New Guinea is perhaps the most linguistically diverse region in the world. Besides the Austronesian languages, there are some 800 languages divided into perhaps sixty small language families, with unclear relationships to each other or to anything else, plus a large number of language isolates. Although there has been relatively little study of these languages compared with the Austronesian family, there have been three preliminary attempts at large-scale genealogical classification, by Joseph Greenberg, Stephen Wurm, and Malcolm Ross. The largest family posited for the Papuan region is the Trans-New Guinea phylum, consisting of the majority of Papuan languages and running mainly along the highlands of New Guinea. Since perhaps only a quarter of Papuan languages have been studied in detail, linguists' understanding of the relationships between them will continue to be revised.
Several languages of Flores and nearby islands, and especially the language of Savu Island (also called Sabu or Sawu) are usually thought to be Austronesian, but are reported to have large numbers of non-Austronesian words in their basic vocabulary. It has been suggested that these may originally have been non-Austronesian languages that have since borrowed nearly all of their vocabulary from neighboring Austronesian languages, but no connection with the Papuan languages of Timor or Halmahera has been found.
The languages of the Andaman Islands may be related to some western Papuan languages, but are not themselves covered by the term Papuan.
According to Ross (see below), the main problem with Wurm's classification is that he did not take contact-induced change into account. For example, several of the main branches of his Trans-New Guinea (TNG) phylum have no vocabulary in common with other TNG languages, and were classified as TNG because they are similar grammatically. However, there are also many Austronesian languages that are grammatically similar to TNG languages due to the influence of contact and bilingualism. Similarly, several groups which do have substantial basic vocabulary in common with TNG languages are excluded from the phylum because they do not resemble it grammatically.
Papuan families proposed by Wurm (with approximate number of languages)
Two of Wurm's isolates have since been linked as the
and since Wurm's time another isolate and two languages belonging to a new family have been discovered,
Malcolm Ross re-evaluated Wurm's proposal on purely lexical grounds. That is, he looked at shared vocabulary, and especially shared idiosyncrasies analogous to English I and me vs. German ich and mich. The poor state of documentation of Papuan languages means that this approach is largely restricted to pronouns. Nonetheless, Ross believes that he has been able to validate much of Wurm's classification, albeit with revisions to correct for Wurm's partially typological approach. (See Trans-New Guinea languages.)
It has been suggested that the families which appear when comparing pronouns may be due to pronoun borrowing rather than to genealogical relatedness. However, Ross argues that Papuan languages have closed-class pronoun systems, which are resistant to borrowing, and in any case that the massive number of languages with similar pronouns in a family like Trans-New Guinea preclude borrowing as an explanation. Also, he shows that the two cases of alleged pronoun borrowing in New Guinea are simple coincidence, explainable as regular developments from the protolanguages of the families in question: as earlier forms of the languages are reconstructed, their pronouns become less similar, not more. (Ross argues that open-class pronoun systems, where borrowings are common, are found in hierarchical cultures such as those of Southeast Asia and Japan, where pronouns indicate details of relationship and social status rather than simply being grammatical pro-forms as they are in the more egalitarian New Guinea societies.)
Ross has proposed 23 Papuan language families and 9-13 isolates. However, because of his more stringent criteria, he was not able to find enough data to classify all Papuan languages, especially many isolates which have no close relatives to aid in their classification.
Ross also found that the Lower Mamberamo languages (or at least the Warembori language; he had insufficient data on Pauwi) are Austronesian languages which have been heavily transformed by contact with Papuan languages, much as the Takia language has. The Reef Islands-Santa Cruz languages of Wurm's East Papuan phylum were a potential 24th family, but subsequent work has shown them to be highly divergent Austronesian languages.
Note that while this classification may be more reliable than past attempts, it is based on a single parameter, pronouns, and therefore must remain tentative. Although pronouns are conservative elements in a language, they are both short and utilise a reduced set of the language's phonemic inventory. Both phenomena greatly increase the possibility of chance resemblances, especially when they are not confirmed by lexical similarities.
Papuan families proposed by Ross
Language isolates proposed by Ross (sorted by location)
Former isolates classified by Ross:
Languages reassigned to the Austronesian languages:
Unclassified due to lack of data:
Greenberg also suggested a connection to the Tasmanian languages. However, the Tasmanian peoples were isolated for perhaps 10,000 years, genocide wiped out their languages before much was recorded of them, and few linguists expect that they will ever be linked to another language family.
William Foley (1986) noted lexical similarities between R.M.W. Dixon's 1980 reconstruction of proto-Australian and the languages of the East New Guinea Highlands. He believed that it was naïve to expect to find a single Papuan or Australian language family when New Guinea and Australia had been a single landmass for most of their human history, having been separated by the Torres Strait only 8000 years ago, and that a deep reconstruction would likely include languages from both. However, Dixon later abandoned his proto-Australian proposal, and Foley's ideas need to be re-evaluated in light of recent research.