Within the framework of the distinction between episodic and semantic
memory, it has been argued that these two memory systems are organised
in a hierarchical way. The hierarchical hypothesis assumes that episo
dic memory is a specific subsystem of semantic memory and therefore im
plies that episodic memory cannot exist without semantic memory. If th
is hypothesis is correct, it should be expected that (episodic) yes/no
recognition performance would improve in patients with preserved sema
ntic memory, following semantic encoding. In the present study we inve
stigated the influence of semantic encoding on recognition memory perf
ormance in a population of 28 aphasic patients (AA) and 14 normal cont
rols (NC). Experiment 1 considered recognition memory for semantically
unrelated items, whereas Experiments 2 and 3 assessed recognition mem
ory for semantically related items. In Experiment 3, but not in Experi
ment 2, subjects were explicitly instructed to make a semantic associa
tion between the items. AA were impaired, compared to NC, only on the
recognition memory performance of Experiment 1. The ability to make a
semantic association between two items was significantly and positivel
y correlated to the ability to recognise, in a subsequent test, those
same items. A further analysis showed that patients who were impaired
on the semantic association task did significantly worse on the recogn
ition task of Experiment 3 than NC and than patients who were unimpair
ed on the semantic association task. These findings are discussed in t
he context of memory deficits in aphasia and interpreted as giving sup
port to the view that episodic memory for an item is affected by the l
evel of semantic awareness of that same item.