0. No evidence for borrowing

• Items reflecting a reconstruction in any one of the protolanguages except Proto Western Bel, except where this reconstruction is itself known to reflect a borrowing.

• Phrasal items for which no evidence is available to date the innovation of the phrase.

1. Very little evidence for borrowing

• Used for items where there is no evidence available about either borrowing or inheritance.

• Used for items reflecting a Proto Western Bel reconstruction, as there is a higher probability than for other reconstructed items that the reconstruction itself represents a borrowing.

• Used for items attributed in W13 to ‘Early Oceanic’ because of their canonic form, noted in W10 ‘Notes on borrowed’.

2. Perhaps borrowed

• Used if the the apparent borrowing source is suspect in some way explained by a comment in W22.

3. Probably borrowed

• Used if there are reasons to think the item is borrowed, but no borrowing source can be identified.

• Used if the possible Bargam source item is formally so different from the Takia item that the borrowing source is probably a cognate of the Bargam item in some language as yet unidentified.

4. Clearly borrowed

• Used if the borrowing source is known and there is no reason to doubt that the item is borrowed.

The orthography used for pre-modern lexical items here and in the database is a phonemic spelling . This differs in two respects from the practical orthography adopted by the Summer Institute of Linguistics in consultation with Takia speakers and used by Ross (2002). Vowels other than -a- which occur in syllables before the stress are inserted by rule: either -i- or -u-. Inserted vowels are often written in the practical orthography, but this results in alternative orthographic representations of the same word (with inserted i or inserted u or neither). For consistency’s sake inserted vowels are not written here. Second, there are several differences involving w and y. The main one concerns the phonological interpretation of wordswith falling phonetic vowel sequences such as the following, where [tigi′yo] ‘we called you’ and [tusu′we] ‘we poked (it)’ are represented phonemically as t-gy-ó and t-swé respectively.That is, the intervocalic glides [y] and [w] are considered to represent phonemes, whilst thefirst vowel of the sequence is interpreted as an insertion. This allows for simpler generalisations about the insertion of pre-stress vowels than in previous work, where these words wereinterpreted as t(i)-gi-o and t(u)-sue respectively.

Morphosyntactic facts that need to be known in order to interpret the wordlist data, and especially the morpheme-by-morpheme glosses, are:

• Verbs always have a prefix coreferential with the subject, e.g.ŋ-aw ‘Igo’, y-aw ‘he/she/it goes’ etc. Their citation forms are shown with an initial hyphen, e.g. -aw ‘go’.

• Nouns may be either alienably or inalienably possessed. Inalienably possessed nouns have a suffix coreferential with the possessum,e.g. bani-g ‘my hand’, bani-n ‘his/her/its hand’. Their citation formsare shown with a final hyphen, e.g. bani-‘ hand’.

• The possessor suffix and subject prefix are omitted in a phrasal citation form and in its a morpheme-by-morpheme gloss if the affixes vary in person and number. Thus the citation form for ‘to love’ is bbe- -pani (heart--give) ‘give (one’s) heart’, with the variants bbe-n i-pani (heart-his 3SG-give) ‘s/he loves’, bbe-g ŋ-pani (heart-my 1SG give) ‘I love’, and so on.

• Many adjectives also take a possessor suffix. Again, the citation form is shown with a final hyphen, e.g. sae- ‘bad’. An adjective of this type always takes the 3SG suffix -n when it is used predicatively. An attributive adjective follows the noun it modifies, and in the past took a possessor suffix agreeing in number and person with that noun. Increasingly in modern Takia the suffix is -n regardless of agreement.

Determinations here are straightforward except for items that include enclitics. Although

enclitics are phonologically bound (to some degree: they do not participate in stress assignment),

they are syntactically independent. Items that include them are therefore labelled ‘Analyzable

phrasal’.

The periodizations and dates require some explanation. Proto Oceanic, the language of the Lapita culture at the time that it expanded eastward around 1200 BC, came into being as the result of contact between the Austronesian speakers who arrived in the Bismarck Archipelago around 1400 BC and speakers of Papuan languages whose ancestors had long been in New Guinea and the Bismarcks. I have labelled this period Early Oceanic. Proto Oceanic was originally posited by Dempwolff (1937) and has since been firmly reconstructed (Lynch et al. 2002:Ch.4).

Pre-Oceanic means‘ Austronesian speaking before 1400 BC’. The early limit of 3500 BC is the earliest data for a culture associated with Austronesian, but it is not really meaningful to set a start date in this way.

The end dates of the Pre-Oceanic, Early Oceanic, Western Oceanic, New Guinea Oceanic, North New Guinea, Ngero-Vitiaz and Bel periods are associated with the reconstructed interstages which we label ‘Proto Western Oceanic’, ‘Proto New Guinea Oceanic’, and so on. Note that these proto languages represent cross-sections of ongoing language change, rather than periods of time. The fact that a lexical item in the Takia database is said to be descended from are constructed etymon in, say, Proto Ngero-Vitiaz says only that an ancestral form was present in Proto Ngero-Vitiaz. It entered the language at the latest during the Ngero-Vitiaz period, and may have entered it earlier, but there is currently no data to support an earlier reconstruction. The protolanguages from Western Oceanic through to Bel are justified by

Ross (1988). There are also reconstructions in the database for Proto Western Bel. Western Bel is the subgroup which includes Takia, Megiar (a dialect of Takia), Bilibil and the various dialects of Gedaged (Gedaged proper, Riwo and Siar). I assign Proto Western Bel to the Early Takia period, as explained below.

Except for the end date for Bel, the dates for the prehistoric periods listed in the previous paragraph are based on the correlation of reconstructed linguistic events with known archaeological events. There is room for error, both in the correlation and in the archaeology. The dates associated with Pre-Oceanic and Early Oceanic are well accepted in the historical linguistic and archaeological literature are quite well founded (Spriggs 1997, Kirch 1997,1998). Western Oceanic was the network of dialects left behind on New Britain and New Ireland after the initial wave of eastward migration. Ross (1988:382–386) proposed that there were two waves of Oceanic settlement eastward into the northwest Solomons area and that the second was associated with the break-up of Western Oceanic, dated to around 900 BC on the basis that there was late Lapita settlement on Buka Island by 800 BC and on New Georgiaby 600 BC (Sheppard and Walter 2006).

The break-up of New Guinea Oceanic, spoken (I assume) in western New Britain, must have occurred around 200 BC, as the archaeology of the south coast of Central Papua suggests that Oceanic speakers had reached there by 100 BC (Vanderwal 1973). The continuing network of dialects on New Britain represents the North New Guinea period, until the Vitiaz Strait area was resettled by Oceanic speakers sometime between 300 and 500 AD following a post-Lapita hiatus in settlement. This represents the beginning of the archaeological Sio pottery period (Lilley 1999, 2000, In press). The Ngero-Vitiaz end date of 1000 AD represents the appearance of new pottery types (Lilley 1999, Specht et al. 2006, Lilley and Specht In press, westward expansion along the coast and offshore islands, and the probable begin¬nings of the Vitiaz Strait trade network, which persisted until after European contact in the twentieth century (Harding 1967).

According to oral tradition, the ancestors of all speakers of Bel languages (Takia is a member of the Bel family) lived on Yomba Island, where Hankow Reef is located, until it disappeared under the waves (Mennis 2006). I estimate that this occurred around 1600, and that the ancestors of the Takia settled on Karkar around 1680 after a tsunami had wiped out the earlier population (for the oral history, see McSwain 1977:24, for dating Ross 2007b). The Early Takia period is the last prehistoric period before the arrival of German missionaries. It is also the period in which I infer that loans from Bargam and Waskia occurred. A number of items are reconstructable for Proto Western Bel on the basis of the cognates given in Mager’s (1952) Gedaged dictionary. I have not assigned these to a separate ‘period’ of Takia history. They belong to a time when the Western Bel languages, listed earlier, were still more or less a unity. I assume this to have been for a relatively short period of time before and after theTakia settled in Karkar Island, and I have assigned such items to ‘Early Takia’.

The history of Takia since European contact is known to us, at least in outline. German Lutheran missionaries were active on Karkar from 1888 to 1895, but made little impact. I have allowed the Early Takia period to extend into the historical period up to 1912, as there is no evidence of borrowing before 1912. A second group arrived in 1912 and were more successful, but were forced out by Germany’s wartime defeat in 1918 (Wagner and Hermann 1986:109–110, 127–129). I have labelled this early period of six years the Early missionary period. I have left a gap in periodization between 1918 and 1935 since I know of no conditions likely to lead to borrowing during that period.

After the transfer of New Guinea (the northern half of modern mainland Papua NewGuinea) to Australian sovereignty, German and Australian Lutheran missionaries continued their activities, and in 1935 Gedaged (or Graged), another Bel language (described in a manuscript grammar by Dempwolff (n.d.)) was adopted as a lingua franca by the mission and was used extensively in education and in church activities. Narer on Karkar Island was an important educational and missionary centre. This period, which I call the Gedaged schools period lasted until 1962, when the mission largely replaced Gedaged with the national lingua franca Tok Pisin (Freyberg 1977, Ross 1996).

Tok Pisin had doubtless already made inroads on Karkar before the second World War, however from about 1935, and was already spoken by most of the population in 1962. Today all Takia speakers are bilingual in Tok Pisin. I call the period during which Tok Pisin has been a lingua franca on Karkar the modern period, as it extends into the present. Recent years have seen a reduction in the influence of the Lutheran Church on Karkar and a growth in a number of smaller Christian denominations. Only one borrowing on the list (‘to fast’) is attributable to this last period, which I call the smaller missions period. The Gedaged schools period and the smaller missions period both overlap with the modern period, but this is necessary if the dating of twentieth-centruty loans is to be as accurate as the data allow.