Training of four Puerto Ricans and four non-Puerto Ricans to assess the portrayal of Puerto Ricans in a sample of ethnic literature using a holistic approach is described. Basic assumptions involved in a holistic reading or scoring are as follows: (1) readers judge each textual unit as a whole and also within the context of the range of categories of a particular instrument; (2) the group develops evaluative criteria from the content; and (3) the analytical task occurs after the readers have formed their initial perceptions. The training had three major objectives: to interpret the categories of the instrument, to develop criteria to determine the positive or negative image of the principal character, and to enable the reviewers to reach a reasonable degree of consensus in their analysis of the literature. An analytic scale provided specific data regarding the character's image. Three different responses to the same content during the early training period indicated the need for adjustment and further content clarification in order to achieve the purpose of the training: group consensus. Generally, the ratings varied more among individuals than between ethnic groups, with one exception. With the exception of a story that concerned indiscriminate sexual behavior, use of drugs, and confrontations with legal authority, the consensus of the readers drew closer, and an analysis of variance of the nine readers' ratings of six stories indicated a significant decrease in variability on each dimension from the onset of the training to its conclusion. Implications of cross-cultural training for the bilingual classroom are addressed. References and instructions for coding the image of principal characters are appended. (SW)