The two groups, with a list of neighbours for each word, are as in the table below.
|deem||deed, deep, deer, seem, teem||bash||base, bask, bass, bast, bath, bosh, bush, cash, dash, gash, hash, lash, mash, rash, sash, wash|
|maul||caul, gaul, haul, mail, mall, marl||mope||cope, dope, hope, lope, mode, moke, mole, mops, more, mote, move, nope, pope, rope, tope|
|romp||pomp, ramp, rome, roup, rump||lust||bust, dust, gust, just, last, lest, list, lost, lush, must, oust, rust|
On the basis of the studies reviewed in the textbook that have included tasks with low frequency target words, then, we would expect a facilitatory effect of neighbourhood size in naming tasks, so that if participants had to read aloud one of the target words, then they should do so faster for bash, mope, and lust, because these have larger neighbourhoods than deem, maul, and romp.
On the other hand, when the task requires participants to uniquely identify the target, as in progressive demasking, then we would predict an inhibitory effect of neighbourhood size, so that the words with smaller neighbourhoods (and fewer close competitors for recognition) will be recognised faster.
These neighbourhoods were obtained from the R statistical programme with the library vwr using the command hamming.neighbors from that library, with the database english.words, as follows: hamming.neighbors('bash', english.words)[1:1]