Skin, Kin and Clan


Genesis of the Trinity: The Convergent Evolution of Trirelational Kinterms

Joe Blythe


While ordinary kinterms encode kinship relations between pairs of individuals, trirelational kinterms are semantically dense expressions that encode kinship relations between three individuals. Several times, these terms have emerged independently on the Australian continent. This emergence is explained as a convergent evolutionary process driven by interactional preferences that shape the design and use of person reference items in conversation. The case in point is the pragmatically motivated lexicalisation of trirelational kinterms in Murrinhpatha.

Trirelational kinterms, also known as triangular (Evans, Johnson & Kohler 1992; Garde 2002; Heath 1982), triadic (Alpher 1991; Garde 2013, 2014), ternary (Green 1998; McGregor 1996) and shared (McConvell 1982; O’Grady & Mooney 1973) kinterms, are typologically unusual among the world’s languages. These complex items are not unique to Australia and have also been attested in the Brazilian Amazon (Lea 2004) and Patagonia (Evans, Golluscio & Mellico 2010). However, why they have flourished so prolifically in Australian languages has been somewhat of an enigma. While occasionally, related forms point to shared inheritance and/or lexical diffusion in a handful of cases—for example, Pintupi, Warlpiri and Gurindji (McConvell 1982, p. 100; 1991) and western Arrernte, Alyawarr and eastern Anmatyerr (Green 1998, pp. 41–5)—the scattered distribution of these terms across different language families suggests multiple independent innovations.

In biology, convergent evolution is the process whereby similar ecological pressures yield similar adaptations in lineages that are unrelated or distantly related. The resultant organisms share similar morphological or behavioural adaptations that suit the ecological conditions they inhabit, despite having potentially different sources. As a result of parallel selective pressures driving convergent structuration within the language domain, I here assume an overarching theory of generalised evolution that subsumes biological evolution, cultural evolution, evolution of concepts and evolution of language (e.g. Croft 2000; Hull 1990; Levinson 2006).

In this chapter, I further an argument presented in Blythe (2013) that many of the unusual kin-based lexical and morphological phenomena identified in Australian languages have emerged as a result of roughly analogous cultural practices and preferences guiding the selection of person reference items in face-to-face conversation. Consequently, approximately equivalent structures have emerged through evolutionary convergence. Australia is a continent characterised by extensive classificatory kinship and widespread taboos that impose limits on the use of personal names. Classificatory kinship and naming taboos jointly exert selective pressures that have seen a variety of highly specialised kin-based referential expressions emerge in languages that are only distantly related. Blythe (2013) presented a case study in pragmatically motivated grammaticalisation of kin-based morphosyntax in Murrinhpatha (kin-based pronouns), as evidenced by diachronic reanalysis and interactional linguistics. The implication of that study is that analogous constraints on language use are behind the convergent evolution of similar kin-based pronoun paradigms across the Australian continent. In this chapter, I argue that the lexicalisation of trirelational kinterms in Australian languages is also driven by analogous constraints on reference, and facilitated by the existence of similar all-encompassing kin-based frameworks that provide circuitous pathways by which interlocutors can comply with those constraints. The second Murrinhpatha case study demonstrates one route by which circuitous reference formulations can become lexically frozen as trirelational structures.1

In describing the lexicalisation process, I reappraise early ethnographic reports into Murrinhpatha kinship to determine if previously attested kinship terminology has persisted into the twenty-first century; and if so, what can be gleaned about its appearance or non-appearance in a corpus of informal Murrinhpatha conversation (Blythe, n.d.). The larger corpus includes more than 60 hours of unprompted face-to-face conversations conducted by male and female speakers of all ages. Of this, four hours have been transcribed and annotated thus far. Most recordings are high-definition video filmed with a wide-angle lens. I set up the recordings and then extracted myself from the scene as the conversations commenced. Some parts of transcripts are included in this chapter and Appendix 4 provides a guide to transcription conventions.

A corpus of spontaneous conversation is invaluable in illuminating the interactional pragmatics of kinship terminology, such as the differential epistemic leveraging of ego versus altercentric kinterm reckoning (Blythe 2010). Additionally, the unsolicited explication of genealogical relations in informal conversation provides an authentic Indigenous metalanguage of kinship, rather than the sorts of purpose-driven metalanguage that emerge within elicitation contexts. Further, when kinterms alleged from earlier sources fail to surface in conversational corpora, an investigation can be instigated into whether the use of the alleged forms has waned, or whether the forms were provided so as to satisfy a particular line of ethnographic questioning. Despite these benefits, complete reliance on conversational corpora for information on low-frequency items is impractical. Studying conversation is an extremely useful addition to the fieldworker’s toolkit and augments conventional ethnographic methods (including elicitation), but it does not replace them.2

All kinterms are deictic expressions that express a relationship between individuals or groups of individuals. Regular kinterms, or ‘binary’ kinterms (McGregor 1996, 2012), are two-place predicates. They express the relationship between a referent (the person being spoken about) and a propositus or anchor (the person[s] to whom the referent is being related). Thus, in the expression ‘your mother’, the kinterm ‘mother’ is grammatically anchored to the addressee by the possessive pronoun ‘your’. Essentially, ‘mother’, in this instance, has an overt second-person propositus. Kinterms can also be covertly anchored. Normally, a covert propositus is pragmatically recoverable through conventionalised connotation. Thus, if a man speaking to his wife uses the term ‘mum’, his wife will probably infer the term as being used for reference to his own mother, and not hers, and that a first-person propositus is being implied. If he then uses the same term ‘mum’ when addressing his child, those present will infer that the term is probably being used for reference to his wife, the child’s mother, and that a second-person propositus is being implied. If he then uses the term ‘mum’ when addressing his brother, the brother will probably infer a covert first-person inclusive propositus (i.e. our mother).

Trirelational kinterms are semantically dense referential items that express relationships between three individuals (i.e. they are three-place predicates). The Murrinhpatha trirelational term yilamarna expresses the relationship between a man and his brother. Unlike the ordinary brother term (ngathan), yilamarna also expresses the relationships between the man and his child (wakal) and between the child and the brother (yile, see Figure 53). The relationships between all three individuals are encoded. As most trirelational terms are anchored in two places, they effectively have two propositi. Thus, if the triad encompasses the speaker, the addressee and the referent, then we have the speaker as propositus and the addressee as propositus. Figure 54 compares a regular Bininj Gunwok term to an approximately equivalent trirelational term. The regular kinterm nakurrng (Figure 54, left) is overtly anchored to the addressee with the possessive pronoun ke, meaning ‘your MoMoBrSo’ (Garde 2002, p. 157). If the individual referred to previously is the nephew of a male speaker, the trirelational term ke nakurrng can also be used (Figure 54, right). Thus, the term would mean the person who is your MoMoBrSo and my ZiSo, given that you are my DaCh (Garde 2002, p. 422). In this case, the relationship of the referent to the speaker (kangkinj, mZiSo) and of the addressee to the speaker (mamamh, DaCh) are inferable because the now-fronted free pronoun ke appears in a more prominent position. Trirelational terms are more specific than their regular counterparts because they have more restrictive denotata. They can be used effectively for reference to only a subset of the individuals that could potentially be referred to with the regular kinterms.

Figure 53: The Murrinhpatha trirelational kinterm yilamarna.

Source: Author’s work.

Figure 54: An ordinary Bininj Gunwok kinterm anchored to the addressee (left) and a trirelational kinterm anchored to both speaker and addressee (right). The term also encodes the relationship between speaker and addressee.

Source: Author’s work.

Within the literature, there is considerable variation in how researchers describe the mapping of the participant roles pertaining to speech events onto the triad of individuals semantically implicated by trirelational kinterms. This is partly due to usage conventions specific to the language in question and partly due to structural variation within the semantics of the terms. A number of authors have described trirelational kinterms as encoding relationships between speaker, addressee and a(n) (external, third person) referent. Thus, McGregor (1996, p. 219) described Gooniyandi’s ‘ternary monadic’ terms as having an Ego (always the speaker), a Propositus (usually the addressee) and a Referent. Merlan (1989) described Jawoyn’s Yenderr terms similarly, but captured this configuration using the term Speaker, rather than Ego. The Murrinhpatha terms do not map as consistently onto participant roles. When used for reference to a third person, they are normally anchored to the speaker and are further anchored externally—not to the addressee. They can also be used as vocatives, whereby the referent is the addressee. They are then anchored both to the speaker and externally.3

The cross-linguistic similarity emerges when the triad of expressed individuals is decoupled from participant roles. Trirelational terms can then be grouped into two basic types. The most common type is that with two propositi and a single referent (see Figure 55)—terms are used to speak about a particular individual, but in a way that explicates (or implicates) the genealogical connection to two other individuals, who may or may not be participants in the actual speech event. The Murrinhpatha terms are of this type.

Figure 55: Dual propositus trirelational kinterms.

Source: Author’s work.

Less common are the trirelational variants of dyadic kinterms.4 These are terms used for reference to a pair of related individuals that are anchored in various ways to a single propositus (as in Figure 56). In Gooniyandi, there are five contrastive terms used for husband-and-wife dyads. Each dyad relates to the speaker in different ways. For example, the term marralangi denotes a husband-and-wife pair, one of whom is the speaker’s opposite-sex sibling or cross-cousin (McGregor 1996, p. 228). The term woordoolangi denotes a husband-and-wife pair, one of whom is the speaker’s same-sex sibling or cross-cousin (McGregor 1996, p. 228). Similar trirelational dyads exist in Banyjima (Dench 1980), Nyangumarta (O’Grady & Mooney 1973), Gurindji (McConvell 1982) and the Mapundungun language of Patagonia (Evans, Golluscio & Mellico 2010).

Figure 56: Trirelational dyadic terms.

Source: Author’s work.

In some respects, this demarcation of types is something of a hair-splitting exercise, because, as McGregor (1996, p. 226) pointed out, dyadic kinterms have an inherent propositus that so happens to be one of the referents. In which case, whether the triad being expressed is conveyed with two propositi or two referents is really a question of the prominence being given to the individuals captured within the triad—thus, it has more to do with the pragmatics of tokens than the semantics of types. In which case, if it is predominantly an individual being spoken about, then the first model would better apply. If a pair is being spoken about, then the second model would better apply.

In the following sections, I examine the semantics of the Murrinhpatha trirelational kinterms and their usage. Strangely, the terms are not necessarily used for a triad of genuine individuals. The additional semantic resources of these terms allow reference to a ‘nameless’ person to be triangulated through a third (perhaps imaginary) individual, which means dealing with a structural ambiguity within the larger Murrinhpatha kinship system. Four of the trirelational kinterms provide structural solutions to pragmatic problems in that they plug functional holes in the array of regular kinterms. In turn, this has allowed another four structurally similar terms to enter the larger kinship lexicon, even though these are not required to fill similar pragmatic holes. The fact that Murrinhpatha forms are only used for kin for whom there are some form of name avoidance is evidence pointing to their evolutionary history.

In the concluding section, I survey the range and usage of trirelational kinterms in other Australian languages and find evidence to support a theory of convergent evolutionary origins springing from person referencing performed within contexts of personal name avoidance. Being lexically compact and semantically precise forms that are not names, trirelational kinterms have evolved to satisfy preferences for minimality, recognisability and circumspection about cultural protocols. As similar interactional conditions apply across the continent, recurrent structures emerge that satisfy what amount to analogous design constraints. The explanation proposed for these multiple innovations is well known in evolutionary biology, but has seldom been invoked in diachronic linguistics.

Trirelational Kinterms in Murrinhpatha

Murrinhpatha is a polysynthetic head-marking language spoken in the coastal region bounded by the Moyle and Fitzmaurice rivers, predominantly in Wadeye and surrounding communities. Prior to the establishment of a Catholic mission in 1935 on Murrinhpatha land, the region had no permanent European population. As the mission became better established, speakers of Marri Tjevin, Marri Amu, Magati Ke, Marri Ngarr and Jaminjung took up residence and began using Murrinhpatha for daily communication. The use of these neighbouring languages has waned drastically, while Murrinhpatha has emerged as the regional lingua franca.

In 1935, the anthropologist W. E. H. Stanner travelled with a group of missionaries to Murrinhpatha country to establish the first mission in the Moyle and Fitzmaurice rivers’ region. Four years later, the mission was relocated to Port Keats, which is now the community of Wadeye. Although he did not fully grasp their semantic complexity, Stanner (1937, pp. 314–15) remarked upon a group of morphologically complex ‘circumlocutory’ terms that ‘make references to or about a person even more indirect by tracing the relationship through an earlier generation’. These circumlocutory terms are only used for kin for whom there is some degree of name avoidance. The Murrinhpatha observe strong name avoidance between actual sons- and mothers-in-law, poison cousins (MoMoBrCh/MoBrDaCh/FaZiDaCh) and opposite-sex siblings. Between same-sex siblings, names can be used for third-person reference, but are seldom used for address—instead nicknames such as tepala (‘deaf one’) are greatly preferred. Names of recently deceased persons are avoided by the entire community, while names of the distantly deceased are avoided by close relatives for considerably longer and sometimes indefinitely.

The trirelational term that a man uses for reference to his WiMo or MoMoBrDa is kawumamnge. This term is transparently composed of a kinterm procliticised to a verb (see Box 1). All Murrinhpatha trirelational terms are composed in this fashion—at least historically.

Box 1: Transcript—incipient trirelational term for ‘wife’s mother’ formed with ‘say’.


kawu=mam                -nge


“’kawu’, he/she says/said to her”

Source: Author’s work.

Kawu is a grandparent term (MoMo), whereas the verb mamnge is normally used to report prior speech directed to a female addressee: ‘he/she says/said [it] to her’. In explicating the term, a female consultant stated that if her son-in-law was to refer to her using the term Kawumamngeka, mamka kardu wakalwa, wakal ngarra nukunuya, ‘[in saying] “kawumamnge”, the child talks, his own child’. The term literally means ‘the female person that he/she calls MoMo’, whereby ‘he/she’ should be understood as the man’s wakal (So/Da). Figure 57 shows that the relationship between the child and the referent (MoMo/wDaCh) is overtly expressed. The man’s child as the person addressing the referent is covertly expressed (inferable). Since two sides of the triangle are clearly understood, the third relationship (WiMo/MoMoBrDa) can also be inferred.

Figure 57: Kawumamnge—(literally) ‘the female person that he/she (ego’s child) refers to as MoMo’.

Source: Author’s work.

In the scenario depicted in Figure 57, the mother-in-law is the referent (R), the speaker is a propositus (P1), and the son/daughter is both the propositus for the embedded term kawu and a second propositus (P2) for the trirelational term. In Murrinhpatha, eight trirelational terms have been attested—all of which are used in situations in which some form of name avoidance is appropriate. The terms are formulaic (see Figure 58) in that all contain an embedded kinterm for whom no avoidance would be expected (i.e. the relationship between P2 and R is ‘in the clear’; see Figure 58). In each of these terms, the relationship between P1 and P2 is invariant as ‘son/daughter’.5 As P2 is always the direct progeny of P1, only the avoidance relationship (P1–R) needs to be inferred. Where the Murrinhpatha trirelational terms perhaps differ from other systems is that an actual son or daughter needn’t exist. Essentially, the terms make an indirect reference to R by triangulating through P2, which is possibly an imaginary descendent of P1.

Figure 58: The Murrinhpatha trirelational terms all contain an embedded kinterm that is presented as if being uttered by P2, the son/daughter of P1.

Source: Author’s work.

Although less intense than the avoidance between a man and his mother-in-law, a woman also avoids the name of her husband’s mother. The trirelational term that a woman uses for her HuMo (or MoMoBrDa) is mangkamamnge—literally, ‘he/she calls her mangka’ (see Box 2). Pragmatically, the term can be understood as ‘the woman that my child calls mangka (FaMo), who I shouldn’t mention by name on account of her being my HuMo (or MoMoBrDa)’.

Box 2: Transcript—incipient trirelational term for ‘husband’s mother’ formed with ‘say’.


mangka=mam                -nge

FaMo =3SG.SB.8say/do.NFUT-3SG.F.IO

“He/she calls her mangka”

Source: Author’s work.

Table 65 lists the eight attested Murrinhpatha trirelational kinterms.6 They are all lexicalised clauses that include an embedded kinterm. There are four terms for spouse’s parents and four terms for siblings, depending on the gender of P1 and the gender of the referent. The terms used for reference to a female all contain the framing speech verb mamnge. As third singular subjects of verbs are unmarked for gender, it does not matter whether the imagined child is the son or daughter. Less morphologically transparent, the terms used for reference to males contain the cranberry element = marna. The expected non-future masculine direct object counterpart to the speech verb mamnge would be mamna ‘he/she said/says to him’.7 As the trirelational terms used for male referents are semantically analogous to those used for females, I presume marna to be derived historically from *mam­­-rna, the trirelational term used by men for reference to a brother (Stanner 1937, p. 314), as exemplified by yilamarna in Box 3. As none of the described morphophonemic processes in modern Murrinhpatha prohibit the nasal cluster /mn/ (Street 1987; Walsh 1976), I presume these terms to have eroded prior to the modern morphophonology.8

Table 65: The eight attested Murrinhpatha trirelational terms.

Trirelational term



P1→ R

P2→ R


pipi (WiMo, MoMoBrDa)

kawu (MoMo)


pipi (HuMo, MoMoBrDa)

mangka (FaMo)


kaka (HuFa)

Kangkurl (FaFa)


kaka (WiFa)

thamuny (MoFa)


munak (Zi)

kale (Mo, MoZi)


munak (Zi)

pipi (FaZi)


ngathan (Br)

kaka (MoBr)


ngathan (Br)

yile (Fa, FaBr)

Source: Author’s work.

Box 3: Transcript—incipient trirelational terms for ‘sibling’ formed with ‘say’.

yilamarna < *yilemamna              (hypothesised surface form)

             yile=mam        -rna (hypothesised underlying form)

             Fa =3SG.SB.8say/do.NFUT-3SG.M.IO

“He/she calls him yile (Fa)”

Source: Author’s work.

In the next section, we investigate the place of trirelational terms within the larger collective of kinterms and how, from a pragmatic point of view, four of the terms plug functional holes in the paradigm of ordinary kinterms.

The Place of Trirelational Terms within the Larger Set of Kinterms

After only a few weeks of fieldwork in 1935, Stanner (1936) published the field report Murinbata Kinship and Totemism. In this report, he claimed the Murrinhpatha were in the process of transforming their kinship system from a simple Kariera type, with two lines of patrilineal descent, to a more complex Aranda type, with four lines of patrilineal descent (Berndt & Berndt 1999; Elkin 1968; Radcliffe-Brown 1930). He thought that this transformation was being driven by the Murrinhpatha’s enthusiastic adoption of the Jaminjung subsections in the first decades of the twentieth century. Norwegian ethnographers Johannes and Aslaug Falkenberg, who had conducted six months of fieldwork at Port Keats in 1950, concurred with Stanner, and suggested that the process of transformation had advanced during the intervening 15 years (Falkenberg 1962, p. 206; Falkenberg & Falkenberg 1981, p. 142).

The passing decades revealed Stanner and the Falkenbergs to have been premature in drawing these conclusions. There is no linguistic evidence to support the borrowing of kinterms from Jaminjung.9 Although there is no doubt that Murrinhpatha men, who ventured into the Victoria River district on account of the pastoral industry, borrowed the Jaminjung subsection terminology, the sociocentric system that each described (Falkenberg 1962, pp. 225–31; Stanner 1936, pp. 211–2) was a (four) section system with two (equivalent) names for each section. In the twenty-first century, this system is but a distant memory. For these ethnographers influenced by Radcliffe-Brown’s structural-functional conceptual frame (Hinkson 2005), Murrinhpatha’s ‘dual augmented’ kinship system (Keen 2013) appeared to be in flux. However, in the last 40 years, the kinship system has proven to be relatively stable, despite enormous population growth (Taylor 2010) and dramatic changes to many other aspects of social organisation (Furlan 2005; Ivory 2009; Mansfield 2013).

Using languages sampled from the AustKin database, Keen (2013) provided a modern classification of Australian kinship typologies that resembled the structuralist-functionalist models (Elkin 1938, 1939; Radcliffe-Brown 1930) because they were also based on lines of descent. Keen’s ‘dual’ and ‘quadruple’ terminologies approximately subsume the Kariera and Aranda models respectively. He classified Murrinhpatha’s terminology as ‘dual augmented with separate cross-cousin terms’ (Keen 2013, p. 15). However, he also noted that dual augmented was perhaps ‘not a unitary type by a cover term for several distinct variants of Dual terminologies’ (p. 28) that typologically ‘move[d] closer to the Quadruple terminologies’ (p. 18). Thus, the Murrinhpatha system is structurally intermediate between the earlier Kariera and Aranda types, but evidently not undergoing radical typological transformation.

Murrinhpatha’s system of ordinary kinterms show a straightforward sibling merger for two of the four grandparent terms: kangkurl (FaFa = FaFaBr = FaFaZi) and kawu (MoMo = MoMoBr = MoMoZi). Thamuny, the term for MoFa, is merged not only for siblings (MoFa = MoFaBr = MoFaZi) but also with FaMoBr. A distinct term mangka exists for FaMo (showing merger only with FaMoZi) (see Figures 58 and 59). In ego’s generation, cross-cousins pugarli (MoBrCh and FaZiCh) are distinguished from parallel cousins (FaBrCh and MoZiCh), which are considered equivalent to brothers (ngathan) and sisters (munak). Preferred marriage is to a matrilateral second cross-cousin (MoMoBrDaCh): purrima (♀) or nangkun (♂). Purrima is also a woman’s husband and a man’s sister-in-law, while nangkun is also a man’s wife and a woman’s brother-in-law.

Figure 59: The Murrinhpatha kinchart for a male ego (trirelational kinterms are not included).

Source: Author’s work.

The –1 generation is a typical Hawaiian pattern in distinguishing only males (muluk) from females (newuy). However, normally, all kin of this generation are referred to simply as wakal (literally, ‘small’), without distinguishing gender. As avoided affines, children of a female pugarli (MoBrDaCh and FaZiDaCh) can also be referred to by the term nginarr.

Figure 60: The Murrinhpatha kinchart for a female ego (trirelational kinterms are not included).

Source: Author’s work.

The +1 generation makes terminological distinctions for gender and for ego’s versus alter’s patrimoiety. The four terms are yile (Fa, MoMoBrSo and MoFaZiSo), kale (Mo, FaMoBrDa and FaFaZiDa), kaka (MoBr, FaMoBrSo, FaFaZiSo and WiFa) and pipi (FaZi, MoMoBrDa and MoFaZiDa). Although terminologically equivalent, the affinal kin in either patrimoiety are by no means socially equivalent to the consanguineal kin. Stanner and the Falkenbergs thought the WiMo (pipi) could be was distinguished from FaZi (also pipi) using the phrasal expression pipi nginarr (Falkenberg 1962; Falkenberg & Falkenberg 1981; Stanner 1936, p. 199). Although Stanner was correct in realising that the term nginarr was also used for WiMo (Stanner 1936, p. 199), he did not appreciate the term’s broader denotation of G±1 affine (male or female) in the MoMoBr’s patriline (see Figures 58 and 59). The Falkenbergs presented WiMo/MoMoBrDa (as pipi nginarr, or bip:i ŋinar in their orthography) as being distinguished from FaZi (as pipi ngutjngen, or bip:i ŋoitnan in their orthography). Although the phrasal expressions pipi nginarr and pipi ngutjngen are grammatically acceptable in Murrinhpatha, a corpus-based examination failed to attest them. Rather, WiMo is referenced regularly either as pipi or as nginarr, but the combination has never been attested. The adjective ngutjngen means ‘ordinary’ and, in certain contexts, might contrast with a pipi in the MoMoBr’s patriline. Thus, these phrasal expressions appear to be ad hoc descriptions of two functionally different types of ‘aunt’, probably produced under conditions of elicitation for explicating their social non-equivalence.10

These early ethnographers were right to expect that MoBrDa/WiMo should be terminologically distinguishable from FaZi, because the highly avoided affines demand very different kinship behaviour from that of a consanguineal ‘aunt’. Although the conversational corpus revealed this ambiguity to be seldom problematic, it is reasonable to assume that an inability to ever make this distinction would indeed be problematic. However, the ad hoc solution provided for their benefit (no doubt produced by Murrinhpatha speakers when speaking predominantly in English) differs from the Indigenous solution—trirelational terms. Kawumamnge and mangkamamnge provide a means for specifying a pipi as being of the avoided nginarr variety (SpMo and MoMoBrDa) and not of the ‘ordinary’ variety (FaZi). In one’s spouse’s patrimoiety, kangkurlmarna and thamunymarna disambiguate the term kaka, effectively specifying a mildly avoidable SpFa, as opposed to a consanguineal MoBr. As such, these terms plug functional gaps in the paradigm of ordinary kinterms, giving the overall system the power to make functional distinctions when necessary.

A Structural Explanation for the Emergence of Trirelational Kinterms

Corpora of natural conversation provide valuable insights into how knowing individuals come to explain kin relationships to less informed individuals. If the person being spoken about is not well known to all present, then explaining that individual’s place within a kinship network is more informative than merely providing the person’s name. The most common device used by Murrinhpatha speakers to explain relationships is to combine a kinterm with a semantically general class 8 ‘say/do’ verb,11 as exemplified in Extract 1. In this extract, Mick, Rob and Dave talk about a woman named Janet who, until then, has only been mentioned by a nickname.12

Extract 1: Ngandimeli (20120715_JB_video_GYHM100_02).

Source: Author’s work.

Prior to this extract, Dave has enquired as to her identity. Mick provides her real name at line 1, which Rob confirms as being correct at line 3. At line 7, Rob explains to his classificatory brother Dave (addressing him with the nickname murnu, ‘bone’) how Janet relates to them. He does this with the kinterm kaka (MoBr/WiFa) plus the class 8 ‘say/do’ verb mamnyewurran: ‘[She] says “kaka” to you and me’ (effectively, ‘she calls us “uncle”’). In doing so, one of the persons in the know associates the referent to himself and his interlocutor, and thus grounds his epistemic authority within a framework of kinship relations (Blythe 2010).

In this extract, and in many others besides, the ‘saying’ verbs used to explicate these relations are the same class 8 verbs from which the trirelational kinterms are composed. In the trirelational terms, these verbs, replete with bound indirect object pronouns, plus accompanying kinterms have become lexicalised as nouns. As we will see below, they are nouns imbued with a capacity to explicate thorny kinship relations.

In many dialects of Dutch, the term neef is ambiguous because it denotes both male cousins (PaSbSo) and nephews (SbSo). Likewise, nicht denotes both female cousins (PaSbDa) and nieces (SbDa). In some dialects of Dutch, the morphologically complex term Oom = zegger and ‘“uncle” = sayer’ specifies a nephew (or a niece), as distinct from a cousin. Oomzegger and related tante = zegger and ‘“aunt” = sayer’ are comprised of embedded kinterms plus the explanatory ‘saying’ expression zegger (which is derivable from the speech verb zeggen ‘to say’).13 The embedded terms focus on the phrasal terms’ propositus (the uncle or aunt), thus disambiguating them from cousins (essentially because the ‘cousin’ sense of neef [or nicht] is a reciprocal relationship). These complex kinterms have a similar morphological structure to the Murrinhpatha terms, but, not being trirelational terms, are less semantically complex. Nevertheless, similar to the Murrinhpatha terms, they also deal with a structural ambiguity within the larger set of kinterms.

In the Murrinhpatha case, the specification problem is essentially that in G+1, within each patrimoiety, affines are not distinguished from consanguineal kin. This Kariera-type patterning fails to replicate descent lines that are (more or less) distinguished in G+2 by the somewhat lopsided Aranda-esque grandparent terminology (see Figures 59 and 60).14 If Murrinhpatha speakers were to disambiguate an avoided nginarr type pipi from a consanguineal pipi (FaZi) by exactly employing the Dutch strategy, they would have to shoot for an expression meaning ‘“son-in-law”-sayer’ or similar. The problem here is that although a woman’s son-in-law can be referred to as muluk or wakal, the Hawaiian patterning in G–1 will not locate the son-in-law accurately within the MoMoBr’s patriline. Nginarr does precisely this but does not specify the ‘sayer’ as male, nor specify him as G–1, as opposed to G+1. By taking the perspective of ego’s son or daughter, the second propositus P2, for practical purposes, is a downward-skewed version of P1. If the propositus is skewed down a generation, the referent is effectively pushed up a generation into that part of the kinchart in which MoMoBr’s patriline is distinguished terminologically from FaFa’s patriline. Kawumamnge is able to specify WiMo/MoMoBrDa because kawu specifies both patriline and generation, whereas wakal, muluk and nginarr only specify either generation or patriline, but not both.

By taking the perspective of her children, a woman is able to use mangkamamnge to specify her HuMo/MoMoBrDa, as opposed to her FaZi. This becomes possible because P2 is effectively a skewed-down version of P1. By skewing down a generation, the referent is pushed up into G+2 as mangka (FaMo), which is the most distinctive of the grandparent terms. By the same mechanism, the two trirelational terms for fathers-in-law specify an affinal kaka (HuFa/FaMoBrSo), as opposed to a consanguineal kaka (MoBr). By skewing down the propositus as P2, the term with a male P1 (thamunymarna) targets a FaMo (thamuny), whereas the term with a female P1 (kangkurlmarna) targets a FaFa (kangkurl).

The four trirelational terms for affines—kawumamnge, mangkamamnge, kangkurlmarna and thamunymarna (see Table 65)—solve the specification problems encountered by Stanner and the Falkenbergs. Although the phrasal expression pipi nginarr might be a perfectly acceptable way to distinguish WiMo/MoMoBrDa from FaZi, it appears not to be the actual solution. The Indigenous solution is to take the stock-standard kin-explaining expressions and use them in a way that takes the perspective of kinsmen (ego’s children) who do not have problems specifying the same referents. The trirelational sibling terms have no such specification issues to resolve. Kakamarna, the special avoidance term used by women for reference to their brothers is, in practical terms, no more precise than the regular brother term ngathan.15 Most likely, these sibling terms piggybacked on the affinal terms in also skewing down a second propositus (P2). As all sibling names are avoided to some degree (at least in contexts of address), the addition of these trirelational terms to the regular sibling terms enriches the range of options for referring to (or addressing) siblings without resorting to their names. In this way, the sibling terms have sneaked through an avoidance window left ajar by the affinal terms.

The Utility-Driven Emergence of a Specialised Class of Words

Most Aboriginal languages of Australia are critically endangered or no longer spoken on a daily basis. The fact that trirelational kinterms are predominantly found in Australian languages makes them a highly endangered class of words. Findings show that semantically complex kinterms are acquired later than simpler terms (Haviland & Clark 1974),16 leading to the prediction that highly complex lexical items that are difficult to acquire might be among the first items to fall into disuse, as minority languages become threatened by dominant languages. Research into the use of these words in naturalistic contexts is urgently required, particularly because their pragmatics can inform their diachronic development.

The Murrinhpatha trirelational kinterms have emerged as a result of usage-based constraints on person reference items. These constraints take the form of conversational preferences for 1) not using personal names under conditions of taboo (Blythe 2013; Garde 2008; Levinson 2007); 2) using recognitionals (reference forms that invite interlocutors to recognise who is being spoken about) (Sacks & Schegloff 1979; Schegloff, 1996, 2007); and 3) using forms that are not more verbose than necessary (Enfield 2013; Hacohen & Schegloff 2006; Levinson 1987). Trirelational terms satisfy these constraints by 1) not being names; 2) being highly specific and semantically dense reference forms—which makes them useful as recognitionals; and 3) being extremely compact.

Yet, by no means are these terms the only available alternatives to names. Ordinary kinterms, nicknames, descriptions and even kin-based pronouns (Blythe 2009, 2013) are also used as devices for introducing new referents into conversation so that one’s interlocutor can recognise who is being spoken about, as well as personal names. As previously discussed, the trirelational terms used for reference to affines are semantically more precise than their ordinary (binary) counterparts; however, this is not true of the sibling terms. If the trirelational sibling terms exist alongside their ordinary counterparts (ngathan, ‘brother’, and munak, ‘sister’) as alternatives to personal names, but are less frequently used than the binary terms, we can surmise that their usage will be pragmatically marked. This is probably true for all trirelational terms, as alternatives to the alternatives for names. At the time of writing, only one trirelational term has surfaced in the annotated (five-hour) corpus of conversation (kalemamnge, the term used for a sister, by a sister). An interactional analysis of the extract containing this term is included as a supplement to this chapter. The extract supports the notion of pragmatically marked referential usage—forms used for doing something special, over and above simple name avoidance.

Languages such as Bininj Gunwok, Mawng and Gurindji have larger collections of trirelational terms than Murrinhpatha, whereas for certain others (Watjarri and Yidiny), only one or two terms are reported (Dixon & Irvine 1991, p. 151; Douglas 1981, p. 251). The more expansive systems are possibly older than Murrinhpatha’s.17 The map in Figure 61 shows trirelational kinterms that have been attested in Australian in 23 languages belonging to seven distinct language families. That they are scattered in the north and the centre is probably testament to these languages being more vital and better described than those of the south and east of the continent, which was where the impact of European invasion was felt earlier. Thus, trirelational kinterms may have been even more widespread than the map would suggest.

Figure 61: Australian languages for which trirelational kinterms have been attested.

Source: Author’s work.

Murrinhpatha trirelational kinterms are used in all circumstances when producing the referent’s name is inappropriate, which speaks to their evolutionary origins. A few researchers have specifically linked the use of these terms to avoidance contexts. For instance, McGregor (1996, p. 220) stated that the Gooniyandi terms ‘are restricted to circumstances in which at least one of the persons, usually the referent, is in a strong in-law avoidance relationship with ego’. Of Mawng’s Kunteypi trirelational terms, Singer et al. (in prep.) stated that ‘nowadays only some terms are used, mainly those that refer to people for whom respect or avoidance is necessary such as opposite-sex siblings, mother-in-law and poison-cousin’.18 For a number of languages,19 it is evident that at least some of the trirelational terms are used when there is an avoidance relationship between two people in the implicated triad, or when one person in the triad is deceased (e.g. Walmajarri, Richards & Hudson 1990, p. 129). Thus, for 15 of the 23 languages in Figure 61, information about their usage in avoidance contexts is consistent with the proposition that name avoidance lies behind their evolutionary emergence. The fact that the remaining languages are agnostic on this issue cannot be taken as evidence to the contrary. Either the researchers did not attest to usage in avoidance relationships because it was not their concern, or the vestigial documentary evidence on these systems did not inform a view of their historical pragmatics.20 The Amazonian data also suggest a correlation between some trirelational kinterms and constrained relationships.21

Conversely, the relatively transparent morphosyntax of the Murrinhpatha trirelational terms plus the view of their position within the larger array of kinterms, along with information about their usage, illuminate a historical picture of the circumstances preceding their development. When Murrinhpatha speakers had specification issues to resolve, they did nothing extraordinary. They dealt with the G+1 ambiguity by taking the most ordinary relationship-explaining clauses, and configured them so as to indirectly target the affine whose name should be avoided, by using kinterms that located these affines within descent lines that were terminologically distinguished at G+2. To achieve this, indirect reference was made from the perspective of Ego’s children. Although a brilliant technical solution, skewing the propositus down a generation was not a result of a teleological mechanism (Croft 2000, pp. 66–71). Rather, in order to have become lexicalised, it would have been the non-intended outcome of recurrent conversational practices (Blythe 2013; Keller 1994). Triangulating through Ego’s children was a logical extension of the altercentric anchoring of kinterms that was characteristic of child-directed speech.22 Thus, they might have originated as ad hoc descriptions of avoided kin produced in the company of Ego’s children (‘the one he/she calls X’), performed, in all likelihood, as multimodal utterances with accompanying points or glances towards the children in question. Through routinised use, these relationship-explaining clauses became lexically fused as nouns. In the case of the terms used for reference to males (whereby = marna < = mam-rna, 3SG.SB.say/do(8).NFUT-3SG.M.IO.), this was accompanied by loss of a segment, which could be construed as demorphologisation (Brinton & Traugott 2005, pp. 52–4). Although these male terms, morphologically, are only partially transparent, as lexical nouns they largely retain the explanatory semantics that have been previously attributed to the clausal domain, and can be parsed as such.23

Exactly how similar the mechanisms by which these semantic structures emerged in other Australian languages remains to be determined. Nevertheless, pan-continental classificatory kinship and practices of name avoidance suggest that the pragmatic motivations driving their development is similar. Conversationalists always need to make reference to others such that their recipients can recognise who they are talking about. Within Australian Aboriginal conversations, depending on who is present at the time, a reasonable proportion of individuals should not be mentioned by name—at least within certain contexts. There are many ways to bypass this problem. Many involve making circuitous reference to the ‘nameless’ individual by triangulating through another individual. Classificatory kinship systems provide a wide range of pathways along which these triangulations can be calculated. It can be done with a regular anchored kinterm; however, a pair of anchored kinterms will calculate the triangulation with greater precision.24 Trirelational kinterms will calculate the triangulation with as much precision as a pair of kinterms—although more compactly. Conversational preference structure provides min–max design constraints that minimise lexical bulk while maximising referential precision. In circumstances in which there is a call for referencing that recipients will recognise but where particular personal names are dispreferred because of taboos, there is a strong motivation for selecting precise-yet-snappy referential expressions. While this motivation is observable within interactional timescales in face-to-face conversation, the fruit of these motivating pressures can also be observed within evolutionary timescales.

Within evolutionary biology, morphological adaptations can enter a population when genomic mutations are replicated vertically or horizontally through virally induced gene transfer. Useful mutations spread throughout a population because they afford the organism selective advantage over members of the same population that lack the mutation. In convergent evolution, analogous ecological constraints drive the selection of similar morphological structures (phenotypes) in lineages that are either unrelated or distantly related. The resultant organisms share similar morphological adaptations that are purpose-built to suit the ecological conditions they inhabit.25 If a population within one of these lineages then becomes separated and begins to diverge, the previously convergent feature becomes a feature shared by sister organisms within a subgroup of the previous lineage. If related yet disparate populations come into contact with each other, genetic material can recombine within a single hybrid population—sometimes giving the impression of a terminated lineage. The evolutionary pictures of an organism’s prehistory are often complicated by the application of several evolutionary processes within the same lineage (King 2013).

As with biological evolution, the prehistoric picture of linguistic structures can be equally complicated. A convergent evolutionary account for the emergence of trirelational kinterms need not exclude shared inheritance, nor diffusion. These processes are clearly also implicated in the evolution of these semantic structures within the Australian context. Much work remains in unpacking how many of the related forms point to shared inheritance or borrowing—or both. However, the fact remains that the distribution of phenotypically similar trirelational structures within Australia is much more widespread than the distribution of trirelational terms displaying genetic relatedness. These distribution patterns clearly point to evolutionary convergence, as implicated in the prehistory of Australian kin-based linguistic structures. The ecological constraints driving the convergence take the form of interactional preferences pertaining to the selection of person reference items. The same sorts of structure emerge because they are recurrently good solutions to the constraining pressures that drive their emergence. Trirelational structures are only one type of kin-based phenomena to emerge in languages that tend to emanate unusual kin-based lexical and morphological structures. From an evolutionary standpoint, the flowering of specialist kin-based/sociocentric lexicon and morphosyntax is absolutely what should be expected within an interactional ecology characterised by expansive networks of social relations and constraints on whether individuals can be referred to by name.


Interactional uses of trirelational kinterms in conversation are hitherto undocumented. Extract 2 demonstrates the vocative use of a trirelational kinterm as an alternative to a name. Located within a generally humorous episode of conversation, the particular token is packaged as part of a turn designed to solicit laughter. In Extract 2, one of the conversationalists uses the term kalemamnge to address her classificatory sister, instead of a personal name, and instead of the ordinary ‘sister’ term munak. As such, the token is part of a lexically exaggerated tongue-in-cheek proposal that is understood to be non-serious.

Research within psychology, conversation analysis and interactional sociolinguistics on joking and teasing has revealed that conversationalists deploy a variety of playful off-record markers, so as to signal that the content of teasing turns should be interpreted non-literally. These markers include smiles and laughter particles, mock aggression, use of nicknames, marked pronoun usage and formulaic expressions, as well as prosodic exaggeration (e.g. amplitude and register shifts, vowel lengthening and singsong intonation) (Glenn 2003; Haugh 2010; Jefferson 1979; Keltner et al. 2001; Lytra 2007, 2010; Miller 1986; Straehle 1993). Relatedly, lexical selection can also be exaggerated (as with ‘extreme case formulations’: allalways, the most, the best, every, never) (Edwards 2000; Pomerantz 1986). In Extract 2, we see a variety of off-record markers employed to indicate that what is being proposed is not entirely serious.

In Figure 62, four women are sitting on a beach on the estate of the Yek Nangu Murrinhpatha patrilineal clan, watching the sun set into the sea. The sun is a totem of the Yek Nangu clan. Alice and Lily, classificatory sisters, are Yek Nangu clanswomen. Rita is Lily’s daughter. Rita and Karen are both Marri Ngarr women of the Rak Wakal Bengkuny and Rak Kungarlbarl clans respectively. Like Rita, Karen’s mother was a Yek Nangu Murrinhpatha woman, which is why both Karen and Rita are visiting their kangatji—their ‘mothers’ country’. Alice’s late father was a prolific composer of djanba songs, while Alice is a singer.

Figure 62: A video still corresponding to Lily’s line 46 in Extract 2—Bere, kalemamnge tepala murriny nartwardangu, ‘Right, deaf-one “sister”, take it away!’

Source: Author.

Extract 2: Nanthak (20110828_JB_video_GYHM100_03_673560_737630).

Source: Blythe video recording and transcription (2012).

At line 1, Alice softly sings a line of a song, then explains (at line 2) that it was the song she and some others had sung (the previous night at church). At line 4, Rita recognises the song as one she has on her mobile phone. In a noticeably loud utterance, Karen teasingly admonishes Rita (at line 6, while pointing at her, line 8) for being flippant about her.26 She then proposes (at lines 10 and 12) that Alice make up a song about Rita (a Rak Wakal Bengkuny ‘cycad’ woman) and herself (Karen, a Rak Kungalbarl ‘crocodile’ woman). Laughing at the suggestion, Alice refuses to comply (line 13). In an extended turn, Karen then embellishes her proposal by suggesting that the song should recount how she and Rita have been enjoying themselves in their respective mothers’ country (lines 14 and 17), while watching the sun set (lines 20 and 23). The humorous proposal is further elaborated over several lines, which, for brevity’s sake, have been removed.

Seemingly as content for this imaginary song, Karen then proposes (at lines 40 and 42) that Rita, as the Rak Wakal Bengkuny clanswoman, and she, as the Rak Kungarlbarl clanswoman, are being refreshed in their respective mothers’ country. Alice and Lily’s overlapped agreement tokens yukuy, ‘that’s right’ (lines 43 and 44), seem to endorse this material as worthy of committing to song. At line 46, Lily instructs her sister to ‘take it away’ (that is, to start singing). The command is issued in the imperative with a second-person singular ‘take’ verb. Recipiency for the command is issued with double-barrelled avoidance address terms, kalemamnge and tepala. Kalemamnge is the trirelational term used by a woman for a sister, ‘the female that “he/she” (my son/daughter) calls kale (Mo/MoZi)’.27 Sisters normally address each other (not by name) with the nickname tepala (‘deaf’ < deaf-fellow). Alice laughs at the command (at line 47) and does not commence singing. The overlapped turns at lines 48 and 49 are difficult to discern. At line 51, Karen suggests that she and Rita might cry, which also prompts laughter from Alice (line 53). Presumably, Karen is suggesting that preserving this delightful scene in song would be so emotive that tears would well up in their eyes.

A single vocative expression is normally sufficient to solicit the attention of a targeted recipient. However, in this instance, kalemamnge co-occurs (at line 46) with another dedicated avoidance address term tepala, ‘deaf-one’. As such, the use of two vocatives makes for a lexically exaggerated formulation of address that signals that the instruction to start singing should not be taken literally (as evidenced interactionally by the laughter and the non-compliance). Given that no song has been composed as yet, the instruction can only be interpreted non-literally.

This particular token surfaces when personal name avoidance is expected. Yet, when coupled with an additional name-avoidance vocative, the non-minimal vocative combination accomplishes something over and above regular name avoidance—namely, the solicitation of laughter. This lexical exaggeration (Edwards 2000; Pomerantz 1986) makes for pragmatically marked name avoidance that contributes to the generally jokey interactional frame.


Many thanks to Aung Si, Jenny Green, Elena Mujkic, Ruth Singer, Lesley Stirling and the editors, Piers Kelly and Patrick McConvell, for commenting on prior versions of this manuscript. The research has also benefited from feedback following presentations given at the Max Planck Institute for Psycholinguistics in 2012, at the 2011 Australian Languages Workshop on Stradbroke Island and the 2012 Australian Linguistic Society Conference in Perth. I am indebted to Phyllis Bunduck, Lucy Tcherna, Gertrude Nemarlak, the late Elizabeth Cumaiyi, Carmelita Perdjert and Ernest Perdjert for advice on Murrinhpatha language and kinship, and to Mark Crocombe at the Kanamkek Yile Ngala Museum and Languages Centre in Wadeye for logistical support. The research was funded by the European Research Council (StG Project 240853) and the Australian Research Council (DP110100961 and DE130100399).


Alpher, B 1991, Yir-Yoront lexicon: sketch and dictionary of an Australian language, Walter de Gruyter.

Bates, HW 1863, The naturalist on the River Amazons: a record of adventures, habits of animals, sketches of Brazilian and Indian life, and aspects of nature under the equator, during eleven years of travel. with a memoir of the author, by Edward Clodd, London: John Murray.

Berndt, RM & Berndt, CH 1999, The world of the First Australians: Aboriginal traditional life, past and present, Canberra: Aboriginal Studies Press.

Blake, BJ 1979, A Kalkatungu grammar, Canberra: Pacific Linguistics.

Blythe, J n.d., The language archive, viewed 9 August 2017,

Blythe, J 2009, Doing referring in Murriny Patha conversation, viewed 9 August 2017,

Blythe, J 2010, ‘Self-association in Murriny Patha talk-in-interaction’, Australian Journal of Linguistics, 30(4), pp. 447–469.

Blythe, J 2013, ‘Preference organization driving structuration: evidence from Australian Aboriginal interaction for pragmatically motivated grammaticalization’, Language, 89, pp. 883–919.

Blythe, J, Nordlinger, R & Reid, N 2007, Murriny Patha finite verb paradigms, Unpublished manuscript.

Brinton, LJ & Traugott, EC 2005, Lexicalization and language change, Cambridge University Press.

Butcher, A 2006, ‘Australian Aboriginal languages: consonant-salient phonologies and the “place-of-articulation imperative”’, in J Harrington & M Tabain (eds), Speech production: models, phonetic processes, and techniques, New York: Psychology Press, pp. 187–210.

Croft, W 2000, Explaining language change: an evolutionary approach, Harlow: Pearson Education.

Dench, A 1980, ‘Kin terms and pronouns of the Panyjima language of northwest Western Australia’, Anthropological Forum, 5(1), pp. 105–20.

Dixon, RMW & Irvine, T 1991, Words of our country: stories, place names and vocabulary in Yidiny, the Aboriginal language of the Cairns-Yarrabah region, St. Lucia, Qld: University of Queensland Press.

Donley, JM, Sepulveda, CA, Konstantinidis, P, Gemballa, S & Shadwick, RE 2004, ‘Convergent evolution in mechanical design of lamnid sharks and tunas’, Nature, 429, pp. 61–5.

Douglas, WH 1981, ‘Wadjarri’, in RMW Dixon & BJ Blake (eds), Handbook of Australian languages, volume 2, Canberra: Australian National University Press.

Edwards, D 2000, ‘Extreme case formulations: softeners, investment, and doing nonliteral’, Research on Language and Social Interaction, 33(4), pp. 347–73.

Elkin, AP 1938, ‘Kinship in South Australia’ Oceania, 8, pp. 419–52.

Elkin, AP 1939, ‘Kinship in South Australia (continued)’, Oceania, 10, pp. 196–234.

Elkin, AP 1968, The Australian Aborigines: how to understand them, Sydney: Angus & Robertson.

Enfield, NJ 2013, ‘Reference in conversation’, in J Sidnell & T Stivers (eds), The handbook of conversation analysis, John Wiley & Sons, Ltd, pp. 433–54.

Evans, N 2003, An interesting couple: the semantic development of dyad morphemes, Köln: Institut für Sprachwissenschaft der Universität zu Köln.

Evans, N & Garde, M 2013, The stance triangle and Kunderbi trirelational terms, presentation at the University of Queensland, 2012.

Evans, N, Golluscio, L & Mellico, F 2010, ‘La construcción diádica en mapudungun y sus implicancias tipológicas’, Llames, 10, pp. 49–66.

Evans, N, Johnson, P & Kohler, M 1992, Kayardild dictionary and thesaurus: a vocabulary of the language of the Bentinck Islanders, north-west Queensland, University of Melbourne, Department of Linguistics and Language Studies.

Falkenberg, J 1962, Kin and totem: group relations of Australian Aborigines in the Port Keats District, Oslo: Oslo University Press.

Falkenberg, A & Falkenberg, J 1981, The affinal relationship system: a new approach to kinship and marriage among the Australian Aborigines at Port Keats, Oslo: Universitetsforlaget.

Furlan, A 2005, Songs of continuity and change: the reproduction of Aboriginal culture through traditional and popular music, PhD thesis, University of Sydney.

Garde, M 2002, Social deixis in Bininj Kun-wok conversation, PhD thesis, University of Queensland.

Garde, M 2008, ‘Person reference, proper names and circumspection in Bininj Kunwok conversation’, in I Mushin & B Baker (eds), Discourse and grammar in Australian languages, Amsterdam: John Benjamins.

Garde, M 2013, Culture, interaction and person reference in an Australian Language, Amsterdam; Philadelphia: John Benjamins.

Garde, M 2014, ‘Shifting relations: structure and agency in the language of Bininj Gunwok kinship’, in RJ Pensalfini, M Turpin & D Guillemin (eds), Language description informed by theory, Amsterdam: John Benjamins, pp. 361–81.

Geytenbeek, J & Geytenbeek, H 1982, ‘Nyangumarta kinship: a woman’s viewpoint’, in S Hargrave (ed.), Language and culture work papers of the SIL-AAB, Darwin: Summer Institute of Linguistics, pp. 19–31.

Glasgow, K 1994, Burarra-Gun-nartpa dictionary with English finder list, Summer Institute of Linguistics, Australian Aborigines and Islanders Branch Darwin.

Gleiss, AC, Jorgensen, SJ, Liebsch, N, Sala, JE, Norman, B, Hays, GC et al. 2011, ‘Convergent evolution in locomotory patterns of flying and swimming animals’, Nature Communications, 2, p. 352.

Glenn, P 2003, Laughter in interaction, Cambridge: Cambridge University Press.

Green, J 1998, Kin and country: aspects of the use of kinterms in Arandic languages, Master’s research thesis, University of Melbourne.

Hacohen, G & Schegloff, EA 2006, ‘On the preference for minimization in referring to persons: evidence from Hebrew conversation’, Journal of Pragmatics, 38(8), pp. 1305–12.

Hansen, KC, Hansen, LE & Tjapaltjarri, T 1974, Pintupi kinship, Institute for Aboriginal Development.

Haugh, M 2010, ‘Jocular mockery, (dis)affiliation, and face’, Journal of Pragmatics, 42, pp. 2106–19.

Haviland, SE & Clark, EV 1974, ‘“This man’s father is my father’s son”: a study of the acquisition of English kin terms’, Journal of Child Language, 1, pp. 23–47.

Heath, J 1982, ‘Introduction’, in J Heath, F Merlan & A Rumsey (eds), Languages of kinship in Aboriginal Australia, Sydney: University of Sydney.

Hinkson, M 2005, ‘The intercultural challenge of Stanner’s first fieldwork’, Oceania, 75, pp. 195–208.

Hull, DL 1990, Science as a process: an evolutionary account of the social and conceptual development of science, University of Chicago Press.

Ivory, B 2009, Kunmanggur, legend and leadership: a study of Indigenous leadership and succession focussing on the northwest region of the Northern Territory of Australia, PhD thesis, Charles Darwin University.

Jefferson, G 1979, ‘A technique for inviting laughter and its subsequent acceptance-declination’, in G Psathas (ed.), Everyday language: studies in ethnomethodology, New York: Irvington, pp. 79–95.

Johnson, SD & Jürgens, A 2010, ‘Convergent evolution of carrion and faecal scent mimicry in fly-pollinated angiosperm flowers and a stinkhorn fungus’, South African Journal of Botany, 76(4), pp. 796–807.

Keen, I 2013, ‘The legacy of Radcliffe-Brown’s typology of Australian Aboriginal kinship systems’, Structure and Dynamics, 6(1), viewed 25 June 2013,

Keller, R 1994, On language change: the invisible hand in language, London: Routledge.

Keltner, D, Capps, L, Kring, AM, Young, RC & Heerey, AA 2001, ‘Just teasing: a conceptual analysis and empirical review’, Psychological Bulletin, 127, pp. 229–48.

King, C 2013, ‘The tangled tree of life: tangled roots and sexy shoots: tracing the genetic pathway from the last universal common ancestor to homo sapiens’, Genotype, 1.

Laughren, M 1982, ‘Warlpiri kinship structure’, in J Heath, F Merlan & A Rumsey (eds), Languages of kinship in Aboriginal Australia, Sydney: University of Sydney, pp. 72–85.

Lea, Vanessa. 2004, Aguçando o entendimento dos termos triádicos Mẽbêngôkre via os aborígenes australianos: dialogando com Merlan e outros. Liames: Línguas Indígenas Americanas 4, pp. 29–42.

Levinson, SC 1987, ‘Minimization and conversational inference’, in J Verschueren & M Bertuccelli-Papi (eds), The pragmatic perspective: selected papers from the 1985 International Pragmatics Conference, Amsterdam/Philadelphia: John Benjamins, pp. 61–129.

Levinson, SC 2006, ‘Introduction: the evolution of culture in a microcosm’, in SC Levinson & P Jaisson (eds), Evolution and culture: a Fyssen Foundation Symposium, Cambridge, Mass.: MIT Press, pp. 1–41.

Levinson, SC 2007, ‘Optimizing person reference—evidence from repair on Rossel Island’, in NJ Enfield & T Stivers (eds), Person reference in interaction: linguistic, cultural and social perspectives language culture and cognition, Cambridge: Cambridge University Press, pp. 29–72.

Liu, Y, Cotton, JA, Shen, B, Han, X, Rossiter, SJ & Zhang, S 2010, ‘Convergent sequence evolution between echolocating bats and dolphins’, Current Biology, 20, R53–R54.

Lytra, V 2007, ‘Teasing in contact encounters: frames, participant positions and responses’, Multilingua, 26, pp. 381–408.

Lytra, V 2010, ‘Constructing academic hierarchies: teasing and identity work among peers at school’, Pragmatics, 19, pp. 449–66.

Mansfield, J 2013, ‘The social organisation of Wadeye’s heavy metal mobs’, The Australian Journal of Anthropology, 24, pp. 148–65.

McConvell, P 1982, ‘Neutralisation and degrees of respect in Gurindji’, in J Heath, F Merlan & A Rumsey (eds), Languages of kinship in Aboriginal Australia, Oceanic linguistic monographs, Sydney: University of Sydney, pp. 86–106.

McConvell, P 1991, ‘Long lost relations: Pama-Nyungan and northern kinship’, in P McConvell & N Evans (eds), Archaeology and Linguistics, Oxford: Oxford University Press, pp. 207–35.

McGregor, W 1996, ‘Dyadic and polyadic kin terms in Gooniyandi’, Anthropological Linguistics, 38(2), pp. 216–47.

McGregor, W 2012, ‘Kin terms and context among the Gooniyandi’, Anthropological Linguistics, 54, pp. 161–86.

Merlan, F 1989, ‘Jawoyn relationship terms: interactional dimensions of Australian kin classification’, Anthropological Linguistics, 31, pp. 227–63.

Merlan, F & Heath, J 1982, ‘Dyadic kinship terms’, in J Heath, F Merlan & A Rumsey (eds), Languages of kinship in Aboriginal Australia, Sydney: University of Sydney, pp. 107–24.

Miller, P 1986, ‘Teasing as language socialization and verbal play in a white working-class community’, in B Schieffelin & E Ochs (eds), Language socialization across cultures, Cambridge: Cambridge University Press, pp. 199–212.

Nordlinger, R 2010, ‘Verbal morphology in Murrinh-Patha: evidence for templates’, Morphology, 20, pp. 321–41.

Nordlinger, R & Caudal, P 2012, ‘The tense, aspect and modality system in Murrinh-Patha’, Australian Journal of Linguistics, 32(1), pp. 32, 73–113.

Norton, HL, Kittles, RA, Parra, E, McKeigue, P, Mao, X, Cheng, K et al. 2007, ‘Genetic evidence for the convergent evolution of light skin in Europeans and east Asians’, Molecular Biology and Evolution, 24, pp. 710–22.

O’Grady, GN & Mooney, KA 1973, ‘Nyangumarda kinship terminology’, Anthropological Linguistics, 15, pp. 1–23.

Pomerantz, A 1986, ‘Extreme case formulations: a way of legitimizing claims’, Human Studies, 9, pp. 219–29.

Radcliffe-Brown, AR 1930, ‘The social organization of Australian tribes’, Oceania, 1, pp. 34–63, 206–46, 322–41, 426–56.

Richards, E & Hudson, J 1990, Walmajarri-English dictionary, Darwin: Summer Institute of Linguistics.

Sacks, H & Schegloff, EA 1979, ‘Two preferences in the organization of reference to persons in conversation and their interaction’, in G Psathas (ed.), Everyday language: studies in ethnomethodology, New York: Irvington, pp. 15–21.

Schegloff, EA 1996, ‘Some practices for referring to persons in talk-in-interaction’, in BA Fox (ed.), Studies in Anaphora typological studies in language, Amsterdam; Philadelphia: John Benjamins, pp. 437–85.

Schegloff, EA 2007, ‘Conveying who you are: the presentation of self, strictly speaking’, in NJ Enfield & T Stivers (eds), Person reference in interaction: linguistic, cultural and social perspectives, Cambridge: Cambridge University Press, pp. 123–48.

Singer, R, Garidjalalug, N, Hewett, H & Mirwuma, P in prep. Mawng dictionary.

Stanner, WEH 1936, ‘Murinbata kinship and totemism’, Oceania, 7, pp. 186–216.

Stanner, WEH 1937, ‘Aboriginal modes of address and reference in the north-west of the northern Territory’, Oceania, VII, pp. 300–15.

Straehle, CA 1993, ‘“Samuel?” “Yes, dear”: Teasing in conversational rapport’, in D Tannen (ed.), Framing in discourse, New York; Oxford: Oxford University Press, pp. 210–30.

Street, CS, 1987, An introduction to the language and culture of the Murrinh-Patha, Darwin: Summer Institute of Linguistics, Australian Aborigines Branch.

Strehlow, C 1907–15, Die Aranda Und Loritja-Stämme in Zentral-Australien, Frankfurt am Main: Joseph Baer & Co.

Taylor, J 2010, Demography as destiny: schooling, work and Aboriginal population change at Wadeye, Canberra: Centre for Aboriginal Economic Policy Research, The Australian National University.

Walsh, MJ 1976, The Murinypata language of north west Australia, PhD thesis, The Australian National University.

Wilkinson, M & Zorc, RD 2010, ‘Djambarrpuyŋu Toolbox Import (2010-04-28)’, AUSTKIN, viewed 7 July 2013,

Appendix 4: A Guide to Transcription Conventions

Abbreviations used in this paper: ANAPH = ‘anaphoric’ demonstrative, DC = ‘daucal’ (the morphological collapse of dual and paucal), DIST = distal demonstrative, DM = discourse marker, DO = direct object, DU = dual, EMPH = emphatic, F = feminine, FOC = focus, FUT = future tense, FUTIRR = Future irrealis, INC = inclusive of the addressee, INTJ = an interjection, INTS = intensifier, IO = indirect object, M = masculine, NC:ANM = nominal animate class, NC:HUM = nominal ‘human’ class (living Aboriginal people), NC:RES = nominal ‘residue’ class, NEG = negation, NFUT = non-future tense, NS = non-singular, NSIB = non-sibling, PC = paucal (several), PIMP = past imperfective, PL = plural, POSS = possessive, PROX = proximal demonstrative, RR = reflexive/reciprocal, SB = Subject, SG = singular, TEMP = temporal adverbial, TOP = topic, * (as in *word) = reconstructed form or posited ancestral form. 1, 2, 3 = first, second, third person. Additional numbers between 1–38 convey verb class. For example, 3SG.SB.19Poke.PIMP expresses the fusion of: third singular subject, 19 ‘poke’ verbal classifier, and past imperfective.

Kinterm abbreviations: Br = brother/brother’s, Ch = child, Da = daughter/daughter’s, Fa = father/father’s, Hu = husband/husband’s, m = man’s [kin], Mo = mother/mother’s, Pa = parent, Sb = sibling, So = son/son’s, w = woman’s [kin], Wi = wife/wife’s, y = younger [kin], Zi = sister, sister’s. For example: mZiDaCh = man’s sister’s daughter’s child.

Symbols relating to the transcription of speech:

[ ]

Overlapping speech.


Silence (i.e., 0.9 seconds)


0.1 seconds of silence.

xxx xx

Indiscernible speech.


Difficult to discern text. Bracketing indicates either a best guess at transcription or text alleged by consultants that I believe to be dubious


Transcriber’s comments


Fully rising terminal intonation.


Fully falling terminal intonation.


Mid-low falling terminal intonation.


lightly rising terminal intonation.

1 In Blythe (2010, p. 451), I stated that trirelational kinterms were unattested in Murrinhpatha. At that stage, I did not realise that these infrequently used complex kinterms were trirelational.

2 Although I might appear to privilege conversational corpora in the study of kinship, I wish to point out that the investigation undertaken here has demanded extensive elicitation of both kinship semantics and kinterm usage. Kinterms were collected from ethnographic sources, working lexicon files, conversational transcripts and field notes. Most transcripts have been interlinearised (a process that demands extensive elicitation). Kinterms attested in conversation were crosschecked against elicited genealogies and genealogies were used to generate and crosscheck kincharts. Figures 59 and 60 represent my current understanding of a kinship system that is almost certainly evolving, but not in the manner suggested by earlier ethnographers (see below). I thank an anonymous reviewer for suggesting these methodological clarifications.

3 An interactional analysis of one of these terms used vocatively in conversation is included as a supplement to this chapter.

4 Dyadic morphemes (Evans 2003; Merlan & Heath 1982) are specialised dual affixes to kinterms that mark that a pair of individuals are related in the manner of the relationship expressed in the base kinterm. For example, if the Gooniyandi term dyadic suffix -langi is attached to goornda (male cross-cousin), then the term goornda-langi would denote a pair of cousins (McGregor 1996, pp. 219–21).

5 For this reason, P1, the person to whom these Murrinhpatha trirelational terms are anchored, is equivalent to (but not synonymous with) Ego. Although these infrequently used terms have only ever been volunteered with the speaker as P1 (Ego), elicitation tests suggest that it is theoretically possible to have P1 as the addressee. As such, P2 would be the addressee’s son or daughter.

6 Language consultants volunteered roughly half of these terms when particular individuals were mentioned in transcription and elicitation sessions. The rest of the paradigm was fleshed out through targeted elicitation. The list appears to be exhaustive. Other avoided kin types predicted to yield trirelational kinterms (e.g. a woman’s daughter’s husband) were not forthcoming (although in the case of wDaHu, a phrasal circumlocution was provided). Although only one term has emerged unsolicited in the four-hour annotated conversational corpus (kalemamnge, 1 token), I expect the larger collection contains further tokens.

7 Mam is a very general non-future ‘say/do’ verb that is underspecified for aspectual viewpoint (Nordlinger & Caudal 2012). Therefore, while it usually means ‘said’, it can also mean ‘says’.

8 A current morphophonemic process not mentioned in Street (1987) or Walsh (1976) is that the apical alveolar versus retroflex distinction is neutralised in nasal clusters.

9 The three terms that Stanner (1936, p. 199) claimed to be borrowings from Jaminjung were pugali (MoBrCh and FaZiCh), lambarra (‘[ZiSo] whose daughter one marries’) and ngaguluk (WiMoBr). Lambarra is a wanderwort, widely attested in Northern Australia, but seldom used by the Murrinhpatha. Stanner’s own comparative chart of kinterms (AIATSIS MS3752 Series 5 Item 23[a]) attests pugarli in nearby Ngan’gi wumirri, Ngan’gikurungkurr, Marri Ngarr, Magati Ke and Wagiman, making the Jaminjung hypothesis difficult to sustain. Stanner claimed that the Murrinhpatha ngaguluk was a changed version of the Jaminjung ngawuluk (MoMoBrSo). If the term was diffused, linguistic evidence (which is weak at best) would suggest that the diffusion was in the opposite direction, given that lenition is more frequent than fortition in intervocalic environments (Butcher 2006).

10 Other such ad hoc phrasal descriptions in the Falkenberg monographs include yile nginarr (WiMoBr and MoMoBrSo) and wakal nginarr (MoBrDaCh), which have not surfaced in the conversational corpus. The former is either yile (and hence not distinguished from Fa using ordinary kinterms) or nginarr (and hence is not distinguished from any other G±1 affine in the MoMoBr patriline). MoBrDaCh is referenced either as nginarr or wakal (G–1), but never in combination, or as muluk (G–1 male) or newuy (G–1 female).

11 Murrinhpatha verbs are generally complex predicates comprised of an inflected classifier stem and an uninflected lexical stem (Nordlinger 2010). There are some 38 classifier paradigms (Blythe, Nordlinger & Reid 2007). Tentative semantic glosses can be provided for some of these, but not all (hence, they are given numeric labels).

12 All names here are pseudonyms.

13 I am grateful to Mark Dingemanse for alerting me to these terms.

14 Aranda systems have four distinct terms in G+2 (FaFa, MoMo, MoFa and FaMo)—each of which show sibling merger. The four terms demarcate distinct patrilineages (FaFa, MoMoBr, MoFa amd FaMoBr).

15 Technically, it is more precise because the gender of P1 is marked. However, when P1 is the current speaker, this information is redundant because recipients are normally aware of the speaker’s sex.

16 This accords with Garde’s (2013, pp. 119–20) observation that Bininj Gunwok Kun-derbi terms are acquired by younger adults and teenagers (rather than children).

17 In the case of Murrinhpatha, the relationship between P1 and P2 is presently fixed as ‘parent of’. If this fixed relationship were to be unlocked such that P1–P2 could become variable, then we might expect a more expansive collection of terms to emerge—one that is less closely tied to avoidance contexts.

18 Cf. McConvell (1982, pp. 104–6) on Gurindji, and Green (1998, p. 41) on Alyawarr and eastern Anmatyerr.

19 Alyawarr and eastern Anmatyerr (Green 1998), Bininj Gunwok (Garde 2013), Dalabon (Evans & Garde 2013), Djambarrpuyŋu (Wilkinson & Zorc 2010), Gooniyandi (McGregor 1996), Gurindji (McConvell 1982), Jawoyn (Merlan 1989), Kayardild (Evans, Johnson & Kohler 1992), Mawng (Singer et al. in prep.), Pintupi (Hansen, Hansen & Tjapaltjarri 1974), western Arrernte (Strehlow 1907–15) and Yir-Yoront (Alpher 1991).

20 Banyjima (Dench 1980), Burarra (Glasgow 1994), Nyangumarta (Geytenbeek & Geytenbeek 1982; O’Grady & Mooney 1973), Iwaidja (Bruce Birch, pers. comm.), Kalkatungu (Blake 1979), Warlpiri (Laughren 1982), Watjarri (Douglas 1981) and Yidiny (Dixon & Irvine 1991).

21 Lea (2004, p. 31) attested 24 Mẽbêngôkre trirelational kinterms associated with both consanguineal and affinal kin relationships, plus 17 associated with formal friendships.

22 I am indebted to Nick Enfield for suggesting this.

23 A man’s WiMoBr is an avoided affine that is normally referred to with the (binary) kinterm ngaguluk. I asked a consultant whether an acceptable way to refer to this man might be as kawumarna (a term that I thought up—potentially decomposable as ‘the male person that one’s son/daughter calls kawu [MoMoBr]’). The consultant had never heard the term but, after a moment’s consideration, stated that it would be an acceptable way to talk about one’s ngaguluk (Carmelita Perdjert, pers. comm.).

24 For example, ‘The person who is my uncle and your cousin’ is more precise than ‘my uncle’ and more precise than ‘your cousin’.

25 A dramatic example of convergent evolution in nature is that of hummingbirds and hummingbird hawk-moths that beat their wings in a similar fashion and at similar speeds. Both occupy the same ecological niche by sucking nectar from tubular flowers using a long proboscis (Bates 1863, pp. 180–92). Other examples include the independent development of echolocation in bats and toothed whales (Liu et al. 2010); thunniform body types in lamnid sharks, tunas, ichthyosaurs and whales (Donley et al. 2004); carrion and faecal scent mimicry in both angiosperm flowers and stinkhorn fungi (Johnson & Jürgens 2010); intermittent energy-reducing locomotory patterns in seals, sharks and migratory birds (Gleiss et al. 2011); and light-skin pigmentation in both European and East-Asian humans (Norton et al. 2007).

26 In the mythical Dreamtime, the sun was a woman. I presume kardu mere pumengewiyewiye nekiyu, ‘we mustn’t be disrespectful of her’, to be a reference to this sun-woman. Although the lexical content overtly labels the situation as serious, the loud and feigned aggression indexes the following proposal as non-serious (Lytra 2007, 2010; Miller 1986).

27 Perhaps coincidentally, between the two sisters sits Lucy’s daughter Rita, who ordinarily addresses Alice as kale, ‘mother’s sister’. Although I’ve been told that an actual son/daughter need not be present to use one of these trirelational kinterms, the presence of one clearly does not preclude their usage.