02.04.2013 Views

Hindko and Gujari. c - SIL International

Hindko and Gujari. c - SIL International

Hindko and Gujari. c - SIL International

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

206<br />

Appendix A Methodologies<br />

Thus the RTT aims to be a closer reflection of a subject’s comprehension of the<br />

language itself, not of his or her memory, intelligence, or reasoning. 3<br />

In order to ensure that the text is a fair test of the intelligibility of the<br />

linguistic variety in focus, other speakers of the same local variety are asked to<br />

listen to the text <strong>and</strong> answer the questions. If they are able to do that, it is<br />

assumed that the story is an adequate sample of local speech, <strong>and</strong> that the<br />

questions are readily answerable by those for whom this speech form is native.<br />

This testing of subjects in their native speech form for the purpose of test<br />

validation is often referred to as hometown testing.<br />

It is possible that a subject may be unable to answer the test questions<br />

correctly simply because he does not underst<strong>and</strong> what is expected of him. This<br />

is especially true with unsophisticated subjects or those unacquainted with testtaking<br />

procedures. Therefore, a very short (pre-test) story with four questions is<br />

recorded in the local variety before beginning the actual testing, in order to<br />

acquaint the subject with the test procedures. If he is able to answer these pretest<br />

questions correctly, it is assumed that he is capable of functioning as a<br />

suitable subject. Each subject then participates in the hometown test in his<br />

native speech form before participating in recorded text tests in non-native<br />

varieties. Occasionally, even after the pre-test, a subject fails to perform<br />

adequately on an already validated hometown test. Performances of such<br />

subjects were eliminated from the final evaluation, the assumption being that<br />

uncontrollable factors unrelated to the intelligibility of speech forms are<br />

skewing such test results. 4 Thus, validated hometown tests are used for subject<br />

screening, in an attempt to ensure that recorded text testing results reflect as<br />

closely as possible the relative levels of comprehensibility of the speech forms<br />

represented.<br />

Test tapes are prepared for each location where a test is to be<br />

administered (test point). The hometown test tape includes (a) a short<br />

introduction in the local speech form to explain the purpose of the test, (b) the<br />

pre-test to orient <strong>and</strong> screen test subjects, (c) the hometown test text in its<br />

entirety followed by a repeat of the text, in short sections, with the relevant test<br />

questions <strong>and</strong> adequate pauses inserted in appropriate locations.<br />

The non-native test tapes are similar, omitting the screening elements<br />

from the hometown test tape. A short introduction in the local speech form<br />

reminding subjects of the test procedures precedes each recorded text. Then the<br />

recorded narrative in the non-native variety is given in its entirety, followed by<br />

the comprehension questions, now translated into the local speech form for that<br />

test point <strong>and</strong> with the relevant part of the non-native text repeated before each<br />

question.<br />

When speakers of one linguistic variety have had no previous contact<br />

with that represented in the recorded text, the test scores of ten subjects tend to<br />

be more similar — especially when scores are in the higher ranges. Such<br />

3 Recorded texts <strong>and</strong> associated comprehension questions will vary in<br />

terms of their relative difficulty <strong>and</strong> complexity or in terms of the clarity of the<br />

recording. Comparisons of RTT results from different texts need to be made<br />

cautiously <strong>and</strong> in the context of other indicators of intelligibility.<br />

4 For the purposes of this research, recorded text test subjects performing<br />

at levels of less than 80 percent on their hometown test were eliminated from<br />

further testing or were excluded from the analysis.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!