Texts were presented sentence by sentence (Experiment 1) or word by wo
rd (Experiment 2) at a fixed rate to subjects high or low in test anxi
ety, under various conditions: no interference, concurrent articulator
y suppression, and concurrent irrelevant speech (presented auditorily)
. High-anxiety subjects produced overt articulation more frequently th
an low-anxiety subjects, especially in the irrelevant speech condition
. The most salient finding was an interaction between anxiety and inte
rference on comprehension performance: under word-by-word-but not unde
r sentence-by-sentence-presentation, anxious subjects showed poorer co
mprehension than non-anxious subjects in both conditions known to inte
rfere with the articulatory loop (i.e. articulatory suppression, and i
rrelevant speech), but equivalent comprehension in the no interference
condition. These findings suggest (a) that the articulatory loop has
a special compensatory role for anxious individuals in reading compreh
ension, and (b) that the importance of this auxiliary mechanism is enh
anced when other strategies, such as regressive fixations and control
of reading speed, cannot be used.