3.2 Informal questions for the subjects of the Recorded Text Test
After the subject had listened to the text from a village other than his or her own, we asked four questions about the person who gave the text. The purpose of these questions was to sound out
attitudes and opinions about the different dialects, concerning prestige, if the dialect is considered easy to understand, etc.:
1 Have you heard someone who speaks like this before? Where did you hear that person? 2 Where do you think the speaker comes from?
3 Does he speak your language well? 4 How much of what he said did you understand - a lot, about half, only a little?
We cannot consider the answers to these questions to be entirely reliable. Some subjects could not or would not answer the questions. Some subjects were given the answer for instance, where the speaker
was from by the person who was translating for us. Other subjects clearly did not have an opinion but felt pressured to give an answer, any answer.
We make here a few observations about the experience of asking these questions. First of all, since we screened out subjects who had had previous contact with speakers of other dialects, questions 1 and 2
were outside their realm of experience, and therefore made the subject uncomfortable because he or she did not know how to respond. This usually served as a verification that the subjects hadnt had previous
contact with speakers of these dialects. Several of the people informally helping us to translate, and anxious to prove their worth as someone knowing something about their language, would answer the
question for the student. Some students did have ideas about where the person on the tape might have been from, which was interesting from the point of view that there are certain conceptions about
speakers of other dialects, even in the minds of those who have not had much contact with them. It was also interesting that the adults who served as translators during the testing were almost always able to
accurately identify where the story-teller was from. Occasionally, we learned that a subject did know someone who spoke that speech variety after all. For example, this is how we discovered that there
were one or two families from Gao living in Sawani.
Question number 3 might have yielded some interesting results for the study of language attitudes; however, it seemed that the students were reticent to say anything that might have been critical of the
person telling the story, and who was clearly older than they; therefore, the students almost always answered that the story-teller spoke the language well. The answers to question 4 seemed to reflect
more the general performance of the student in the classroom, rather than on that particular text. This was revealed by observing the interaction between the student and the teacher who often acted as the
translator, or sat next to the translator out of curiosity or to ensure sterling performance on the part of his students.
3.3 Results of lexicostatistical analysis
Nicolaï 1979b has already carried out cognate studies of Songhay to draw certain conclusions about the relatedness of the Songhay dialects, their classification into language families, and the predicted
comprehension between those dialects. Our purpose in gathering word lists was first of all, to add to the body of data collected, and secondly, to have lexical data from each test site of the Recorded Text
Test. The analysis resulting from the comparison of apparent cognates is useful as a check on the results of the Recorded Text Test. We must underline the fact that these figures are not from studies of
cognates, but from words which in their surface form appear to have a common root, that is, they are lexically similar.
Table 3.3 shows the lexical similarity percentages for word lists elicited in the villages visited. To find the lexical similarity between word lists from two villages, scan down the column of figures directly
below the name of the first village until you reach the number in the row of the second village. These figures were calculated using the WordSurv program, Version 2.4, which counts the number of
lexically similar words in two word lists and divides by the total number of words common to both lists. The numbers 4 through 10 and words which appear to have been borrowed from other languages
were excluded from this calculation. See Appendix E for the entire word list.
Table 3.3 - Percentages of lexical similarity
Tanda
Dendi
93 Niki Beri
Zarma, Dosso
92 95
Bardouga
Zarmaganda
90 93
94 Goria
Zarma, W. Bank
89 92
93 96
Maloum Beri
Kurtey
89 92
93 93
91 Namarigoungou
Songhay, E. Bank
88 89
91 94
92 93
Dibilo
Songhay, Tera
87 87
89 94
92 90
93 Fantio
Songhay, Kaado
85 86
86 91
90 91
89 90
Sawani
Wogo
Joseph Grimes 1988 points out that while there is a correlation between low lexical similarity 60 and low intelligibility, high lexical similarity is not an accurate predictor of high inherent intelligibility.
The lowest percentage of lexical similarity that we measured was 85 between the speech of Tanda and the speech of Sawani. This simply confirms the need for the intelligibility testing which we
performed.
3.4 Dialect preference