In search of (lost?) contact-induced change in phonology

by Pavel Iosad

Slide 1

Slide 1 text

In search of (lost?) contact-induced change in phonology Pavel Iosad Fourth AMC Symposium, University of Edinburgh 1

Slide 2

Slide 2 text

Outline • What is phonology? And how does it participate in language contact? • Applying sociolinguistic typology to phonology • Phonological convergence in recipient-agency situations: beyond transfer of matter • Source-language agency and substrates in phonology: does it work? 2

Slide 3

Slide 3 text

Slide 4

Slide 4 text

Slide 5

Slide 5 text

Slide 6

Slide 6 text

Language contact and phonological change

Slide 7

Slide 7 text

Slide 8

Slide 8 text

Slide 9

Slide 9 text

Slide 10

Slide 10 text

Slide 11

Slide 11 text

Slide 12

Slide 12 text

Slide 13

Slide 13 text

Slide 14

Slide 14 text

What counts as a phonological pattern? For phonologists • Phonetic categories • Phoneme inventories • Distributions and phonotactics • Patterns of allophony • Morphophonological alternations • Featural representations For typologists • Phonetic categories? (Maddieson 2018) • Phoneme inventories? Sure! (Moran & McCloy 2019, Moran 2019) • Distributions and phonotactics? Getting there (Macklin-Cordes, Bowern & Round 2021, Joo & Hsu 2024) 3

Slide 15

Slide 15 text

Slide 16

Slide 16 text

Slide 17

Slide 17 text

Slide 18

Slide 18 text

Slide 19

Slide 19 text

Slide 20

Slide 20 text

What can be transferred? In the widely cited distinction between transfer of matter and transfer of pattern (Sakel 2007), phonological material sits rather uneasily: • Introducing new segments that weren’t there before, with phonemic status: widely accepted and widely attested (Stolz & Levkovych 2021), often in connection with lexical borrowing • Promotion of existing distinctions to phonemic status Fricatives in English (Lass 1987) • Old English wul[f] but wul[v]as /f/ ≠ o[fː]rian /fː/ • Middle English • French borrowings: victory, veal, zeal… • Degemination: o[v]er ≠ o[f]er • Apocope: wul[f] ≠ lo[v]e < OE lufu • Southern English Fricative Voicing: vixen, vat 4

Slide 21

Slide 21 text

Division of labour and the life cycle An observed ‘sound pattern’ can have multiple aetiologies (Bermúdez-Otero 2015, Natvig 2019): • Universal phonetic pattern, outwith cognitive control Phonetics Outwith cognitive control Phonetic rule Language- specific phonetics Postlexical level Word level Stem level Lexicon Morphology Domain narrowing Domain narrowing Stabilization Phonologization Phonology Rule death Morphologization The life cycle of phonological processes 5

Slide 22

Slide 22 text

Division of labour and the life cycle An observed ‘sound pattern’ can have multiple aetiologies (Bermúdez-Otero 2015, Natvig 2019): • Universal phonetic pattern, outwith cognitive control • Language-specific phonetic rule, under cognitive control but outwith phonological computation Phonetics Outwith cognitive control Phonetic rule Language- specific phonetics Postlexical level Word level Stem level Lexicon Morphology Domain narrowing Domain narrowing Stabilization Phonologization Phonology Rule death Morphologization The life cycle of phonological processes 5

Slide 23

Slide 23 text

Division of labour and the life cycle An observed ‘sound pattern’ can have multiple aetiologies (Bermúdez-Otero 2015, Natvig 2019): • Universal phonetic pattern, outwith cognitive control • Language-specific phonetic rule, under cognitive control but outwith phonological computation • Phonological rule, possibly within a stratal architecture Phonetics Outwith cognitive control Phonetic rule Language- specific phonetics Postlexical level Word level Stem level Lexicon Morphology Domain narrowing Domain narrowing Stabilization Phonologization Phonology Rule death Morphologization The life cycle of phonological processes 5

Slide 24

Slide 24 text

Division of labour and the life cycle An observed ‘sound pattern’ can have multiple aetiologies (Bermúdez-Otero 2015, Natvig 2019): • Universal phonetic pattern, outwith cognitive control • Language-specific phonetic rule, under cognitive control but outwith phonological computation • Phonological rule, possibly within a stratal architecture • Morphological exponent Phonetics Outwith cognitive control Phonetic rule Language- specific phonetics Postlexical level Word level Stem level Lexicon Morphology Domain narrowing Domain narrowing Stabilization Phonologization Phonology Rule death Morphologization The life cycle of phonological processes 5

Slide 25

Slide 25 text

Division of labour and the life cycle An observed ‘sound pattern’ can have multiple aetiologies (Bermúdez-Otero 2015, Natvig 2019): • Universal phonetic pattern, outwith cognitive control • Language-specific phonetic rule, under cognitive control but outwith phonological computation • Phonological rule, possibly within a stratal architecture • Morphological exponent • Historical remnant Phonetics Outwith cognitive control Phonetic rule Language- specific phonetics Postlexical level Word level Stem level Lexicon Morphology Domain narrowing Domain narrowing Stabilization Phonologization Phonology Rule death Morphologization The life cycle of phonological processes 5

Slide 26

Slide 26 text

An example: stop preaspiration Stage UR SR Phonetics Example Phonologization /p t k/ [p t k] Nʰp ʰt ʰkN Faroese, Ulster Ir- ish, Sea Sámi 6

Slide 27

Slide 27 text

Slide 28

Slide 28 text

An example: stop preaspiration Stage UR SR Phonetics Example Phonologization /p t k/ [p t k] Nʰp ʰt ʰkN Faroese, Ulster Ir- ish, Sea Sámi Stabilization /p t k/ [hp ht hk] Nhp ht hkN Icelandic, North- ern Sámi, Argyll Gaelic Lexicalization /hp ht hk/ [hp ht hk] Nhp ht hkN South Sámi, Här- jedalen Swedish 6

Slide 29

Slide 29 text

Slide 30

Slide 30 text

Phonological change and sociolinguistic typology

Slide 31

Slide 31 text

Where does phonology sit? • Can phonological patterns transfer? 7

Slide 32

Slide 32 text

Where does phonology sit? • Can phonological patterns transfer? Yes! (Thomason & Kaufman 1988) 7

Slide 33

Slide 33 text

Slide 34

Slide 34 text

Slide 35

Slide 35 text

Where does phonology sit? • Can phonological patterns transfer? Yes! (Thomason & Kaufman 1988) • How does phonology interact with the two different modes of agentivity? (van Coetsem 1988, Winford 2005) • L1 agentivity: yes, although the degree of integration matters. • L2 agentivity: yes, sound patterns are generally involved, but social evaluation seems to matter a lot. 7

Slide 36

Slide 36 text

Slide 37

Slide 37 text

Phonology and language contact in the past • The sociohistorical circumstances should match the transfer mechanism, but phonology can be involved in both modes • Phonological change by itself is not easily diagnostic for recovering the type of contact. 8

Slide 38

Slide 38 text

Slide 39

Slide 39 text

Contact-induced change and transfer As with other abstract, tightly integrated subsystems, not all change in sound patterns that is induced by contact can be conceptualized as transfer or copying • Compromise and interlanguage systems (Kehoe 2015, Andersson, Sayeed & Vaux 2017) • Loss of marked structures / reversion to the mean • Simultaneous innovation And conversely, some changes that involve sharing of patterns are not necessarily contact-induced • True parallels (Poplack, this conference) • Endogenous developments downstream of contact events (Blevins 2017) • Drift (B. D. Joseph 2013) 9

Slide 40

Slide 40 text

The problem with phonology We don’t know enough about possible and/or probable phonological changes, but we know something, and tend to have strong intuitions (Kümmel 2007, Cser 2015) 10

Slide 41

Slide 41 text

The problem with phonology We don’t know enough about possible and/or probable phonological changes, but we know something, and tend to have strong intuitions (Kümmel 2007, Cser 2015) Phonological change and grounding Phonology is about externalization and subject to extremely strong substantive biases. 10

Slide 42

Slide 42 text

Slide 43

Slide 43 text

Key points of the Trudgill conjecture (as I understand them) • The social ecology determines the mechanism of acquisition involved 11

Slide 44

Slide 44 text

Slide 45

Slide 45 text

Key points of the Trudgill conjecture (as I understand them) • The social ecology determines the mechanism of acquisition involved • Some mechanisms of acquisition promote complexification, or at least maintenance of complexity; others support simplification • Complex structures need L1 learners to perpetuate them 11

Slide 46

Slide 46 text

Slide 47

Slide 47 text

Slide 48

Slide 48 text

Chasing complexity in phonology • A definition of complexity from first principles is, of course, very elusive (J. E. Joseph 2021) • In sociolinguistic typology, complexity is primarily conceptualized as L2 learning difficulty • Many candidate measures involve semantics, pragmatics, and discourse factors, which are challenging to apply in phonology • Transparency in form-meaning mapping (Trudgill 2011) • Integration with interface modules (Sorace 2011) • Nature of features involved (Walkden & Breitbarth 2019) 12

Slide 49

Slide 49 text

Slide 50

Slide 50 text

An approach to complexity Two views of complexity (Dahl 2004, Nichols 2020) Inventory complexity the number of distinctive elements in the system Descriptive complexity the amount of information required to describe the system A note on phonology • Phonologists and typologists alike love measuring inventory complexity in phonology, because we all know what a phoneme is and how to count them. • Theoreticians have been looking for rigorous ‘evaluation metrics’, not completely without success (Rasin et al. 2021), but this is difficult to scale up for typological enquiry 13

Slide 51

Slide 51 text

Canonical typology and markedness Canonical typology (Corbett 2005, Round 2023) offers one possible way forward 14

Slide 52

Slide 52 text

Slide 53

Slide 53 text

Canonical typology and markedness Canonical typology (Corbett 2005, Round 2023) offers one possible way forward A (modest?) proposal • Although canonicity does not equal measurable descriptive complexity (Audring 2019), it provides some approximation (Nichols 2020) • For phonology, this first approximation is good enough to build arguments utilizing the Trudgill conjecture • Phonologists know descriptive complexity as ‘(descriptive) markedness’ (Hume 2011) NB! We should resist easy elisions between different aspects of ‘markedness’ (Rice 2007) • In particular, we do not expect a universal drive towards ‘the unmarked’. 14

Slide 54

Slide 54 text

Approaching contact-induced change in phonology How do we look for contact-induced change in phonology? There are heavy caveats, but the hope is that this approach can operationalize complexity in phonology without reference to either meaning or ‘difficulty’ 15

Slide 55

Slide 55 text

Approaching contact-induced change in phonology How do we look for contact-induced change in phonology? Convergence (or non-divergence?) • Straightforward transfer of matter • Less straightforward, but perhaps possible: transfer of pattern There are heavy caveats, but the hope is that this approach can operationalize complexity in phonology without reference to either meaning or ‘difficulty’ 15

Slide 56

Slide 56 text

Approaching contact-induced change in phonology How do we look for contact-induced change in phonology? Convergence (or non-divergence?) • Straightforward transfer of matter • Less straightforward, but perhaps possible: transfer of pattern Canonicity manipulation • Growth (or maintenance) of non-canonical patterns ≈ growth or maintenance of complexity • Increasing canonicity ≈ decrease in complexity There are heavy caveats, but the hope is that this approach can operationalize complexity in phonology without reference to either meaning or ‘difficulty’ 15

Slide 57

Slide 57 text

Types of contact-induced phonological change

Slide 58

Slide 58 text

Convergence under L1 multilingualism? Perhaps the most straightforward type of contact-induced change in phonology, commonly held responsible for many well-known ‘areal sound patterns’ • South Asian retroflexes (Emeneau 1956) • Jewish North-Eastern Neo-Aramaic in contact with Gorani and Kurdish (Khan & Mohammadirad 2024) • Slavic and Baltic in the Great Duchy of Lithuania convergence zone (Sudnik 1975, Erker & Wiemer 2011) 16

Slide 59

Slide 59 text

Slide 60

Slide 60 text

Slide 61

Slide 61 text

Long-term convergence: a closer look An areal pattern in Ireland • Southern Irish English: /t d/ tin den ≠ /t̪ d̪/ thin then (Kallen 2013) • Irish: /t d/ team deck [tiːmʲ dɪkʲ] ≠ /t̪ d̪/ tinn ‘ill’ [t̪ʲiːɲ] doigh [d̪ʌ] ‘pain’ (Ó Curnáin 2007) • Pre-/r/ dentalization: [t̪r d̪r], *[tr dr] • Irish English, north and south (Kallen 2013, Maguire 2020) • Irish: [t̪ d̪] in trail, motor, history… • English [θ ð] > Irish English [t̪ d̪] usually analysed as L2 imposition (Filppula 1999) • English [t d] was borrowed as [t̪ d̪] prior to the growth of English competence in the community: separate L1-actuated transfer • Dentalization: not Irish > English substrate (Maguire 2020) 17

Slide 62

Slide 62 text

Slide 63

Slide 63 text

Problems with directionality: the Balkan schwa • Many languages of the Balkan Sprachbund have undoubtedly phonemic central non-low vowels • Romanian câmp < campum ‘field’, fără < forās ‘without’, văzduh < Slavic *vъzduxъ ‘sky’ • Bulgarian zəb < *zǫbъ ‘tooth’ • Albanian këngë < Latin canticam ‘song’ • Macedonian (Vidoeski 1999) • Northern dialects: sən ‘dream’ < *sъnъ, dən ‘day’ < *dьnь • South-eastern dialects: vək ‘wolf’ < *vl̥kъ • Often listed as a shared sound pattern, one of the few phonetic ‘Balkanisms’, noted already in the earliest accounts of the Sprachbund (Seliščev 1925). 18

Slide 64

Slide 64 text

Slide 65

Slide 65 text

Shared sound change under L1 agency Some possible sources of convergence: 19

Slide 66

Slide 66 text

Shared sound change under L1 agency Some possible sources of convergence: • Perceptual magnet effects (Blevins 2017) 19

Slide 67

Slide 67 text

Shared sound change under L1 agency Some possible sources of convergence: • Perceptual magnet effects (Blevins 2017) • Shared sound change 19

Slide 68

Slide 68 text

Shared sound change under L1 agency Some possible sources of convergence: • Perceptual magnet effects (Blevins 2017) • Shared sound change Co-territorial vernaculars with parallel outcomes of the nasal schwa (Sawicka 2000, Marković 2007, Friedman 2018): • *ə̃ > ɔ in SE Macedonian and Meglenoromanian • > ə(N) in Albanian, Aromanian, W Macedonian • > ɔ̃ > ɔN in Albanian, SW Macedonian 19

Slide 69

Slide 69 text

Slide 70

Slide 70 text

Slide 71

Slide 71 text

How pervasive is shared change? • On the one hand, examples are not very difficult to find • Long vowel diphthongizations in Polabian and Wendland/Altmark German (Wiesinger 2004) • Pharyngealization segmentation in Neo-Aramaic, Gorani and Kurdish (Khan & Mohammadirad 2024) • Long-term balanced bilingualism that enables such change does not seem as common in (ahem) Western Eurasia as elsewhere • Even for established Sprachbünde like the Balkans the situation may need some nuance (Sobolev 2021) • We need much more work on diverse contexts that centres the multilingual repertoire 20

Slide 72

Slide 72 text

Convergence of systems • Shared sound changes explain synchronically converged systems primarily through diachrony • This is in line with much current thinking in phonology (Blevins 2004) and typology (Cristofaro 2019) • Can this explain all instances of phonological convergence? 21

Slide 73

Slide 73 text

Slide 74

Slide 74 text

Slide 75

Slide 75 text

Homoplasy What is shared is not the changes but the system that the changes bring about (Hock 2022: 682 on Indo-Aryan retroflexes) • Homoplasy: different diachrony, convergent synchrony (Lass 1997, Van de Velde & van der Horst 2013) • Possible explanations: • Importation via lexical borrowings • Shared sound change 22

Slide 76

Slide 76 text

Slide 77

Slide 77 text

Slide 78

Slide 78 text

An example: High Latvian vowel ‘reduction’ • East Slavic akan’e: *a [traˈva] ‘grass’ ∼ [ˈtravi̵] ‘grass.pl’ *o [vaˈda] ‘water’ ∼ [ˈvodi̵] ‘water.pl’ • Usually treated as CSl *o > a in an unstressed syllable • High Latvian (Seržant 2010) *a [ˈvodu͡ot] ‘drive.inf’ ∼ [ˈpovada] ‘reins’ *o absent in native vocabulary • Diachrony of High Latvian 1. *a > o except before front vowels/palatal consonants 2. Allophonic alternation, with /o/ in the elsewhere context: gods ‘year.nom.sg’ ∼ gadi ‘year.nom.pl’ (Standard Latvian gads ∼ gadi) 3. *ɛ > a in palatal contexts leads to phonemicization: /a/ ≠ /o/ 23

Slide 79

Slide 79 text

Slide 80

Slide 80 text

Slide 81

Slide 81 text

Contact-induced non-divergence Varieties that exist in contact with other languages can fail to participate in innovations found in non-contact varieties • Pyrenean Romance pleká ‘fold’ < plĭcare, saper ‘know’ < sapēre vs. Spanish llegar, saber: cf. Basque katea ‘chain’ < catēnam, bake ‘peace’ < pācem (Jungemann 1950) • Breton hañv ‘summer’, deñved ‘sheep.pl’ vs. Welsh haf, defaid < *samos, *damatī (Jackson 1967): cf. nasal vowels in Gallo-Romance • English is the only Germanic language to have maintained both PGmc [θ] and [w], both segments also present in Welsh (Tolkien 1963) 24

Slide 82

Slide 82 text

Slide 83

Slide 83 text

Dealing with retentions • In general, the answer has to be aggregating the data with typological methods (Sinnemäki et al. 2024) 25

Slide 84

Slide 84 text

Dealing with retentions • In general, the answer has to be aggregating the data with typological methods (Sinnemäki et al. 2024) • The specific problem in phonology is the high probability of parallel developments: 25

Slide 85

Slide 85 text

Slide 86

Slide 86 text

Slide 87

Slide 87 text

L1 agentivity and contact in phonology: summary • Many aspects where the framework is a good fit • L1 agentivity conducive to maintaining or increasing complexity/non-canonicity • Non-canonical structures can be acquired/transferred via L1 learning mechanisms • Perpetuation of structures via shared sound change • Issues that require more work • Contact-driven non-divergence • Homoplasy • Do we have enough stable multilingualism to go around to explain the convergences? 26

Slide 88

Slide 88 text

Shift-induced interference in phonology • This sounds like it should be unproblematic: subversion effects under conditions of language shift • Especially phonetic detail is widely understood to be L2-hard and subject to imposition • No shortage of proposals in the literature ascribing sound change to substrates/language shift 27

Slide 89

Slide 89 text

Slide 90

Slide 90 text

How does phonological imposition happen? • Three mechanisms for propagation of L2-driven phonological change • Ethnolectalization: contact-influenced variety stabilizes as distinct • Users of contact-influenced variety are a majority in the community • De-ethnolectalization and spread of originally contact-influenced features via community-internal dynamics • Does this happen? • Ethnolectalization: Hebridean English, post-Sámi Northern Norwegian • Numerical preponderance: requires particular sociohistorical situations: Southern Irish English (Filppula 1999) • De-ethnolectalization 28

Slide 91

Slide 91 text

Slide 92

Slide 92 text

Phonologically invisible language shifts • There are plenty of arguments in the literature reconstructing past contacts from phonological evidence, but do we have documentation? 29

Slide 93

Slide 93 text

Slide 94

Slide 94 text

Phonologically invisible language shifts • There are plenty of arguments in the literature reconstructing past contacts from phonological evidence, but do we have documentation? • In documented cases of complete language shift, phonological influence seems quite elusive • Latin: plenty of evidence for individual bilingualism (Adams 2003), but clear substrate effects in phonology are rare to non-existent! (Adams 2007) 29

Slide 95

Slide 95 text

Slide 96

Slide 96 text

Slide 97

Slide 97 text

What’s going on? • In these well-documented cases, language shift is slow (See Maguire 2020: for detailed argumenation for Ulster) • The community shifts over time, but at any given stage the proportion of shifters is low • Phonetic-phonological non-target forms may attract a social penalty in a way that grammatical features do not (On this difference see Eckert & Labov 2017) • Social dynamics rarely conducive to propagation of contact-induced features 30

Slide 98

Slide 98 text

Slide 99

Slide 99 text

Slide 100

Slide 100 text

Areal effects and drift Some ‘areal’ effects in northern Europe • Preaspiration • Sonorant preocclusion: *nn > dn • Tonal accents • Initial stress (Salmons 1992) • Contrastive quantity (Ewels 2009) • The languages are in contact, but the sociohistorical situation does not allow for L2-actuated phonological change: instead, many of the parallels emerge from the operation of the life cycle on similar starting points • This is a theory of drift (B. D. Joseph 2013) 31

Slide 101

Slide 101 text

Slide 102

Slide 102 text

Simplification in phonology? • If we are anywhere near the right track, we should see examples of ‘simplification’ (increase in canonicity) in L2-actuated phonological change… • … or perpetuation of complexity (non-canonicity) in the absence of L2-driven learning • One example from the literature: glide hardening in European vernaculars (Andersen 1988, Trudgill 2011) • Romansh flukr vekr for flūr ‘flower’ vejr ‘see.inf’ • Jutland Danish bic byc for biˀ ‘bee’, byˀ ‘town’ • Franconian German tsikt for Zeit ‘time’ • Norwegian dialectal beksel for beisl ‘bridle’ • Latvian dialectal juks kreikt for jū̀s ‘you.pl’, krī̀t ‘fall.prs.3sg’ • This change does, however, have at least phonetic grounding in the devoicing of high vocoids (Mortensen 2012) 32

Slide 103

Slide 103 text

Another (better?) example • Another such feature is vowel excrescence in sonorant + consonant clusters (Iosad & Maguire in preparation) • Variable, partly lexicalized • West Germanic: English, Scots, Dutch, much of High German (with Luxembourgish and Yiddish) • North Germanic: (historical) Danish • Brythonic Celtic • Variable, undergoing stabilization • Northern Sámi, Lule Sámi, other Western Sámi varieties • Finnish dialects • Currently productive in the phonology • Scottish Gaelic • Historically regular, now morphologized/lexicalized • Irish, Manx • East Slavic 33

Slide 104

Slide 104 text

Social dialectology and the non-canonicity of epenthesis • The excrescent vowel is non-canonical: arises from the misalignment of C- and V-gestures in the sonorant (Hall 2006), assuming a canonical segment is characterized by tight gestural coupling (Round 2023) • Why doesn’t it get into standard languages easily? • The social dynamics of LOL (Literate, Official, Lots of Users: Dahl 2015) languages are simplification-friendly • Large and loose social networks, historically most users learn the standard as a second language • Possible sociolinguistic typology account of how a pervasive feature like epenthesis flies so entirely under the radar • Are there more features like this? 34

Slide 105

Slide 105 text

L2 agentivity and contact in phonology: summary • L2-driven phonological change clearly exists at the individual level • Community-wide phonological imposition requires very specific sociohistorical circumstances, which may be relatively rare • Even for languages in contact, ‘substrate’ influence may not be visible in phonology 35

Slide 106

Slide 106 text

Some conclusions • The study of language contact is very exposed to the lack of communication between theoretical phonology and historical-typological linguistics 36

Slide 107

Slide 107 text

Slide 108

Slide 108 text

Slide 109

Slide 109 text

Some conclusions • The study of language contact is very exposed to the lack of communication between theoretical phonology and historical-typological linguistics • Phonology is special in ways that create challenges for this research programme • Both of these are crucial to future progress: • Theoretically informed approach to what a sound pattern is • Serious engagement with the sociohistorical context • Some important things that I think are true • Contact-induced change in phonological systems is rarer than we think • The place to look is often very localized contact between vernaculars • … and we need a much broader view of what is possible in terms of social interaction in diverse communities 36

Slide 110

Slide 110 text

Thank you! pavel.iosad@ed.ac.uk https://www.anghyflawn.net Bluesky @anghyflawn.net Some of this presentation builds on joint work with Warren Maguire and Dmitry Nikolaev 36