2.1.1 Language change and types of subgroups
When considering how languages change over time, an analogy can be drawn with the two
fundamental processes of biological evolution, that of divergent and convergent evolution. That is, much like in the process of divergent evolution in biology, where a single population splits into two or more populations by evolving an increasing number of independent traits over time, a linguistic ancestor gives rise to two or more distinct descendant varieties, each defined by the number of independent traits they evolved over time in relation to their ancestral form. Likewise with the process of convergent evolution in biology, where two or more genetically distinct populations evolve an increasing number of shared traits over time, giving rise to analogous structures in organisms, a set of unrelated languages may evolve an increasing number of analogous traits over time. But there are also limitations to this analogy. Unlike organisms, linguistic traits may freely converge across both related and unrelated language varieties by means of language contact, forming “hot spots” for convergence, often called “diffusion areas”. The issue with classifying languages in such areas have gained much attention in Oceanic linguistics, and researchers have since introduced two types of subgroups in linguistic classification, so-called “innovation-defined”
subgroups, and “innovation-linked” subgroups (Pawley & Ross, 1995).
A so-called innovation-defined subgroup is a subgroup in which all members share a set of innovations that evolved in their immediate ancestor and are reflected in each of the daughter languages. This kind of subgrouping is what is represented in the traditional family-tree model, and what I will refer to using the term “subgroup” in this thesis.
A so-called innovation-linked subgroup, often called a “linkage” (Lynch, 1999), is a subgroup in which all the members form a network of languages with overlapping linguistic innovations, where no single innovation is shared throughout the entire network, and the languages as such cannot be defined by any single trait like with proper subgroups. A linkage may be a result of diffusion across languages that share more than one immediate ancestor, or it may be a result of
an earlier dialect continuum, in which case the languages share a single immediate ancestor, but still cannot be assigned any exclusively shared innovations because the parent did not exist long enough as a unit to evolve any defining innovations of its own (Ross et al., 2016, p. 12).
2.1.2 External and internal relations
The earliest evidence of human settlement on New Caledonian are remnants of stoneware associated with the Lapita cultural tradition. This archaeological tradition appeared first in the Bismark Archipelago around 1600 BCE, and rapidly spread outward, its bearers reaching New Caledonia, Vanuatu, Fiji, Tonga, and Samoa in Remote Oceania by boat around 1300–1000 BCE (Pawley & Ross, 1995, p. 64). The earliest archaeological sites of the Lapita tradition from the northern part of the New Caledonian mainland date back to around 1000 BCE. Judging from the archaeological record, the mainland appears to have been settled first, with the neighboring smaller islands being settled shortly thereafter (Sand, 1999, p. 142).
Around 600–800 BCE, Lapita associated stoneware is replaced by novel traditions throughout the archipelago. This divergence of archaeological traditions in the region is already observed within the first centuries after settlement, and by the common era, separate archaeological traditions had taken hold in the north and south of the mainland respectively (Sand, 1999, p. 155).
Haudricourt first explored the internal relationship of the Kanak languages, working under the hypothesis that these languages descended from a single Austronesian language that arrived to the archipelago before the common era (Haudricourt, 1971, p. 383). The closest relatives of the languages of New Caledonia are now believed to be the Austronesian languages of north, central, and south Vanuatu (Lynch, 1999; Lynch & Ozanne-Rivierre, 2001), which together are
hypothesized to form an innovation-linked subgroup, reflecting an early dialect continuum that has been named the Southern Oceanic linkage (Lynch, 1999). This subgroup forms one of multiple higher order subgroups within the Oceanic group, as shown in Figure 1. Note that each node in figure 1 represents a split in the family tree which reflects a hypothesized linguistic ancestor6, where proposed intermediary subgroups are marked in parenthesis. There is today a strong consensus that the spread and diversification of the last common ancestor of all Oceanic Austronesian languages, what is known as Proto-Oceanic, was directly associated with the spread of Lapita pottery into Remote Oceania (Pawley & Ross, 1995).7
6 Hypothesized ancestors are marked with the prefix proto-, e.g. Proto-Oceanic (POc), etc.
There are currently two proposals for the internal classification of the languages of New Caledonia.
Haudricourt’s first suggestion for an internal subgrouping of the Kanak languages organized them into five groups on the mainland, Far north, North, Center, South, and Deep south, to which he added the three Kanak languages of the Loyalty Islands, without further specification about their internal relationships (Haudricourt, 1971, p. 359), as illustrated in Figure 2 below. The two subgroups of the southern mainland, which Haudricourt (1971) named the South and Deep south groups respectively, are I call the Mid and Far-Southern subgroups respectively in this thesis.
Within the Mid-Southern subgroup, Haudricourt (1971) organized the language varieties into three additional subgroups, named after the primary settlements around which they are spoken (p. 372).
As before, note that each node in the family tree also represents a hypothesized linguistic ancestor.
Figure 1: Major Oceanic subgroups (Ross et al., 2016).
Later research has concluded that the first split in the group must have been between the ancestor of the Kanak languages of the mainland and those of the Loyalty Islands (Lynch & Ozanne-Rivierre, 2001; Lynch, 2003). Haudricourt’s classification (see Figure 2) was later elaborated by Ozanne-Rivierre, who demonstrated that the languages classified as Far north, North, and Center are linked by a number of phonological innovations, and together form a Northern branch of the Mainland group (Ozanne-Rivierre, 1995). This has led to the Mid and Far Southern languages to be later classified in a joint Southern branch (Lynch & Ozanne-Rivierre, 2001), but so far there is little evidence in support of this branch as a proper, innovation-defined subgroup. In this regard, Haudricourt’s classification remains the most reliable model to date.