Phonological grammar and uncovering past language contacts

In this paper I discuss methodological issues relevant to using phonological phenomena to uncover language contact in the past. Specifically, I argue that abstract patterns of segmental phonology are significantly less prone to contact transfer than often assumed, and call for an exploration of their contact ecology that is more embedded in current approaches to structural and sociohistorical aspects of language contact, including sociolinguistic typology (Trudgill 2011), current approaches to phonological architecture (Natvig 2019), and the life cycle of phonological processes (Bermúdez-Otero 2015). The role of phonology in understanding language contact and areal effects has historically been ambiguous. Some phonological phenomena are widely recognized to be prone to horizontal transfer, at least under certain conditions: examples include phonemic inventories (e. g. Stolz & Levkovych 2021), stress systems (Salmons 1992), and syllable structure (Napoleão de Souza & Sinnemäki 2022). However, phonological phenomena that do not fall under these rubrics are also sometimes cited as indicative of areal effects, often with an appeal to substrate mechanisms. Classic examples relevant in the context of north-eastern Europe include tonal accents and palatalization contrasts (Jakobson 1931; but see Koptjevskaja-Tamm 2006) and distinctive consonant quantity (e. g. Ewels 2009; Daugavet 2013). In this paper, I argue that a proper evaluation of such proposals depends on operationalizing them in terms of current approaches to phonological architecture and sociolinguistic typology, and that not all cases of past contact posited on the basis of phonological evidence stand up to closer scrutiny.

Specifically, I examine the role of phonological patterns in language contact under different modes of agentivity (van Coetsem 1988; Winford 2005). Under recipient-language agentivity (“borrowing”), close contact can lead to convergence (up to complete isomorphy) in areas such phonemic inventories and syllable structure, an outcome that is attested in the Lithuanian-Polish-Belarusian contact zone (e. g. Sudnik 1975). I argue that whilst commonly recognized as an empirical possibility, such changes have an unclear status in contact ecologies: whilst lexical borrowing is a possible vector, it is not always plausible. Instead, I suggest that we sometimes need to reckon with shared sound changes: rather than involving a “source” and “recipient” of transfer, a change affects all varieties in a multilingual repertoire simultaneously, a situation that is attested, for instance, in the Balkan Sprachbund with pre-nasal raising and the reflexes of nasal vowels, for example with *ǫ > ɔ in co-territorial Macedonian and Meglenoromanian (Joseph 2009). This phenomenon represents a “source-oriented” account, emphasizing diachrony as an explanatory factor in convergent developments (Blevins 2004), and remains under-theorized. Most current views of phonological convergence under recipient-language agentivity provide some account of transfer of phonological matter such as phonemes, but they fare less well with relatively abstract phenomena that are better understood as transfer of pattern (Sakel 2007). In particular, cases like those described by Seržant (2010), where languages in contact end up with similar alternations that do not submit to source-oriented accounts, remain an important challenge.

In examining phonological change under source-language agentivity (usually linked to language shift), a key consideration is the demographic and social structure of populations in contact. It is widely recognized that phonetic and phonological interference is pervasive in the speech of individuals acquiring a second language as adults (e. g. Kehoe 2015; Andersson, Sayeed & Vaux 2017), which might lead us to expect substrate effects to be common. However, an examination of relatively well-documented language shift situations such as the Latin West (Adams 2007; Filipponio 2015) or the shift to English in the British and Irish Isles, including Ulster (Maguire 2020), the Isle of Man (Lewin 2017), and Cornwall (Wakelin 1975), turns up surprisingly few clear instances of such substrate interference in phonology. A key reason, I argue, that language shift proceeded too slowly for the individuals showing interference to represent a sufficient share of the overall language community at any given time. I suggest this consideration should lead us towards more stringent evidential standards in using phonology to examine past linguistic contacts.

To illustrate the point, I provide an extended report of the application of the methodology to a posited areal feature of northern Europe, namely stop preaspiration in North Germanic, Celtic, and Sámi languages (Hansson 2001; Rießler 2008). I argue that neither structural nor social considerations support the hypothesis that it spread via language contact, and offer an alternative account in terms of “drift” (Joseph 2013). I also offer some preliminary considerations of how the method can be extended to examining contact and drift in more closely related languages, with a case study of the e~’o alternation in Slavic (Andersen 1998).

← Enhancement of quantitative contrast by place as fortition

|  All presentations