A Different Method of Approaching the Problem of Interlopers in TOT States

Everyone is familiar with the experience of not being able to recall a word which is felt sure to be known. This experience, known as the Tip of the Tongue (TOT) phenomenon, has intrigued psychologists for decades; however, there is still a deficit in research on the topic. My intent is to add slightly to the general pool of knowledge on TOT phenomena, specifically with regards to phonological blocking.
As early as 1893, James was describing the TOT phenomenon. Others also started to become interested in the topic during the first half of this century (e.g. Woodworth 1938, 1940). But despite this early treatment, the TOT phenomenon enjoyed no systematic exploration until the ground-breaking study by Brown & McNeill in 1966 (work prior to that time was largely anecdotal and unanalyzable, and consequently will not be treated here). The experimenters abandoned the �naturalistic� method of collecting TOTs, and instead developed a technique called �prospecting� (Brown & McNeill 1966). In order to elicit TOT states in the lab, they read definitions of rare English words to subjects, who indicated on a special questionnaire whether they knew the word, did not know the word, or were in a TOT state. If Ss were in the final condition, they proceeded to answer questions about their knowledge of the word, such as the number of syllables they thought it contained, the first letter, and other words with related sound or meaning (Brown & McNeill 1966). The aim of the study was to see what attributes about words in an unrecalled TOT state were accessible, and to postulate a theory about the nature of lexical storage based on the data.
One large problem with the method noted by the authors themselves was that it produced much fewer incidences of TOT than had been hoped or expected. Nevertheless, their study has become key in any investigation of TOT phenomena, one of the greatest lasting effects of it being the procedure. This method of prospecting has been favored by nearly all who set out to do TOT research (Kohn et al. 1987; Koriat & Lieblich 1974, 1975; Rubin 1975; Gardiner, Craik & Bleasdale 1973). Precious few daring souls have tried alternate methods, such as Yarmey (1973) who presented Ss with pictures of famous people whose names were to be recalled, or Grunberg, Smith, & Winfrow (1973) who had Ss produce their own blockages by talking on different categories, or the verbal paired associate tasks employed by Ryan, Petty, & Wenzlaff (1982), or Smith, Brown, & Balfour (1991).
In the last experiment, in fact, an entirely new approach was taken to the study of TOT phenomena. The experimenters fabricated animals which they called TOTimals, and had Ss learn a list of these, to be cued by each one�s picture. Later, they tested subjects on their recall of these TOTimals� names. This provided an exceptionally high yet accurate number of TOT states among Ss, who said that they felt no different from natural TOTs. Other benefits which they noted with their procedure included the fact that it allowed the control of otherwise unmanipulable variables that were present in the word definition procedure, such as recency/frequency of encounters with the target, associations made to the target by the subject, and prior knowledge of the target. In addition, they felt it �bridged the gap� between the artificial environments of other experiments and the �real world� (Smith, Brown, & Balfour 1991).
One focus of TOT research has been to find out what types of attributes are retrieved in partial or generic recall, and in so doing to develop a model of how lexical or other items are stored. Studies have generally shown that initial letters are among the most common items retrieved, often in a morpheme-like string (Rubin 1975). Another common attribute is the number of syllables (Lovelace 1987), although some suggest that this, as well as syllabic stress retrieval is more a result of educated guessing on the part of the subject based on his or her knowledge of English (e.g. a word is more likely to have two syllables than ten) (Perfect & Hanley 1992; Koriat & Lieblich 1974).
Two of the most prominent theories are the Incomplete Activation theory and the Blocking theory (Brown 1991). The former maintains that links between attributes in a lexical entry become weakened, and so certain ones do not become activated. This is helped by phonologically similar words which enable these missing attributes to be retrieved (Meyer & Bock 1992). The latter says that interlopers or blockers are wrong words that �get in the way� of retrieval by drawing attention away from the target (Jones & Langford 1987). And Ryan, Petty, & Wenzlaff (1982) have shown, by using a dual task setup, that TOT states do disrupt attention.
The effects and phenomena of so-called "interloper" words (here I use Langford & Jones� (1989) terminology despite the negative connotations) are a common point of examination. Opinions differ as to whether these interlopers block retrieval of TOT words or in fact facilitate retrieval (Brown 1991). Meyer & Bock (1992) theorized that the apparent blocking effect was the result of spreading word form activation, by suppressing phonological competitors as they become activated. Empirical data have been ambivalent (c.f. Meyer & Bock 1992; Perfect & Hanley 1992; Jones 1989; Jones & Langford 1987). One study (Perfect & Hanley 1992) even presented the neutral position that interlopers neither block nor facilitate retrieval, but rather show the processes at work.
Perfect & Hanley, in an attempt to test the two competing theories mentioned above, ran three experiments, the first two of which varied slightly from the initial studies of Jones (1989) and Langford & Jones (1987). These yielded dramatically different results in favor of incomplete activation. The third experiment, however, used the procedure of the previous experimenters and resulted in data which support blocking. This suggests that the means by which these interlopers are studied is extremely important to the outcome, and that the results may not necessarily be generalizable. The experimenters decided that this was caused by the very words used by Jones, however (Perfect & Hanley 1992).
Current debate notwithstanding, there are some aspects of interlopers which are universally agreed upon: if an effect is caused, interlopers phonologically related to the target have greater effect, and semantically-related interlopers show very little effect, if any at all (Perfect & Hanley 1992; Brown 1991; Maylor 1990; Jones & Langford 1987). Another consistency among studies of interlopers is the general method used. Langford & Jones (1987) essentially used a modified version of the much-favored Brown (1966) method-- before or after definitions of rare words were read, a cue word (interloper) was also read to Ss. This method and minor permutations thereof seems to be the only one which has been used in studies about interlopers (c.f. Meyer & Bock 1992; Perfect & Hanley 1992; Maylor 1990; Jones 1989).
I propose to test the effects of interlopers in a radically different way than what has thus far been done. By using a paired associate set-up much like that in the TOTimal experiment, the same factors which benefited their experiment can again be utilized. That is, the possibility of prior relationships the S has with the target can be controlled. In addition, and, I feel, possibly even more important, this set-up is more like real-world experience. The list to be learned serves as a type of mini mental lexicon, and would function in much the same way. When one searches memory to retrieve a known word to match a given concept, one may instead come upon related words which are also known and held in the lexicon (Meyer & Bock 1992). This is the case with the learned list. The predominant method, however, forces a possible interloper word upon the subject, unlike the reality in which many words can possibly serve as interlopers for a given target. This method simply strikes me as inherently unappealing.
The list employed will consist of Russian words (targets) paired with English equivalents (the concepts for which an item must be searched in the lexicon). The reason for this choice is my own familiarity with the language, coupled with the fact that it should be sufficiently abstruse to preclude former knowledge of elements on the part of the subject, including knowledge of stress and syllabification. Elements of the list will have phonological, semantic, both, or neither similarities, just as did cues in previous work. An example of a possible target set of words for the list is:
1
2
3
4 		�fifteen�
�fifty�
�to show�
�to display� 		[piat �nad cat]
[piat di �ciat]
[p ka �zat]
[�vi sta vit]
where 1 and 2 are both semantically and phonologically similar, 2 and 3 are phonologically similar only, 3 and 4 are semantically similar only, and 4 and 1 are neither phonologically nor semantically similar. Rather than presenting to the S one interloper, as is usually done, all the words on the list have the potential to be interlopers to each other. Theories can be drawn about lexical storage based on which words elicit TOTs, which words serve as interlopers, and for which targets, and the effects on each.
Ss will be chosen on the basis that they have no prior knowledge of Russian. They will be required to learn the list by means of a paired associate task to a certain criterion. After a delay, the subjects will be tested on the list. They will be instructed to report if they are in a TOT state for any word, and encouraged to try and work out the target aloud, and use clues such as, �It sounds like...� S�s will be required to repeat the test until they have resolved all TOTs.
Some very simple hypotheses come forth from this. The blocking theory predicts that the most TOT states will be elicited on words which are the most phonologically similar, that is, are similar in initial letter, stress pattern, and syllabification. In addition, these will have the strongest blocking effect, so consequently will take the greatest number of test trials to resolve. It is doubtful that there will be any interlopers of semantic similarity, but if any should arise, they will not contribute any discernible effects. The incomplete activation theory again predicts that there will be more incidence of TOTs on most phonologically similar words. However, it also predicts that these words will be the first to be resolved, the converse of the above. The same condition remains regarding semantically similar interlopers.
Brown (1991) makes a point that normal statistical analyses are often difficult to apply to TOT studies because of a �fragmentary data problem�. That is, it is often difficult to elicit TOT states in different subjects with the same targets. However, most experimenters are able to find some satisfactory means of computation.

References

Brown, A. 1991. A review of the tip-of-the-tongue experience. Psychological Bulletin 109.204-223.

Brown, R. & D. McNeill. 1966. The �tip of the tongue� phenomenon. Journal of Verbal Learning and Verbal Behavior 5.325-337.

Gardiner, J., F. Craik, & F. Bleasdale. 1973. Retrieval difficulty and subsequent recall. Memory and Cognition 1, 213-216.

Gruneberg, M., R. Smith, & P.Winfrow. 1973. An investigation into response blocking. Acta Psychologica 37.187-196.

James, W. 1893. The principles of psychology, Vol 1. NY: Holt.

Jones, G. & S. Langford. 1987. Phonological blocking in the tip of the tongue state. Cognition 26.115-122.

Kohn, S. et al. 1987. Lexical retrieval: the tip-of-the-tongue phenomenon. Applied Psycholinguistics 8.245-266.

Koriat, A. & I. Lieblich. 1974. A study of memory pointers. Acta Psychologica 41.151-164.

Lovelace, E. 1987. Attributes that come to mind in the TOT state. Bulletin of the Psychonomic Society 25.370-372.

Maylor, E. 1990. Age, blocking, and the tip of the tongue state. British Journal of Psychology 81.123-134.

Meyer, A. & K.Bock. 1992. The tip of the tongue phenomenon: blocking or partial activation. Memory and Cognition 20.715-726.

Perfect, T. & J. Hanley. 1992. The tip-of-the-tongue phenomenon: do experimenter-presented interlopers have any effect. Cognition 45.55-75.

Rubin, D. 1975. Within word structure in the tip-of-the-tongue phenomenon. Journal of Verbal Learning and Verbal Behavior 14.392-397.

Ryan, M., C. Petty, & R. Wenzlaff. 1982. Motivated remembering efforts during tip-of-the-tongue states. Acta Psychologica 51.137-147.

Smith, S., J. Brown, & S. Balfour. 1991. TOTimals: a controlled experimental method for studying tip-of-the-tongue states. Bulletin of the Psychonomic Society 29.445-447.

Woodworth, R. 1938. Experimental Psychology. NY: Holt.

Woodworth, R. 1940. Psychology. NY: Holt.

Yarmey, A. 1973. I recognize your face but I can�t remember your name: further evidence on the tip-of-the-tongue phenomenon. Memory and Cognition 1, 287-290.

Back
Models of Human Behavior