You are on page 1of 43

Parallel variation

in two speech
a comparison of final /n/ elision
in Idi and Ende

Dineke Schokkin Kate L. Lindsey

The Australian National University Stanford University
Big questions (cf. Evans 2018)
❖ Is there something special about the patterning of
variation in small-scale speech communities (Stanford
& Preston 2009)?

❖ To what extent can results of variation studies be

applied to societies in different times or places?
❖ On the one hand, the ‘uniformitarian principle’ (Labov
1972) is assumed to apply to historical contexts
❖ On the other, we must be ‘wary of extrapolating
backward in time to neolithic preurban societies’
(Labov 1994)
Why do it?
❖ There is evidence that we cannot assume the
“uniformitarian principle” to hold:
❖ Rates of linguistic change are different between
WEIRD and small-scale societies like in southern New
❖ Rates and types of multilingualism are very different
between societies
❖ Conditions of language acquisition are different
between societies
Advantages and
❖ Advantages
❖ Sample size as a proportion of total population
❖ Faster rates of linguistic change

❖ Disadvantages
❖ Ethnographic and social context less accessible
❖ No access to historical data
❖ Lack of comparative data
Idi and Ende
Idi Ende
Pahoturi River family
~1600 speakers in 6 settlements ~1000 speakers in 2 settlements
Settlements map
Täme ❖ This research: four Idi and
two Ende settlements
❖ Primary contact lgs.
through intermarriage:
Idi: Nen, Täme
Ende: Täme, Kawam,
Bitur, Gogodala
Background: SNG region
(cf. Evans et al. 2018)
❖ Very diverse: multiple
unrelated phyla

❖ High degree of egalitarian

(Haudricourt 1961; François

❖ Subsistence agriculture
+ hunting

❖ Small settlement groups

(tens – low 100’s
until 1950s-1960s)
Idi and Ende
Idi Ende
Pahoturi River family
~1600 speakers in 6 settlements ~1000 speakers in 2 settlements
Four nasal phonemes: m n ɲ ŋ
Prenasalisation of stops (with variation too)

❖ Word-finally, nasals are often deleted…

❖ Where does it happen? Which lexical items? What part of
the utterance?
❖ Who does this?
❖ Can we find a pattern?
❖ For this talk: focusing on verbs
Independence day procession in Dimsisi village, 2014
Verbal inflection (cf. Gast 2013; 2014)
❖ Suffix -(V)n: very frequent
❖ 1|3sg subject in remote past, recent past; present tense
❖ 3sg|2nsg subject in future
❖ underspecified (i.e. occurring with most
person/number values of subject) in intransitive
prefixing verbs, some copulas

❖ Overall, verb-final /n/ is dropped about 12% of the

❖Remote past lexical (suffixing) verb

nyonggo=a dämä be-a\nggas/en

path=CORE there 3sgU.REMPST-AUG-make-1|3sgA
‘He made a road there.’

❖ Recent past lexical (suffixing) verb

Gwadang ydi da n-a\nggas/an

‘What did Gwadang do (just now)?’
❖ Remote past prefixing verb

Emmy bi d-ya\r/ge-n
Emmy 1nsg.N O M REMPST-nsgU-go-nplS>plO-1|3sgA

river-A L L
‘Emmy and us went to the river.’

❖ Present progressive transitive

yau yndhpä y\r/an

NEG see 3sgU-A U X -1|3sgA.R E C P S T
‘I don’t see him.’
❖ Present progressive intransitive

Idi yeka yeka w\l/an

Idi language speak INTR-AUX-1|3sgA.RECPST
‘I am speaking Idi.’

❖ Future lexical (suffixing) verb

oil=a b-a\nggas/en
oil=CORE FUT-AUG-make-3sgA
‘(S)he will make oil (with the coconuts).’
Examples: wlan 1|3SG
1. Senior male, without final N deletion

2. Senior male, with final N deletion

3. Middle-aged female, without final N


4. Middle-aged female, with final N deletion

❖ Corpus of ~ 50,000 words, naturalistic speech
❖ ‘Coconut interviews’, procedurals, personal narratives, legends

❖ After data cleanup: 3,284 tokens of verbs ending in /n/

❖ Coded for
❖ Nasal-drop (y or n)
❖ # of syllables (2, 3, 4+)
❖ Preceding segment (low/mid/high full vowel; reduced vowel)
❖ Root and token frequency (high, middle, low)
❖ Following segment (vowel, consonant, pause)
❖ TAM value (remote past, recent past, present, future)
❖ Verb class (lexical, auxiliary, copula, prefixing)
❖ Subject person/number and expression (NP, pro, zero)
❖ Disregarded cases e.g.:
❖ Verbs in utterance-initial position: this is rare as Idi has
basic SOV order, and copulas/auxiliaries never occur
in this position
❖ The verb ibäny ‘plant’
❖ 2nsg.F U T inflections: categorical use of /n/, no elision
~ 20-40 y/o ~ 40-60 y/o ~ 60+ y/o Total
Female 6 4 6 16
Male 8 9 6 23
Total 14 13 12 39
❖ Age groups roughly correspond to ‘generations’:
parents, grandparents, great-grandparents (very few
young unmarried speakers in sample)

❖ Speakers were also coded for

❖ Clan affiliation
❖ Current place of residence
❖ Language of father/mother/spouse (if

❖ Only data from speakers that had more than 30 tokens

of verbs ending in /n/ were included
Classification tree

(see e.g. Tagliamonte & Baayen 2012)

Age effect
20-40 yo 40-60 yo 60+ yo Total
Future 0% 4% 10% 7%
Present 45% 61% 66% 60%
Recent past 0% 5% 16% 9%
Remote past 4% 2% 9% 5%
Total 8% 9% 17% 12%
Following Segment
20-40 yo 40-60 yo 60+ yo Total
C 10% 13% 21% 15%
P 8% 7% 16% 11%
V 5% 3% 3% 4%
Total 8% 9% 17% 12%
Logistic regression analysis
❖ Independent factors included
❖ Following Segment
❖ Verb class
❖ Age group
❖ Sex

❖ Speaker and word were included as random effects

(Johnson 2009)
Df Sum_sq F p
Fol_Seg 2 30.033 15.0166 < 0.00001 ***
Sex 1 0.367 0.3673 0.544522 n.s.
Age_Group 2 7.601 3.8006 0.022456 *
TAM 3 92.409 30.8028 < 0.00001 ***
V_class 3 2.123 0.7077 0.547327 n.s.
❖ TAM is single most important predictor of nasal elision: in
present, /n/ is elided in almost 60% of cases

❖ In the other tenses, following segment is a strong predictor:

/n/ is elided more often when the following segment is a
consonant or pause, as opposed to a vowel

❖ Age is a significant social factor: 60+ people show a

tendency to drop the nasal

❖ Women show a slightly higher tendency to drop the nasal

than men across all age groups (n.s.)
Letae Jerry beating sago in Limol village, 2018
❖ Conducted 73 interviews in Limol village in 2018

❖ Average interview:
❖ 18.3 minutes
❖ 127 /n/ final tokens

❖ Corpus of ~ 86,000 words (all sociolinguistic

questionnaire interviews)

❖ 9,120 tokens ending in /n/

~ 20-40 y/o ~ 40-60 y/o ~ 60+ y/o Total
Female 12 12 8 32
Male 17 11 8 36
Total 29 23 16 68
*5 speakers under 18 were not included to provide a better comparison with Idi
for this presentation
❖ Age groups match Dineke Schokkin’s study

❖ Speakers were also coded for

❖ Clan affiliation
❖ Hometown
❖ Language of father/mother/spouse
❖ Selected most frequent lemma dan ‘sing. sub. pres.
copula’ (N = 2134)
Uses of dan
❖ Two general types of copular uses

❖ With descriptive arguments

❖ Modifier: ‘He is big.’
❖ Nominal: ‘He is a big man.’
❖ Locational: ‘He is in Limol.’
❖ Dummy: ‘He is like this.’
Copulas with modifiers
Syntactic construction
❖ With non-descriptive arguments
❖ Knowledge: ‘My knowledge is writing.’ = ‘I know how
to write.’
❖ Can: ‘I am good I write.’ = ‘I can write.’
❖ Possessive: ‘My children are not.’ = ‘I don’t have
❖ Purposive: ‘I am for hunting.’ = ‘I will go hunting.’
Copulas in knowledge constructions
❖ Selected most frequent lemma dan ‘sing. sub. pres.
copula’ (N = 2134)

❖ Three variants: dan ~ da ~ danän

❖ Coded for
❖ Final nasal (0, n, nän)

❖ Following segment (vowel within 0.4s, consonant

within 0.4s, pause (#), intonation break(##))
❖ Syntactic construction
Classification Tree
Classification Tree
(without danän tokens)
❖ When the innovative, yet rare, danän variable is removed,
the non-realization of /n/ can be predicted by linguistic
variables alone: following segment (C or ##; p<0.001) and
construction type (non-descriptive; p<0.001).

❖ The non-realization of /n/ in the present copula dan

appears to be stable across the community from this
apparent time sample.

❖ The dan token, however, is perhaps most comparable to the

present tense auxiliary tokens in Dineke’s study of Idi,
which also shows community stability and very high rates
of n-dropping.
❖ The presence of the third variable, danän, can be
predicted by social variables: age (18-40; p<0.001) and
sex (female; p<0.001).

❖ BUT very low tokens (N=24)

❖ The sequence -än is homophonous with the 3rd

singular subject agreement suffix in non-copular verbs.

❖ 91% of the time it accompanies a descriptive


❖ Also appears on copulas of other types.

Next Steps
❖ Expand Ende dataset to include non-present tense

❖ Expand Ende dataset to include more copulas,

including those with a third danän type variable.

❖ Look at cognate Idi copula da ~ dand (not n-final, but

also showing variable realisation)
❖ In Idi, the oldest speakers are leading in the use of the
/n/-less form.

❖ In Ende, the youngest women are leading in the use of

the danän form.

❖ How can results of WEIRD variation studies inform

our understanding of Idi and Ende variation?

❖ Can we determine whether these are cases of age-

gradation or a change in progress?
Evans, Nicholas 2018. The dynamics of language diversity. In Mesthrie, Rajend and David Bradley (Eds.), The Dynamics of Language. Plenary and focus
lectures from the 20th International Congress of Linguists. Cape Town: UCT Press.
François, Alexandre. 2012. The dynamics of linguistic diversity: egalitarian multilingualism and power imbalance among northern Vanuatu languages.
International Journal of the Sociology of Language 2012(214): 85-110.
Gast, Volker, 2013. Issues in Idi verb morphology. Unpublished manuscript.
Gast, Volker. 2014. Verb morphology. Unpublished manuscript.
Haudricourt, André-Georges. 1961. Richesse en Phonèmes et Richesse en Locuteurs. L'Homme 1:1.5-10.
Johnson, Daniel E. 2009. Getting off the GoldVarb Standard: Introducing Rbrul for Mixed Effects Variable Rule Analysis. Language and Linguistics Compass
3(1): 359-383.
Kollwam, Aitru. 2018. SE_PI116: Sociolinguistic Questionnaire – Aitru Kollwam.
Labov, William. 1972. Sociolinguistic Patterns. Philadelphia, PA: University of Pennsylvania Press.
Labov, William. 1994. Principles of Linguistic Change, Vol. 1: Internal Factors. Oxford: Wiley-Blackwell.
Meyerhoff, Miriam. 2015. Turning variation on its head: Analysing subject prefixes in Nkep (Vanuatu) for language documentation. Asia-Pacific Language
Variation 1(1): 78-108.
Meyerhoff, Miriam. 2016. Borrowing from Bislama into Nkep (East Santo, Vanuatu): Quantitative and qualitative perspectives. Language and linguistics in
Melanesia, available from
Meyerhoff, Miriam. 2017. Writing a linguistic symphony: Analyzing variation while doing language documentation. Canadian Journal of Linguistics/Revue
Canadienne De Linguistique: 1-25. doi:10.1017/cnj.2017.
Nagy, Naomi. 2009. The challenges of less commonly studied languages: Writing a sociogrammar of Faetar. In James N. Stanford and Dennis R. Preston
(eds.), Variation in Indigenous Minority Languages, pp. 397-417. Amsterdam: John Benjamins.
Sowati (Kurupel), Skola. 2018. SE_PI066: Sociolinguistic Questionnaire – Skola Sowati Kurupel.
Stanford, James N. 2007a. Lexicon and description of Sui adjective intensifiers. Linguistic Discovery 5(1): 1-27.
Stanford, James N. 2007b. Sui adjective reduplication as poetic morpho-phonology. Journal of East Asian Linguistics 16(2): 87-111.
Stanford, James N. 2008. A sociotonetic analysis of Sui dialect contact. Language Variation and Change 20(3): 409-50.
Stanford, James N. & Dennis Preston. Variation in Indigenous Minority Languages. Amsterdam: John Benjamins.
Tagliamonte, Sali A. & R. Harald Baayen. 2012. Models, forests and trees of York English: Was/were variation as a case study for statistical practice.
Unpublished ms., available from
❖ Data collection

Puli Ämädu, Simon Bagi, Christian Döhler, Birke Eka, Nick Evans,
Carls Gana, Kaune Gana, Masa Gegera, Kmonde Gigu, Magham Greh,
Judy James, Qandro Kaeko, Tobias Maletz, Titi Masa, Sawe Masro,
Paul Mikuku, Bess Purge

❖ Data analysis

Danielle Barth, Kate Lindsey, Mark Ellison, Eri Kashima, Catherine

Travis & other members of the Wellsprings team

❖ Funding

ARC Laureate project ‘The Wellsprings of Linguistic Diversity’