1. Collège d'enseignement secondaire is the middle school of the French educational system. Students from the C.E.S. in this study include ages 11 to 16 with a mode age of 14. Adults are defined as anyone who has left the lycée. In this study, ages of adults range from 20 to 72 years with a mode age of 24. Lycée is the French high school. Students from the lycée in this study include ages 15 to 20 with a mode age of 16. The research design was constructed to investigate both age and region as independent variables. No significant difference was found attributable to region.

2. No differences were suggested for the variable sex in the post-hoc analysis. Delisle (1982) also noted no difference based on sex. It should be noted that in both cases this referred to the sex of the native judge. Research is needed into the evaluation of errors made by learners of different sexes. Certain errors may be tolerated more when made by males than by females, for instance.

3. SPSS is a package computer program: Statistical Package for the Social Sciences. NY: McGraw-Hill, 1975.

4. When running numerous correlations on the same date, there is, of course, an increased possibility of error, in that at the .01 level, significance will be found falsely in 1 out of 100 cases. The research reported here may be criticized on this basis, however, the patterns of similarities and differences observed in the ordering of errors and the replication of findings of the pilot study in the main study make this weakness appear less troublesome. Furthermore, the main research questions were examined through only a few correlation procedures. Additional correlations were then used for post-hoc analyses as attempts to gain insights into possible variables underlying the difference attributed to age. The patterns in sensitivity to gender error revealed in this research are intriguing and certainly merit further controlled investigation.

5. Since the voice on the tape was that of an American female, any expectations for learner speech must be considered expectations for American females. It is not unlikely that different expectations would exist for different ethnic groups and perhaps for different sexes.

6. Lack of correlation does not prove difference. Examination of the hierarchies and the scatterplot does, however, reveal a certain difference in pattern. Difference in pattern (noted here and elsewhere in this article) is not to be taken as indicative of a difference that is statistically significant.

7. This interpretation was first suggested to me by Samuel N. Rosenberg, Indiana University-Bloomington, during a discussion of this research.

8. For example, the profession of a retired worker in his seventies must be rated for prestige at the time when the work was being performed. The job might have much less prestige at present than formerly.

9. This opinion was expressed to me by Maurice Gamier, French sociologist, Indiana University-Bloomington.

10. I would like to thank Professor Maurice Gamier for his help with this classification.

11. The relationship between adult occupation and education: for males r_s =.79**, r =.71**, for females r_s =.44**, r =.42**. (Correlation coefficient of age left school with socio-economic index from the general French population, according to the results of ongoing research by Maurice Gamier, Indiana University, males =.49, females =.53). The relationship between father's occupation and mother's occupation was r_s =.37**. For students between class of origin and class of destination: father's occupation with student's career aspiration, r_s =.36**, mother's occupation with student's career aspiration, r_s =.48**. Students were thus considered on the whole to be from families of homogeneous social class and to aspire to occupations in that social level.

12. A good introductory discussion of these requirements can be found in Robach (1974).