Abstract: Proto-Oceanic *R was irregularly lost or merged with another liquid in several branches of Oceanic. In Southern Oceanic, the likelihood of *R loss increases from north to south. This squib examines the merger and loss of *R in Micronesian, where the pattern is most similar to the southernmost languages of North Central Vanuatu. I discuss scenarios of lexical diffusion that could have led to the observed state of affairs, proposing an interpretation where structured variation along the hypo/hyperarticulation continuum is inherited from a proto-language, resulting in parallel developments in distant branches.

Abstract: Tomoip is an Oceanic language spoken on the island of New Britain in the Bismarck Archipelago of Papua New Guinea. Although previously thought by some to be a non-Austronesian language, Tomoip is undeniably a member of the Oceanic branch of the Austronesian family. However, its precise classification has remained a challenge, largely because its lexicon seemingly contains a high percentage of non-Austronesian content and its Austronesian forms exhibit several complex sound correspondences. This paper provides a historical phonology of Tomoip, detailing the sound changes that have led from Proto-Oceanic to contemporary Tomoip. Based on these sound changes and morphological evidence, we propose that Tomoip is most closely related to the New Ireland subgroup of the Meso-Melanesian cluster. We further hypothesize that Tomoip and Proto-New Ireland are coordinate branches of a single proto-language, which we call Proto-Tomoip–New Ireland.

Abstract: The lexicon of the Timoric language Tetun Terik exhibits a notable typological oddity in that a vanishingly small number of verbs (<1%) are vowel-initial, despite there being no such paucity of vowel-initial adjectives or nouns. Etymological investigation reveals that these 'lost' *#V verbs have prothesised an /h/ of unclear origin. At the same time, Tetun Terik attests two vowel hiatus repair strategies: an epenthetic [h] at certain morphological prefix boundaries, alongside the default and more areally consistent [ʔ]. This paper presents a diachronic account of these three interconnected puzzles, namely the #V gap, prothetic /h/, and epenthetic [h], as the outcome of morphophonological reanalysis and phonological exaptation. In particular, it proposes that epenthetic [h] was innovated on the basis of fossilized and reinforced causative morphology, whereas prothetic /h/ on etymologically #V verbs arose due to analogical backformation and frequency-based categorial asymmetries in vowel hiatus contexts.

Abstract: This paper examines the fate in Formosan languages of the Proto-Austronesian voiced stop and affricate phonemes. It argues that two sets with different airstream mechanisms should be distinguished: an implosive set consisting of *b and *Z (in Ross's notation), little subject to devoicing, and a plain-voiced set, prone to devoicing, comprised of the more retracted stops *D [ɖ] and *g [g], and two affricates *d [ʤ] and *z [dz]. Outside of lenitions, the plain-voiced set remains voiced only in the southern Formosan languages Paiwan, Puyuma, and Rukai. This general pattern is obscured in the modern languages by the widespread mergers affecting the coronals. In the present analysis, the Proto-Austronesian alveolar stop—implosive [ɗ]—is thought to be the phoneme symbolized *Z by Ross, not his *d, which I take to have been [ʤ]. This paper also argues that *D [ɖ], *d [ʤ], and *z [dz] became implosive in Bunun through merger with *Z; and that the more retracted articulation of /d/ compared with /t/ in Pazeh and Bunun comes from being 'zombie implosives', that is, plain-voiced sounds having evolved out of an implosive.

Abstract: This paper examines the factors that influence the choice between actor voice and undergoer voice in the Oceanic language Äiwoo from two perspectives: a quantitative study comparing 80 actor-voice clauses with 80 undergoer-voice clauses and a qualitative study looking at all occurrences of actor voice in the available data. The results show, on the one hand, that the factors influencing voice choice in Äiwoo largely overlap with those described for symmetrical voice languages in the western Austronesian region. On the other hand, the usage patterns of the actor voice also show affinities with the transitivity alternations found in canonical Oceanic languages: the undergoer voice is by far the most frequent in discourse, suggesting that it may be on its way to being reanalyzed as the default transitive construction, and the predominant context for the use of the actor voice is with an undergoer argument that is low in referentiality, similar to the contexts that favor "semi-transitive" constructions in many Oceanic languages. The Äiwoo data thus point to a possible path of reanalysis from a symmetrical voice system to an Oceanic-type system of transitivity alternations.

Abstract: Many Austronesian languages employ similar words for the concepts 'hand' and 'five'. Indeed, (partial) colexification of these two meanings is generally reconstructed back to Proto-Austronesian. However, a number of Austronesian languages do not colexify 'hand' and 'five', raising the question of what drives such lexical splits. Based on a sample of 812 Austronesian languages, I identify 465 languages (57 percent) as not exhibiting colexification of 'hand' and 'five'. I find that terms for 'hand' are often subject to replacement, whereas terms for 'five' are generally stable throughout the family. The replacement of 'five' has occurred primarily in languages in which the inherited decimal counting system has been lost and the numerals 'six' through 'nine' are no longer unanalyzable monomorphemic words. This suggests that lower numerals like 'five' are less likely to be stable lexical items when they do not lie somewhere in the middle of a series of underived number terms.

Abstract: In this paper, I argue for the existence of a three-way place contrast (dental, alveolar, retroflex) in the coronal stop inventory of the protolanguage of the Pahoturi family of Southern New Guinea. This is based on a correspondence set of over a dozen words in which one of the languages, Kawam, has coronal reflexes distinct from the rest of the family, pointing to the existence of a phonemic distinction between dental and alveolar stops in the protolanguage. This distinction has only left a trace in Kawam and has been lost entirely elsewhere in the family. Next, I consider the consonant inventories of nearby language families and suggest that the dental-alveolar contrast was one of several are-ally diffused phonological features that the Pahoturi family once had in common with the Yam languages to its west and Australian languages to the south. As previous work has suggested, it appears that in the past, this set the Pahoturi and Yam families apart from the other Papuan languages nearby, but due to convergence, Pahoturi and Yam varieties today look more similar to their neighbors. Finally, to better understand the cultural landscape of the greater region, I review the known archaeological history of the greater Torres Strait region and conclude that the Torres Strait Islands were once more linguistically diverse. I suggest that during such a time, contact with the Australian linguistic area played a role in the development of Proto-Pahoturi and Proto-Yam phonology.

Abstract: To date, no detailed analysis has been carried out on the verb classes or on the functions and distribution of the Santa Cruz valency-changing morphology from a diachronic perspective. This study aims to fill the gap by presenting new data and comparing two valency-increasing devices of three Santa Cruz languages, namely, Nalögo, Natügu, and Engdewu. We focus on the functions and distribution of the causative (v)a - and the transitivizer - ti . When compared to Proto-Oceanic, as is to be expected in historical change, Santa Cruz languages show points of both continuity and change. We show that the Santa Cruz causative and transitivizer are reflexes of the Proto-Oceanic causative *pa[ka]- and transitivizing *-i, respectively. The prefix (v)a - retains all the functions of Proto-Oceanic *pa[ka]-, while the suffix - ti retains most functions of Proto-Oceanic *-i, albeit with many in a state of decay or near decay. In addition, we hypothesize that the decline in the use of - ti as a transitivizer likely accounts for the emergence of the prefix (v)ö -, a reflex of Proto-Oceanic *paRi-, as a productive detransitivizer. This shift makes the valency-orientation profile of Santa Cruz languages less canonically Oceanic.

Abstract: In this paper, I reconstruct the word-prosodic system of Proto-Maˈya-Salawati, the parent of several Austronesian languages of the Raja Ampat archipelago in northwest New Guinea. The aim of this reconstruction is twofold. First, word-prosodic changes in the daughter languages provide evidence for the subclassification of this branch, thus contributing to our understanding of language history in this little-known area. Second, the synchronic word-prosodic systems of the daughter languages are genealogically and typologically unusual in two ways: (i) they are all tonal, which is highly uncommon in the Austronesian language family, and (ii) several additionally have contrastive stress, an extremely rare combination crosslinguistically. This paper thus contributes to our theoretical understanding of mechanisms for tone change, as well as how orthogonal systems of lexical tone and lexical stress can emerge and decline. I will conclude that Proto-Maˈya-Salawati itself can be reconstructed with both tone and lexical stress and that several typologically uncommon phonological changes have occurred in the daughter languages—in particular, tone changes conditioned by vowel height and prosodic developments conditioned by syllable loss.