ArticlePDF Available

Abstract and Figures

Old Rapa, the indigenous Eastern Polynesian language of the island of Rapa Iti, is no longer spoken regularly in any cultural domains and has been replaced in most institutional domains by Tahitian. The remaining speakers are elders who maintain it only through linguistic memory, where elements of the language are remembered and can be elicited but they are not actively used in regular conversation. Reo Rapa, a contact language that fuses Tahitian and Old Rapa, which has developed from the prolonged and dominant influence of the Tahitian language in Rapa Iti since the mid nineteenth century, has replaced the indigenous Old Rapa language at home and between most people in regular social interaction. This article analyzes Reo Rapa through an examination of its genesis and its structure. This article furthermore defines Reo Rapa as a unique contact variety, a shift-break language: a language that resulted from stalled shift due to a collective anti-convergence sentiment in the speech community. This article further discusses a variation of Reo Rapa speech, New Rapa, which presents important questions for the natural-ness of language change and the visibility of actuation.
Content may be subject to copyright.
     () -
©   , , | ./-
Ocial census data puts the Rapa Iti population at approximately 500 people (Challier, 2012),
however, based on my observations, there are not more than about three hundred people
Reo Rapa: A Polynesian Contact Language
Mary Walworth
Max Planck Institute for the Science of Human History
Université de la Polynésie française
Old Rapa, the indigenous Eastern Polynesian language of the island of Rapa Iti, is no
longer spoken regularly in any cultural domains and has been replaced in most insti-
tutional domains by Tahitian. The remaining speakers are elders who maintain it only
through linguistic memory, where elements of the language are remembered and can
be elicited but they are not actively used in regular conversation. Reo Rapa, a contact
language that fuses Tahitian and Old Rapa, which has developed from the prolonged
and dominant inuence of the Tahitian language in Rapa Iti since the mid nineteenth
century, has replaced the indigenous Old Rapa language at home and between most
people in regular social interaction. This article analyzes Reo Rapa through an exami-
nation of its genesis and its structure. This article furthermore denes Reo Rapa as a
unique contact variety, a shift-break language: a language that resulted from stalled
shift due to a collective anti-convergence sentiment in the speech community. This
article further discusses a variety of Reo Rapa speech, New Rapa, which presents impor-
tant questions for the natural-ness of language change and the visibility of actuation.
Rapa Iti – Polynesia – contact languages – mixed languages – actuation
1 Introduction
On the island of Rapa Iti, one of the least populated islands in French Polynesia,
a century and a half of indirect socio-political pressure from Tahiti, combined
 :    
journal of language contact 10 (2017) 98-141
permanently living on the island. The ocial census number is likely inated due to two
main factors: (1) students who board nine months of the year on other islands for school; (2)
permanent residents of Tahiti with Rapa heritage claiming residency in Rapa Iti to maintaina
cultural connection with the island.
At about 350 nautical miles from its nearest neighbor, Ra’ivavae, Rapa Iti is one of the most
isolated islands in the Pacic. In addition to its geographic isolation, Rapa is socially isolated
from other islands; due to the lack of an airport on the island, travel to and from Rapa is pos-
sible only by a several-day-long boat trip.
Reo Rapa translates literally to ‘language of Rapa’ in both Tahitian and Old Rapa. I have cho-
sen this name because it seems suitable for the language most widely used on Rapa Iti, and
spoken by nearly the entire population.
with physical and social isolation from the rest of the region has resulted in
a complex linguistic situation. Old Rapa, the indigenous Eastern Polynesian
language of Rapa Iti, is no longer spoken regularly in any cultural domains and
has been replaced in most institutional domains by Tahitian (Walworth, 2015).
The remaining speakers are elders who maintain it only through linguistic
memory, where elements of the language are remembered and can be elicited
but they are not actively used in regular conversation (Walworth, 2015). Reo
Rapa, a contact language that fuses Tahitian and Old Rapa, which has devel-
oped from the prolonged and dominant inuence of the Tahitian language in
Rapa Iti since the mid nineteenth century, has replaced the indigenous Old
Rapa language at home and between most people in regular social interac-
tion. In this article, I describe Reo Rapa through an examination of its genesis
and its structure. Furthermore, I explore how Reo Rapa can be dened in the
existing framework of contact languages, highlighting it as a unique contact
variety. Section 2 investigates how and why Reo Rapa came into existence,
through a discussion on the dominant inuence and prestige of Tahitian in
Rapa Iti. Section3 examines precisely how Reo Rapa is mixed, highlighting
which Old Rapa elements have been maintained (identifying tokens) and
where shifts to Tahitian have occurred. Section 4 addresses uniformity of
speaker choice between generations. Section 5 discusses how Reo Rapa ts
into the contact language framework, and section6 investigates its variant,
New Rapa.
2 Genesis of Reo Rapa: Tahitian Prestige
Reo Rapa has emerged from a prolonged dominance of Tahitian in Rapa Iti.
Tahitian has long been viewed as a more prestigious language among Rapa
people, maintained in all major cultural institutions since the introduction of
journal of language contact 10 (2017) 98-141
See Walworth 2015 for more on the inuence of Tahitian in various cultural domains.
Because these two languages are so closely related, they share a number of lexical and gram-
matical features.
= adjunctive . See Grammaire de l’Académie tahitienne (2009: 224) for the function
of this  in Tahitian, which is the same function here in Reo Rapa.
In Tahitian, ‘food’ is ’a (Fare Vāna’a, 1999). However, in Reo Rapa, both vowel segments
have become short resulting in ma’a.
Christianity in the mid-nineteenth century. Historically, the prestige of Tahi-
tian has been explicit: it has been associated with new religious ideals, forced
Tahitian instruction in early schooling, and use of Tahitian in the government.
The prestige of Tahitian still is evident in these areas today, but is now more
implicit than previously. Whereas the use of Tahitian was once more overtly
enforced, today the language is implicitly required for participation in any
socio-cultural domain, and to assimilate to what has become the Tahitian-
centered culture of French Polynesia.
3 Composition of Reo Rapa
Tahitian has been inuential in Rapa Iti to the point of bilingualism and subse-
quent dominance in nearly all cultural domains. From this situation, Reo Rapa,
the variety of speech made up of Old Rapa and Tahitian, has developed.
Most Reo Rapa content words come from Tahitian, save for a special set of
Old Rapa words, or Old Rapa tokens, that come from traditional activities and
practices. Grammatical words, on the other hand, are more evenly sourced
from both Tahitian and Old Rapa. Examples (1–4) demonstrate this varied level
of mixing. In these examples, words in bold are Tahitian derived, underlined
words indicate elements shared by both Old Rapa and Tahitian, and plain
italicized words are Old Rapa components.
() ’a haere mai -ku fare
 go  ART.PossO-S house
‘Come to my house.
() e ’āpā i -ku hoa e haere
 embrace  ART.PossA-S friend  go
i te fare tunu -ku ma’a
  house cook ART.PossA-S food
 :    
journal of language contact 10 (2017) 98-141
e tāma’a café ka oti
 dine dinner  nish
‘I’ll go kiss my friends go to my house cook my food eat dinner – done.
() kāre -na taringa e faa-ro’o
 ART.PossO-S ear  -listen
‘He does not listen.’ (lit. ‘his ears lack listening’)
() tinaki te mei’a ’e te eika
eat.with.hands  banana   sh
‘Eat the banana and sh with your hands.
In order to more precisely measure the division of features from Tahitian and Old
Rapa, I analyzed both casual speech data and the results of a cross- generational
language test that included 127 words and 22 sentences. This test was admin-
istered primarily in November and December 2013, with some remaining par-
ticipants tested in April 2014. I chose the test words and sentences based on my
analyses of casual speech observed in 2012 and 2013. I noted frequent and seem-
ingly consistent use of particular words in mixed speech by a number of speak-
ers and so selected these terms as test items. The lexical list (table1a and1b) was
meant to identify which Old Rapa token vocabulary was consistent. The test
sentences (table2) allowed me to examine the use of these terms in a more con-
nected speech (rather than as words in isolation) in order to assess approximate
use of vocabulary and grammatical words, as well as to observe syntactic pat-
terns. This second test was done because I observed that in some cases, speakers
were aware of the Old Rapa terminology and in the word-list elicitation would
produce an Old Rapa term, while in regular speech they would not use these
terms. One example is the words used to mean ‘go’. In word list elicitation, about
55 percent of participants produced the Old Rapa term, naku. However, in the
phrasal elicitation, 91 percent of participants used Tahitian haere to mean ‘go’.
I observed that haere was also used by everyone in casual speech. For example, a
common way to end an evening between friends is to say:
() haere e komo
go  sleep
‘Go and sleep.
Phrases were tested rst, then terms. I would say the sentence or word
in French (and in Tahitian for elder consultants who did not understand
French, with the help of a Tahitian-speaking translator), and participants were
journal of language contact 10 (2017) 98-141
For example, a common occurrence in word-list elicitation was for participants to state an
initial response and then immediately after remember the Old Rapa word and retract. The
following is an example from a male participant (), 44 years old, interviewed on November
23, 2013: {: rouge (‘red’) [listing], :’ute’ute [pause] non, non – kurakura}. The Tahitian
form ’ute’ute, instead of the Rapa form kurakura, was recorded as his response in this case.
instructed to translate the sentence into their “Rapa” language. Participants
were surveyed alone in a semi-private space and were asked to not disclose
the questions to others. Initial responses from test participants were con-
sidered over secondary responses to accurately reect their intuitive choice
(in natural settings).
Participants in this survey were from all adult age groups and both sexes,
and represent about 15 percent of the actual population. The goal was to survey
ve or six people from every age group, in order to see what patterns were con-
sistent across all age-groups that speak Reo Rapa. This was also done to check
 Elicited vocabulary for Reo Rapa test
Body Parts Island Life Kinship/People Environment Animals Space/Time
head house woman sky small
hair canoe man sun octopus night
beard oven parent moon, month dog morning
chin king brother fresh water goat
nose chief sister river shark
mouth school grandparent sea water sh
ear weed (the
taro bed)
family calm sea eel
neck peel taro with
children mountain turtle
stomach pound taro group of people hill tentacles
ngers taro bed generation island lobster
buttocks paddle person wave pig
tear name stone
skin Singular wind
cheeks Dual wave
teeth Plural rain
intestines re
 :    
journal of language contact 10 (2017) 98-141
 Elicited vocabulary for Reo Rapa test
Flora Verbs Adjectives Numbers Nouns Colors
taro look good  words yellow
know skinny  speech red
tree learn hard/rough/spiny  food black
come/go small  thought
eat large  egg
talk hot small bundle
divide cold  rewood
run  path
climb cave
cover (close) 
sleep 
work 
for possible variability between generations. The age groups are organized as
follows: age group 1: 18–29; age group 2: 30–39; age group 3: 40–49; age group 4:
50–59; age group 5: 60–69; age group 6: 70+. Table3 presents the age and sex of
each participant, by age group. These participants were chosen based on the
following factors: (1) they grew up/spent their childhood in Rapa; (2) they cur-
rently live in Rapa, or have spent most of their lives there; (3) they self-identify
as Rapa locals.
3.1 Phonological Prole of Reo Rapa
Reo Rapa exhibits a mix of phonological features from both Tahitian and Old
Rapa, where lexical items in Reo Rapa are phonologically marked for their
journal of language contact 10 (2017) 98-141
 Test participants’ age and sex, by age group
Age group 1 Age group 2 Age group 3 Age group 4 Age group 5 Age group 6
M,  M,  F,  M,  M,  M, 
F,  F,  M,  F,  F,  F, 
F,  F,  M,  F,  M,  F, 
F,  M,  M,  F,  F,  F, 
F,  F,  F,  F,  M,  M, 
F,  M,  M,  F,  F,  F, 
 Elicited sentences for Reo Rapa test
I went to look for my brother.
The boy was sleeping at school.
The teacher woke the boy up.
What are you doing?
The woman is in the taro bed.
This fort is bigger than that one.
I am taller than she is.
I wish you two happiness.
That man has three children.
 The shark bit the child.
 Here is my house.
 She is angry because of me.
 I cook taro every morning.
 This sh is bigger than the other sh I had the other day.
 There’s a woman over there.
 All of the women are over there.
 A few women are over there.
 I did not go to the house.
 He is jealous.
 Your book is on the table.
 That’s not my book.
 That’s mine.
 :    
journal of language contact 10 (2017) 98-141
respective source languages (apart from those contributions that are evi-
denced in both Old Rapa and Tahitian). As discussed in the introduction, there
is an active awareness that Reo Rapa mixes Old Rapa and Tahitian, but what’s
more is that speakers demonstrate some awareness of what sounds are from
Tahitian or Old Rapa. As shown in table4, phonemes that are absent in Ta-
hitian, though present in Old Rapa, are the velar nasal /ng/ and the velar stop
/k/; phonemes of Tahitian that are not found in Old Rapa are /h/ and /f/. Reo
Rapa speakers generally understand that where there is a glottal stop in Tahi-
tian, there “should” be a /k/ or /ng/ in Old Rapa, and where there is an /h/ or /f/
in Tahitian, there “should” be a glottal stop in Old Rapa. As a result, most Reo
Rapa speakers can usually identify which components of Reo Rapa are Tahi-
tian or Old Rapa, based purely on phonological forms that demonstrate these
contrasting sound correspondences. On account of the general phonological
awareness of speakers for each lexeme, I analyze the Reo Rapa phonological
system as layered or stratied, rather than demonstrating an adaptive phonol-
ogy. In Reo Rapa, all consonant phonemes from both Old Rapa and Tahitian
have been maintained. Table4 shows the consonant sound correspondences
of Proto Polynesian (), Proto Eastern Polynesian (), Tahitian (), Old
Rapa (), and Reo Rapa (), further demonstrating the layered consonant
Regarding vowels, Tahitian and Old Rapa share the same ve-vowel system
(a, e, i, o, u; with surface length contrast), thus Reo Rapa also exhibits ve
vowel phonemes. The precise articulatory nature of Reo Rapa’s vowels in con-
trast to those of Old Rapa and Tahitian is not examined in the current study,
but is an important topic for future investigation.
 Consonant relexes of  and  in Tahitian, Old Rapa, and Reo Rapa
 *p *t *k *m *n *ng *f *s *h *q *w *l *r
 *p *t *k *m *n *ng *f *s ø *q *w *r *r
 p t  m n  f h ø ø v r r
 p t k m n ng ø ø v r r
 p t k /  m n ng / f /  h /  ø ø v r r
This is particularly notable in the emerging new variety of Reo Rapa, New Rapa, discussed
in section6.
 For more on Old Rapa’s vowel system, see Walworth, 2015.
journal of language contact 10 (2017) 98-141
3.2 Morpho-syntactic Prole of Reo Rapa
Table5 provides an overview of the primary morpho-syntactic components
of Reo Rapa, by source language. Content words (lexicon) are mostly Tahi-
tian and grammatical words come from both source languages. Derivational
morphemes also come from both, with the slight dominance of Tahitian. The
sources of syntactic structures are much harder to identify, as these languages
are so closely related and have very similar syntax.
Some features such as imperfective , imperative , adjunctive ,
conjunctions, and the prenominal article are not included in the table because
these features are realized by (morphologically) identical items in both Tahitian
and Old Rapa (e.g., the prenominal article te is used in both languages; the
imperfective Tense-Aspect-Mood () marker is e in both languages; the im-
perative  marker is a in both languages; the possessive markers are and
in both languages; all prepositions except ‘instrumental’ and all preposition-
al location nouns except ‘above’ are the same in both languages). Additionally,
because Tahitian and Old Rapa are closely related, they have ahigh lexical sim-
ilarity due to their many cognate reexes from . Thus, a clear identication
 Summary of source language contributions in Reo Rapa
Old Rapa Tahitian
Lexicon culturally associated lexical
most of the rest of the
Grammatical Words perfective
lexical adverbs
denite article
S, S bound pronouns
all free pronouns except S
past negative
non-past negative
interrogative words
grammatical adverbs
locational noun for ‘above’
instrumental preposition
S free pronoun
existential negative
Derivational Axes passive nominalizing
 All Tahitian forms and their functions were observed with Tahitian speakers and con-
rmed in Fare Vāna’a (2009).
 :    
journal of language contact 10 (2017) 98-141
of the source language for a particular Reo Rapa lexical item is only possible
where either source language has developed separate innovations.
3.2.1 Lexicon
Although much of the lexicon is the same in both languages, most of unshared
lexical items come from Tahitian. There is only a small amount of Old Rapa
vocabulary in the Reo Rapa lexicon that is diferent from Tahitian forms. These
lexical items tend to be unique Old Rapa innovations, specialized vocabulary
for traditional Rapa Iti activities, and some basic vocabulary. Table6 presents
the results of the word-list elicitation test. Here, I have presented each word
elicited (in the same order as on the test), the Tahitian form, the Old Rapa
form, the prevailing choice, and the percentage of participants who chose
the prevailing form. Entries marked with an asterisk are those that showed
 Percentage of participants who chose Reo Rapa form in elicitation
English Gloss Tahitian Form Old Rapa Form Reo Rapa Form
head upo’o eipoko eipoko %
hair rouru rauka’a rauka’a %
beard huruhuru ta’a kumikumi kumikumi %
chin ta’a tanga ta’a %
nose ihu pitā’u pitā’u %
mouth vaha ngutu ngutu %
ear tari’a taringa taringa %
neck ’a’ī kakī kakī %
stomach ’ōpū kōpū kōpū %
 ngers manimani maiki’o manimani %
 buttocks ’ōhure komi komi %
 woman vahine p ē’ā vahine %*
 man tāne rua tāne %
 look hi’o/’ite noko noko %*
 know ’ite kite ’ite %
 learn ha’api’i ’āikete ha’api’i %*
 All Tahitian forms were taken from Fare Vāna’a’s Dictionnaire Tahitien-Français (1999).
 Many participants who did not use kakī instead used the word for throat in Tahitian,
journal of language contact 10 (2017) 98-141
 I have included the full list of numbers 1–10, although some of these terms are clearly
identical in Tahitian and Old Rapa.
 come/go haere naku haere %
 eat ’amu kai kai %*
 parent metua karakua metua %
 brother tu’āne tungāne tu’āne %
 sister tuahine tua’ine tuahine %
 grandparent rū’au ’ina’ina ’ina’ina %
 sky ra’i rangi ra’i %*
 day mahana ao mahana %*
 sun pake pake %
 moon, month ’āva’e kāvake ’āva’e %
 fresh water vai kōta’e kōta’e %
 river ’ānāvai mangavai mangavai %
 sea water miti kara miti %
 taro taro mīkaka mīkaka %
 good/well maita’i maitaki maitaki %
 one ’ē ta’i hō’ē %
 two piti rua piti %*
 three toru toru toru %
 four maha ā maha %
 ve pae rima pae %*
 six ono ono ono %
 seven hitu ’itu hitu %*
 eight va’u varu va’u %
 nine iva iva iva %
 ten ’ahuru rongouru ’ahuru %
 calm sea mania karamate mania %
 tear (n) roimata karavai karavai %
 name i’oa eingoa eingoa %
 hundred hānere rau hānere %
 thousand tauatini mano tauatini %
 mountain mou’a mounga mou’a %
 hill ’āivi taratika taratika %
 talk paraparau ’akaero paraparau %
English Gloss Tahitian Form Old Rapa Form Reo Rapa Form
 Percentage of participants who chose Reo Rapa form in elicitation (cont.)
 :    
journal of language contact 10 (2017) 98-141
 No one uses kono’u in regular speech.
 words, speech parau koai parau %*
 children tamari’i tamariki tamariki %
 group of
pupu puke pupu %*
 generation u’i kopanga u’i %
 divide vāhi panga’a vāhi %*
 stone ’ōfa’i kōni’i ’ōfa’i %
 wind mata’i matangi mata’i %
 skinny pārarai moko’i moko’i %
 nd singular ’oe koe koe %
 nd dual ’ōrua korua korua %
 nd plural ’outou koutou koutou %
 skin ’iri kiri kiri %
 hard, rough pa’a paka paka %
 yellow re’are’a rengarenga re’are’a %
 red ’ute’ute kurakura ’ute’ute %
 black ’ere’ere kerekere ’ere’ere %
 house fare ’are fare %*
 household ’utuāfare ngutuā’are ’utuāfare %
 small iti koio koio %
 small animal fanau’a kororio kororio %
 run horo ’or o horo %
 octopus fe’e ’eke fe’e %
 large rahi ngare rahi %
 dog ’ūrī kurī kurī %
 climb a’e kake kake %
 canoe va’a kami’a kami’a %
 cover tāpo’i tāpuki tāpuki %
 wave ’are ngaru ngaru %
 island fenua ’enua fenua %
 oven umu ko’otu ko’otu %
 cheeks pāpāri’a pāpāringa pāpāringa %
 sleep ta’oto komo komo %
 work ’ohipa ’anga ohipa %
 rain ua kono’u ua %
English Gloss Tahitian Form Old Rapa Form Reo Rapa Form
journal of language contact 10 (2017) 98-141
 This was a particularly interesting contrast as 100% of participants stated niho for teeth
and puaka ni’o for goat, lit. ‘pig with teeth’. It may be of interest here to note that, in Old
Rapa, puaka appears to signify any land mammal that has been recently introduced.
 night ru’i %
 re ahi ngara’u ngara’u %
 person ta’ata tangata tangata %
 teeth niho ni’o niho %
 goat pua’aniho puaka ni’o puaka ni’o %
 hot ve’ave’a veravera ve’ave’a %
 cold to’eto’e toketoke to’e to ’e %
 Cordyline
kaukaro kaukaro %
 intestines ’ā’au nakau nakau %
 food mā’a kaikai m ā’a %
 shark ma’o mango ma’o %
 sh i’a eika eika %
 weed taro bed ūtaru ūtaru %
 paddle (n) hoe ’oe hoe %
 morning po’ipo’i pōpongi po’ipo’i %
 eel puhi takaviri puhi %
 turtle honu ’onu honu %
 thought (n) mana’o manako mana’o %
 egg huero moa ’ua huero moa %
 tree tumu tumu rākau tumu rākau %
 stroll ori kāuari kāuari %
 swallow horomi’i omiri omiri %
 small bundle pū’ohu tākaravai %
 rewood vahie vahie %
 king ari’i ariki ari’i %
 chief ra’atira rangatira ra’atira %
 armpit ’ē’ē ketekete ketekete %
 assemble ha’aputu ’akaumu ha’aputu %
 gather plants ’ohi ko’i ko’i %
 path ’ē’a kēka ’ē’a %
English Gloss Tahitian Form Old Rapa Form Reo Rapa Form
 Percentage of participants who chose Reo Rapa form in elicitation (cont.)
 :    
journal of language contact 10 (2017) 98-141
evidence of almost full replacement by Tahitian in regular casual speech, even
though in the elicitation test many speakers were able to identify the Rapa
forms. Light shading indicates old Rapa lexical tokens. Additional Old Rapa
tokens observed in regular speech are: karā ‘basalt stone used for pounding
pōpoi’; most plant names; and most sh names.
3.2.2 Grammatical Words
Grammatical words come from both source languages. Old Rapa contributes
the following elements to Reo Rapa: perfective  marker ka; denite ;
question words a’awhat’, ’ea ‘where’, a’ea ‘when’, nā ’ea ‘how’; ’ia ‘how many’;
negatives ki’ere past negative, kāre non-past negative; adverbial comparative
ake; and all pronouns except for the rst-person singular free pronoun. Some
of these are presented in examples (6–12).
Animals that were brought originally have their own unique names: kurī ‘dog’, kiore ‘rat’,
manu ‘bird’, moa ‘poultry, puaka ‘pig’. Mammals that were introduced post-settlement
take puaka and an identier: puaka ni’o ‘goat’, puaka toro ‘cow’, puaka ’oro ’enua ‘horse’.
 dive hopu tīpoko hopu %
 school fare ha’api’ira’a ’are ’āikete’anga fare ha’api’ira’a %
 cry ta’i tangi tangi %
 smile ’ata kata kata %
 tentacles ’ave kakave kakave %
 lobster ’ōura kōura kōura %
 wet taro bed roki roki %
 sit pārahi no’o pārahi %
 peel taro
with nger &
’oni ’oni %
 pound taro tu’i tuki tuki %
 pig pua’a puaka puaka %
 ght moto tāmaki moto %
 bite hohoni kati kati %
 cave po’o ana po’o %
English Gloss Tahitian Form Old Rapa Form Reo Rapa Form
journal of language contact 10 (2017) 98-141
 In Reo Rapa, the lexeme rahi appears to be a Tahitian input, however the lexeme’s syn-
tactic function and semantic value in Reo Rapa is not always consistent with how it us
used in Tahitian. According to an anonymous reviewer, in Tahitian, rahi is used as a post-
nominal intensier where its semantic value is more quantitative than qualitative. For ex-
ample: ‘Ua para rahi te taofe ‘The cofee ripened in great quantity.’ To further demonstrate
the contrast between Reo Rapa and Tahitian, this reviewer suggested that the Tahitian
equivalent of the translation in example (6) would be the following,‘Ua para roa te taofe
which employs roa, a quantitative intensier, instead of rahi: ‘Ua para roa te taofe.
() perfective  ka:
ka rahi para te taofe
 much ripe  cofee
‘Some cofee was really ripe.
() denite tō:
e hina’aro na vau mei’a ra
 like  S  banana 
‘I would like those bananas (you mentioned).
() question words:
a. e a’a tō-koe huru
 what ART.PossO-S state
‘How are you?’ (lit. ‘what’s your state?’)
b. haere na koutou i ‘ea
go  Pl  where
‘Where are you all going?’
() past negative:
ki’ere vau i haere i te fare
 S  go   house
‘I did not go to a house.
() non-past negative:
kāre -koe puta
 ART.PossA-S book
‘You don’t have your book.’ (lit. ‘your book doesn’t exist’)
() adverbial ake:
me rahi ake teie eika i
thing big   sh 
 :    
journal of language contact 10 (2017) 98-141
tā-ku eika tō mahana ra
ART.PossA-S sh  day 
‘This sh is bigger than my sh the other day.’
() rst-person bound pronoun:
e pohehae ’ōna i a-ku
 jealous S  PER-S
‘She is jealous of me.’
In Reo Rapa, Tahitian inputs for grammatical words include: the subjunctive
 marker ’ia; preposition i; negatives aita ‘no’, ’eiaha ‘prohibitive’; preposi-
tional locative noun ni’a ‘above’; plural mau; demonstratives teie ‘this’, terā ‘that’;
rst person singular vau; adverbs noa ‘continually’ and iho ‘indeed, absolutely’;
and quantiers pauroa ‘all’ and rahi te mau ‘most’. Most of these grammatical
words function in the same way as their corresponding forms in Old Rapa.
Subjunctive  ’ia functions in the same way as Old Rapa subjunctive kia.
() ’ia rekareka kōrua
 happy Du
‘I wish you two happiness.
 prepositions *ki ‘direction, instrument, goal’ and *i ‘location, source,
cause’ have reexes in Old Rapa in two separate forms: ki and i. The two par-
ticles have merged in Tahitian (Fare Vāna’a, 2009; Lazard and Peltzer, 2000),
and therefore, the distinction between the two  prepositions has been lost.
Reo Rapa also shows a merger of these particles, taken from Tahitian.
() haere tāua i Hiri e ko’i te tuvava
go DuIncl  Hiri  gather  guava
‘Go (you two) to Hiri and gather guavas.’
() ’ea koe komo ai
where S sleep 
i roto i -koe fare
 in  ART.PossO-S house
‘Where is it that you sleep in your house?’
Tahitian contributes two negative particles, aita and ’eiaha. ’aita is used
in Reo Rapa as a simple “no” response. In Tahitian, ’aita is also used in exis-
tential negation, for which Old Rapa typically uses non-past negative kāre.
journal of language contact 10 (2017) 98-141
 The  forms in this section are taken from Greenhill and Clark (2011).
 Both forms are reexes of  *lunga ‘above’, however theTahitian form, ni’a , demon-
strates two sporadic sound changes that contrast with Old Rapa’s form, runga, which fol-
lows the regular sound changes that occurred in the language.
The Tahitian prohibitive negative ’eiaha functions in the same way as Old Rapa
eia’a. Most Reo Rapa speakers employ the Tahitian form.
() imperative and corresponding prohibitive:
a. ’a haere mai i -ku fare ……
 go   ART.PossO-S house
‘Come to my house.
b. eiaha ’a haere mai i -ku fare
  go   ART.PossO-S house
‘Don’t come to my house.’
Most prepositional locative nouns are shared by Tahitian and Old Rapa. Both
languages exhibit cognate reexes of  *loto ‘inside’; *lalo ‘below’; *muri
‘behind, with, after’ *muqa ‘ahead, in front, before’.  *fafo ‘outside, out’
has been replaced with an innovation rāpae in Tahitian. This form has been
borrowed into Old Rapa, in which we do not nd a reex of *fafo. Thus, the
source of this form in Reo Rapa is assumed to be Old Rapa rather than Tahitian.
This Tahitian innovation has fully replaced  *fafo ‘outside’ in Old Rapa. As a
result of these shared prepositional locatives, Reo Rapa also exhibits roto, raro,
muri, mua, and rāpae. The only contrasting prepositional locative between the
source languages is ‘above’: runga in Old Rapa and ni’a in Tahitian. Reo Rapa
exhibits Tahitian ni’a.
() -koeputateini’aihoi te’amu-ra’a
ART.PossA-Sbook above   eat-
‘Your book is on a table.
Old Rapa marks general plurality with anga, a reex of  *nga ‘plural’. Tahi-
tian, on the other hand, employs an innovated term mau to indicate plurality.
Reo Rapa speakers use the Tahitian form, mau.
() ka ’ite vau te mau vahine ti ko ra
 see S   woman here place 
‘I see some women just over there.
 :    
journal of language contact 10 (2017) 98-141
 The variation of vau is based on personal observations of Tahitian, but it is also reported
in Fare Vāna’a’s (2009) Grammaire de la Langue Tahitienne.
Example (12) demonstrates that Reo Rapa employs the Old Rapa rst person
bound pronoun. However, for the rst person independent pronoun, Reo Rapa
uses Tahitian vau rather than Old Rapa ou. It is important to mention that in
Tahitian, vau exhibits an allomorph au when following a front vowel. Reo
Rapa does not exhibit this allomorph and vau is used in all environments.
() e mana’o vau François i ’āfa’i
 think S  François  bring
mei’a ra i te fare
 banana    house
‘I think it was François who brought those bananas to a house.
Tahitian and Old Rapa share the demonstrative forms terā ‘that (far from
speaker and addressee)’ and tenā ‘that (far from speaker)’; thus, these elements
are used in Reo Rapa. A third demonstrative meaning ‘near (to speaker)’ is evi-
dent in both source languages, although with diferent forms: in Old Rapa, te
nei; in Tahitian teie. Tahitian teie is the form used in Reo Rapa.
() e mānea teie mea i a kōrua
 pretty this thing   Du
‘You two are glowing.’ (lit. ‘this thing is beautiful on you two’)
Finally, Tahitian contributes a number of adverbs to Reo Rapa. Tahitian noa is
used in lieu of Old Rapa ta’anga to mean ‘continuously’; Tahitian iho marks ‘in-
deed, absolutely’ and functions as a reexive, as well as indicating an upward di-
rection or that something has just occurred. This is used in place of Old Rapa noti.
() noa
e fa’aea noa vau i Rapa nei atu
 stay  S  Rapa  
te ’āva’e tītema
 month December
‘I am staying in Rapa from now until December.’
() iho
Tevaitau terā ai pāre tei iho i te oire
Tevaitau   fort     town
‘Tevaitau is the fort that is up from town.
journal of language contact 10 (2017) 98-141
 In Rapa Iti, when men catch sea urchins, they dump them on the dock and women gather
together to process them. This involves hitting the top of the urchin gently with a medium
sized stick to crush the spines and break the top of the shell.
3.2.3 Derivational Axes
Reo Rapa employs the causative and nominalizing axes from Tahitian. The
causative prex in Tahitian can be ha’a- or fa’a- and serves the same functions
as Old Rapa ’aka-. Reo Rapa uses both Tahitian forms of the causative. This
appears to be more lexical than grammatical, however, as I have found no in-
stance of either of the Tahitian prexes being used with an Old Rapa base.
() kāre e ua fa’a-hou
  rain CAUS-again
‘It’s not raining anymore.’ (lit. ‘There is no more rain.’)
() e fa’a-’amu na Marie i te puaka ni’o
 CAUS-eat  Mary   pig teeth
‘Mary is feeding a goat.
() ha’a-vitiviti koe ’ohipa
CAUS-fast PossA S work
‘Do your work faster.’
For nominalization, Reo Rapa uses Tahitian -ra’a instead of Old Rapa -’anga.
This sux is productive and is also used with  bases, as shown in (26).
() ka oti te ko’i-ra’a
 done  gather-
‘Gathering is nished.
In passive axation, Reo Rapa has maintained Old Rapa -’ia rather than Tahi-
tian -hia. The Old Rapa form is used with Tahitian bases as well, as shown in
() ka oti i tupa’i-’ia te vana te rākau
 done  hit-PASS  urchin   wood
‘I’m done hitting an urchin with a stick.’
() ka rave-’ia -na poti te ra
 take-PASS ART.PossO-S boat   
 :    
journal of language contact 10 (2017) 98-141
 This term was coined by Yaron Matras based on his observations of the language’s source
language contributions during a personal discussion with me in January 2015.
ka fati -ku poti
 break ART.PossO-S boat
‘We took his boat because mine is broken.’
3.3 Discussion of Source Language Contributions
The distribution of features from the source languages is consistent, however
it is not obviously systematic, i.e. the lexicon from one language and the gram-
mar from the other. The lexicon is sourced mostly from Tahitian, but grammar
is sourced from both Old Rapa and Tahitian. Grammatically, the features are
not clearly divided either; there is a mix of nominal and verbal morphology
from both source languages, as well as derivational and inection morphology.
The only visible distinction in the source language contributions is that the
Old Rapa features all contain a phoneme that contrasts with a corresponding
Tahitian phoneme, in a given morpheme. Thus, source language division ap-
pears to be phonologically motivated, based on phonological assimiliphobia,
where the Old Rapa contributions all contain an Old Rapa sound correspon-
dence that contrasts with the Tahitian sound correspondence in the cognate
form (i.e.,  ng ,  k , and   h, f).
4 Uniformity of Speaker Choice
This section considers the general consistency of Reo Rapa’s composition
through the lens of generational variation. Here, I answer the question: do Reo
Rapa speakers consistently pull the same components from the same source
languages? The responses from the sentence elicitation portion of my multi-
generational test indicate that Reo Rapa is very consistent. (29–35) are sev-
eral examples of the sentences elicited. Here again, terms in bold are Tahitian,
plain italicized terms are Old Rapa, and underlined terms represent identical
forms in both languages. These examples include a top line with percentages
indicating how many participants chose a particular morpheme. Percentages
are generally high, thus signaling the stability of Reo Rapa.
() % %-% % % %
ka fa’a-ara ’orometua ha’a-pi’i
 -awake  master -call
journal of language contact 10 (2017) 98-141
% % % %
i tō tamariki ra
  child 
‘The teacher woke the student.
() % % % % % % %
e toru tamariki a tō tangata ra
 child   person 
‘That man has three children’
() % % % % % % % %
ka kati-’ia tamariki ra e te m a’o
 bite-PASS  child    shark
A shark bit the child.’
() % % % % % %
me rahi ake vau i a-na
thing tall  S  PER-S
‘She is taller than I.’
() % % %
’ia rekareka kōrua
‘I wish you two happiness.
() % % % % % % %
ki’ere vau i haere i te fare
 S  go   house
‘I did not go to any house.
() % % %
ka riri ’ōna
 angry S
‘She is angry.’
5 Dening Reo Rapa
There is no question that Reo Rapa represents language mixing of some sort
due to contact. It is necessary now to explore what kind of language mixing is
 :    
journal of language contact 10 (2017) 98-141
occurring in Reo Rapa, and precisely what kind of contact language Reo Rapa
is. Before attempting to dene it, I will rst summarize Reo Rapa’s identifying
Reo Rapa was born from the introduction of the Tahitian language into a
monolingual community.
• The community did not require new communicative means, but rather
came to this new variety of speech as a result of bilingualism and subse-
quent shift due to the dominance and prestige of Tahitian.
Tahitian and Old Rapa, the source languages of Reo Rapa, are both Eastern
Polynesian languages and share many linguistic features including a large
proportion of their lexical items (both content words and grammatical
words) as well as nearly identical syntactic structure.
There is consistency in source language mixing among all ages of speakers,
indicating that Reo Rapa is stable and furthermore is inter-generationally
There is clear uniformity in source language choice at the morpheme level,
however the division in source language contribution is not divided system-
atically (i.e., grammar from one language and lexicon from the other; nomi-
nal morphology from one language and verbal from another; inectional
morphology from one language and derivational from another).
The source language choice is instead based on phonological assimiliphobia.
Given the above characteristics of Reo Rapa, how might we dene it in the con-
text of other contact languages and new varieties that arise from contact? Is
the mixing of Tahitian and Old Rapa just code-switching between two separate
languages? Is it simple borrowing from Tahitian, or even heavy borrowing from
continuing shift to Tahitian? Is Reo Rapa perhaps a koine or a mixed language?
Or is it a previously un-described type of language mixture due to language
5.1 Code-switching
Perhaps the strongest argument against Reo Rapa as just code switching, is
consistency and predictability in speaker choice from the source languages.
Tahitian forms and Old Rapa forms are not interchangeable for a given speaker.
In fact, the distribution of Tahitian forms and Old Rapa forms is consistent
across the entire speaker-base. Were it a case of frequent code switching, we
would expect a more random distribution of Tahitian and Old Rapa forms for
a given set of speakers.
journal of language contact 10 (2017) 98-141
5.2 Simple Borrowing
Borrowing refers to incorporating lexical items of one language into another
language’s vocabulary. Speakers typically consider the borrowed forms to be
part of their language and often adapt the pronunciation of these forms to t
the native phonological system. This is not what we nd in Reo Rapa. Rath-
er, the speakers are aware that certain forms belong to Old Rapa or Tahitian,
based on their phonological form. Additionally, Reo Rapa includes signicant
mixing of grammatical elements from Tahitian. The high level of grammatical
input from Tahitian indicates that this mixing is not simply lexical borrowing.
5.3 Continuing Shift
Once again, the evidence of high consistency between generations provides a
strong argument against Reo Rapa as a phase in continuing shift to Tahitian.
The consistency between generations indicates stability and inter-generational
transmission of Reo Rapa, which would not occur in continuing shift.
It is true that Reo Rapa was created through a “borrowing approach” (Meak-
ins 2013:188), where shift occurred from bilingualism and overt social pressures
to use Tahitian. In this way, the mixing of source languages in Reo Rapa has
occurred via steady borrowing from Tahitian over time. However, the shift that
was at one time progressive has stalled, as evidenced in the consistency of the
language’s components. In the model in gure1, taken from Meakins (2013: 182),
Reo Rapa would fall into the second category, “shift by degree,” where mixing
occurs by gradual shift from an ancestral language to an introduced language
resulting from a change in dominance (183). According to Meakins, in the “shift
by degree” category, the process of language shift “does not go to completion
ncestry language
ncestry language
ncestry language
Fusion of structural and lexical
Borrowing of lexical and
grammatical material
Re-introduction of vocabulary
Introduced language
Introduced languag
Introduced languag
Mixed marriage
Shift by degree
Reversal of shift
 Direction of shift in mixed language genesis (Meakins, 2013: 182)
 :    
journal of language contact 10 (2017) 98-141
 Yaron Matras used the term anti-convergence in personal conversations with me about
Reo Rapa, in January 2015.
and what remains is the mixed language” (183). The resulting language can
thus have varying amounts of material from the introduced language, but does
not represent complete language replacement. This is precisely what we see
evidenced in Reo Rapa.
Meakins (2013: 183) suggests two reasons for a halt in shift: (1) speakers do
not have full access to the introduced language, and/or (2) remaining parts
of the ancestral language may be a marker of social identity. The latter is true
for Reo Rapa, and is evident in two ways. First, as the majority of Old Rapa
lexical elements in Reo Rapa have to do with traditional activities and Rapa Iti
practices, they likely represent a unique Rapa Iti identity and have thus been
maintained. These tokens of Rapa identity are reminiscent of what Matras
and Bakker (2003: 7) refer to as an “inherited special lexicon”, or a selective
retention of an ancestral language’s vocabulary after language shift, which
“diachronically represents the selective retention of vocabulary, following lan-
guage shift.” Second, as a result of an anti-convergence sentiment, wherein
speakers want to retain Old Rapa in attempts to not speak Tahitian, the gram-
matical contributions of Old Rapa in Reo Rapa are consistently morphemes
that contain some phonemic diference from Tahitian. The resistance to speech
assimilation, to sounding Tahitian, has stopped the shift process from going to
full completion.
5.4 Koine
Reo Rapa cannot be considered a koine because it did not emerge through
koineization, a process in which “new varieties of language are brought about
as a result of contact between speakers of mutually intelligible varieties of that
language” (Kerswill, 2002: 669). According to Trudgill, koineization usually oc-
curs when people from diferent parts of a single language area settle in a new
place together (1986). Furthermore, koineization is the consequence of speech
accommodation, the adaptation of speech between speakers (Trudgill, 1986;
Kerswill, 2002), that results from “intimate and prolonged social interaction
(Siegel, 2001: 6–7). A koine is therefore a language that mixes features of mutu-
ally intelligible dialects due to the speech accommodation occurring between
speakers of those dialects who have found themselves living together in a new
journal of language contact 10 (2017) 98-141
 A reviewer of an earlier version of this paper suggested that Reo Rapa might be in a pre-
koine phase of koineization, but I do not believe this to be the case. Siegel dened the
pre-koine phase as “the unstabilized stage at the beginning of koineization…in which
various forms of the varieties in contact are used concurrently and inconsistently” (1985:
373). Siegel’s pre-koine coincides with Trudgill’s “Stage ” of koineization (Trudgill, 1998
in Kerswill, 2002: 686). In Trudgill’s Stage , the speakers are migrants and are no rst-
language speakers. Under these denitions, Reo Rapa cannot be considered a pre-koine.
I have demonstrated in this paper that Reo Rapa is stable across generations, that it is in-
tergenerationally transmitted, and that the source language contributions are consistent.
Under the above denition, Reo Rapa cannot be considered a koine.
First, Reo Rapa’s source languages are Tahitian and Old Rapa, which are not
mutually intelligible. These languages are members of the same subgroup of
the Polynesian Language family (Eastern Polynesian), but they are not dialects
of each other. Second, the contact situation between Old Rapa and Tahitian
speakers was indirect, and the two communities were never in prolonged, lo-
cal contact. Instead, Reo Rapa emerged from bilingualism in a community
that was previously monolingual, and then shifted to the dominant Tahitian
language. Reo Rapa is thus the result of shift in one speech community, not
compromise between two communities, and it did not come about through
speech accommodation.
5.5 Mixed Language
If Reo Rapa did not emerge from accommodation or a need for mutual com-
munication, perhaps then, Reo Rapa is a mixed language. In the simplest
terms, a mixed language is dened as “the result of the fusion of two iden-
tiable source languages, normally in situations of community bilingualism”
(Meakins, 2014: 392). Additionally, “what distinguishes mixed languages from
other contact varieties is that they emerge as expressions of identity rather
than as a result of a communicative need” (Meakins, 2013: 186). According to
these basic criteria, Reo Rapa shows characteristics similar to other mixed lan-
guages: it emerged as a result of socio-cultural pressure, which lead to bilin-
gualism and a subsequent fusion between Tahitian and Old Rapa. Reo Rapa
was born from the introduction of the Tahitian language into a monolingual
community; the community did not require new communicative means, but
rather came to a new variety of speech as a result of bilingualism and domi-
nance of one language over the other.
Matras and Bakker (2003: 1) further dene mixed languages as “varieties that
emerge in situations of community bilingualism, and whose structures show
an etymological split that is not marginal, but dominant, so that it is dicult
 :    
journal of language contact 10 (2017) 98-141
to dene the variety linguistic parentage as involving just one ancestral lan-
guage.” The fundamental element of an ambiguous “linguistic parentage” is fre-
quently referenced in literature on mixed languages (Matras and Bakker, 2003;
Meakins, 2013; Thomason, 1995: 16, 2003: 21). Matras and Bakker go so far as to
write that mixed languages are radically diferent languages from their ances-
tral languages and that they are “substantially diferent from earlier stages of the
language before the incorporation of structures from a second source” (2003:
11–12). This factor is not clear-cut in Reo Rapa. It does apply to Reo Rapa in one
sense, as Tahitian has replaced certain features unique to Old Rapathrough
mixing. This has led to classication of the ancestral language of Rapa Iti as
Tahitic and even as a dialect of Tahitian, which is an inaccurate classication
(cf. Walworth,2015). However, because the two source languages of Reo Rapa,
Old Rapa and Tahitian, are so closely related, they already share a number
of features, most notably their syntactic structure. With source languages so
closely related, and thus having many cognate morphemes, there is no way to
determine the source language for all of Reo Rapa’s features. This is perhaps the
most signicant problem for dening Reo Rapa as a mixed language.
Another potential diculty in labeling Reo Rapa as a mixed variety of
speech is that contributions from the source languages are not split system-
atically. Many mixed languages demonstrate mixed systems where the gram-
mar is predominantly derived from one source language and the lexicon from
another (Golovko, 2003: 191; Meakins, 2013: 179; Meakins, 2014). While there is
signicant variation in the degree to which mixed languages fuse grammatical
features form their source languages (Meakins, 2014: 393; 2013: 179; Matras p.c.
2015), mixed languages are typically categorized as either Grammar-Lexicon
(-) mixed languages or Verb-Noun (-) mixed languages (Meakins, 2013:
179). - mixed languages are more structurally mixed, where the mixed lan-
guage combines the nominal system of one source language and the verbal
system of the other (173). Reo Rapa does not t either of these categories. At
an item level, the mixing is consistent, but at the system internal level (e.g.,
lexical vs. grammatical/syntactic; nominal vs. verbal; inectional vs. deriva-
tional), there is no stark diferentiation of source languages. There is no clear-
cut division in source language contribution at a system level. Although the
contributions from each source language are clear and consistent in Reo Rapa,
they can only be observed at a unit level.
 Matras, in personal discussions with me about Reo Rapa, echoed Meakins and expressed
that mixed languages are extremely varied in composition which makes it dicult to
make a clear structural denition for contact varieties categorized as “mixed languages”.
journal of language contact 10 (2017) 98-141
5.6 Shift-break Language
Reo Rapa does not t perfectly into any of the criteria for the existing catego-
ries of contact languages. Reo Rapa thus represents a type of language mixing
that has not yet been dened in the current framework for contact languages.
It is perhaps best described as a shift-break language: a language that resulted
from stalled shift due to a collective anti-convergence sentiment in the speech
community. In Reo Rapa, anti-convergence has manifested in unique Old Rapa
vocabulary and phonological assimiliphobia. Anti-convergence not only has
caused the halt in shift, but has also prompted further changes toward a rever-
sal in shift, leading to the variant of Reo Rapa, called New Rapa, which is fur-
ther examined, in section6. Figure2 demonstrates the ow of language shift
in Rapa Iti.
As indicated by gure2, Reo Rapa is a language born from bilingualism
in Rapa Iti of Tahitian and Old Rapa, and then developed out of language
shift and subsequent fusion of linguistic features. This shift has stalled due
to anti-convergence sentiments, and Reo Rapa has become intergeneration-
ally stable. The halt in language shift due to anti-convergence represents a
new kind of contact language, a shift-break language. The shift-break has re-
cently evolved into a reversal of shift and a new variety of Reo Rapa speech,
New Rapa.
6 New Rapa
People under the age of 50 sometimes use a variety of Reo Rapa, which I have
termed New Rapa. New Rapa represents an attempt by younger age groups
in Rapa Iti to reverse the shift to the Tahitian language. Whereas in Reo Rapa
Old Rapa features that phonologically contrast with Tahitian are retained, in
NewRapa, Tahitian elements of Reo Rapa are phonologically modied to re-
ect what speakers assume sounds more like Old Rapa. This desire to make
their language sound more like Old Rapa stems from a covert prestige in being
a “true local” Rapa person.
Old Rapa
New Rapa
Reo Rapa
[Reversal of Shift]
[Language Shift] [Language Shift] Ta
 Evolution of language shift in Rapa Iti
 :    
journal of language contact 10 (2017) 98-141
6.1 Reverse Shift
If we view Reo Rapa as a language in which “shift by degree” has occurred, New
Rapa could be considered a “reversal of shift.” In the New Rapa speech style,
younger Reo Rapa speakers are adding back what they assume to be lexical
and phonological features of Old Rapa, as a means of language creation. These
reverse additions to Reo Rapa sometimes coincide with legitimately Old Rapa
features; however, oftentimes the additions appear to be historically inaccu-
rate, as discussed in section6.5. Therefore, it would appear that the goal of
the speakers is to create speech that sounds like Old Rapa, rather than truly
returning to Old Rapa; they adopt an Old Rapa style of speaking, rather than
speaking the Old Rapa language. This language creation, because it is centered
on conscious generalizations about the Old Rapa language, is rarely used in
casual or spontaneous speech. It is used more frequently in contexts where
language can be planned (e.g., formal speeches, elicited speech, chants, and
written materials). For this reason, it also emerges in newer forms of media,
such as Facebook and popular music.
The intentional use of modied speech to express a certain group identity
mirrors what Thomason has referred to as “deliberate language diferentiation”
(1995: 30). In deliberate language diferentiation, a speech group will modify
language in order to dene a distinct cultural identity. New Rapa represents,
then, a sort of linguistic “u-turn,” (Boretzsky and Ilga, 1994 in Meakins, 2013:
182), which Meakins (2013) described as “an intentional undoing of shift to-
ward an outside language, where a speech group tries to reclaim their ancestral
language.” In Rapa Iti, however, the creation and use of New Rapa is not sim-
ply an attempt to reclaim the ancestral language; it is a form of resistance to
Tahitian linguistic and cultural assimilation, and a reection of an attempt to
return to a unique Rapa Iti identity.
6.2 Cultural Nostalgia and Rapa Insider Identity
In recent years, a motivation to form a unique Rapa identity apart from the
rest of French Polynesia, most specically Tahiti, has developed. Culturally,
a Rapa-insider identity is being amplied. Those who are from Rapa Iti and
do not move away from the island feel they have more knowledge of the past
and therefore identify themselves as “more” Rapa than Rapa people who have
moved away or outsiders who have moved to the island permanently.
Rapa Iti is caught between a desire to preserve its older traditions in or-
der to be diferent from other islands in the French Polynesian Territory and
 The lyrics of a very popular song among Rapa people, “Pito,” which was written in New
Rapa, can be found in Appendix D of Walworth (2015).
journal of language contact 10 (2017) 98-141
the material desire of youth to move toward a more Tahitian, and increasingly
French (Western), way of living. To this move by the youth, there is radical pull
back by the elders. This is manifested less in actual preservation and perpetua-
tion of traditional activities and rather in a sort of cultural nostalgia – a longing
for old ways that is expressed but not necessarily acted upon. This cultural nos-
talgia invokes positivity for the Rapa way of life by reecting negatively on how
things are done on other islands, particularly in Tahiti. Furthermore, there is a
sense among the Rapa people that “outsiders” from Tahiti cannot understand
how life is carried out on Rapa, and most importantly that they will not be able
to cope with, or will not want to participate in, the more traditional lifestyle.
One woman explained, “All these young girls coming from Tahiti, they won’t go
in the roki (‘wet taro bed’). They don’t want to get muddy and they are scared
of eels.” Another woman consultant of mine routinely said of her daughter-
in-law, who grew up in Tahiti, She’s lazy [because] she’s from Tahiti. She won’t
learn how to make pōpoi or help with making bread” (though in reality, the
daughter-in-law does try frequently to learn and help).
6.3 Linguistic Nostalgia and the Covert Prestige of “localness”
The cultural nostalgia that has developed out of pride in being a Rapa insider
and the appreciation of Rapa ways of life expressed through the depreciation
of Tahiti are paralleled in the creation of New Rapa. New Rapa thus reects
a linguistic nostalgia, a longing for the old ways of speech (Walworth, 2015).
Where Reo Rapa developed from an overt pressure to use Tahitian in multiple
ocial domains, New Rapa is developing from a less-explicit desire to mark
an insider Rapa Iti identity. Linguistic nostalgia has led to a covert prestige,
where localness is dened by how “Rapa” one sounds. Old Rapa’s special inher-
ited lexicon and the Old Rapa phonemes k and ng have become attributed to
“localness.” These linguistic marks of being local in New Rapa can be viewed as
a reection of covert prestige (in contrast to Tahitian’s overt prestige in Rapa
Iti society). Covert prestige generally refers to the hidden values that are as-
sociated with a non-standard form of speech (Labov, 1972: 249; Trudgill, 1972:
183). In this case, there is a hidden value in being more local, and localness is
attributed to non-standard, Old Rapa sounding lexemes and sounds.
6.4 New Rapa’s Genesis
The attachment of a “Rapa identity” to Old Rapa sounding forms can be almost
entirely attributed to one man, Pierrot Faraire. Pierrot Faraire is the founder of
 This particular situation is likely also related to the general treatment of in-laws in Rapa
(see Walworth, 2015).
 :    
journal of language contact 10 (2017) 98-141
the Tomite Reo Rapa (Rapa Language Committee), a former principal of the
Rapa elementary school, and the leader of Tamariki Oparo (a dance group that
performs and competes in a territory-wide competition, the Heiva, every July
in Tahiti). Faraire founded the Tomite Reo Rapa in the late 1990s, in order to
assemble the elders to re-discover the older form of the language.
As the founder and longtime leader of the Rapa dance group, Tomite Reo
Rapa, and a religious leader as well, Faraire has been able to inuence Rapa
Iti society and in particular the teenage and young adult students from Rapa
Iti who are going to school in Tahiti. For them, he perhaps represents a con-
nection to their home island, and as the leader of the dance group Tamari-
ki Oparo, he is a Rapa Iti celebrity. In his role as the dance group leader, he
also writes chants and songs to accompany the dancing, using what he feels
are sounds and words of an authentic Old Rapa language. According to him,
“The chants that I write are a way to renew our language.” These chants have
a direct inuence not only on younger Rapa people in Tahiti, but on people
in Rapa, because the dance competitions are televised, and everyone in Rapa
watches them and hears the new Tamariki Oparo songs. Faraire’s innovations
thus quickly permeate the Rapa Iti society. Many elders on Rapa disagree with
the word creation exhibited in his chants. They believe Faraire’s language to be
fabricated, and some even refuse to use words that have become commonplace
now in Rapa. Among these are the common greeting aronga and a term mean-
ing ‘thank you’, tongia. In fact, non-elder members of the Rapa community
admit to never having heard “Pierrot’s language” in the past; however, they do
not challenge his authority and assume that he has information about the past
that they do not have. The younger age groups, particularly those who have
participated in the cultural dance group, or have simply admired the dance
group (it is quite popular due to its success in the annual French Polynesia-
wide competition), have begun incorporating New Rapa, or la langue de Pierrot
(‘Pierrot’s language’) into their speech. They learn his new terms in the perfor-
mance chants for the cultural festival, and believe them to be some older form
of the language. They then use them to assert their Rapa Iti localness. This is
precisely how words like aronga and tongia have permeated the everyday life
of Rapa people.
 According to my elder consultants, there were no Old Rapa terms for ‘hello’ or ‘thank
you’. These concepts were brought with Tahitian, and so the Tahitian words ia ora na and
māuruuru were used for ‘hello’ and ‘thank you’, respectively.
 Whenever I engaged with non-elders about language, they would routinely ask me, “Have
you talked with Pierrot? He probably knows.
 An example of this use comes from a discussion over dinner with my host family. My host
father was very involved with the church, and knowledgeable in Old Rapa. Becauseof
journal of language contact 10 (2017) 98-141
As a result of his inuence, “Pierrot’s language” is becoming a part of the
Rapa speech today and is forming a new speech style on the island, New Rapa.
For example, aronga, whether historically accurate or not, is now a common
greeting used by everyone in Rapa Iti, save for a few elders who vehemently
disagree that it is, or ever was, an Old Rapa word. Tongia is equally common as
a way of saying ‘thank you’. It is perplexing that these words have been able to
enter into regularly used speech in Rapa Iti in spite of being disputed by elders.
I suggest that it has been possible because Faraire has told people that the
words were recorded a long time ago. Speakers have no reason not to believe
that they are truly ancient Old Rapa words, and words perhaps so far gone that
even their elders do not remember them.
6.5 Processes of New Rapa’s Creation
New Rapa is growing beyond the isolated innovations of Faraire and is evident
in more formal language use of younger age groups in Rapa Iti. New Rapa ex-
hibits an attempt to “Rapanize” the Tahitian elements of Reo Rapa in order
to make Reo Rapa sound less like Tahitian. The New Rapa style of speech
includes mostly the same distribution of source language features as Reo Rapa.
The major diference in New Rapa is that it also contains the following: (1)
Tahitian lexemes that have undergone Rapanization, by being phonologically
modied to reect Old Rapa’s consonant phoneme inventory; (2) borrowed
terms from other languages that have the same consonant phonemes as Old
Rapa; and (3) Faraire’s creations (e.g., aronga).
6.5.1 Rapanization of Tahitian Lexemes
Much of New Rapa relies on the process of Rapanization. As a result, the speech
style has developed from speakers creating their own words based on what
they think sounds more like Old Rapa. It is an educated guessing game, based
on what people generalize to be the phonological diferences between Old
Rapa and Tahitian. Recall from previous sections that there is some speaker
this, I was asking him one evening about the meaning of a word in a prayer that I had
heard at church that morning. The word was karanga, in this context meaning ‘speech’
or ‘the word’. He said that he did not recognize that meaning of the term, which, for him,
meant ‘world’. I made a note and as I was writing, my host sister (age 30), who had danced
in Tamariki Oparo, said, “No, that’s right. That word is in one of Pierrot’s songs.
 Kieviet and Kieviet (2006) briey mentioned that some words used in Rapa Iti appeared
to be Tahitian words that had been rapanisé ‘Rapanized’. They did not explore this obser-
vation further, but they should be acknowledged as the rst to use the term “Rapanized”
in this context.
 All Tahitian terms referenced in this paper were taken from Fare Vāna’a (1999) unless oth-
erwise indicated.
 :    
journal of language contact 10 (2017) 98-141
awareness of which particular sounds exist exclusively in Old Rapa or in Tahi-
tian. Speakers generalize that the velar nasal ng and the velar stop k are “Old
Rapa sounds” because they are present in Old Rapa, but absent in Tahitian.
Similarly, “Tahitian sounds” are generalized to be h and f, as these are present
in Tahitian but absent in Old Rapa. Based on these generalizations, speakers
conclude that where there is a glottal stop in Tahitian, there must be a k or ng
in Old Rapa; where there is an h or f in Tahitian, the corresponding form in Old
Rapa has a glottal stop in its place. This awareness is used in the process of Ra-
panization to target what a speaker believes to be a Tahitian sound and replace
it with an Old Rapa sound.
As Tahitian and Old Rapa share many reexes from , a speaker’s intu-
ition for Rapanization is often historically accurate; that is, it adheres to regular
sound correspondences. The examples in table7 are this kind of Rapanization.
In these cases, there is no way of identifying if a form is actually a form that
was used in Old Rapa, or if it is a modern Rapanization of a Tahitian form. In
fact, the only way to determine this is when the two languages do not exhibit
cognate forms (due to innovation in one or the other language). This is where
the language modication is more visible and the process of reanalysis can be
detected. Table 7 shows examples of Rapanized Tahitian forms that contrast
with Old Rapa forms, due to innovation either in Old Rapa or in Tahitian.
In other cases, as mentioned, it is not as clear if the New Rapa term is a Rapa-
nization of a Tahitian form, or is a re-introduction of an Old Rapa form. These
are instances where the Old Rapa form has been lost and so there is no way to
compare the New Rapa form with the Old Rapa form, or instances where the
Old Rapa form and the New Rapa form are identical. Table8 shows examples
of when the Old Rapa form has been completely lost; elders no longer remem-
ber it. Furthermore, my elder consultants denied that these Rapanized forms
existed in Old Rapa. We can therefore safely assume that the Rapanizations are
not re-introductions of Old Rapa forms.
There are two additional situations in which we can more clearly see a mod-
ern modication of terms. The rst is when replacement of a Tahitian sound
occurs in words or concepts that were borrowed into Tahitian after Western
contact. In most cases, these represent concepts that were introduced through
the missionaries or the Bible. Because Christianity was transmitted to Rapa
Iti islanders through Tahitian, we can safely assume that the Tahitian forms
for Christian concepts were the original forms in Rapa Iti. The two primary
examples of this are Tahitian himenesing’, hepetomaweek’ and puta ‘book’ to
New Rapa ’imene, ’epetoma and puka, respectively. My elder consultants attest
that these three terms are newer forms and that only the Tahitian forms were
used previously. Consequently, we can safely assume that puka, ’epetoma and
’imene are modern Rapanizations.
journal of language contact 10 (2017) 98-141
 Greenhill and Clark (2011).
 This is a possibly related proto-form, which meant ‘handsome lothario’ in  Greenhill
and Clark, 2011).
  *fera ‘spread wide open’ (Greenhill and Clark, 2011).
 Old Rapa () puke means ‘group of children’.
 Rapanized forms in New Rapa with conlicting Old Rapa forms
Old Rapa Tahitian New Rapa  Gloss
ka ’ua kua *kua ‘perfective’
 ’anga ’ohipa ’o’ipa *sanga ‘work’
no’o pārahi pāra’i *nofo ‘sit down; stay’
 takaviri puhi pu’i *pusi ‘eel’
ao mahana ma’ana *qaho ‘day’
’ā vahine va’ine *fane ‘woman’
 tīpoko hopu ’opu *sopu ‘dive’
mānea nehenehe ne’ene’e *mana-qia ‘pretty’
 pō ru’i ruki *po ‘night’
 ngare rahi ra’i *lasi ‘large, many, much’
 ā maha ma’a *fa ‘four’
 ketekete ’ē’ē kēkē  *keke ‘armpit’
 paruparu rohirohi ro’iro’i  *ru ‘tired’
 araara fera ’era *fera ‘wide-eyed; revelation’
 ou vau/au vou *ou ‘S’
 tūra pi’i piki *kalanga ‘to call out’
 Examples of Rapanized Tahitian lexemes rejected by elders
Old Rapa Tahitian New Rapa (Rapanized)  Gloss
ø’ori kori *koli ‘dance’
øhere ’ere *sele ‘love; beloved’
øpu’e puke *puke ‘pile’
øhuero ’uero *fua ‘egg’
øfa’a’apu ’akakapu ø ‘plantation’
øfatu ’atu  *fatu ‘master; chief
øfati ’ati *fati ‘break’
øhoa ’oa *soa ‘companion, friend’
 :    
journal of language contact 10 (2017) 98-141
The second situation in which we can more readily identify Rapanization
is when only some of the Tahitian sounds are replaced. An example of this is
Tahitian ’ohipa ‘work’, which becomes New Rapa ’o’ipa. If there were an Old Rapa
cognate for the Tahitian form, based on sound correspondences of Old Rapa
and Tahitian, we would expect the Tahitian initial glottal stop to be replaced,
in addition to the Tahitian h. However, the New Rapa form only replaces the Ta-
hitian h. As a result, I suggest that o’ipa is a reanalyzed Tahitian borrowing and
not an Old Rapa term. This partial replacement is also observed frequently in
complex words, both compounds (example 36) and axed forms (examples
() ‘suppress’
 ’oroma’i >  ’oromaki
’Oro is a reex of  *koro ‘intend; desire’ and ma’i/maki are reexes of 
*maki ‘sickness, illness, a sore’ (Greenhill and Clark, 2011). Based on regular
sound correspondences, we would expect to see something like koromaki in
Old Rapa. However, this is not the form that is exhibited and, as a result, we
can presume that oromaki is not an Old Rapa term, but is instead a Rapanized
Tahitian form.
() ‘obey’
 ha’apa’o >  akapa’o
() ‘gather together’
 ha’apu’e >  akapu’e
() ‘learn’
 ha’api’i >  ’akapi’i
() ‘reliance’
 turu’ira’a >  turu’i’anga
() ‘reading (n)’
 t ai’ora’a >  tai’o’anga
() ‘celebration’
 turu’ira’a >  turu’i’anga
In these examples, only the Tahitian sounds in the ax are replaced, but those
in the root are not. This is due to a basic linguistic awareness that the Tahitian
journal of language contact 10 (2017) 98-141
causative axes fa’a - and ha’a- correspond with Old Rapa aka- (examples
37–39), and that the Tahitian nominalizing sux -ra’a corresponds to Old
Rapa-’anga (examples 40–42).
6.5.2 Historically Inaccurate Rapanization
There are many instances in which Rapanization does not follow regular
sound correspondences; that is, the sound used in New Rapa is not expected
based on regular sound correspondences between Old Rapa and Tahitian.
Cases such as these are usually called “hypercorrection. In New Rapa, there
are two types of historically inaccurate Rapanization: (1) a historically unex-
pected “Old Rapa sound” replaces a “Tahitian sound” (e.g., ng is used where k
might be expected); or (2) an “Old Rapa sound” hyper-inserts, thus creating a
new syllable segment. This occurs with both content words and grammatical
The most prominent example of historically inaccurate Rapanization is
aronga, which translates to both a greeting and ‘love’. This word is used fre-
quently on Rapa, but it is a “new” term that Rapa people recognize as not used
until about ten years ago. Anyone who is familiar with regular sound corre-
spondences in Polynesian would notice that this term is a result of an attempt
to provide a reex of Proto Polynesian () *qarofa, also meaning ‘love’
(Greenhill and Clark, 2011). One would also note the peculiar presence of the
velar nasal. Based on regular sound correspondences, one would expect aro’a
in Rapa as the reex of  *qarofa, cognate with Tahitian aroha ‘compassion,
love, salutation’ (Fare Vāna’a 2008). Here, in order to sound more like Old Rapa,
the velar nasal is used in lieu of the expected glottal stop.
Similarly, in the lexical test described in section2.2.1, many of my consul-
tants under the age of 50 provided the form tukāne for ‘brother or woman’s
brother’. The form in Tahitian is tu’āne. Based on other cognate forms in
Eastern Polynesian languages (reexes of*tuŋane) and regular sound cor-
respondences, we would expect ng to occur in Old Rapa, in the place of
Tahitian’s glottal stop, resulting in tungāne. Here, however, the velar stop is
  *faka- (Greenhill and Clark, 2011).
  *-Canga (Greenhill and Clark, 2011).
 Reexes of *qarofa in many  languages are used as a greeting. These reexes are listed
in Greenhill and Clark, 2011 under  *qarofa.
 Consciousness of this choice was explained to me by Pierrot Faraire, in 2014.
 Greenhill and Clark (2011).
 :    
journal of language contact 10 (2017) 98-141
A slightly diferent instance of historically unexpected replacement is in the
rst person singular pronoun. In Old Rapa, rst person singular is ou. InReo
Rapa, rst person singular is vau, one of the Tahitian forms. In New Rapa,
speakers often use a created form, vou, which exhibits retention of the Tahi-
tian labiodental fricative from vau, and a replacement of Tahitian au with Old
Rapa ou. This is demonstrated below, in example (43), taken from a New Rapa
translation of a Tahitian poem, “Te Poreho”.
() ’aka-rongo ake ra vou
-hear   S
ki te ngarungaru o te tai
  waves PossO  sea
‘Then, I listened to the waves of the sea.”
A fairly consistent example of hyper-insertion of an “Old Rapa sound” is the
insertion of k in certain grammatical markers. Among Old Rapa’s prepositional
markers are i and ki (Walworth, 2015). In Tahitian (and in Reo Rapa), these two
markers have merged to i. In New Rapa, k is inserted in all Reo Rapa instances
of preposition i, resulting in ki throughout. In this case, sounding more like
Old Rapa means sounding “not Tahitian”; thus, both the i and ki prepositions
in Old Rapa are expressed as ki in New Rapa. Similarly, the accusative marker
becomes ki in New Rapa. In Old Rapa, as well as in Tahitian (and Reo Rapa) the
accusative marker is i. Examples (44) and (45), elicited from a male speaker
(age 29) and a female speaker (age 24) demonstrate these hyper-insertions.
Hyper-insertions are in bold, and the reader may refer to section4 for the stan-
dard Reo Rapa utterances corresponding to these examples.
() kua fa’a-ara te orometua ha’a-pi’i
 -wake  master -call
ki tamariki ra
  child 
A teacher woke the student.
() me rahi ake vau ki a-na
thing tall  S  -S
‘I am taller than she.’
 ake ra functions in a narrative as a transition in time, ‘then’.
journal of language contact 10 (2017) 98-141
The elders have little tolerance for the hyper-insertion of k in these types of ut-
terances. One of my elder consultants said, “They are putting k in everywhere.
It sounds like k-k-k-k-k, caca.”
6.5.3 Borrowing from Polynesian Languages
Some terms have been borrowed from other Eastern Polynesian languages that
are not Tahitian. For the most part, these terms appear to have been borrowed
from Rarotongan, in-line with Faraire’s comment that he uses a Rarotongan
dictionary to nd Old Rapa words. These borrowings are typically found in
prayers that were composed by Faraire. One such example is kōpapa ‘body;
corpse’ (also heard as akakōpapa ‘lie down; genuect’). My elder consultants
did not agree with this, reporting to have never heard this term and to have
only ever heard tino, the Old Rapa reex of  *tino ‘body’. Rarotongan does
have a term kōpapa ‘body’, which appears to be an innovation. Most other 
languages have a reex of  *tino for ‘body, including Tahitian. It is likely
that this term was borrowed by Faraire from Rarotongan into Rapa because it
is clearly distinct from the Tahitian form.
6.5.4 Re-introduction of Old Rapa Lexemes
Not all of New Rapa is word creation. It does appear that some Old Rapa vo-
cabulary has been reintroduced. It is clear that these words have been reintro-
duced because the Old Rapa form is not used in standard Reo Rapa, where the
Tahitian form is used instead. I suggest that they have been reintroduced via
Faraire through his chants and translated prayers. Some examples of this are:
anga ‘plural classier’, ’enua ‘island’, ngutua’are ‘household’, ta’unga ‘leader’;
’ā ‘woman’, and karakua ‘parent’. These terms were all heard in prayers and
sermons/convocations given by Rapa Iti deacons at church services (examples
46–49). ’enua and ngutua’are have also appeared in Facebook posts by Rapa
islanders as well as in song lyrics. It should be noted that the Old Rapa mean-
ings of ta’unga ‘leader, karakua ‘parent’, and ngutuā’are ‘household’ have been
expanded in the religious context to include ‘deacon’, ‘God (as the ultimate
father)’, and ‘congregation’, respectively. In the following examples, New Rapa
words are in bold. Examples (46–48) are taken from the prayers and sermons
at church services; example (49) is from a FaceBook post by a 30 year old Rapa
() te karakua nō te ’ere ’e te aronga
 parent   love   compassion
 Caca is French slang for excrement.
 :    
journal of language contact 10 (2017) 98-141
aka-kōpapa atu na mātou i mua
CAUS-lay.down   PlExcl  in.front
i a-koe
 PER-S
‘Father, for your love and compassion, we lay down before you.
() ’aka-tura tō karakua tāne ’e
CAUS-upright  parent man 
karakua pē’ā
 parent woman
‘Respect the father and the mother.’
() aronga te anga ta’unga
greeting   leader
‘Hello, deacons.
() te ’apa maitaki atu nei vau
 embrace well   S
ki tō-ku ngutuā’are ki te ’enua Rapa
 ART.PossO-S family   island Rapa
‘I am embracing my family there in Rapa.
6.6 Non-uniformity of Utterances
It is very important to note that New Rapa is not consistent between speakers;
rarely do any two speakers provide an identical utterance for the same mean-
ing in elicitation. New Rapa is not uniform because so much of it is based on
the individual speaker’s assumption of what is more Old Rapa sounding, which
in turn is often based on an inconsistent knowledge of sound correspondences
and a lack of knowledge of Old Rapa forms. Building on their understandings
of the diferences between Old Rapa and Tahitian, speakers “test out” various
combinations to see what will be accepted by their listeners as local “Rapa
speech. Over time, perhaps certain features of New Rapa will become more
stable, but presently, they are inconsistent. For example, the phrase ‘that wom-
an in the taro bed’ has one consistent Old Rapa form, and one consistent Reo
Rapa form, but New Rapa has multiple possibilities (example 50). In fact, the
same speaker might produce all the variations on separate occasions. There
does not appear to be any pattern in which variation is used when, nor are
choices clearly based on social situations or present company. Variations do
not appear to be related to context, gender, presence of other speakers, or age
group of speaker. It seems to be simply language experimentation. The only
clear motivation is to use forms that are diferent from Tahitian.
journal of language contact 10 (2017) 98-141
() ‘that woman in a taro bed’
Old Rapa: pē'ā ra ki roto ite roki
 woman   in   taro.bed
Reo Rapa: vahine ra iroto ite roki
 woman   in   taro.bed
New Rapa (bold items here are those that vary):
pē'ā/va'ine ra ki/i roto ki/i te roki
6.6.1 Doublets in Reo Rapa
Doublets are a common occurrence in language contact situations where sig-
nicant borrowing has occurred (Blust, 2011). Reo Rapa also exhibits doublets
due to the contact situation, but they are a unique type of doublet that has
resulted from the re-nativization of borrowings rather than through imple-
mentation of a borrowed form. These doublets difer from those described
by Blust (2011) in that they are not lexical replacements that contradict native
phonology. They are instead re-nativized forms of lexical replacements. These
doublets are made up of a Tahitian sourced Reo Rapa form and a Rapanized
form. Examples of some of Reo Rapa’s doublets are given in table9.
 Doublets resulting from Rapanization
Reo Rapa (Tahitian) New Rapa Gloss
’ori kori ‘dance’
’ohipa o’ipa ‘work’
pārahi pāra’i ‘sit down; stay’
puhi pu’i ‘eel’
mahana ma’ana ‘day
māhina mā’ina ‘moon
vahine va’ine ‘woman’
hopu ’opu dive’
nehenehe ne’ene’e ‘pretty’
 ru’i ruki ‘night’
 rahi ra’i ‘large’
 maha ma’a ‘four’
 ’ē’ē kēkē ‘armpit’
 huero ’uero ‘egg’
 fa’a’apu ’akakapu ‘plantation’
 :    
journal of language contact 10 (2017) 98-141
6.6.2 Homonyms
Finally, in the creation of New Rapa, homonyms are forming between Tahitian
components of Reo Rapa and Rapanized forms. Examples (51–54) demonstrate
some new homonyms that have resulted from Rapanization.
() ra’i
 ‘sky’ <  *langi ‘sky’
 ‘large’ <  *lasi ‘numerous, large, great’
() ’are
 ‘wave’ <  *kale wave that ripples or breaks’
 ‘house’ <  *fale ‘house’
() ma’a
 ‘food’ <  *maqanga ‘mouthful of food’
 ‘four’ <  *fa ‘four’
() ’ere
 ‘black’ <  *kele ‘dark, black’
 ‘love, beloved’ <  *sele ‘beloved, prized’
6.7 Broader Implications of New Rapa
The type linguistic innovation occurring in Rapa Iti is important to high-
light as it demonstrates historical changes that are intentionally initiated by
individuals in a speech community, and which do not follow regular sound
correspondences. This kind of deliberate change provides evidence for socially
motivated, rather than linguistically motivated, sound change and furthermore
 All  forms in these examples are from Greenhill and Clark (2011).
Reo Rapa (Tahitian) New Rapa Gloss
 ’ite kite ‘to know’
 to’eto’e toketoke ‘cold’
 rohirohi ro’iro’i ‘tired’
 fera ’era ‘wide-eyed’
 hoa ’oa ‘friend; companion’
journal of language contact 10 (2017) 98-141
supports theories that sound change is not always linguistically motivated
(Blust, 2005; Milroy, 2003). Blust wrote (2005: 221) that “social forces are widely
recognized as the engine driving the implementation of some sound changes,
but until recently these have not been implicated at all in the actuation of
sound change.” In the creation of New Rapa there are clear “exogenous motiva-
tions” (Milroy, 2003), system external motivations, for the changes occurring in
the language. While a small example, New Rapa provides proof that actuation
of sound change can be motivated by purely social forces.
The notion of altering one’s speech in order to associate with (or disassoci-
ate from) a particular social group is not at all unusual. However, clearly identi-
fying the source for the linguistic features that come to mark a particular social
group is typically not possible; the precise source of actuation is rarely known
(Labov, 1972: 317). In 2001, Labov (89–119) discussed actuation of language
change in terms of a “triggering event”, or cause event, which sparks system
internal change, or in his words, “a bend in the chain of causality”. However,
as he shows through his many examples, while we can identify the change, the
“bend”, it is dicult to specically identify the moment of actuation. Labov
(2001) and Baker (2008) have furthermore proposed the idea of a linguistic
leader, a speaker who observes a relationship between sound and identity and
adapts their speech accordingly. This marked speech then spreads to other
speakers and consequently afects a sound change (Baker, 2008: 29). However,
there are few real-time examples that ofer evidence of linguistic leaders.
Whether through a triggering event or via a linguistic leader, the implemen-
tation of a linguistic feature that later becomes associated with a particular
group’s identity may go completely unnoticed, according to Labov (1972: 319).
He wrote specically: “The change rst appears as a characteristic feature of a
specic subgroup, attracting no particular notice from anyone” (319). New Rapa
contrasts with this description, as both the inception and implementation of
its characteristic features are clearly visible. In this way, New Rapa provides a
rare opportunity to identify the clear and deliberate speech changes a speech
group is making in order to establish cultural distance from another group.
7 Conclusion
Reo Rapa presents a new type of contact language—one born from bilingual-
ism and developed out of language shift. The stall of the language shift has
resulted in a fusion of linguistic features that have become intergenerationally
stable. The halt in language shift due to anti-convergence represents a new
kind of contact language, a shift-break language. The shift-break has evolved
 :    
journal of language contact 10 (2017) 98-141
into a reversal of shift and a new variety of Reo Rapa speech has emerged as
New Rapa. The changes that have developed in New Rapa and the visibility of
actuation present a unique opportunity. While there are many examples of
language contact afecting language change, there are very few that can point
to the precise moment or exact person that initiates a change. In fact, most
of the discussion on possible language external forces being responsible for
the actuation of language-internal change is speculative or anecdotal at best
(Golovko, 2003: 184–185; Kulick, 1992: 2–3; Laycock, 1982: 36). New Rapa there-
fore represents one of the few cases available for study that provides concrete
evidence demonstrating non-linguistically motivated actuation of language
change. Finally, and fundamentally, Reo Rapa’s unique development and New
Rapa’s unusually visible genesis ofer an excellent example of the importance
of studying smaller and particularly endangered languages.
I wish to acknowledge Yaron Matras for taking time to discuss the Rapa Iti
language situation with me and for providing comments and helpful sugges-
tions on a previous version of this paper. Many thanks to Felicity Meakins,
Lyle Campbell, and two anonymous reviewers for providing suggestions for
improvement on earlier drafts. My additional thanks are due to Katie Drager,
Yuko Otsuka, Robert Blust, and Ken Rehg for their comments on the data (orig-
inally written-up as part of my dissertation). Finally, I am indebted to the many
people in Rapa Iti who ofered their time to participate in my language sur-
vey; to Pierrot Faraire for taking time to talk with me about language creation
and conservation; and for my primary consultants (Te’a Tamata, Teuira Vahine,
Ma’urei Angia, Takura Angia, and Lionel Watanabe) whose expert knowledge
of Old Rapa made it possible for me to compare it with the other varieties of
speech I found spoken on Rapa Iti. The research for this paper was funded by
the Chateaubriand Fellowship in the Humanities and the Bilinski Foundation.
All errors in the citation or interpretation of the data are my own.
Baker, Adam. 2008. Addressing the actuation problem with quantitative models of
sound change. Working papers in linguistics 14(1): 29–4.1.
Blust, Robert. 2005. Must sound change be linguistically motivated? Diachronica, 22(2):
journal of language contact 10 (2017) 98-141
Blust, Robert. 2011. The problem of doubletting in Austronesian languages. Oceanic
Linguistics 50: 399–457.
Challier, Steven. 2012. Atlas démographique de la Polynésie française: Recensement de la
population. Institut Statistique de Polynésie française: Pape’ete.
Fare Vāna’a. 1999. Dictionnaire Tahitien-Français. Tahiti: Académie Tahitienne.
Fare Vāna’a. 2009. Grammaire de la langue tahitienne. Tahiti: Académie Tahitienne.
Golovko, Evgeniy V. 2003. Language contact and group identity: The role of “folk”
linguistic engineering. In The mixed language debate: Theoretical and empirical
advances, Yaron Matras and Peter Bakker (eds.), 177–208. Berlin: Mouton de
Greenhill, Simon J. and Ross Clark. 2011. POLLEX-Online: The Polynesian Lexicon Proj-
ect Online. Oceanic Linguistics 50(2), 551–559.
Kerswill, Paul. 2002. Koineization and accommodation. In The handbook of language
variation and change, J.K. Chambers, Peter Trudgill, and Natalie Schilling-Estes
(eds.), 669–702.
Kieviet, Paulus and Antje Kieviet. 2006. Lexique Rapa-Francais. SIL International.
Kulick, Don. 1992. Language shift and cultural reproduction. Cambridge: Cambridge
University Press.
Labov, William. 1972. Sociolinguistic Patterns. Philadelphia: University of Pennsylvania
Labov, William. 2001. Principles of linguistic change, Volume 2: Social Factors. Massachu-
setts: Blackwell Publishers.
Laycock, Donald C. 1982. Melanesian Linguistic Diversity: A Melanesian choice? In
Melanesia: Beyond diversity, Ronald James May and Hank Nelson (eds.), 33–38. Can-
berra: The Australian National University.
Lazard, Gilbert and Louise Peltzer. 2000. Structure de la langue tahitienne. Paris:
Matras, Yaron and Peter Bakker. 2003. Introduction. In The mixed language debate: the-
oretical and empirical advances, Yaron Matras and Peter Bakker (eds.), 1–20. Berlin:
Mouton de Gruyter.
Meakins, Felicity. 2013. Mixed languages. In Contact languages: A comprehensive guide,
Peter Bakker and Yaron Matras (eds.), 159–228. Boston: Mouton de Gruyter.
Meakins, Felicity. 2014. Language contact varieties. In The languages and linguistics of
Australia: A comprehensive guide, H. Koch and R. Nordlinger (eds.), 365–416.
Milroy, James. 2003. On the Role of the speaker in language change. In Motives for lan-
guage change ed. by Raymond Hickey, 143–157. Cambridge: Cambridge University
Siegel, Jef. 1985. Koines and koineization. Language in Society 14(3): 357–378.
Siegel, Jef. 2001. Koine formation and creole genesis. In Language contact and creoliza-
tion, N. Smith and T. Veenstra (eds.), 175–197.
 :    
journal of language contact 10 (2017) 98-141
Thomason, Sarah G. 1995. Language mixture: Ordinary processes, extraordinary re-
sults. Spanish in four continents: Studies in language contact and bilingualism, Car-
men Silva Corvalán (ed.), 15–33. Washington, D.C.: Georgetown University Press.
Thomason, Sarah G. 2003. Social factors and linguistic processes in the emergence of
stable mixed languages. In The mixed language debate: Theoretical and empirical
advances, Yaron Matras and Peter Bakker (eds.), 21–40. Berlin: Mouton de Gruyter.
Trudgill, Peter. 1972. Sex, covert prestige and linguistic change in the urban British Eng-
lish of Norwich. Language in Society 1(2): 179–195.
Trudgill, Peter. 1986. Dialects in contact. Oxford: Basil Blackwell Inc.
Walworth, Mary. 2015. The language of Rapa Iti: Description of a language in change.
PhD Dissertation. Honolulu: University of Hawai‘i at Mānoa.
... where an entire subset of a speech community might use a particular sound to reinforce ingroup membership. This kind of relationship between identity and specific features of speech has been observed in numerous speech communities throughout the world (Labov 1972;Trudgill 1972;Walworth 2017) and has also been suggested as a historical motivation for linguistic divergence in Vanuatu more specifically (François 2012). It is therefore entirely possible that bilabial trills form an integral part of the linguistic identity of speakers, which would support their persistence over time. ...
Full-text available
This paper demonstrates that unexpected sound changes are best explained by an approach that accounts for different motivations: phonetic, structural and social. Here, we focus on a multifaceted investigation of the cross-linguistically uncommon bilabial trills to show the complex interaction between different drivers of sound change. In this paper, we highlight and examine the prenasalized voiced bilabial trill mʙ and plain voiceless bilabial trill P [ʙ̥ ] found in a number of Oceanic languages spoken on Malekula Island in Vanuatu. We offer a comparative-historical analysis, and we identify the various forces that led to the emergence and persistence of mʙ and P in Malekula languages: the historical articulatory environments, the particular make-up of these languages' consonant inventories, complementary sound changes and phonological processes, contact with non-Austronesian languages, and in-group identity attachment. Furthermore, we offer a hypothesis for the relative timing of these factors on the historical pathway of Malekula's bilabial trills.
Full-text available
Relationships between universally recognized language families represent a hotly debated topic in historical linguistics, and the same is true for correlation between signals of genetic and linguistic relatedness. We developed a weighted permutation test and applied it on basic vocabularies for 31 pairs of languages and reconstructed proto-languages to show that three groups of circumpolar language families in the Northern Hemisphere show evidence of relationship though borrowing in the basic vocabulary or common descent: [Chukotko-Kamchatkan and Nivkh]; [Yukaghir and Samoyedic]; [Yeniseian, Na-Dene, and Burushaski]. The former two pairs showed the most significant signals of language relationship, and the same pairs demonstrated parallel signals of genetic relationship implying common descent or substantial gene flows. For finding the genetic signals we used genome-wide genetic data for present-day groups and a bootstrapping model comparison approach for admixture graphs or, alternatively, haplotype sharing statistics. Our findings further support some hypotheses on long-distance language relationship put forward based on the linguistic methods but lacking universal acceptance. Significance statement Indigenous people inhabiting polar and sub-polar regions in the Northern Hemisphere speak diverse languages belonging to at least seven language families which are traditionally thought of as unrelated entities. We developed a weighted permutation test and applied it to basic vocabularies of a number of languages and reconstructed proto-languages to show that at least three groups of circumpolar language families show evidence of relationship though either borrowing in the basic vocabulary or common descent: Chukotko-Kamchatkan and Nivkh; Yukaghir and Samoyedic; Yeniseian, Na-Dene, and Burushaski. The former two pairs showed the most significant signals of language relationship, and the same pairs demonstrated parallel signals of genetic relationship implying common descent or substantial gene flows.
Full-text available
This specially commissioned volume considers the processes involved in language change and the issues of how they can be modelled and studied. The way languages change offers an insight into the nature of language itself, its internal organisation, and how it is acquired and used. Accordingly, the phenomenon of language change has been approached from a variety of perspectives by linguists of many different orientations. This book, originally published in 2003, brings together an international team of leading figures from different areas of linguistics to re-examine some of the central issues in this field and also to discuss new proposals. The volume is arranged into sections, including grammaticalisation, the typological perspective, the social context of language change and contact-based explanations. It seeks to cover the subject as a whole, bearing in mind its relevance for the general analysis of language, and will appeal to a broad international readership.
Mixed Languages are speech varieties that arise in bilingual settings, often as markers of ethnic separateness. They combine structures inherited from different parent languages, often resulting in odd and unique splits that present a challenge to theories of contact-induced change as well as genetic classification. This collection of articles is devoted to the theoretical and empirical controversies that surround the study of Mixed Languages. Issues include definitions and prototypes, similarities and differences to other contact languages such as pidgins and creoles, the role of codeswitching in the emergence of Mixed Languages, the role of deliberate and conscious mixing, the question of the existence of a Mixed Language continuum, and the position of Mixed Languages in general models of language change and contact-induced change in particular. An introductory chapter surveys the current study of Mixed Languages. © 2003 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin. All rights reserved.
Introduction: internal and external factors in change It is true, I think, that in what might be called the dominant tradition in historical linguistics, it has been assumed that languages change within themselves as part of their nature as languages. The ‘external’ agency of speaker/listeners and the influence of ‘society’ in language change have tended to be seen as secondary and, sometimes, as not relevant at all. Roger Lass has been a prominent, but balanced, defender of the traditional view. He has correctly pointed out (1980: 120) that in the tradition, it has been assumed that it is languages that change and not (necessarily) speakers who change languages. More recently (1990: 370), he has commented that language change is not something that speakers ‘do’ to their language, and that ‘endogenous change is part of the nature of the beast’ (1997: 208). He has also (largely correctly) suggested in various publications that speaker-based explanations have been unsatisfactory because of the attribution to speakers of qualities that they may not actually have. In some such accounts the speakers appealed to are disembodied abstractions who can be made to ‘do’ almost anything the researcher wants them to. Much more generally, however, the idea of endogenous or internally triggered change is so deeply embedded in our subject that it feeds into what can be called the discourse of historical linguistics. In this discourse, individual languages are typically presented as changing within themselves rather than being changed through the agency of speaker/listeners. © Cambridge University Press 2004 and Cambridge University Press, 2009.