Priming experiments have demonstrated the powerful, and automatic, nature of spreading activation. Subjects are shown prime/target pairs. Targets, in this instance, are either non-words, unrelated real words or related words. A prime, for example, might be HAMMER and the associated target either CAILS, SAILS or NAILS, or the prime might be DOCTOR and the target either DURSE, PURSE or NURSE. Subjects have, as usual, to make a lexical decision and in this case to indicate, as fast as they can, whether the target is a real word or not. In this instance, if the prime was DOCTOR and the target NURSE, or the prime was HAMMER and the target NAILS they will make their decision significantly faster than in either of the other two options in each pair test. This is taken to indicate that sight of the word DOCTOR, for example, excited all related items so that NURSE was already somewhat activated even before it was seen, making its recognition easier, or at least faster, compared, for example, to PURSE, which was not activated by DOCTOR.

Priming has also been used to demonstrate that we hold ‘free’ and ‘bound’ morphemes separately, as whole items, in our semantic lexicons. (A free morpheme may be a word – for example book, enjoy, free, pain, ground. A bound morpheme is a morpheme which must be bound to a free morpheme to make sense, for example, -ed, -ing, -ment, un-, -ly.) An experimental technique to investigate the management of bound and free morphemes might go like this: Subjects (these patient people are often psychology students at the mercy of their professors, and fodder for their endless experimentation) see primes which are free morphemes and targets which may be these words with bound morphemes attached to them, or unrelated words, or non-words. As so often, subjects must decide whether targets are real words or not. A prime might be PAIN, for example, and targets either PAINED, PAINT or TPINA, or a prime might be SICK and target either SICKLY, SICKLE or SELCKI. Subjects, it is found, are faster when identifying the root + affix (eg PAIN - - - PAINED or SICK - - - SICKLY) than the real but unrelated word alternative target if that is shown (eg PAIN - - - PAINT or SICK - - - SICKLE). Similarly, CAR will prime CARS but not CARD; SEE will prime SEEN but not SEED. This is in spite of their orthographic and phonological similarities. In other words we are not ‘primed’ by previous sight of a similar letter string but we are significantly primed by sight of a morpheme (PAIN as in PAINED but not as in PAINT; CAR as in CARS but not as in CARD). (e.g. Murrell & Morton 1974)

Interestingly, the phenomenon of ‘floating morphemes’ confirms the conclusion that we hold morphemes as such within our mental lexicon and only concatenate, or unite, them in the post-lexical assembly phase of constructing language. In other words we find the relevant roots and bound morphemes separately in the lexicon, fetch them out separately and put them together appropriately outside the lexicon as we assemble our utterances. We don’t, in other words, store compounded words like, for example, ‘meaningful’ or ‘meaningless’ in our lexicons – we store ‘meaning’, ‘-ful’ and ‘-less’ separately and put them together as necessary en route to saying (or writing) the whole word. Sometimes we don’t quite get this right and may say things like: ‘a plumful of hats’ [a hatful of plums] ‘he was cooked spotting dick’ [he was cooking spotted dick] or ‘she writes her slanting’ [she slants her writing]. I once announced to a roomful of goat breeders ‘The winner numbing is …’. We do not usually transpose random bits of words, but whole morphemes. We do not tend to do things like ‘he was cooted spocking dick’.

The question of whether reading is visually or phonically managed has been extensively researched by priming methods, as we have already glimpsed (& see Taft 1991). Morton (1979) used priming to explore the independence from each other of possible auditory and visual access systems to the semantic system (and see appendix 1) as follows: Subjects heard some prime words spoken but read others from the screen before them. They also heard some target words spoken but read others on the screen. Some targets were real words unrelated to the prime words and some were non-words, but some target words (the ones the experimenter was really interested in) were exactly the same as the prime. Subjects had to indicate whether they thought targets were real words or not. Morton found that when prime and target were identical and either both auditorily presented or both visually presented reaction time was less than when auditory prime presentation was followed by visual target presentation or vice versa. In other words subjects were ‘primed’ when identical prime and target were both spoken, or both read, but not primed when one was spoken but the other written or vice versa. Morton concluded that there must be separate and independent visual and auditory systems accessing the semantic system. Priming experimentation enabled him to infer that the black box contains visual pathways from text to meaning, and phonological pathways from sound to meaning, and that these pathways are distinct and, at least in ‘pure’ circumstances, function completely separately.