Encoding and retention of semantic and phonemic information in short-term memory

A probe recognition task was used to evaluate the relative effectiveness of semantic and phonemic encoding in STM. On each trial a list of 10 words was presented at a rate of either 350, 700, or 1400 msec per word. Recognition was tested with a probe which could be a homonym, a synonym, or identical to one of the words in the list. The retention functions for all three probe types were similar in shape, supporting the hypothesis that semantic encoding is possible in STM. An interaction between type of encoding and rate occurred, indicating that encoding is a time-dependent serial process.