From the historical viewpoint, cognates are forms in related languages that descended from a single form which existed at a previous point in time when the languages were one. Cognates are the daughter forms of the same parent form. Cognates can be determined accurately only through application of the comparative method. The approach taken here is to first classify cognates based on their appearance. The term apparent cognate better reflects the inaccuracies of this inspection method of determining cognation. Then the COMPASS module may be used to apply some insights of the comparative method to converge on a more historically accurate analysis.
The cognate analysis performed by WORDSURV is simply a tally of the decisions made by the linguist in the DATABASE module of the program. It is up to the linguist to group the forms into cognate sets; the computer Chapter 5: Data Analysis 59 simply compares cognate set labels to add up how many times each pair of word lists shares a cognate set.
Wimbish, John S. 1989. WORDSURV: A Program for Analyzing Language Survey Word Lists, pages 58-59. Occasional Publications in Academic Computing, number 13. Dallas, TX: Summer Institute of Linguistics.