A set of 150 experimental personality items was constructed for an alternate form of the word association personality worksample developed by the Johnson O'Connor Research Foundation. The items were intended to possess several semantic properties hypothesized to facilitate discrimination between objective and subjective examinees. Specifically, the items were intended to be items that were contrasts, were concrete, had a physical or sensational reference, and were capable of generating alternative responses other than the contrast response. The items were also expected to elicit the keyed response from at least 15% of the examinees. Each of the experimental items was administered to approximately 100 examinees; a total of 1,541 examinees participated in the study. The importance of each of the subject property to an item's performance was examined by having three raters indicate the degree to which each stimulus-keyed response word pair possessed each of the properties. The mean ratings were then correlated with the items' correlations with total scores on Worksample 35 A (referred to as item-total correlations). Interactions among the ratings were also analyzed. Correlations revealed that the contrast property was the best predictor of item-total correlation with Worksample 35 A scores. A semantic analysis of Worksample 35 A items by M. Daniel, and administration instructions for the worksample are appended. Six data tables and one figure are included. (TJH)