1.1 What Are Heritage Languages?
In 1992 John “Giovann” Carosiello came to the University of Pennsylvania Linguistics Department, hoping to convince Professor William Labov to create a documentary about lu franciaj [lu frantʃáj], his family’s home language. Instead, I, a graduate student just finishing coursework in sociolinguistics and formal linguistic theory, was invited to meet with him. You might not know what lu franciaj is, and neither did I. Here is what Giovann [dʒuán:] told me. He was born and raised in Philadelphia. He grew up trilingual, speaking lu franciaj at home with his family, Italian in his South Philly neighbourhood, and English at school and then for years at work as a metallurgist at the Philadelphia Naval Base. When we met in 1992, he was sixty-nine and retired. Having heard that studying languages was a good way to keep your brain active as you age, he had enrolled in a French class. He quickly discovered that he and the French course instructor did not speak the same language – they could not even understand each other. His sister, Antonietta “Dunet” [dunɛ́t:] Falgie, remembered that their language was sometimes called lu prowensal (Francoprovençal) and she had heard that it was a language that had never been written down and was spoken only in two tiny villages, perched on mountaintops in Apulia, southern Italy, where their parents were born.
Giovann and Dunet were heritage language (HL) speakers: people who picked up a language at home, as their first language, that is not an official language of the country they grew up in. In comparison to national and official languages, HLs are not well supported by educational, governmental, or media institutions. If speakers have reference documents (e.g., dictionaries, grammar books, textbooks), they are often for a variety of the language spoken in another country, what we will refer to as the homeland variety in this book. The homeland variety is the (or a) language as spoken in the country from which their families immigrated, sometimes several generations earlier. In his homeland, Giovann’s variety has multiple names. It is known there as lu faidar, “Faetar,” and lu cellaj, “Cellese,” because the two villages where it is spoken are Faeto and Celle di San Vito, both in the province of Foggia. It is also sometimes referred to there as lu prowensal or lu francoprowensal.
Faetar is a HL twice over. Some 800 years ago, a group from the French Alps moved 1,200 km south to Apulia, Italy. In the Alps, where France, Italy, and Switzerland’s borders meet, Francoprovençal dialects have been spoken for centuries. These dialects form a set of language varieties that developed in parallel with French and Italian, spoken in the area between langue d’oc (southern French and Occitan varieties) and langue d’oïl (northern French varieties) (Ascoli, Reference Ascoli1878; Kasstan & Nagy, Reference Kasstan and Nagy2018). People moved from that homeland south to Apulia and settled in two villages, Faeto and Celle di San Vito. Their descendants continue to speak this Francoprovençal HL (Zulato et al., Reference Zulato, Kasstan and Nagy2018). To this day, it remains distinct from neighbouring regional Italian and Apulian dialects.
After 800 years of language contact, Faetar exhibits similarities to both French and Italian and is not completely mutually intelligible with Francoprovençal varieties in the Alpine Francoprovençal region that intersects the borders of France, Italy, and Switzerland. In a nutshell, Faetar is a left-branching subject–verb–object language, like other Romance languages. Its complex system of optional subject pronouns is more like Italian than French. However, negation is post-verbal, making it more like French than Italian, and there is much syncretism in verb suffixes – again, more like French. Its phonemic system resembles that of many southern Italian varieties, including geminates (word-medially and, reportedly, word-initially; cf. Nagy [Reference Nagy1994]), apocope, and the reduction of unstressed vowels to schwa, like southern varieties of French (Nagy, Reference Nagy2000a).
The fact that it retains some similarities with other Francoprovençal varieties but has evolved in other ways, and not only in ways that increase parallels with Italian, made me wonder how HLs work. Do variation and change operate the same way in HLs as in more frequently studied majority languages? After some 800 years, differences were evident between Homeland (Alpine) Francoprovençal and Heritage (Apulian) Faetar, but when did these differences emerge? Were they continuing to evolve? Were some speakers more innovative than others in terms of changing the language? The certainty of speakers in Faeto that they were still speaking Francoprovençal increased my curiosity. After all, they didn’t have regular contact with Francoprovençal speakers elsewhere.
Today, the people of Faeto and Celle are bilingual in this Francoprovençal variety and Italian. Many of those who emigrated kept up the languages they brought with them. They, and their children, are HL speakers in North America. In Philadelphia, Giovann and Dunet were heritage speakers of both Faetar and Italian. They are part of a large group of North American heritage speakers from Italy.
Italy unified as a country in 1861 and has experienced much migration since. While precise counts of immigrants from a particular Italian village are unavailable, in 1901 alone, half a million people emigrated from Italy. In the period from the 1940s to the 1970s, seven million immigrated to Canada. In the 1950s and 1960s, thousands more people left southern Italy every year and migrated to North America. By 1971, Ontario had nearly half a million Italians (Canadian Museum of Immigration at Pier 21, n.d.; Ramirez, Reference Ramirez1989). Many people from Faeto and Celle settled in the Greater Toronto Area, creating a group of speakers of Heritage Faetar. I use the label Faetar because Giovann and Dunet, who introduced me to the language, came from Faeto. Although I have worked primarily with speakers of Faetar in Italy, it is important to note that many Toronto families came from the smaller neighbouring village of Celle. They refer to their language variety as Cellese. However, no systematic differences have been reported between the two villages’ varieties. This highlights the fact that linguistic differences and glossonyms don’t covary systematically – different names for varieties sometimes correspond to important differences in grammar and/or vocabulary but not always. And sometimes enough differences exist between two groups of speakers that their language varieties might be expected to have different names, but they don’t.
What about heritage speakers in the North American context? When does a heritage variety differ enough from its homeland counterpart to acquire a distinct label? This question is motivated by labels such as American English, Canadian English, and le(s) français canadien(s) for varieties that developed from European homeland languages. These are readily recognized today but evolved during the early phases of emigration from England and France (Nagy, Reference Nagy, Côté, Knooihuizen and Nerbonne2016b). Do heritage varieties continue to evolve? In parallel with homeland varieties? Are some speakers more innovative than others in terms of changing the language? How might this differ depending on the ethnolinguistic vitality of the community?
Countries with a strong history of immigration have many HLs. Toronto is thus an excellent site for addressing these questions, having speakers of nearly 200 HLs. Figure 1.1 lists the top fifteen HLs in Canada (Statistics Canada, 2016a). (Faetar, of course, has too few speakers to appear in such a list.) Similar data for Toronto appears in Table 3.3.
Figure 1.1 Canada’s fifteen biggest mother tongueFootnote 1 languages (Statistics Canada, 2016a)
More generally, then, HL speakers acquire their family’s language from birth, but that language is not an official language of their country of residence. Thus, we use “heritage language” in Rothman’s (Reference Rothman2007, p. 360) sense:
Like all monolingual and childhood bilingual learners, heritage speakers are exposed naturalistically to the heritage language; however, this language is by definition a nonhegemonic minority language within a majority-language environment.
The official/majority language may be acquired either simultaneously with the family language (simultaneous bilingualism) or later (sequential bilingualism), or not at all. Thus, we shy away from definitions of HL speakers that include references to degree of fluency, such as: [a] language student who is raised in a home where a non-English language is spoken, who speaks or at least understands the language, and who is to some degree bilingual in that language and in English (Valdés, Reference Valdés, Peyton, Ranard and McGinnis2001, p. 38). Other labels for “heritage language” include home language, mother tongue, community language, international language, allophone language, ethnic language, minority language, ancestral language, non-official language, langue d’origine, and langue patrimoniale.
In research, HL speakers are often positioned between first language (L1) acquirers and second language (L2) learners. Like L1 acquirers, HL speakers are exposed to the family language in the home. To a greater extent than in monolingual L1 acquisition, language input is variable across HL speakers, at least in quantity, as noted by Putnam and Sánchez (Reference Putnam and Sánchez2013, p. 482). The age at which the majority language (in Giovann’s and Dunet’s case, English) is introduced to the speaker depends on many factors, including family composition, family linguistic and educational preferences, age, and circumstances of immigration. (Additional sociolinguistic issues are discussed in Chapter 4.) Many families who have immigrated feel pressure to assimilate to the mainstream culture and help their children succeed in the new community, resulting in a gradual decrease in the use of the HL at home.
An important difference between HL speakers and L2 speakers is related to status. While people whose L1 is a majority language are often lauded for studying or speaking other languages, people whose L1 is a HL are often criticized for both the way they speak their mother tongue and the way they speak their L2 (the majority language of the community). Kirk Hazen (pers. comm., March 2022) drew an analogy between HL speakers today and speakers of African American Vernacular English (AAVE), or other varieties of non-standard English, in the 1960s. No credit was accorded such speakers for knowing two language varieties. Rather, students who used non-standard features of English were often diagnosed as having learning disorders or speech disorders, or just being dumb. However, by the 1980s, ASHA (the American Speech-Language-Hearing Association) changed their testing protocol, having gained understanding that the language of African American students is a distinct, systematic variety. For setting up understanding of this “difference not deficit” approach, we can thank scholars of language variation such as William Labov, Walt Wolfram, and their colleagues and students (cf. Godley et al., Reference Godley, Sweetland, Wheeler, Minnici and Carpenter2006, Reference Godley, Reaser and Moore2015; Labov, Reference Labov and Lanehart2001a; Labov & Baker, Reference Labov and Baker2010; Reaser & Wolfram, Reference Reaser and Wolfram2005). The field of education also switched perspective, though more gradually. Teachers now get taught about “dialect diversity NOT deviance” and help students acquire “formal genres” in addition to their vernacular. In this regard, HLs are now where non-standard Englishes were fifty years ago. Researchers of HLs can help support such change by the terms we use in our own research interactions and reporting, as discussed in Section 1.1.
The findings in this book are offered to help us move forward, building on the lessons offered by work with other minoritized varieties.
1.2 What Are the Goals of This Book?
In light of the need just outlined, this book is intended for researchers and graduate students, particularly those interested in conducting research in the comparative variationist framework in languages other than the majority, national, or official languages of their countries. It should motivate critical reconsideration of research findings in all frameworks as we continue to expand and improve the empirical approaches applied in the pursuit of knowledge about (more of) the world’s languages and the field of linguistics.
Equally, the book is meant to serve as a tool for those who wish to revitalize or retain the vitality of their languages. Over and over, in discussions with speakers of many HLs, we hear the same thing: Speakers stop speaking their language due to the discouragement of being told (by family, friends, or teachers) that they don’t speak very well or they don’t speak the “right” language. As noted by Beaudrie (Reference Beaudrie2023, p. 1),
At play are widespread language ideologies of monolingualism and the supremacy of the standard language that privilege monolingual, educated varieties of HLs (deemed standard). These ideologies, typically reinforced in educational settings, devalue multilingual and bilingual practices as well as local varieties of the HL, often undermining students’ ethnolinguistic identity.
When a language doesn’t serve to represent an identity that speakers value, its use will decline. In contrast, Philipp-Muller (Reference Philipp-Muller2023) finds that “when one is proficient in their HL, there is an increased sense of connectedness with one’s linguistic and cultural community, and an increased sense of authenticity and expressivity afforded by the HL.” Thus, if people want to maintain a language, not criticizing speakers for linguistic variation is an important step. As noted in Grenoble (Reference Grenoble, Grenoble and Furbee2010) and Nagy (Reference Nagy, Preston and Stanford2009b), we can contribute to this effort by showing community members that:
(1) variation exists in all languages, not just minoritized ones, thus it is not a sign of language loss;
(2) variation is the inevitable indicator of language change, again, in all languages; and
(3) the types of variation and change that we find in HLs strongly resemble those found in majority and hegemonic languages and are therefore clearly not inherent signs of a language’s (or a speaker’s) weakness.
To these ends, this book describes the goals, methods, and findings of Heritage Language Variation and Change in Toronto (HLVC), the first large-scale project investigating variation and change in a range of HLs using one unified set of methods. The project began in 2009. Its primary goal is to document HLs spoken by immigrants and two generations of their descendants in Toronto, as well as a homeland comparator sample. The project created a corpus of time-aligned, transcribed spontaneous speech in ten languages, sampling from some 400 speakers. It is available for research on a variety of topics.
This book presents analyses of patterns in several HLs spoken in Toronto, in which different generations since immigration are compared to each other and to homeland comparators (monolinguals who have not emigrated). The comparative variationist sociolinguistic framework (cf. Labov, Reference Labov, Baugh and Scherzer1984; Poplack & Levey, Reference Poplack, Levey, Auer and Schmidt2010) is employed. These analyses are based on spontaneously produced speech in ecologically valid environments, that is, settings where the speakers typically use their HL, interacting with another HL speaker. They paint a picture of HL speakers that contrasts strongly with what we have learned from many experimental and psycholinguistic studies. The latter approach often depicts HL speakers as quite distinct from monolinguals, exhibiting incomplete acquisition or language loss. In contrast, the analyses conducted here do not produce evidence to support such characterizations. Rather, in spontaneous speech, HL speakers use language in very much the same ways as homeland speakers.
The second goal is description of linguistic variation in these varieties, based on these collections of spontaneous speech (and two other tasks; see Chapter 4). Documenting such variation is the first step in understanding linguistic changes in progress, providing data to determine whether differences between homeland and heritage varieties, when found, are best attributed to internal change, different identity-marking practices in the two locales, language contact effects, incomplete acquisition, or attrition.
This set of analyses of a range of languages paves the path to the third goal, pushing variationist sociolinguistic research beyond its monolingually oriented core and its majority-language focus. Chambers (Reference Chambers, Britain and Cheshire2003, p. 97) notes that “interlanguage exhibits orderly heterogeneity as outlined by second-language acquisitionists but has so far been under-exploited by variationists.” While the topic of the chapter in which he introduces this concept (which he refers to as the “Language Gap”) suggests that he is thinking about interlanguage varieties of the dominant language of the society, one might say the same about heritage varieties. To address this gap, we examine languages that have received very little attention within variationist sociolinguistics (cf. Meyerhoff & Nagy, Reference Meyerhoff and Nagy2008; Stanford & Preston, Reference Stanford, Preston, Stanford and Preston2009) to see if they behave differently from more frequently studied languages. Nagy and Meyerhoff (Reference Nagy, Meyerhoff, Meyerhoff and Nagy2008) reported that English is the target language for some 80 percent of variationist sociolinguistic publications in two journals that represent the field, and two Romance languages are the target of 18 percent of the publications. Fagyal and Davidson (Reference Fagyal, Davidson, Gabriel, Gess and Meisenburg2022, p. 344) note that “third wave studies currently represent a minority in Romance sociophonetics,” yet they are better represented in sociolinguistics than all the other languages and families of the world, which share the remaining 2 percent of the publications. Examining smaller languages, without official status in the place where they are studied, helps sociolinguistics offset the intertwined ideologies of monolingualism and standard languages, discussed in Section 2.2.3.
The extent to which outcomes of contact among languages can be predicted by combinations of social and linguistic factors, and the relative importance of each, remains an open and critical question in linguistics. Given that over half of the world’s population is multilingual from childhood (Tucker, Reference Tucker1999), it is unfortunate that in variationist sociolinguistics, the dominant trend is decidedly to examine one language at a time, essentially treating speakers as monolingual and focusing on a few global languages, primarily English (Nagy & Meyerhoff, Reference Nagy, Meyerhoff, Meyerhoff and Nagy2008; Ravindranath, Reference Ravindranath2015; Smakman & Heinrich, Reference Aikio, Arola, Kunnas, Smakman and Heinrich2015), Stanford (Reference Stanford2016). This contrasts notably with formal linguists’ efforts to develop a Universal Grammar of rules, constraints, or parameters that are cross-linguistically relevant.
The languages examined in this book are typologically diverse: Asian (Cantonese, Korean), Austronesian (Tagalog), Romance (Faetar, Italian), and Slavic (Polish, Russian, and Ukrainian). Hungarian and (Iberian) Portuguese data collection is in progress, further expanding this diversity, but is not included in this volume.
This expansion into languages and, therefore, linguistic variables that have not previously been investigated prepares us, as a field, to better engage formal linguistic theory, as “theory can (and should) explain the variable aspects of [language] along with the categorical facts” (Guy, Reference Guy, Bayley and Lucas2007, p. 5). While sociolinguists have contributed to the development of formal theory by offering amendments based on variable data (cf. Anttila & Cho, Reference Anttila and Cho1998; Guy, Reference Guy1991; Guy & Boberg, Reference Guy and Boberg1997; Labov, Reference Labov1969; Reynolds, Reference Reynolds1994, p. 19), this has been based primarily on monolingual speech data. Few examples of efforts to fit multilinguals’ speech to the predictions of formal linguistic theory exist outside of characterizations of code-switching conditioned by syntactic structure (cf. Myers-Scotton, Reference Myers-Scotton1993a, Reference Myers-Scotton1993b; Poplack, Reference Poplack1980; Poplack & Meechan, Reference Poplack and Meechan1998). As more robust findings, based on multiple languages, become available, ongoing collaboration between formal theory and variationist linguistics will be critical. As HLs amplify tendencies present in homeland varieties (Laleko, Reference Laleko2010; Polinsky & Kagan, Reference Polinsky and Kagan2007), rich HL data will allow us to test phenomena heretofore unnoticed, informing our understanding of acquisition, loss, and use. Thus, a book that encompasses the goals, methods, and findings of such work is a crucial tool for the field of linguistics. Documentation of undescribed homeland and HL varieties contributes to linguistic knowledge, tests the universality of theoretical constructs, and aids the sociolinguistic goal of defining grammars that account for both variable and categorical patterns.
This approach promotes HL vitality through research, training, and knowledge mobilization that is shared among linguistic scholars, students who are HL speakers, and their communities. Without supports and scaffolding to enable communication and cooperation between minority-language speakers and linguists, most importantly, by exchanging information in a range of contexts, such work would not be possible. This detailed report on the project thus serves the fourth goal, supporting the expansion of multilingual variationist sociolinguistics with the explicit intent of promoting HL vitality through research, training, and knowledge mobilization both in and out of the classroom (cf. Nagy, Reference Nagy, Trifonas and Themistoklis2018). To support this, the book includes activities and research prompts to help readers pursue such research (see Chapter 9). Such training materials will help increase our capacity by bridging the gap between HL speakers and sociolinguistic scholars, which, in turn, expands the set of languages and communities that our field can report on, as further discussed in Chapter 9.
With its goal of offsetting the “monolingual bias,” the HLVC project is a unique attempt to provide a more comprehensive view of linguistic diversity. This volume reports on what we have learned about variation and change in HLs, focusing on production in HLs, expanding the basis on which we can formulate and test hypotheses about linguistic variation. These analyses represent HL speakers in a way that contrasts strongly with previous reports from experimental psycholinguistic studies. In those, HL speakers are often depicted as exhibiting “deficient” language, attributed to incomplete acquisition or language loss. In contrast, the analyses reported here produce little evidence to support such characterizations. Rather, in spontaneous speech, HLs and their corresponding homeland varieties differ minimally. This is an important finding to share with community members and researchers.
1.3 Why Study Heritage Languages?
There are many motivations for studying HLs. Examining how languages in contact with each other, such as a HL and a national language, affect each other provides valuable evidence of how language is stored in the mind. What parts are malleable? What parts resist contact-induced change? How do we distinguish contact-induced variation from other types (e.g., internally motivated, identity-related)? Answers to these questions inform formal linguistic theory: Units of change and units of stability need to be accounted for in the grammar. Variation, whether or not it leads to change, provides evidence of how language is used for marking social identity in real-world contexts, placing it also in the domain of sociolinguistic theory. Generalizations about how languages change in particular contexts can aid language policy and planning to account for future stages of evolving languages. Documenting current usage practices of HLs can facilitate language teaching, offering locally relevant examples and generalizations (cf. Kisselev et al., Reference Kisselev, Dubinina and Polinsky2020). Other potential applications of analysis of a wider circle of languages include speech recognition and speech synthesis, tools that are becoming increasingly important and are thus widening the gap between speakers of well-documented and less-well-documented languages. From a social justice perspective, it’s important to produce studies of newer, less standardized dialects (Meyerhoff & Nagy, Reference Meyerhoff and Nagy2019) to support their ethnolinguistic vitality (Giles et al., Reference Giles, Bourhis, Taylor and Giles1977) and for practices such as Language Analysis for Determination of Origin, “the application of linguistic analysis to languages spoken by asylum seekers applying for refugee status ‘to ascertain whether they speak the language of a group they say they belong to’” (Patrick & Eades, Reference Patrick and Eades2011).
In terms of ethnolinguistic vitality, heritage varieties are much like other endangered varieties, including indigenous languages such as Canada’s First Nations languages, in at least the following ways:
(1) small numbers of speakers and lower transmission rates than national languages;
(2) lack of institutional support (existing educational supports are often in a variety that doesn’t match that of heritage speakers); and
(3) speakers may have negative attitudes to their HL.
Thus, another motivation for studying any HL is that it may empower its speakers by validating them as language experts and illustrating that their language is worthy of academic scrutiny. Such work can improve the status of the languages, as is evident through the growing pride and extension of domains of use in Faetar since documentation began (Nagy, Reference Nagy2000b). Accessible, objective descriptions of HLs can help offset the stigma sometimes associated with them, especially for later generations, whose varieties diverge from remembered homeland models. This is highlighted in the following astute observation in Fagyal (Reference Fagyal2010, p. 23):
Certes, au vu des statistiques, on peut voir le verre à moitié vide ou à moitié à plein. On peut chercher à prédire l’érosion inévitable et complète d’une langue d’héritage à la génération suivante à partir du moment où elle n’est plus la langue dominante des bilingues. Ou alors, on peut refuser de considérer une langue d’héritage comme morte simplement parce qu’elle est orale et réservée à la sphère familiale.
[Certainly, from a statistical perspective, we can see the glass as half empty or half full. We can predict the complete and inevitable erosion of a HL in the next generation from the moment when it is no longer the bilingual speakers’ dominant language. Or, instead, we can refuse to consider a HL as dead simply because it is [only] spoken and reserved for the familial sphere.]Footnote 2
Methods introduced in this book enable students to contribute efforts to offset stigma via presentations in community centres, schools, and cultural organizations about the inherent nature of variation and change in language, emphasizing the similarities between heritage and other languages. This research may directly help Toronto’s 2.8 million HL speakers, but there is also an international audience for this work: Many countries are developing interest in bilingual language use and look to Canada for guidance.
Let’s return to Giovann and Dunet and their relatives in Faeto, Italy. I recorded more than eighty speakers in Faeto in the early 1990s, with additional samples (including additional recordings of some of the same speakers) collected in 2005 and 2019. On all these occasions, I observed a high rate of use of Faetar, from speakers aged four to ninety-four. During my visits to Faeto, I found that many people express pride in having a language that is distinct from the non-standard regional Italian variety spoken around them. Dialects of southern Italy have been imbued with negative prestige as a result of long-standing stereotypes of southern Italians as lazy, slow, and uneducated (Di Ferrante, Reference Di Ferrante2007; Maria Catricalà & Di Ferrante, Reference Catricalà, Di Ferrante, Agresti and Bienkowski2010, p. 253). Several times, however, people in Faeto remarked to me that (a) they knew their language was originally spoken in France, (b) France is, in some ways, better than Italy (and those ways don’t include the cuisine, they will tell you!), and that these two items are why the Faetar dialect has survived for over 800 years. This shows that they recognize their language as an important marker of their distinct group identity. And they attribute enough power to this distinction to keep a minority language, which never had more than 5,000 speakers in Apulia at any one time, alive despite it not being supported by educational institutions. However, Giovann and Dunet, having lived their whole lives in Philadelphia, felt that they did not know their language well. They had not passed it along to their children and did not know of any way to help their language survive. A sense of helplessness had driven Giovann to seek the support of a professional linguist to create a documentary about his language. Yet the material that he could offer to support such a project consisted only of village maps and genealogy trees. With my guidance, he and a small group of friends and family in Philadelphia were able to produce a rudimentary set of language lessons – enough to prepare me to go to Faeto and conduct fieldwork with speakers that the Philadelphians felt were better (“real”) speakers of their language. Their expertise as HL speakers was invaluable – it formed the entire platform on which my research developed.
Their perspective raised a series of questions, each building on the previous. Do HLs develop differences from their homeland counterparts (and if so, how quickly)? Is it true that attitudes can boost a language’s vitality? Under what conditions? Is there any way to prove that attitudes affect vitality? If it is true, would it also be true for other heritage speakers in Canada? How would we show that one language variety has survived “better” than others? What factors would we need to consider or control to provide convincing evidence? What rates of inter-generational transmission would “count” for “keeping a language alive” and resisting shift to the majority language? How could we consider the varying rates of use of the language by language or by speaker?
Despite the many reports about language change in language-contact contexts, our theories have not progressed to the point of being able to make widely valid generalizations across already-studied contexts, let alone make predictions for future contexts, about how languages vary and change due to contact. This is due, in part, to the differences in methods of data selection, collection, and analysis, as well as differences in the language sets and the contact contexts that are compared. To move forward, it is imperative to apply consistent methods at all stages of research that examines multiple languages and contexts – the more diverse the better – to document HLs and how they behave.
So, this is yet another reason to study HLs: A diverse set of minority languages spoken in the same context (one large, highly multilingual Canadian city) provides a fine “lab” in which to compare trends across languages and see what answers emerge.
From this impetus developed the HLVC project (http://ngn.artsci.utoronto.ca/HLVC). As a first step, I recorded about twenty Faetar speakers in Toronto between 2009 and 2011, with the goal of comparing their language forms and usage patterns with those in Faeto. As much as possible, I applied the same methods for recruiting, recording, and analyzing their speech as I had used in Faeto. However, I quickly recognized benefits of improvements in technology (digital recordings), transcribing and coding practices (ELAN, a software tool produces time-aligned transcription that can be integrated with coding of all variable patterns, Brugman & Russel, Reference Brugman and Russel2004), analysis (Mixed Effects Models that reduce Type I errors by reducing the influence of individuals on group patterns), and workflow (processes that support collaboration among students, researchers, and HL speakers in order to facilitate the exchange of knowledge and smooth the process). Most of all, I delighted in being able to collaborate with students and colleagues along every step of the way, a process considerably more fruitful than the necessarily solitary work required for producing a dissertation.
This has led to the HLVC team applying consistent methodology across ten languages (as of 2022) in contact with English in one metropolitan context, with the goal of advancing our understanding of contact-induced change. The project examines a range of sociolinguistic variables in three generations of heritage speakers as well as homeland speakers. Of course, projects examining other contexts and other contact languages will only further our understanding of language contact effects.
There are three areas to which this work may be especially beneficially applied: language documentation, language acquisition, and variationist sociolinguistics, or the study of connections between language patterns and membership in social groups. These will be discussed in more detail next.
1.3.1 Documenting Linguistic Diversity
To balance out the view of HL speakers evoked through the description of Giovann’s and Dunet’s self-proclaimed ignorance about their mother tongue, we can contrast them with another speaker, Archangelo Martino. Archangelo, a speaker of Cellese, came from Celle to Brantford, Ontario, as a young man. He became a French teacher and a diplomat and made it a life-long project to develop and share knowledge about his HL, culminating in the publication of a description of the language, including etymological analysis (Martino, Reference Martino2011). This is but one example of language documentation work that would not be possible without the knowledge of speakers of a little-known language. The need to document a diversity of languages, and particularly heritage varieties, has been articulated in Aalberse et al. (Reference Aalberse, Backus and Muysken2019) and Evans and Levinson (Reference Evans and Levinson2009). Grenoble (Reference Grenoble, Grenoble and Furbee2010, p. 83) notes that documenting diversity (within languages) can “help educate speakers in dialect/language awareness, to understand that variation is the natural result of language change and is [also] found in vital languages which are robustly spoken.” This is critical for language maintenance, which, in turn, is critical to provide languages that we can document as well as use for communication, particularly of culturally specific knowledge.
1.3.2 Better Understanding of Language Acquisition
The field of HL studies developed in response to interest in understanding how language is acquired and stored in the mind. The acquisition of HLs is understood to overlap only partially with both first (L1) and second language (L2) acquisition. That is, like L1s, HLs are acquired naturalistically at home, rather than primarily through explicit instruction at school. However, like L2s, they are often developed from less input than L1s. Furthermore, their use often decreases as speakers enter school. Meisel (Reference Meisel2020) illustrates, however, that the amount of input cannot be the only important factor. It is also the case that HLs receive little or no attention in the school system which may also affect acquisition and maintenance of the language, especially genre and register variation. As many HL speakers are bi- or multilingual, turning our attention to these speakers, and the factors that influence their language, improves our understanding of how most of the world’s population acquires, maintains, and uses their languages.
In some instances, acquisition of HLs seems to be very much like acquisition of the single language of a monolingual speaker. For example, we will see that homeland and heritage speakers differ only in minor ways in their usage patterns for (PRODROP), (R), (VOT), and (CASE). (It is a convention in variationist sociolinguistics to mark variable names with parentheses, much as slashes are used to identify phonemes and square brackets to identify allophones.)
The possibility of equivalent acquisition is noted by Polinsky and Scontras (Reference Polinsky and Scontras2020a), although it is not the most frequent outcome in experimental studies. As noted by Gürel (Reference Gürel2020), these authors
do not seem to address the question of how the basic characteristics of economy of language processing resources differ in monolinguals and heritage speakers: that is, if economizing in online resources is a basic trait of a human processor, (how) do heritage speakers differ from monolinguals in terms of what counts as a processing pressure and how it is resolved in two populations?
Rather, Polinsky and Scontras, unlike many HL scholars, allow for the possibility of parallel paths of acquisition for HL and other languages. However, there are certainly places where heritage and homeland speakers differ. For example, the same HL speaker groups noted earlier for parallelism with homeland (PRODROP) patterns differ from homeland speakers in their production of (VOT) and (R). Guijarro-Fuentes and Schmitz (Reference Guijarro-Fuentes and Schmitz2015), in an excellent overview of HL research, note that there are bigger gaps between homeland and heritage varieties spoken in the US and Canada than in Europe, so the homeland/heritage similarities reported here are of particular interest. We still do not fully understand where HL grammar is resilient (i.e., converges with the homeland grammar) and where it is vulnerable (does not converge). Further studies to allow for generalization are necessary to flesh out a theory of HL acquisition (Gürel, Reference Gürel2020). We additionally require better understanding of which aspects of language diverge already in the first generation of heritage speakers (immigrants) and which diverge only in later generations. That is, what is the structure of the input received by the later generations, and is it faithfully reproduced?
Existing studies indicate that there are many potentially relevant external variables (social predictors) and that many of these are collinear (correlate with each other). Any study may include only a subset of possible predictors, risking a lack of internal validity. That is, the predictors that were measured may not be directly responsible for language variation but rather correlate with other factors that do affect language production (or perception or judgments) (cf. Aalberse et al., Reference Aalberse, Backus and Muysken2019). Thus, studies such as Montrul and Sánchez-Walker (Reference Montrul and Sánchez-Walker2013), which incorporated many factors, are essential for furthering our understanding.
1.3.3 Better Understanding of Variationist Sociolinguistics by Looking at Heritage Languages
The intersection of HL studies and language variation is understudied. Most of what we know about HL speakers has been gleaned through experimental tasks that focus on a very narrow genre of language use, often not truly communicative, and do not reflect variation among speakers that is linked to standard sociolinguistic predictors. Similarly, most of what we have learned about sociolinguistic variation is based on studies of monolinguals using the well-documented national languages of the US, Canada, and other English-speaking countries (Nagy & Meyerhoff, Reference Nagy, Meyerhoff, Meyerhoff and Nagy2008). In this way, sociolinguistics can contribute to efforts to offset the “fantasme de l’unilinguisme” (fantasy of monolingualism, Juillard, Reference Juillard and Ledegen2007, p. 47) an ideology that remains dominant in many societies (cf. Fagyal, Reference Fagyal2010; Kasstan, Reference Kasstan, Harrison and Joubert2019; Kircher & Fox, Reference Kircher and Fox2021).
Languages in multilingual settings have instead been examined in other perspectives including speech perception and production (Au et al., Reference Au, Knightly, Jun and Oh2002; de Leeuw et al., Reference de Leeuw, Schmid and Mennen2010; Oh et al., Reference Oh, Jun, Knightly and Au2003; Khattab, Reference Khattab2002; Knightly et al., Reference Knightly, Jun, Oh and Au2003; Saadah, Reference Saadah2011; Schmid, Reference Schmid2002); morphology and syntax (Fenyvesi, Reference Fenyvesi, Fenyvesi and Sándor2000; de Groot, Reference de Groot and Fenyvesi2005; Håkansson, 1995; Kim et al., Reference Kim, Montrul and Yoon2009, Reference Kim, Montrul and Yoon2010; Laleko, Reference Laleko2010; Mahajan, Reference Mahajan2009; Montrul et al., Reference Montrul, Foote, Perpiñán, Almazán, de Garavito and Valenzuela2008; Montrul & Bowles, Reference Montrul and Bowles2010; Polinsky, Reference Polinsky1995, Reference Polinsky and Browne1997, Reference Polinsky2006; Ravid & Farrah, Reference Ravid and Farah1999; Rothman, Reference Rothman2007); semantics (Kupisch & Pierantozzi, Reference Kupisch, Pierantozzi, Franich, Iserman and Keil2010; Montrul & Bowles, Reference Montrul and Bowles2010; Montrul & Ionin, Reference Montrul and Ionin2010; Polinsky, Reference Polinsky2006); language processing (Hulsen, Reference Hulsen2000; Montrul, Reference Montrul2009), and contact-induced variation (cf. Thomas & Kaufman, Reference Thomason and Kaufman1988; Léglise & Chamoreau, Reference Léglise and Chamoreau2013). Many of these articles are reviewed in Benmamoun et al. (Reference Benmamoun, Montrul and Polinsky2010, Reference Benmamoun, Montrul and Polinsky2013b). Variation in HLs is often associated with attrition, the loss of competence in aspects of a language, or incomplete acquisition, when compared to a homeland baseline (Laskowski, Reference Laskowski2009; Montrul, Reference Montrul2008). This book adopts the perspective that these are not the only sources of variation in HLs. Rather they are full varieties in their own right, exhibiting internal change, identity-marking variation and change reflecting changes in contexts of use, like all languages, as well as contact-induced influence (interference) (cf. Flores, Reference Flores2015; Nagy, Reference Nagy, Côté, Knooihuizen and Nerbonne2016b; Pires, Reference Pires and Flores2011).
Additionally, we know that speakers who control two or more languages may use items (such as words) or features (such as sounds or grammatical structures) of one language while speaking the other. We expect that speakers’ patterns of language use and their attitudes toward each of the languages (and toward the speakers of these languages) will influence their language mixing. Aalberse et al. (Reference Aalberse, Backus and Muysken2019) propose testable hypotheses regarding connections between language maintenance versus shift, code-switching, and linguistic interference. One goal of this book is to provide evidence to support such claims. We examine variation that suggests instances of potential change in the grammar of the language, specifically in the rule-governed structure at phonemic, morphological, and syntactic levels. We do not examine vocabulary variation in this book. Vocabulary differences when a language is spoken in different locales is unsurprising (Fagyal, Reference Fagyal2010, p. 201; van Coetsem, Reference van Coetsem2000) and has not been shown to be particularly rule-governed.
Thomason and Kaufman (Reference Thomason and Kaufman1988) note that proper assessment of language change and attributing its source to language contact requires documenting differences between the varieties in contact that existed prior to contact and that diminished during contact. Poplack et al. (Reference Poplack, Zentz and Dion2012) and Thomason (Reference Thomason2001) further note that the co-occurrence of language contact and linguistic change is not sufficient to conclude that an observed change is due to contact. For these reasons, a key feature of the methodology adapted by the HLVC project is to test correlations between degrees of contact with the majority language and with the HL and the quantified patterns of linguistic variation, to provide evidence of which instances of language change in HL communities can accurately be attributed to language contact. The speakers with higher rates of contact with the majority language and/or lower rates of use of their HL, are expected to show higher rates of use of the innovative patterns if their source is language contact. For this purpose, we compare homeland speakers, immigrants (Gen1), and descendants of immigrants (Gen2 and Gen3). We find that HLs provide insight into grammar in so far as they exhibit patterns and structures that are not easily explained by simplification or borrowing – the anticipated effects of language contact. The presence of parallel structures and directions of change across HLs can then inform our ability to develop generalizations, even universals, about HL grammars.
This set of analyses of a range of languages paves the path to the third goal, pushing variationist sociolinguistic research beyond its monolingually oriented core and its majority language focus – here we examine languages that have received little attention within variationist sociolinguistics. Such scrutiny is key for understanding whether the principles that have been established in the field of variationist sociolinguistics are indeed universal or rather are majority-language- or even English-specific. This is an area of expanding interest, as seen in several chapters of Smakman and Heinrich’s (2015) edited volume, which shows how many broadly accepted sociolinguistic constructs do not hold up when tested against a wider set of languages. These include ethnolinguistic vitality (Barasa, Reference Barasa, Smakman and Heinrich2015) and its tenuous connection to literacy (Shiraishi, Reference Shiraishi, Smakman and Heinrich2015); code-switching models that only account for two languages at a time when, in reality, more are often used (Barasa, Reference Barasa, Smakman and Heinrich2015); standard language and standardization (Barasa, Reference Barasa, Smakman and Heinrich2015; Bassiouney, Reference Bassiouney, Smakman and Heinrich2015; Mesthrie, Reference Mesthrie, Smakman and Heinrich2015; Satyanath, Reference Satyanath, Smakman and Heinrich2015b); prestige (Mesthrie, Reference Mesthrie, Smakman and Heinrich2015); style-shifting (Satyanath, Reference Satyanath, Smakman and Heinrich2015b); and indeed the very existence of individual languages (Anderson & Ansah, Reference Anderson, Ansah, Smakman and Heinrich2015; Greenberg, Reference Greenberg, Smakman and Heinrich2015), the last two of which have led to increased interest in the documentation and analysis of translanguaging (Canagarajah, Reference Canagarajah2011; Lewis et al., Reference Lewis, Jones and Baker2012; Wei, Reference Wei2018), as discussed in Nagy (Reference Kang and Nagy2016a).
Although there is growing interest, little progress toward generalizing about how languages vary and evolve can be made while disparate methodologies are employed across the languages studied. While, for example, a general hierarchy of borrowability (lexical > phonological > syntactic) exists, cross-linguistically valid finer levels of discrimination have not been firmly established (cf. Matras & Sakel, Reference Matras and Sakel2007), nor is it clear how this corresponds to transfer during the acquisition process. Generalizations are not possible until multiple languages and contexts are examined with identical methods (including methods of analysis and reporting).
1.4 How Do Heritage Languages and Their Speakers Behave?
In this book, we address gaps left in our understanding of HL behaviour as a result of disparate methods being applied to a wide range of contexts. The HLVC project compares the types of inter-speaker and cross-generational variation that occurs in different parts of the grammar across a diverse set of HLs but constrains the context and the methods of investigation. Within the context of Toronto, the project compares large languages such as Italian and Cantonese (each claiming more than 3 percent of Toronto’s population as mother tongue speakers) to smaller languages like Ukrainian (less than 0.5 percent of the population) and Faetar (less than 0.01 percent). As is discussed further in Chapter 3, the languages differ in terms of many other aspects of ethnolinguistic vitality: institutional support, transmission rates, and attitudes of and toward speakers, inter alia. This allows us to grapple with the long-range goals of the project: to better understand what cross-linguistic generalizations are possible about the types of features (or structures or rules or constraints) that are borrowed earlier and more often, and to understand the roles of social factors, on the individual and community levels, in these contact-induced changes.
To summarize, the HLVC project’s approach allows us to address many of the questions raised in this chapter, rephrased and grouped here by topics.
(1) About HLs
1.1 Do variation and change operate the same way in HLs as in more frequently studied majority languages?
1.2 How do we distinguish contact-induced variation from internal change and identity-related variation?
1.3 How do heritage varieties evolve? In parallel with the related homeland variety?
1.4 When does a heritage variety differ enough from its homeland counterpart to acquire a separate label?
(2) About the features and structures of the HL
2.1 What features and structures are malleable? What parts resist contact-induced change?
2.2 How consistent are the patterns across languages?
(3) About HL speakers
3.1 Are some speakers more innovative than others in terms of changing the language?
3.2 Can attitudes affect a language’s ethnolinguistic vitality?
(4) About measuring language-internal change for the purposes of cross-variety comparison
4.1 How can we measure relative rates of usage of the HL among the communities we compare?
4.2 How can we measure relative rates of usage of the HL among the speakers we compare?
1.5 What Is in This Book?
In this book, we take close looks at several variable patterns within a language to illustrate the HLVC methodology and to better understand the conditioning of each variable and what that tells us about the types of language change in that language. We then examine some of these variables across several languages as a way of illustrating the type of analysis that we can do with large corpora of small languages and the types of contributions to linguistics that such analyses make. We consider what generalizations are possible about the types of language change found in HL varieties, by considering questions such as:
(1) Is it even possible to generalize about the types of features or structures or rules or constraints that change under contact, or the types that are borrowed earlier or more often than others? Are there some parts of language that are more malleable and more susceptible to change and others that are more fixed in our minds and that resist any kind of influence?
(2) Can we find clear evidence for language change by looking at such under-documented language varieties, represented by speech that is spontaneously produced? Can it help us to understand what kind of conditioning influences these processes? To what extent do the principles and “universals” that have been proposed by variationist sociolinguists hold up in a broader set of languages and variables? In this multilingual and minority language context, what social factors are relevant? Who will be the innovators and who will be more conservative? Is that the same in a language contact situation versus a monolingual context where all change must be considered as language internal? How do the typically considered factors (e.g., age, sex) behave in tandem with all the additional factors relating to language use and preference that are relevant only to multilingual speakers?
(3) The interaction between style and social status may differ too: We know that HL speakers generally use their language in a smaller set of registers than monolingual speakers, reserving the majority language for more formal and institutionalized contexts. How does this influence variability? If you can switch between codes to signal affinity or solidarity, that sends a clear message. Does that mitigate the need for intralinguistic variation to serve those same goals? Or not?
Although many heritage speakers believe that they speak a simplified or defective version of their languages, and experimental results (as reviewed in Benmamoun et al., Reference Benmamoun, Montrul and Polinsky2013a, b) support that view, this body of research reveals vanishingly few instances of systematic distinction between heritage and homeland varieties in spontaneous speech. This adds to a growing body of assessments of heritage varieties as full languages (cf. Flores et al., Reference Flores, Rinke and Rato2017; Guijarro-Fuentes & Schmitz, Reference Guijarro-Fuentes and Schmitz2015; Nodari et al., Reference Nodari, Celata and Nagy2019; Pires, Reference Pires and Flores2011). Overall, the grammatical systems in these heritage varieties are robust – there is little sign of change as a result of being heritage varieties. Rather, much of the observed variation can be attributed to patterns found in all languages: internal change, identity-marking variation, and change due to changing contexts of use.
