Comparison with Lithgow Comparison with Engkvist and Troolin

states that “[n]ormally two speech communities are considered to be separate languages and mutually unintelligible if the percentage of cognates between lists of vocabulary items is less than 81 percent” Lithgow and Staalsen 1965:4. However, the idea that two varieties with less than 81 percent shared cognates are two separate languages must be taken as a guideline rather than a rule. While two speech forms are considered the same language when they have 81 percent or more shared cognates, different guidelines for determining what constitutes a language apply when using lexical similarity figures. Blair writes that varieties with lexical similarity as low as 60 percent may be considered dialects of the same language, where intelligibility is above 80 percent Blair 1990:23, and this guideline is more commonly used within SIL. As pointed out by Engkvist and Troolin 2002:20, Wurm and Laycock emphasised that it was a “common phenomenon in New Guinea that languages sharing less than 81 percent of basic vocabulary cognates are mutually intelligible” Wurm and Laycock 1961. Therefore, taking into account the difference between lexical similarity and shared cognates, as well as the Papua New Guinean context as described by Wurm and Laycock, it makes sense to use Blair’s standard rather than the 81 percent guideline. Considering Blair’s guideline, given that all the wordlists compared share more than 60 percent lexical similarity, the question then becomes whether all of the dialects are mutually intelligible. Blair’s guideline is that among similar dialects, or dialects of the same language, there should be intelligibility such that a person in one dialect could score 80 percent on a comprehension test of a recorded text in the other dialect. No comprehension testing has been done between the dialects in this survey, so no percentage can be given to quantify the intelligibility between them. The opinion of the survey team, based on what people reported regarding intelligibility see section 2.3, is that there is some doubt that people would score 80 percent on a recorded text comprehension test of the dialects that are farthest from their area; it does seem possible that Basima is intelligible enough to the other dialects that people from every dialect would score 80 percent on a test of the Basima dialect.

2.4.4 Comparison with Lithgow

Comparing the current similarity percentages with those calculated by Lithgow see table 3 reveals that all the similarity percentages recorded by Lithgow are higher than those from the current study. This may indicate that different synonyms were elicited, but is most likely a reflection of different methods of comparison. Table 11 shows the percentages obtained when several of Lithgow’s wordlists 32 were compared using the current study’s similarity criteria as described in 2.4.2. Table 11. Lithgow’s 1964 wordlists reanalysed Degimwanea Basima 98 Basima 79 76 Ulua 77 80 88 Gameta The recalculated percentages above do not differ significantly from the results of the present survey. In fact, the relative differences even when using Lithgow’s percentages are similar, so that Gameta and Ulua are shown to be closer to each other than to Basima, as are Galeya Momo’awa and Sebutuya. Lithgow’s results do, however, show Basima as being significantly more lexically similar to Ulua and 32 See table 30 in Appendix C.1. Wordlists by Lithgow collected in Sebutuya were not reanalysed, since the only wordlist to be found had only fifty words recorded, and with such a list it would be necessary to have a lexical similariy difference of over 10 in order to give any significant results. The range of error with a list of fifty words is too great. Gameta than to Galeya and Sebutuya, which presumably led to his distinction between northern and southern dialects, with Basima belonging to the northern dialect along with Gameta and Ulua. 33

2.4.5 Comparison with Engkvist and Troolin

Engkvist and Troolin 2002:6 suggest only a tentative table representing the dialect situation in the area. When considering their percentages, it is interesting to note the following: • The similarity between Ulua and Gameta is lower 75 percent than that of the current survey 88 percent. • The similarity between Yaya and Gubiayawe Basima villages, and Gameta is lower 73 percent and 78 percent, respectively than that of the current survey 80 percent. • The similarity between the Basima villages and Gameta is not significantly different from the similarity of Basima to Sebutuya, implying that Basima is a central dialect. The differences that exist between Engvist and Troolin’s results and the results of the current study may be due to different synonyms being elicited. Also, none of the wordlists, with perhaps the exception of Yaya, were elicited in exactly the same hamlets, so the comparisons are not being made between the exact same locations. Because this is a dialect chain, it is necessary to compare wordlists from the same locations in order for a second comparison to be accurate.

2.4.6 Summary of lexicostatistical analysis