General analysis techniques and definitions

For interviews with church and local government authorities, individual answers were recorded. In cases where the highest level leaders for particular locations were unable to participate in the research, persons filling in for their positions were asked to participate in the research. We aimed at interviewing one leader a pastor or evangelist in the case of a ProtestantEvangelical church, and a priest or a catechist in case of Roman Catholic churches out of all churches represented in a village. We also interviewed Muslim leaders in the community. There were no Tongwe among the Christian church leaders. The interviewed Muslim leaders were generally Tongwe. The selection of participants for collecting wordlists was made with the assistance of local government leaders who, according to our requests, appointed and asked for participation from members of the local community.

4.4 General analysis techniques and definitions

The SRT scores were converted to RPE levels using the technique described below. The mean values of the RPE levels were then calculated, along with the standard deviations, for all the participants, as well as for each subset of the sample. While the validity of taking the mean of ordinal values such as RPE levels is debatable, it was felt that the mean was at least a helpful indicator of the approximate average Swahili proficiency of the group. The various factors—age, gender, education and travel—were then tested using a chi-square goodness of fit test, to see if, at the 95 level, they were significant factors in Swahili proficiency. WORDSURV version 2.5 was used to calculate the percentage matrix and the variance matrix for comparing the lexical similarity of the wordlists Wimbish 1989. The term “significance,” when used without qualification in this report, refers to statistical rather than substantive significance. Whenever the difference is described as “significant” without qualification, greater than or equal to 95 statistical significance is meant. Conversely, a difference described as “not significant” refers to less than 95 statistical significance. Whenever the words “average” or “mean” are used without qualification, the arithmetic mean is meant. The term “educated” refers to a person who was educated at least up through Standard Four of primary school or participated in education for adults, provided they learned to readwrite reported data. The term “uneducated” refers to a person who had no formal education above Standard Four or anyone who had some formal education but was unable to readwrite. Travel categories are defined as follows: 0 – no travel 1 – few days 1–3 2 – many days 3, few weeks 1 month 3 – many weeks, few months 4 – many months, few years 5 – many years. The term “travelled” refers to a person who has travelled more than three days—categories 2 to 5, while the term “untravelled” refers to a person who has travelled a few days or not at all. 5 Test results The results of the tests and questionnaires have been organized according to the appropriate research questions.

5.1 SRT results of measured Swahili bilingual ability