Языки стран Дальнего Востока, Юго-Восточной Азии и Западной Африки

15 Tsankova A. D. | Quantitative Survey on the Correlation of Basic Factors for the Functioning... Stage IV Correlational analysis The presented linguistic factors with their exponents were defined as independent, or the predictor variables, and the degree of regularity in the usage of the aspect-tense markers was defined as the dependent, or the outcome, target variable. Before implementing the statistical analysis, the independent variables were tested for outliers, residuals and multicollinearity , and proved to be eligible for carrying out correlational analysis. We applied SPSS Statistics multiple regression method to evaluate the influence of these factors on the explication of the aspect-tense markers. Series of analyses were carried out, exploring different combinations of the predictor variables to find out the most determinative models of correla- tion for every separate marker 了 , 过 and 着 . Stage V Interpretation of the research results: I. Variation analysis of the markers 了 1 and 了 2 The marker 了 is presented in Modern Chinese by two homonymous variants — the so-called 了 1 (basically a perfective marker), used in post- verbal position, and 了 2 (basically a marker of currently relevant state), used in phrase-final position. In cases when the verb is in final position without any argument, the markers overlap structurally and sometimes functionally, depending on the semantics of the predicate and the situation of speech. Accordingly, we apply separate analyses for the two different variants of 了 . 1. The appearance of 了 1 (post-verbal position, followed by arguments) covers 340 positions, which were analyzed as a separate part of the whole data. Combining the factors in different correlational models, we defined a representative correlational model with relatively high degree of determination to the resultative variable. The most important statistical values, highlighted in the tables, are as follows: R — The cumulative correlation coefficient. It measures the degree of prediction of the dependent variable. There are three standard levels of prediction (i.e. dependence on the factors) — low, medium and high.