Abstract
Aims:
This article investigates English lexical insertions in a corpus collected with first- and second-generation Italian migrants in two English cities, namely Bedford and Cambridge. It aims to establish whether, and to what extent, two Italian heritage communities share the same patterns of language alternation.
Methodology/approach:
I describe the frequency and type of lexical insertions to verify whether the study of lexical insertions or insertions of multiword units can shed light on the linguistic repertoire of each of the communities. The two Italian communities share many social features but differ in the ways Italians integrated or not into the host society.
Data analysis:
The dataset used for this contribution was collected through qualitative interviews with Italian migrants belonging to three different generations: first, second, and intermediate generations. The corpus used in this paper consists of approximately 50 hours of speech (35 hours for Bedford and 15 hours for Cambridge). I correlated the insertions of single lexemes with semantic and pragmatic specificity from a quantitative and qualitative perspective.
Conclusions/results:
The results provided evidence that insertion is first influenced by specificity as already Backus argued. External parameters such as migrants’ employment, ethnic composition of the social network, the size of the migrant community, and its social visibility do not, therefore, affect the borrowability of the lexemes belonging to specific domains, the frequency of which leads to their entrenchment as English lexemes even when speakers adopt the monolingual mode in Italian or dialect. The two communities do, however, differ in terms of the matrix language into which the English insertion is embedded: Italians prefer Italian as the matrix language in Cambridge but not in Bedford.
Originality:
The study confirms how similar Italian communities can be characterized by different linguistic usage and by a different linguistic repertoire.
Introduction
In a recent volume on heritage languages, Aalberse et al. (2019, p. 22) assert that the social features of immigrant communities can influence the outputs of language contact. Specifically they write:
Consider the daily life of a heritage speaker: at work, at the store, on television or the internet, the speaker operates mainly in the dominant language of the speech community; it is primarily at home with family that the heritage language gets used. This situation will become more extreme as the heritage language community shrinks in size and prominence. What results are very different types of exposure to the two languages. The heritage speaker is likely to encounter much greater variety in the dominant language: more speakers, different accents, more topics, etc. In contrast, the heritage speaker’s input in the heritage language is circumscribed to a small set of speakers (e.g., their immediate family) and the topics common to the situations in which the heritage language is used.
This article investigates the role of the social features of the immigrant communities adopting a comparative approach, distinguishing it from the majority of sociolinguistic studies. Earlier comparative studies of bilingual communities have largely focused on language contact and code-switching (henceforth CS): In such studies as that of Poplack (1985), CS indicates all intersentential switchings and implies a structural (grammatical) approach in schedule alternation. Code-mixing, by contrast, has been less investigated by means of a comparative approach. This is true also for lexical insertions, which are generally associated with the need to fill a semantic gap in the matrix language (“Theoretical framework” section). Lexical insertions in fact have mainly been investigated from a structural and functional approach but not from a comparative perspective.
Traditionally many studies have focused on the continuum between loanwords and CS. This perspective stems from an opposition between synchrony and diachrony, on the one hand, and between the dynamism and stability of linguistic contact on the other (Lantto, 2021; Treffers-Daller, 2009). In this context, it is worth highlighting two discrete views, that of Muysken (2000) versus that of Backus (2001). According to Muysken (2000), the investigation of code-mixing starts from a structural point of view, even if it has been deemed a result of language contact in specific types of bilingual communities (e.g., communities whose speakers have mastery of the two languages in contact). Backus suggests the importance of distinguishing loanwords from lexical insertions: the former is established in a given linguistic community and should be studied from a diachronic perspective; the latter is synchronic and not established, and, as such, inter-individual variation can be documented.
Following the approach of Muysken (1997, 2000) and Backus (2001) this article investigates lexical insertions to establish whether, and to what extent, two Italian heritage communities share the same patterns of language alternation. In particular, I describe the frequency and type of lexical insertions to verify whether the study of lexical insertions or insertions of multiword units can shed light on the linguistic repertoire of each of the communities.
Theoretical framework
Although studies of lexical insertion in Italian communities abroad date back to the work of Vaughan (1926), Menarini (1947), Villata (1980, 1981), and more recently Di Salvo (2011) and Del Vecchio (2023), some aspects remain poorly described: comparative studies offer the opportunity to examine the level of permeability to the dominant language of various heritage languages, or of the same heritage language across different migrant settings. In these studies, scholars who have traditionally focused on borrowings (see Menarini, 1947; Villata, 1980, 1981) adopt instead a language contact approach. More generally, in fact, studies on language contact seek to understand why lexical insertions (Backus, 2001; Muysken, 1997, 2000) are more frequent in particular domains. As early as 1953, Weinreich (p. 59) affirms that:
a bilingual is perhaps even more apt than the unilingual to accept loanword designations of new things because, through his familiarity with another new culture, he is more strongly aware of their novel nature.
According to Weinreich’s approach, semantic specificity influences the frequency of loanwords from the dominant language while Thomason and Kaufman (1988), take an approach based on the grammatical classification of the inserted pattern: they enumerate all the possibilities in a borrowing scale (see Figure 1) with content words at the top. The type of insertion depends on the level or phase of language contact: content word insertions are typical of an early stage of language contact, while major changes in morphosyntax occur at a more intense contact stage.
Phases of language contact in Thomason and Kaufman (1988) model.
Backus (2001, p. 126) argues that as per this approach by Thomason and Kaufman (1988):
The literature on codeswitching tends to deal more with the morphosyntactic integration of elements from another language than with the motivation for using exactly these elements and not others.
Weinreich’s (1953) statement gave rise to a huge body of literature which I will briefly touch on here. Matras (2009), for example, broadens the range of motivations leading to lexical borrowing: in addition to the lexical void (a), Matras (2009, p. 152) highlights sociolinguistic motivations linked to the prestige of the embedded language (b):
Underlying all the different motivation for borrowing is the bilingual’s need to negotiate a complex repertoire of linguistic structures and to balance effectiveness and precision of the expression against the social demand on complying with the norm to select only context-appropriate structures. The are three types of conflict that can arise between effectiveness of expression and compliance with context-appropriateness: (a) when certain meaning can be expressed in one language or set of contexts, but not in another (“gap”); (b) when the social associations of one set of contexts need to be replicated (“prestige”); and (c) when it is inconvenient to maintain the separation of the two languages (“cognitive pressure”).
As such, Matras ventures beyond the exclusive structural dimension of language contact with the inclusion of extralinguistic motivations, ranging from prestige to cognitive pressure, which comes to occupy a central position in subsequent theoretical models (Backus, 2015; Backus & Demirçay, 2021; Hakimov & Backus, 2021). Matras (2009) expands yet further (p. 154):
. . . borrowing depends on extralinguistic factors such as the duration and intensity of cultural contacts, the role and status of participating languages, language attitudes, and the degree of institutional support enjoyed by the languages (e.g. the presence of literacy).
Myers-Scotton (2002, p. 239) distinguishes cultural borrowings, which are those “words for objects new to the culture . . ., but also for new concepts,” from core borrowings, which are “words that more or less duplicate already existing words in the L1.” Lexical insertions are principally associated with the need to fill a semantic gap in the matrix language.
In this article, I will adopt Backus’ (2001) concept of specificity. In Backus’ (2001) words:
“embedded language elements in codeswitching have a high degree of semantic specificity” and “codeswitching is likely for embedded language words that are high in specificity, where highly specific means both that the word has a highly specific referential meaning, and that its matrix language equivalent, if there is one, conjures up quite different connotations” (p. 133).
Backus also argues that “embedded language elements in codeswitching have a high degree of semantic specificity”.
Backus (2001, p. 131) suggests that:
If semantic domain is an important predictor of switches, then it must be part of our definition of specificity: being saliently connected with the embedded language culture enhances a word’s specificity. Such topics have been experienced and talked about in the embedded language most of the time, and are therefore identified with the embedded language. Embedded language vocabulary is better developed in these fields.
It is worth noting that specificity can be explored from at least three perspectives: the semantic, the lexical and the pragmatic. As such, it is important to clarify which aspect I consider in my framework. As per research by Backus (2001), I am considering the semantic level and thus adopt his definition of semantic specificity. However, just as Backus (2001) did in the aforementioned study, I also consider some pragmatic aspects linked to insertion: among which, the cultural meaning of the inserted lexemes, which will be analyzed in terms of not only their referential meaning but also their connotations (see the “The two communities under investigation” section).
This approach, based on his case study of a Turkish-speaking community in the Netherlands, bridges the gap between the functional and structural approaches: referring to the pragmatic, semantic, and lexical dimensions that trace lexical insertions to gaps of lexical units in the matrix language, Backus evidences the importance of memorization and the use of specific lexical units in the embedded language, which would mean they become entrenched and, therefore, borrowable.
Backus (2001) demonstrates that full lexical elements (the content words of Thomason and Kauffman’s approach) can be embedded in a different language because they are memorized and entrenched: if memorized in a given embedded language, some lexical items are more often inserted into the Matrix language (p. 131).
Although there has been a focus on specificity in recent years (Backus & Demirçay, 2021; Hakimov & Backus, 2021), some aspects remain unexplored: first, there are yet to be comparative studies between similar communities. The comparative approach would highlight common trends and differences across similar migrant contexts; second, it would be at least theoretically possible to envisage that, in communities with different patterns of assimilation into the host society (Berry, 2003), there would be different semantic fields affected by contact. Finally, as Backus highlights, the aforementioned comparative approach could also evidence the presence of the same (different) behaviors among speakers of different generations: Del Vecchio (2023), for example, demonstrates that lexical insertions principally affect the behavior of first-generation migrants. But might this pattern exist in other communities or does it exclusively affect the Italian community of Bletchley? Backus (2001, 2015) and Del Vecchio (2023) both base their conclusions on qualitative research, while quantitative studies are still needed.
Aims
This article attempts to describe the lexical insertions from the dominant language (English) in a corpus collected from Italian migrants belonging to three different generations settled in two Italian communities in the United Kingdom, namely, Bedford and Cambridge. The analysis, from a predominantly sociolinguistic perspective, has two principal aims:
The first is to verify whether in the Italian communities of Bedford and Cambridge semantic and pragmatic specificity influences the dynamics of linguistic contact with particular reference to lexical insertion processes.
The second aim is to study the insertion of single lexemes or lexical units composed of multiple lexemes (multiword units) in speakers’ repertoire: I argue that the study of these lexical insertions could help to shed light on the plurilingual linguistic repertoire of the two communities under investigation.
As such, I aim to establish (a) to what extent the cases of insertion are specific according to Backus’s (2001) approach, (b) whether there are shared tendencies between the two communities, and, finally, (c) whether the specificity is conditioned by some external variables (the gender or migration generation of the speakers, although the latter can only be evaluated in a small number of speakers).
The research questions I intend to answer are:
are the contact phenomena concerning highly specific lexemes (Backus, 2001) similar in communities where the integration process (Berry, 2003) of migrants is different?
what are the similarities and differences in the frequency of lexical insertions in communities with different social characteristics in terms of the work carried out by migrants, the presence/absence of interpreters and mediators in the employment domain, the shape of the social network, the numerical size of the migrant community and its social visibility?
Do these communities also differ in their linguistic repertoire?
And, finally, from a more general perspective, is it possible to study speakers’ repertoire starting with lexical insertions?
These questions demand a sociolinguistic approach. I selected two Italian communities in the United Kingdom which, based on previous studies (Di Salvo, 2011, 2012, 2018b), are different in terms of integration and social networks: Italians in Bedford live within a very close Italian network, they feel no need to be integrated into the host society and instead exhibit a strong sense of belonging to Italy, while in Cambridge Italian migrants tend to be assimilated into English society, in part due to the frequent and stable connections they make with English people. I thus intend to verify whether the lesser degree of integration experienced by the Italians of Bedford can be correlated to the higher frequency of lexical insertions that are attributable to highly specific domains such as work and food, and whether, in Cambridge, with the higher assimilation of the Italians, insertions occur across a greater number of semantic domains.
To test this hypothesis, a quantitative approach is required which can highlight the regularity of the presence of stabilized lexical insertions about specific semantic fields. According to this framework, the study of the semantic fields most affected by contact with the dominant language of the country of immigration is essential to reconstruct how cultural contact with the host society occurs in different Italian communities abroad: From this perspective, analysis of the semantic fields subject to linguistic contact can shed light on the complex dynamics of integration in the country of immigration.
With regard to research questions 2 and 3, a comparison of the insertional code-mixing in two communities is made using a quantitative approach, which also factors in the matrix language. I investigate whether lexical insertions from English are embedded in Italian or in dialect, further clarifying variable patterns of insertion and the linguistic repertoires of the two communities.
The two communities under investigation
Italians arrived in Bedford and Cambridge following the signing of bilateral agreements between the Italian and the British governments to encourage Italian immigration to the United Kingdom. Italians were recruited to work in the industrial sector, in factory production of sweets and as cleaners in public offices and hospitals in Bedford, in commercial greenhouses, and as domestic staff for middle-class families in Cambridge.
In Bedford, male migrants were initially forced to live in hostels, housing provided by local brick manufacturers, on the outskirts of the city. The Italian presence was thus barely visible prior to the mid-1950s when migrants moved to Victorian terraces close to the railway station, an area that soon became a predominantly Italian neighborhood. Here, Italians opened the first Italian club, an Italian church, and some continental shops, which strengthened the internal ties of the community which today remains conspicuously conservative, thanks to the presence of a strong Italian network. Italian and even more localized networks were maintained through migratory chains, which supplanted the official recruitment system.
Cambridge, by contrast, did not lend itself to the genesis of ethnic neighborhoods, and the Italian minority was immediately dispersed across the city, without a concentration in any neighborhood. This condition encouraged the learning of English and integration into the host society, as summarized in Tables 1 and 2.
Variation in perceived competence in Italian, dialect, and English (percentage).
Source. Di Salvo (2012).
Percentage of use of Italian vs. dialect vs. English in Bedford (Sample 150) and in Cambridge (Sample 70).
Source. Di Salvo (2012).
As per Table 1, the two communities evidence a similarity in the perceived competence in Italo-romance varieties (both Italian and Italian dialects) but a difference in the ability to understand and speak English since first-generation migrants settled in Cambridge claim to have a better competence in the dominant language of the host society. It is worth highlighting that two communities with a similar migration experience can nevertheless have different outcomes in perceived competence in the dominant language: in this case, in particular, the (perceived) competence in English reported by Italians in Cambridge may be due to their social network: I have detailed in previous studies how Italians in Bedford live in a very close Italian network, while in Cambridge Italian migrants have more contact with English speakers (see also Di Salvo, 2012). In addition, Table 1 would appear to show that the differing level of competence would encourage a more frequent use of English in Cambridge than in Bedford.
The data in Table 2 provide evidence of the structural differences in repertoire of migrants in the two communities. With Italian friends, dialect is more frequently used in Bedford, while in Cambridge Italian is predominant. In more formal domains (at the Consulate, for example), the two communities are similar in their use of Italian, Italian dialect, and English. In their use of Italian, Italian dialect, and English at home, the use of dialect appears predominant in both contexts, but the use of English is greater in Cambridge. The higher percentage of use of English in the family could potentially lead to a language shift toward English and impact the competence in heritage languages by 2nd generation migrants. Investigation of this does not fall within the scope of this paper, but the comparison is useful to highlight the differences in competence and use of the varieties in their repertoire between the two communities. It suggests that even where two communities share the same history of migration, they can diverge from a sociolinguistic point of view.
Method
The dataset used for this contribution was collected through qualitative interviews with Italian migrants belonging to three different generations: first, second, and intermediate generations.
The first generation includes all those who migrated as adults; the second generation, instead, includes all those who were born in the country of immigration of their parents or migrated before the age of 4 years old. 1 In Bedford and in Cambridge, 1st generation migrants included in the study were all born in a southern Italian region and emigrated to England for work reasons or for family reunification. They all have a medium-low level of education and emigrated between the 1950s and the following decade. The second generation, by contrast, has the dialect and/or a regional variety of Italian as a heritage language and English as the dominant language.
The last group deals with the so-called “intermediate” generation or 1.5 generation (Backus & Demirçay, 2021; Del Vecchio, 2023; Rumbaut, 1994, 1997; Scaglione, 2000): this category includes all those who have received their primary socialization in the country of birth and, subsequently, emigrated during their childhood or pre-adolescence. In sociolinguistic research, this category generally includes children who migrated following their parents aged between 4 and 12. In the communities under investigation, the members of this generation emigrated following a family reunion: they joined their father or both parents who already resided in Great Britain for work reasons. These migrants, after their arrival in England, were included in the local school, a situation which caused them to become dominant in English, unlike their parents who are still predominant in dialect or Italian.
The corpus used in this article consists of approximately 50 hours of speech (35 hours for Bedford, 15 hours for Cambridge).
Table 3 provides the overall picture of the informants involved in the study based on the main independent variables that will be discussed in the analysis:
Speakers involved in the study per gender, generation, and city.
The number of speakers in the two cities and by generation is not identical. With regard to the distribution of speakers by city, the higher percentage of speakers resident in Bedford reflects, although not in a statistically significant manner, the higher concentration of Italians in this migratory context: the Italian community of Bedford is larger than in Cambridge.
In terms of the distribution of speakers in the three generations involved in the study (1; 1.5; 2), I mainly focused on first-generation speakers who are strongly affected by the kind of mixing investigated here (Del Vecchio, 2023; Del Vecchio & Di Salvo, 2022).
Although the speaker sample sizes of the two communities differ, I do not view this as a limit to the study.
The analysis followed two complementary methodologies.
I first opted for a quantitative study in which I correlated the insertions of single lexemes with semantic and pragmatic specificity: I thereby tried to understand whether specificity was a relevant parameter for lexical insertions as already evidenced by Backus (2001) for the Turkish-speaking community in the Netherlands. Consequently, I distinguished between specific and not-specific insertions. For this classification, I kept in mind Backus’ (2001) argument (p. 129):
Specificity is best cast in gradient terms, since it makes more sense to say that one word is more or less specific than another, than to say how specific it is in isolation. When a word is said to be highly specific, it cannot be replaced by something else that is even more specific, except when it is paraphrased.
Considering specificity as a gradient term means a given lexeme does not have to be classified as either specific or nonspecific; as such, I focus on the possibility of replacing a given lexeme with an alternative. In particular, I count as specific insertions all those cases where the English lexeme or unit cannot be replaced by an Italian or dialectal lexeme. If there is an equivalent in the matrix language, I investigate the connotations of the English lexeme to establish whether there is a difference between the English and its equivalent. In this case, I deem the lexeme to be specific as the equivalent has the same referential meaning but a different connotation. In so doing, I also consider the culture-bound connotations, as I will discuss in my analysis of the English lexeme “breakfast.”
Subsequently, I evaluate the concentration of specific and nonspecific insertions in relation to some external variables (gender, generation, type of community, native language, and presence of trilingual turns), according to a qualitative and sociolinguistic approach. Such analysis aims to shed light on differences between the two communities, especially in relation to the presence of insertions of specific lexemes. In my discussion of the results in 6, I will also provide an overall reflection of the various aspects that emerged during the analysis.
Results
For this article, I analyzed 950 cases of insertional code-mixing, of which 75.4% were found in the sample collected in Bedford and the remaining 24.6% were collected in Cambridge.
The results of the quantitative study demonstrate the relevance of the hypotheses formulated. First, the high concentration of highly specific lexemes is common to both migratory contexts under analysis, as summarized in Table 4:
Percentage of specific insertional mixings in the two cities.
The concentration of insertions of specific lexemes (as per Backus, 2001) is independent of the migratory scenario investigated, the generation of speakers and their gender, as summarized in the tables below: English-specific lexemes embedded in an Italo-romance variety are in fact prevalent both in Cambridge and Bedford and, in each context, across all three generations (Table 5) and across both men and women (Table 6).
Percentage of specific English lexemes embedded in an Italo-romance variety by generation.
Percentage of specific English lexemes embedded in an Italo-romance variety by gender.
The semantic fields most affected by insertional code-mixing are those summarized in Table 7:
Semantic fields of English lexemes embedded in an Italo-romance variety (percentage).
The data thus reinforces what Backus (2001, p. 131) argues and, in particular, the incidence of lexical insertions that fall within certain highly specific semantic fields. The picture presented of Turkish migrants in the Netherlands is, therefore, extendable to other migrant communities and is an independent immigration context, as the comparison between Bedford and Cambridge further demonstrates.
The communities differ in terms of how and the extent to which the two migrant groups are integrated, the social networks they have established in the migratory context, and their practice of alternating codes with the dominant language of the host society as evidenced in previous studies (Di Salvo, 2012, see also the “The two communities under investigation” section). However, the data on the relative frequency, in terms of insertions, of semantically, pragmatically, and connotatively specific lexical units, highlights a trait the two groups have in common.
I offer here some examples of specific lexemes by city and generation:
(a) Common lexemes in the two communities:
Breakfast, building, cancer, fish and chips, catering, club, college/collegio.
(b) Examples for generations:
Bambrick setter, fork lift, brick yard, bottom, sun facing, gold (I generation).
Cakes, coach, charity shop, chicken liver, fish finger, milk (II generation)
(c) Examples from Bedford
Bambrick setter, fork lift, brick yard, bottom, sun facing, gold.
(d) Examples from Cambridge
Bed and breakfast, puddini < “puddin.”
The qualitative analysis and, in particular, the examples above highlight some differences between the two migratory contexts: although, in both cities, insertions of specific lexemes are predominant and the semantic fields affected by contact with the dominant language of the society are the same (see Table 4), the forms of contact are different and depend on the history of the Italians in each community.
The forms brick, fork lift, on one hand, are present in first-generation speakers living in Bedford, where Italians were employed in brick manufacturing. Thus, the presence of insertions linked to this manufacturing is unique to this community: Italians were employed in a brick company in Bedford but not in Cambridge. This demonstrates that, although the two migratory contexts are united by the tendency of migrants to carry out insertional code-mixing, the qualitative analysis evidences differences in the labor, human and professional history of the migrants.
This exemplifies what Matras (2009, p. 151) states:
the goals that speakers pursue when interacting foreign vocabulary items in conversation are oriented toward the communicative interaction and the effect that language use will have on the interlocutor. These are shaped by the range of speakers’ associations with each of the languages involved, and these associations in turn are determined by the roles and functions that the languages have in the social routines of the speech community. It is in this somewhat indirect and mediated way that social factors are involved in borrowing.
Matras (2009), in particular, underlines the importance of the social customs of each community: In other words, the migratory scenario, to adopt the terminology of the Aalberse et al. (2019) model, conditions the forms of contact. From a more general perspective/outlook, it could also be inferred that the communicative practices of a given community contribute to the consolidation of certain linguistic habits, and lead to the entrenching of particular lexical units of the dominant language that would usually be embedded in the mother tongue. The English lexical units whose inclusion in an Italo-romance variety do not seem, at first glance, to be correlated to the model of semantic specificity (Backus, 2001) deserve further attention / consideration: among these, for example, breakfast, a form that does not fill a lexical gap in Italian, but does not seem to correspond, in terms of connotation, to the Italian “colazione.” The inclusion of such lexemes indicates rather the differences inherent in the two cultural systems. From this perspective, many of the cases of insertion could be described using the aforementioned category of cultural borrowing proposed by Myers-Scotton (2002, p. 239).
A particularly interesting case is constituted by the insertions linked to disease, such as cancer, Alzheimer’s disease, and diabetes: these forms do not fall within the lexical gaps, nor in the cultural borrowings nor do they even appear linked to particular connotations; their inclusion seems related to issues that these migrants encountered following their move to England. As such, if we take up Matras’ categorization of the reasons for which it is possible that lexical insertion occurs (“(a) when certain meaning can be expressed in one language or set of contexts, but not in another (“gap”); (b) when the social associations of one set of contexts need to be replicated (“prestige”); and (c) when it is inconvenient to maintain the separation of the two languages (“cognitive pressure”, p. 151), work-related insertion cases fall into the second case (b), while the lexicon related to disease can be traced back to the cognitive pressure (c) of the dominant cultural system and the language that is its expression.
I should also mention that, according to Backus (2001), the fact that migrants predominantly experienced illness following their arrival in the United Kingdom means that they would discuss this topic in English: the use of English, moreover, may lead to some lexemes becoming entrenched in migrants’ knowledge which, in turn, encourages insertional code-mixing.
From insertion to matrix language
A further difference between the two contexts concerns the code of the matrix language into which the English insertions are introduced: English lexemes can be incorporated either into bilingual (Italian and English; dialect and English) or trilingual (Italian, dialect and English) shifts. By turn, I am referring to a unit of analysis of different lengths (both in relation to its duration and to the number of words). While a speech turn is a unit that varies in size and, consequently, in a number of words, it can however function as an object of investigation to verify the different concentrations of bilingual and trilingual turns within the two communities under analysis (see Del Vecchio, 2023; Torres Cacoullos, 2020).
The data relating to the distribution of code-mixing towards English, with particular reference to the insertional forms analyzed here, highlight that in both migratory contexts bilingual shifts prevail, as summarized in Table 8.
Number of languages per speech turn in Bedford (BED) and Cambridge (CAM) (percentage).
The greater concentration of bilingual turns (English + Italian-Romance variety) both in Bedford and Cambridge may be due to the greater concentration, in the sample, of first-generation speakers whose linguistic behavior, as demonstrated by numerous empirical studies (Aalberse et al., 2019; Del Vecchio, 2023; Di Salvo, 2012; Rubino, 2014), is characterized by the higher concentration of contact outcomes involving referential lexemes, as highlighted in the borrowability scale developed by Thomason and Kauffman (1988), highlighted in the “Theoretical framework” section.
It is noteworthy that, although the tendency to make bilingual shifts is common to both contexts, it is also evident that trilingual shifts in Bedford are more frequent in percentage terms. To understand the significance of these data in terms of linguistic repertoire, we need to consider the language used in alternation with English in the cases of bilingual shifts (Table 7):
The variety alternating with English (in bilingual shifts) is Italian in more than 91% of the analyzed shifts in Cambridge; in the nearby community of Bedford, by contrast, although Italian is prevalent, shifts to dialect constitute nearly 44%. These data represent, in my opinion, further confirmation of the conclusions of a previous study focused on code-switching within the two communities (Di Salvo, 2012): that study demonstrates that trilingualism is more widespread in Bedford. Alternating between Romance varieties is, by contrast, a less widespread practice in Cambridge, since dialect is used sparingly by migrants living in Cambridge. In terms of linguistic repertoire, therefore, the data included in Table 9 indicate that in Cambridge dialect is considered a low variety in comparison with Italian, and speakers generally use Italian as the base language and prevalent (almost exclusive) mother tongue. They also claim to transmit Italian but not dialect to the 2nd generation (Di Salvo, 2018a).
Matrix language in bilingual speech terms in Bedford and Cambridge (percentage).
In Bedford, however, dialect is the code used more often in informal domains, while Italian is preferred with interlocutors outside of the in-group. Italo-romance dialect therefore occupies various domains in Bedford but not in Cambridge,
Discussion and conclusion
The comparison between Bedford and Cambridge allows us to address the paper’s research questions. First, the role of semantic specificity in lexical insertion processes is confirmed: it is independent of the immigration context since, although the Italian communities of Bedford and Cambridge present significant differences in terms of social structure, the quantitative data reveals marked similarities in the contact processes. These similarities pertain to both the greater frequency of specific embedded lexemes and the semantic fields affected by linguistic contact. The similarities between the two communities are also evident in the comparison of the different generations given that similar behaviors were detected across all three generations analyzed.
External parameters such as migrants’ employment, the ethnic composition of the social network, the size of the migrant community, and its social visibility do not therefore affect the borrowability of the lexemes belonging to specific domains, the frequency of which leads to their entrenchment as English lexemes even when speakers adopt the monolingual mode in a Romance variety (Italian vs. dialect).
All these parameters (social network, the size of the migrant community, and its social visibility) which have repeatedly been reported to encourage and/or discourage code-switching do not seem to affect specificity here given that specificity appears decisive at least for lexical insertions.
However, one aspect worthy of reflection is migrants’ type of employment: for the Italian community of Bedford, I was able to highlight the high incidence of English lexemes which refer to the semantic domain “brick industry.” This stems from the fact that in Bedford (but not in Cambridge) migrants were mainly employed in this industrial sector. In Cambridge, by contrast, where migrants were employed across different economic sectors, and thus this particular focus is missing, although work, as a broader semantic sphere, provides numerous cases of insertion. Furthermore, I should emphasize that, in the case of Bedford, only members of the first generation were employed in the brick-making sector, and following generations entered more a diversified employment landscape.
This case exemplifies the correlation between specificity and semantic field, a topic discussed by Backus (2001, p. 132):
If semantic domain is an important predictor of switches, then it must be part of our definition of specificity: being saliently connected with the embedded language culture enhances a word’s specificity. Such topics have been experienced and talked about in the embedded language most of the time, and are therefore identified with the embedded language. Embedded language vocabulary is better developed in these fields. However, with increasing levels of bilingualism, these domain effects get weaker.
The data from Cambridge and Bedford confirm both that “semantic domain is an important predictor of switches” but also that this is especially relevant to the first generation when linguistic contact is still in an early stage. Therefore, the presence of semantic fields mostly affected by contact does not depend on the phases of linguistic contact (early vs. intense) only.
If we also consider the social life of the different generations, the history of the community, and its progressive insertion also in contexts more differentiated working and social play a role in this kind of language contact output.
The two communities do, however, differ particularly in terms of matrix language, into which the English insertion is embedded: the data demonstrate, on the one hand, the greater incidence of trilingual shifts in Bedford relative to Cambridge; on the other hand, in bilingual shifts the predominance of Italian in Cambridge versus Bedford These data align with those discussed in an earlier study on code-switching alternation, which highlights the more deep-rooted tendency to switch to English in Cambridge but above all the higher incidence of Italian there compared with that of nearby Bedford (Di Salvo, 2018b). This study’s data, therefore, further confirm how, despite the same age at migration, the same number of years spent in England and social characteristics of the migrants, otherwise similar Italian communities can be characterized by different linguistic usage and, in particular, by a different position of Italian and dialect within the linguistic repertoire: While in Cambridge the choice of Italian is near ubiquitous (see Table 7) so much so that it is almost considered an unmarked choice, in Bedford a more balanced co-existence of the two Romance varieties is evidenced. It is thus possible to graphically represent the repertoires of the two communities as in Figure 2 and Figure 3 below:
Repertoire of the Italian community of Bedford.
Repertoire of the Italian community of Cambridge.
As Weinreich (1953, p. 11) states:
In speech, it [scilicet interference] occurs anew in the utterances of the bilingual speaker as a result of his personal knowledge of the other tongue. In language, we find interference phenomena which, having frequently occurred in the speech of bilinguals, have become habitualized and established.
The behavior of bilinguals in some communities becomes habitual and stabilized, even in synchrony, to the extent that certain English lexical units are regularly used in English by community members, and their use, through daily repetition, becomes consolidated.
This development highlights the difficulty of identifying, for Italian communities abroad, a single heritage language (HL), since at least those migrants who left at the end of the Second World War (see Vedovelli, 2011), started with the dialect they had learned as their mother tongue and varying degrees of knowledge of Italian, which itself has, in some cases, become the code used mainly with Italian interlocutors outside the community. As such it seems necessary to re-evaluate the notion of heritage language in terms of linguistic repertoire when describing Italian communities across the world today (Di Salvo, 2019; Goria & Di Salvo, 2023).
Footnotes
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
