2019;;Root, 2018).
In some cases, it was not possible to tell whether or not the Guernésiais verb had been integrated morphologically since any inflectional morphology would be phonetically null (cf.Poplack, 2016:391) 77)- (80).Even though the adjectives of Guernésiais are marked for (masculine/feminine) agreement, gender was not detectable on any of these forms (( 74)-( 85) and specifically ( 81)-( 85)) (cf.Poplack, Sankoff and Miller, 1988:68).The fact that many attributive adjectives are pre-posed may also be attributable to English influence (cf.The only example in the corpus of a lone English-origin discourse marker was the consecutive conjunction so, which was present in the speech of 14 interviewees (97)-(100).So is also borrowed in the variety of Insular Norman spoken in Jersey (Jèrriais) (Jones, 2015:147).
Mougeon and Béniak describe the use of so in Ontarian French as an example of core lexical borrowing occurring in a setting of intensive language contact.They suggest it "may serve to symbolise the advanced state of acculturation of bilingual speakers who experience high levels of contact with a superordinate language" (1991:212).Put another way, so is used by active bilinguals who speak both varieties equally in a context which they term "unpatterned bilingualism" (1987:40).Thomason and Kaufman (1988:74) consider such usage to be evidence that contact has intensified beyond the casual level (cf.Dawkins, 1916;Sitaridou, 2013).

Motivation and usage
Having examined the lone English-origin lexical items found in the different parts of Guernésiais speech, this discussion concludes by examining how differently motivated English-origin lexical items are used within the contemporary speech community.Of the 30 most frequently used lone English-origin lexical items, 18 (17 nouns, 1 verb) filled a referential gap and 12 (9 nouns, 1 verb, 2 adjectives) were used despite the existence of an indigenous word (Table 4).In other words, lone Englishorigin lexical items with both types of motivation are common in Guernésiais.
Even when looking at the corpus more broadly, a particular type of motivation did not seem to correlate at all with the production of widespread lone Englishorigin lexical items, which occurred in 7% of "referential gap" contexts (19/288 types), compared to 9% of "prestige" contexts (34/364 types).The percentages of nonce lone English-origin lexical items with both motivations also proved similar (188/288 "referential gap" types (65%) and 250/364 "prestige" types (69%)).A lone English-origin lexical item's degree of linguistic integration seemed equally unaffected by motivation, with one or more token of 121 of the 288 "referential gap" types in the corpus (42%) undergoing phonological and/or morphosyntactic assimilation compared to 160 of the 364 "prestige" types (44%).
As discussed in §3.1.,154 of the 652 lexical types (24%) were listed in the DAG.Of these, 64 (42%) filled a referential gap.Unlike motivation, listedness did seem to show some correlation with integration, with 115 of the 154 listed types (75%) showing phonological assimilation in at least one of their tokens compared to only 74 of the 498 unlisted types (15%).Interestingly, the 13 speakers making least use of lone English-origin lexical items (defined as producing fewer than 20 such lexical items during the first 20 minutes of the interview) tended to confine their usage to listed forms and usually assimilated them to the phonology and morphosyntax of Guernésiais (in other words, they treated them like "native" words).The 4 speakers who used most lone English-origin lexical items (defined as producing more than 60 such lexical items during the first 20 minutes of the interview) all used large numbers of nonce lexical items, seeming to consider any word of English as having the potential to be used in Guernésiais. 19These speakers also displayed more of a tendency to treat lone English-origin lexical items phonologically as words of English.

Conclusion
Lone English-origin lexical items are present in Guernésiais for all the parts of speech analysed, with an overall "hierarchy" similar to that observed in other situations of language contact, namely that "content-heavy" lone English-origin lexical items (nouns and verbs) are more common than lone English-origin adjectives, adverbs and discourse markers.The fact that both phonologically assimilated and non-assimilated forms of the same lone English-origin lexical item are present in the corpus is probably because all speakers of Guernésiais are active bilinguals who speak both languages on a frequent basis.Speakers do not therefore have to rely on recalling the more established form of lone English-origin lexical items in order to be understood by their interlocutor and can instead "source" items spontaneously from their English mental lexicon.Nouns are assigned gender and plural morphologyalthough here again some variability is present, even with frequent lexical itemsand most verbs acquire Guernésiais morphological endings.
In contrast, English-origin adjectives, adverbs and the discourse marker so usually keep their "English shape".The presence of these lexical items is commonly motivated by a referential gap but they also often occur as synonyms of Guernésiais words.Insofar as the items in the corpus are concerned, motivation has no apparent bearing on a lone English-origin lexical item's likelihood of being listed in the DAG, suggesting that inclusion is based on a lexical item's perceived sociolinguistic acceptance within the speech community rather than on being the only term available to denote a particular referent.
Competing interests.The author declares none.

Table 2 .
Gender assignment of the 20 most frequent lone English-origin nouns in the Guernésiais corpus

Table 4 .
Motivation of the 30 most frequent lone English-origin lexical items in the Guernésiais corpus