Three studies were carried out in order to account for development of word definitions. Study 1 was aimed at analysing the role of class inclusion skills and age (5-and 7-year-olds and adults) in production of definitions containing superordinate categorical terms. No differences were found between 7-year-olds who had passed a class inclusion task and those who had not passed it as regards number of definitions containing superordinates, while differences were found between younger and older children and between children and adults. Study 2 was aimed at obtaining some normative criteria on the 'goodness' of definitions provided in Study 1, by using adult judges. It was found that the best definitions are those which contain both categorical terms and specific information about the to-be-defined object. It was also found that adults may adjust their standard definitional criteria to the kind of interlocutor (i.e. a child or a Martian). Study 3 confirms that younger children's definitions fall far short of the adult informativeness and completeness criteria while, by the age of ten, such criteria are met. Overall results were interpreted as conforming to a progressive conventionalization of children's strategies for defining objects.

There is increasing interest for the use of surrogate end points in the evaluation of treatments in patients with liver disease, but adequate validation is seldom available. This study aimed to describe the different course of galactose elimination capacity in patients with alcoholic cirrhosis who continued to drink or abstained from alcohol consumption during follow-up, and to validate changes in galactose elimination as a surrogate end point for death from liver-related causes. Forty-five patients with alcoholic cirrhosis (22 who continued drinking throughout the study period, and 23 who stopped drinking and were abstinent throughout the study period) were retrospectively selected among patients who had galactose elimination capacity measured at 6-month intervals. During follow-up 10 drinkers and 3 abstainers died of liver-related causes (P = .025). Abstainers showed a transient improvement in galactose elimination capacity, followed by a decrease. Continuous drinkers showed a reduction from the beginning. According to Cox's regression analyses, persistent alcohol abuse and galactose elimination capacity were separately related to the risk of death, but, when a time-dependent model was fitted containing galactose elimination capacity and persistent alcohol abuse, only the former remained significant. This implies that variations in the risk of death occurring as a consequence of abstinence from alcohol consumption may be predicted from changes in galactose elimination capacity, and that the mechanisms through which abstinence influences survival are strictly linked to the mechanisms responsible for the changes in the test. Because of the strict association of decrease in galactose elimination capacity and short survival, as proved in several series, this observation represents adherence to the criteria requested for adequacy of a surrogate end point. In conclusion, in alcoholic cirrhosis the decrease in galactose elimination capacity is an adequate surrogate end point for death from liver-related causes, which is worth testing in other conditions and in response to other treatments.