On experimental exploration of the mental lexicons.
These notes will, I hope, go some way to dissipating any lingering distrust of or distaste for cognitive psychology you may still harbour. These notes describe some of the more classic and historic priming experiments, and some experiments using techniques similar to priming. I hope they will entertain. I hope you will feel able to say after reading these notes; truly, cognitive psychology does reach the parts other disciplines cannot reach. It reaches, in fact, right into our ‘black box’.
As we have already hinted, priming (and similar techniques) have given us much of the most basic information we have as to how reading may actually be done inside our inscrutable unconscious. It must be admitted that radical critics of experimental techniques such as we are about to explore have claimed that these techniques are irredeemably artificial and reductionist, focussing, as they usually do, on the reading or manipulation of single words outside of any meaningful context. Such rarified and abstracted circumstances, it has been said, cannot model such a complex behaviour as reading meaningful text. Considerable similar work, though, has been done with whole sentences precisely to explore whether this is likely to be true or not (this work is wonderfully described in Taft 1991). As a result, Taft has concluded that:
‘Lexical access does not appear to be fundamentally different when a word is contained in a sentence compared to when it is presented in isolation’ and that ‘… decoding mechanisms involved in accessing words in context are essentially the same as those involved in accessing isolated words…’ (Taft 1991 p. 57).
Thus emboldened we will examine some historic experiments and the conclusions drawn from them. Examples are mainly taken from Ellis 1984, Rayner & Pollatsek 1989, Taft 1991 & Underwood & Batt 1986.
To get us into the mood we start with the entertaining ‘Stroop effect’. This is not a priming experiment as such, in that only ‘targets’ are presented. The long-suffering subject is asked to make a decision about a string of targets without first seeing a prime stimulus. Otherwise the technique is much the same, in that the time subjects take to make their decisions, and the accuracy of their responses, is measured. In one Stroop experiment subjects were asked to report only the colour of the target (which might be a word or a letter string or a drawing) and to ignore everything else. They might be told to indicate whether the target was yellow, blue, green or red, for example, as fast as they could. Among the targets would be the name of one colour, written in another – for example subjects might see the word RED, but written in green. They are supposed to ignore the word and report the colour in which the target is written, so should report GREEN. They mostly do this correctly, but it takes them significantly longer and the error rate is increased. In another Stroop experiment subjects were shown line drawings and asked verbally to report what each drawing represented. Among the target drawings were some which contained words; thus the subject might see a drawing of a dog with the word CAT written inside it, or a boat with the word HOUSE within it. In these circumstances subjects generally report the identity of the drawing correctly, but again it takes significantly longer and more errors are made.
This is interpreted as indicating that we automatically, and always, process words we see clearly. We do this, in fact, even when we have been specifically instructed to attend exclusively to another property of the target altogether, such as its colour or what it represents, and to ignore all other information on the screen. We can probably say, as a result of these findings, that, in the psycho-jargon, word processing is mandatory; it is an involuntary thing all fluent readers do unconsciously whether they will or no. Under the conditions of the Stroop experiments the processing of the words seen meant that, in effect, two pieces of information (RED and GREEN, or CAT and DOG) were arriving simultaneously. Subjects were momentarily confused and it took them an extra couple of hundred milliseconds to sort out which item was to be reported. The important conclusion was that both items were automatically processed, despite subjects trying not to do so.