【作者】 戚志敏；

【导师】 陈宏；

【作者基本信息】 北京语言大学 ， 语言学及应用语言学， 2009， 硕士

【摘要】 本研究旨在运用多维多值潜在特质模型(MPLT模型)考查改进版HSK[初级]口语考试的潜在能力维度。首先使用联合极大似然估计法估计出模型的难度参数和被试能力值,并对这些值进行描述性统计分析和正态分布检验。最后采用残差分析和卡方检验两种方法对模型的拟合度进行检验。研究结果表明,模型在单个题目上和整体上都拟合良好,能够证实我们的研究假设,即改进版HSK[初级]口语考试测出了两种能力:听力理解力和口语能力。更多还原

【Abstract】 This thesis aims at analyzing the latent traits of HSK[Elementary] Spoken Test. Research data is drawn from the HSK[Elementary] Spoken Test held in November 2007. We choose the first two parts of the spoken test, 40 items in all, and 78 examinees. Our research hypothesis that there are two latent traits behind the spoken test. In order to verify this, we apply the Multidimensional Polytomous Latent Trait model, which was raised by Kelderman and Rijkes in 1994 to estimate the item-parameters and the two abilities. The Joint Maximum Likelihood Estimation method is applied to estimate the item parameters and the value of abilities. The fit-of-goodness is checked in two different ways: residuals and chi-square test. Results show that the model fits the data well and our hypothesis is basically supported.更多还原