JOURNAL OF VERBAL LEARNING AND VERBAL BERAVIOR 5, 3 2 5 - 3 3 7 ( 1 9 6 6 )

The "Tip of the Tongue" Phenomenon

ROGER BROWN AND DAVID McNEILL 1
Harvard University, Cambridge, Massachusetts
The "tip of the tongue" (TOT) phenomenon is a state in which one cannot quite
recall a familiar word but can recall words of similar form and meaning. Several hundred
such states were precipitated by reading to Ss the definitions of English words of low
frequency and asking them to try to recall the words. It was demonstrated that while in
the TOT state, and before recall occurred, Ss had knowledge of some of the letters in the
missing word, the number of syllables in it, and the location of the primary stress. The
nearer S was to successful recall the more accurate the information he possessed. The re-
call of parts of words and attributes of Words is termed "generic recall." The interpreta-
tion offered for generic recall involves the assumption that users of a language possess the
mental equivalent of a dictionary. The features that figure in generic recall may be
entered in the dictionary sooner than other features and so, perhaps, are wired into a
more elaborate associative network. These more easily retrieved features of low-
frequency words may be the features to which we chiefly attend in word-perception.
"lnae features favored by attention, especially the beginnings and endings of words,
appear to carry more information than the features that are not favored, in particular
the middles of words.
Wi l l i am James wrote, i n 1893: "Suppose
we t ry to recall a forgotten name. The state
of our consciousness is peculiar. There is a
gap therein; but no mere gap. I t is a gap
t hat is i nt ensel y active. A sort of wrai t h of
the name is in it, beckoni ng us in a given
direction, maki ng us at moment s tingle with
the sense of our closeness and then l et t i ng
us sink back wi t hout the longed-for term. If
wrong names are proposed to us, this singu-
l arl y definite gap acts i mmedi at el y so as to
negate them. They do not fit into its mould.
And the gap of one word does not feel like
the gap of another, all empt y of cont ent as
bot h mi ght seem necessarily to be when de-
scribed as gaps" (p. 251).
The "t i p of the t ongue" ( TOT) state in-
volves a failure to recall a word of which one
has knowledge. The evidence of knowledge is
either an event ual l y successful recall or else
an act of recognition t hat occurs, wi t hout
a Now at the University of Michigan.
1966 by Academic Press Inc.
addi t i onal training, when recall has failed.
The class of cases defined by the conj unct i on
of knowledge and a failure of recall is a
large one. The TOT state, which James de-
scribed, seems to be a small subclass in which
recall is felt to be i mmi nent .
For several mont hs we watched for TOT
states i n ourselves. Unabl e to recall the name
of the street on which a relative lives, one
of us t hought of Congress and Corinth and
Concord and then looked up the address
and learned t hat it was Cornish. The words
t hat had come to mi nd have certain proper-
ties in common with the word t hat had been
sought (the "t arget wor d") : all four begin
with Co; all are two-syllable words; all put
the pri mary stress on the first syllable. After
this experience we began put t i ng direct ques-
tions to ourselves when we fell into the TOT
state, questions as to the number of syllables
in the target word, its initial letter, etc.
Woodwort h (1934), before us, made a
record of dat a for nat ural l y occurring TOT
325
326 BROWN AND MCNEILL
s t at es and Wenzl ( 1932, 1936) di d t he same
for Ge r ma n wor ds. The i r r esul t s ar e s i mi l ar
to t hose we obt a i ne d and cons i s t ent wi t h t he
f ol l owi ng pr e l i mi na r y char act er i zat i on. Wh e n
compl et e r ecal l of a wor d is not pr e s e nt l y
possi bl e but is f el t t o be i mmi nent , one can
of t en cor r ect l y r ecal l t he gener al t ype of t he
wor d; generic r ecal l ma y succeed when par -
t i cul ar r ecal l fai l s. The r e seem t o be t wo
common var i et i es of gener i c r ecal l . ( a )
Somet i mes a p a r t of t he t ar get wor d is
r ecal l ed, a l et t er or t wo, a s yl l abl e, or affix.
Pa r t i a l r ecal l is neces s ar i l y al so generic si nce
t he cl ass of wor ds def i ned b y t he possessi on
of a n y part of t he t a r ge t wor d wi l l i ncl ude
wor ds ot her t ha n t he t ar get . ( b) Somet i mes
t he a bs t r a c t f or m of t he t a r ge t is r ecal l ed,
pe r ha ps t he f act t ha t i t was a t wo- s yl l abl e
sequence wi t h t he p r i ma r y st r ess on t he f i r st
s yl l abl e. The whol e wor d is r epr es ent ed i n
abst ract f or m recall but not on t he l et t er -
by- l e t t e r l evel t ha t cons t i t ut es i t s i de nt i t y.
The r ecal l of an a b s t r a c t f or m i s al so neces-
s ar i l y generic, si nce a ny such f or m def i nes a
cl ass of wor ds ext endi ng be yond t he t ar get .
We nz l and Woodwor t h had wor ked wi t h
smal l col l ect i ons of da t a for na t ur a l l y oc-
cur r i ng T OT st at es. Thes e d a t a were, for
t he most pa r t , pr ovi de d b y t he i nves t i gat or s ;
wer e col l ect ed i n an uns ys t e ma t i c f ashi on;
and were a na l yz e d in an i mpr es s i oni s t i c non-
qua nt i t a t i ve way. I t seemed t o us t ha t such
d a t a l ef t t he f act s of gener i c r ecal l i n doubt .
An occasi onal cor r es pondence bet ween a re-
t r i eved wor d a nd a t ar get wor d wi t h r es pect
t o number of s yl l abl es , st r ess p a t t e r n or i ni -
t i al l et t er is, a f t e r al l , t o be expect ed b y
chance. Sever al mont hs of " s e l f - obs e r va t i on
and a s ki ng- our - f r i e nds " yi e l de d fewer t han a
dozen good cases a nd we r eal i zed t ha t an
i mpr ove d me t hod of d a t a col l ect i on was
essent i al .
We t hought i t mi ght p a y t o " p r o s p e c t "
for T OT s t at es by r e a di ng t o S def i ni t i ons
of uncommon Engl i s h wor ds a nd as ki ng hi m
t o s uppl y t he wor ds. Th e pr oc e dur e was
gi ven a pr e l i mi na r y t es t wi t h ni ne Ss who
were i ndi vi dua l l y i nt er vi ewed for 2 hr s each. 2
I n 57 i ns t ances an S was, i n f act , " s e i z e d"
b y a T OT st at e. The si gns of i t wer e un-
mi s t a ka bl e ; he woul d a ppe a r to be i n mi l d
t or me nt , s omet hi ng l i ke t he br i nk of a
sneeze, and i f he f ound t he wor d hi s r el i ef
was cons i der abl e. Whi l e s ear chi ng for t he
t ar get S t ol d us al l t he wor ds t h a t came t o
hi s mi nd. He vol unt eer ed t he i nf or ma t i on
t ha t some of t hem r es embl ed t he t ar get i n
s ound but not in meani ng; ot her s he was
sur e wer e s i mi l ar i n meani ng b u t not i n
sound. The E i nt r ude d on S' s a gony wi t h
t wo ques t i ons : ( a ) How ma n y s yl l abl es has
t he t a r ge t wor d? ( b) Wh a t is i t s fi rst l et t er ?
Answer s t o t he fi rst ques t i on wer e cor r ect
i n 47% of al l cases and answer s t o t he
second ques t i on wer e cor r ect i n 51% of t he
cases. The s e out comes encour aged us t o be-
l i eve t ha t gener i c r ecal l was r eal and to de-
vi se a gr oup pr oc e dur e t ha t woul d f ur t her
speed up t he r at e of da t a col l ect i on.
METHOD
Subj ect s
Fifty-six Harvard and Radcliffe undergraduates
participated in one of three evening sessions; each
session was 2 hrs long. The Ss were volunteers
from a large General Education Course and were
paid for their time.
Word List. The list consisted of 49 words which,
according to the Thorndike-Lorge Word Book
(1952) occur at least once per four million words
but not so often as once per one million words.
The level is suggested by these examples: apse,
nepotism, cloaca, ambergris, and sampan. We
thought the words used were likely to be in the
passive or recognition vocabularies of our Ss but
not in their active recall vocabularies. There were
6 words of 1 syllable; 19 of 2 syllables; 20 of 3
syllables; 4 of 4 syllables. For each word we used
a definition from The American College Dictionary
(Barnhart, 1948) edited so as to contain no words
that closely resembled the one being defined.
Response Sheet. The response sheet was laid
off in vertical columns headed as follows:
Intended word (d- One I was thinking of).
(-- Not).
2 We wish to thank Mr. Charles Hollen for
doing the pretest interviews.
a TI P OF THE TONGUE" PHENOMENON 327
Number o] syllables (1-5).
Initial letter.
Words oJ similar sound. (1. Closest in sound )
(2. Middle )
(3. Farthest in Sound)
Words oJ similar meaning.
Word you had in mind i] not intended word.
Pr oc e dur e
We i ns t r uc t e d Ss t o t he f ol l owi ng effect.
I n t hi s e xpe r i me nt we ar e c onc e r ne d wi t h t h a t
s t at e of mi n d i n whi c h a per s on is una bl e t o
t h i n k of a wo r d t h a t he is cer t ai n he knows , t he
s t a t e of mi n d i n whi c h a wo r d s eems t o be on
t he t i p of one' s t ongue. Our t e c hni que f or pr e -
ci pi t at i ng s uc h s t at es is, i n gener al , t o r ead deft -
ni f i ons of u n c o mmo n wor ds a n d a s k t he s ubj e c t
to recal l t he wor d.
(1) We wi l l f i r st r e a d t he def i ni t i on of a l ow-
f r e que nc y wor d.
(2) I f you s houl d h a p p e n t o k n o w t he wo r d
a t once, or t hi nk y o u do, or, i f y o u s houl d s i mpl y
n o t k n o w i t , t he n t he r e is n o t h i n g f ur t he r f or
you to do a t t he mo me n t . 5 u s t wai t .
(3) I f y o u ar e una bl e to t h i n k of t he wo r d
b u t feel s ur e t h a t y o u k n o w i t a n d t h a t i t is on
t he ver ge of c omi ng back to y o u t he n y o u ar e
i n a T OT s t a t e a n d s houl d begi n a t once to fill
i n t he c o l u mn s of t he r es pons e sheet .
(4) Af t e r r e a di ng each def i ni t i on we will as k
wh e t h e r a n y o n e is i n t he T OT s t at e. An y o n e wh o
is i n t h a t s t a t e s houl d rai se hi s h a n d . Th e r e s t of
us wi l l t h e n wa i t unt i l t hos e i n t he T OT s t a t e
h a v e wr i t t e n on t he a ns we r s heet al l t he i n f o r ma -
t i on t he y ar e abl e to pr ovi de.
(5) Wh e n e ve r yone wh o ha s been i n t he TOT
s t a t e ha s si gnal l ed us t o pr oceed, we wi l l r e a d
t he t a r ge t wor d. At t hi s t i me, e ve r yone i s t o
wr i t e t he wo r d i n t he l e f t mos t c ol umn of t he
r es pons e sheet . Th o s e of y o u who h a v e k n o wn
t he wo r d si nce f i r st i t s def i ni t i on was r e a d ar e
a s ke d n o t t o wr i t e i t unt i l t hi s poi nt . Thos e of
y o u wh o s i mpl y di d n o t k n o w t he wo r d or wh o
h a d t h o u g h t of a di f f er ent wo r d will wr i t e n o w
t he wo r d we r ead. For t hos e of y o u who h a v e
been i n t he T OT s t a t e t wo e ve nt ua l i t i e s ar e pos -
sible. Th e wo r d r e a d ma y st r i ke y o u as def i ni t el y
t he wo r d y o u h a v e been seeki ng. I n t h a t case
pl ease wr i t e ' + ' a f t e r t he wor d, as t he i ns t r uc -
t i ons a t t he h e a d of t he c o l u mn di rect . Th e ot he r
possi bi l i t y is t h a t y o u will n o t be s ur e wh e t h e r
t he wo r d r e a d is t he one y o u h a v e be e n s eeki ng
or, i ndeed, y o u ma y be s ur e t h a t i t is not . I n t hi s
case y o u ar e a s ke d t o wr i t e t he si gn ' - - ' a f t e r
t he wor d. Some t i me s wh e n t he wo r d r e a d o u t
is n o t t he one y o u h a v e been s eeki ng y o u r a c t ua l
t a r ge t ma y c ome to mi nd. I n t hi s case, i n a ddi -
t i on t o t he mi n u s si gn i n t he l e f t mos t c ol umn,
pl ease wr i t e t he a c t ua l t a r ge t wo r d i n t he r i ght -
mo s t c ol umn.
( 6) No w we c ome t o t he c o l u mn ent r i es
t hems el ves . Th e fi rst t wo ent r i es, t he gues s as t o
t he n u mb e r of syl l abl es a n d t he i ni t i al l et t er , ar e
r equi r ed. Th e r e ma i ni ng ent r i es s houl d be filled
o u t i f possi bl e. Wh e n y o u ar e i n a T OT s t at e,
wor ds t h a t ar e r el at ed to t he t a r ge t word. do
a l mo s t a l wa ys c ome t o mi nd. Li s t t h e m as t h e y
come, b u t s e pa r a t e wor ds whi c h y o u t h i n k r e-
s embl e t he t a r ge t i n s o u n d f r o m wo r d s whi c h y o u
t h i n k r es embl e t he t a r ge t i n me a ni ng.
(7) Wh e n y o u h a v e f i ni shed al l y o u r ent r i es,
b u t bef or e y o u si gnal us to r e a d t he i n t e n d e d
t a r ge t wor d, l ook a ga i n at t he wo r d s y o u h a v e
l i st ed as ' Wo r d s of s i mi l ar s ound. ' I f possi bl e,
r a n k t hese, as t he i ns t r uc t i ons a t t he h e a d of t he
c ol umn di r ect , i n t e r ms of t he degr ee of t hei r
s e e mi ng r es embl ance t o t he t ar get . Thi s mu s t be
done wi t h o u t knowl e dge of wh a t t he t a r ge t ac-
t ual l y is.
(8) Th e s ear ch pr oc e dur e of a per s on i n t he
TOT s t a t e will s ome t i me s s er ve t o r et r i eve t he
mi s s i ng wo r d bef or e he has f i ni shed fi l l i ng i n t he
c ol umns a n d bef or e we r e a d o u t t he wor d. Wh e n
t hi s h a p p e n s pl ease ma r k t he pl ace whe r e i t
h a p p e n s wi t h t he wor ds " Go t i t " a n d do not
provide any more data.
RESULTS
Cl asses of Da t a
The r e wer e 360 i ns t ances , acr os s al l wo r d s a n d
al l Ss, i n whi c h a T OT s t a t e was si gnal l ed. Of
t hi s t ot al , 233 wer e pos i t i ve TOTs . A pos i t i ve T OT
i s one f or whi c h t he t a r ge t wo r d is k n o wn and,
c ons e que nt l y, one f or whi c h t he d a t a obt a i ne d can
be scor ed as a c c ur a t e or i naccur at e. I n t hos e cases
whe r e t he t a r ge t was n o t t he wo r d i nt e nde d b u t
s ome ot he r wo r d whi c h S f i nal l y r ecal l ed a n d wr ot e
i n t he r i g h t mo s t c ol umn hi s d a t a wer e checked
a ga i ns t t h a t wor d, hi s effect i ve t ar get . A ne ga t i ve
T OT is one f or whi c h t he S j udge d t he wo r d r e a d
o u t n o t t o ha ve been hi s t a r ge t a nd, i n addi t i on,
one i n whi c h S p r o v e d una bl e t o recal l hi s o wn
f unc t i ona l t ar get .
Th e da t a pr ovi de d by S whi l e he s ear ched f or
t he t a r ge t wo r d ar e of t wo ki nds : expl i ci t guesses
as t o t he n u mb e r of syl l abl es i n t he t a r ge t a n d t he
i ni t i al l et t er of t he t a r ge t ; wor ds t h a t c a me t o mi n d
whi l e he s e a r c he d f or t he t ar get . Th e wo r d s t h a t
c a me t o mi n d wer e classified b y S i nt o 224 wor ds
s i mi l ar i n s o u n d t o t he t ar get ( he r e a f t e r cal l ed " SS"
wor ds ) a n d 95 wor ds si mi l ar i n me a n i n g t o t he
328 BROWN AND MC NEILL
target (hereafter called "SM" words). The S's in-
formation about the number of syllables in, and
the initial letter of the target may be inferred from
correspondences between the target and his SS words
as well as directly discovered: from his explicit
guesses. For his knowledge of the stress pattern of
the target and of letters in the target, other t han
the initial letter, we must rely on the SS words
alone since explicit guesses were not required.
To convey a sense of the SS and SM words we
offer the following examples. When the target was
sampan the SS words (not all of them real words)
included: Saipan, Siam, Cheyenne, sarong, sanching,
and sympoon. The SM words were: barge, house-
boat, and junk. When the target was caduceus the
SS words included: Casadesus, AescheIus, cephalus,
and leucosis. The SM words were: fasces, Hippo-
crates, lictor, and snake. The spelling in all cases
is S's own.
We will, in this report, use the SM words to
provide baseline data against which to evaluate
the accuracy of the explicit guesses and of the SS
words. The SM words are words produced under
the spell of the positive TOT state but judged by
S to resemble the target in meaning rather than
sound. We are quite sure t hat the SM words are
somewhat more like the target than would be a
collection of words produced by Ss with no knowl-
edge of the target. However, the SM words make
a better comparative baseline than any other data
we collected.
General Problems o/ Analysis
The data present problems of analysis t hat are
not common in psychology. To begin with, the
words of the list did not reliably precipitate TOT
states. Of the original 49 words, all but zither suc-
ceeded at least once; the range was from one success
to nine. The Ss made actual targets of 51 words
not on the original list and all but five of these
were pursued by one S only. Clearly none of the
100 words came even close to precipitating a TOT
state in all 56 Ss. Furthermore, the Ss varied in
their susceptibility to TOT states. There were nine
who experienced none at all in a 2-hr period; the
largest number experienced in such a period by one
S was eight. In out data, then, the entries for one
word will not usually involve the same Ss or even
the same number of Ss as the entries for another
word. The entries for one S need not involve the
same words or even the same number of words as
the entries for another S. Consequently for the tests
we shall want to make there are no significance tests
that we can be sure are appropriate.
In statistical theory our problem is called the
"fragmentary data problem. '',~ The best thing to do
with fragmentary data is to report them very fully
and analyze them in several different ways. Our
detailed knowledge of these data suggests that
the problems are not serious for, while there is
some variation in the pull of words and the sus-
ceptibility of Ss there is not much variation in the
quality of the data. The character of the material
recalled is much the same from word to word and
S t oS.
Number of Syllables
As t he ma i n i t em of evi dence t ha t S i n a
TOT s t at e can recal l wi t h si gni f i cant success
t he numbe r of syl l abl es i n a t a r ge t wor d he
has not yet f ound we offer Ta bl e 1. The
ent r i es on t he di agonal ar e i ns t ances i n whi ch
guesses wer e cor r ect . The or der of t he means
of t he expl i ci t guesses is t he s ame as t he
or der of t he act ual numbe r s of syl l abl es i n
t he t ar get wor ds. The r a nk or der cor r el at i on
bet ween t he t wo is 1.0 and s uch a cor r el at i on
is si gni f i cant wi t h a p < .001 ( one- t ai l ed)
even whe n onl y fi ve i t ems ar e cor r el at ed. The
modes of t he guesses cor r es pond exact l y wi t h
t he act ual numbe r s of syl l abl es, for t he
val ues one t hr ough t hr ee; for wor ds of f our
and five syl l abl es t he modes cont i nue t o be
t hr ee.
Whe n all TOTs ar e combi ned, t he cont r i -
but i ons t o t he t ot al effect s of i ndi vi dual
Ss a nd of i ndi vi dual wor ds ar e unequal . We
have ma de an anal ys i s i n whi ch each wor d
count s but once. Thi s was accompl i s hed by
cal cul at i ng t he me a n of t he guesses ma de by
al l Ss for whom a pa r t i c ul a r wor d pr eci pi -
t at ed a TOT s t at e and t a ki ng t ha t me a n as
t he score for t ha t wor d. The new me a ns cal -
cul at ed wi t h al l wor ds equal l y wei ght ed wer e,
i n or der : 1. 62; 2. 30; 2. 80; 3. 33; a nd 3. 50.
Thes e val ues ar e cl ose t o t hose of Ta bl e 1
a nd rho wi t h t he act ual numbe r s of syl l abl es
cont i nues t o be 1.0.
We al so ma de a n anal ys i s i n whi ch each S
count s but once. Thi s was done by cal cul at -
i ng t he me a n of a n S's guesses for al l wor ds
3 We ~4sh to t hank Professor Frederick Mostel-
ler for discussing the fragmentary data problem
with us.
"TI P OF THE TONGUE" PHENOMENON 329
TABLE 1
ACTUAL NUMBERS OF SYLLABLES AND GUESSED
NUMBERS FOR ALL TOTs n r aE MA~N EXPERn~E~rr
Guessed numbers No
1 2 3 4 5 guess Mode Mean
1 9 7 1 0 0 0 1 1.53
"~ 2 2 55 22 2 1 5 2 2.33
= 3 3 19 61 10 1 5 3 2.86
=
"d 4 0 2 12 6 2 3 3 3.36
5 0 0 3 0 1 1 3 3.50
of one syllable, the mean for all words of
two syllables, etc. In comparing t he means
of guesses for words of different length one
can only use those Ss who made at least
one guess for each actual length to be
compared. I n the present dat a only words
of two syllables and three syllables pre-
cipitated enough TOTs to yield a sub-
stantial number of such matched scores.
Ther e were 21 Ss who made guesses for bot h
two-syllable and three-syllable words. The
simplest way to evaluate the significance of
the differences in these guesses is with the
Sign Test. I n onl y 6 of 21 mat ched scores
was the mean guess for words of two syl-
lables larger t han t he mean for words of
three syllables. The difference is significant
with a p ~---.039 (one-tailed). For actual
words t hat were only one syllable apart in
length, Ss were able to make a significant
distinction in the correct direction when the
words themselves could not be called to mind.
The 224 SS words and the 95 SM words
provide supporting evidence. Words of sim-
ilar sound (SS) had the same number of
syllables as the t arget in 48% of all cases.
This value is close to the 57% t hat were
correct for explicit guesses in the main ex-
peri ment and still closer to the 4 7 ~ correct
al ready reported for the pretest. The SM
words provide a clear cont rast ; onl y 20%
mat ched the number of syllables in the tar-
~get. We conclude t hat S in a positive TOT
state has a significant ability to recall cor-
rect l y the number of syllables in the word
he is trying to retrieve.
I n Tabl e 1 it can be seen t hat the modes
of guesses exactly correspond with the ac-
tual numbers of syllables in target words for
the values one through three. For still longer
t arget words (four and five syllables) the
means of guesses continue to rise but the
modes st ay at the value three. Words of
more t han three syllables are rare in En-
glish and the generic ent ry for such words
may be the same as for words of three
syllables; something like "t hree or more"
may be used for all long words.
Initial Let t er
Over all positive TOTs, the initial letter
of the word S was seeking was correctly
guessed 57% of the time. The pret est re-
sult was 51% correct. The results from the
main experiment were analyzed with each
word counting just once by entering a word' s
score as "cor r ect " whenever the most com-
mon guess or the onl y guess was in fact
correct; 62% of words were, by this reckon-
ing, correctly guessed. The SS words had
initial letters matching the initial letters of
the t arget words in 49% of all cases. We do
not know the chance level of success for this
performance but with 26 letters and many
words t hat began with uncommon letters
the level must be low. Probabl y the results
for the SM words are bet t er than chance
and yet the outcome for these words was
onl y 8% matches.
We did an analysis of the SS and SM
words, with each S counting j ust once. Ther e
were 26 Ss who had at least one such word.
For each S we calculated the proportion of
SS words matching the t arget in initial letter
and the same proportion for SM words. For
21 Ss t he proportions were not tied and in
all but 3 cases the larger value was t hat of
the SS words. The difference is significant
by Sign Test with p ~--- .001 (one-tailed).
The evidence for significantly accurate
generic recall of inital letters is even stronger
than for syllables. The absolute levels of suc-
cess are similar but the chance baseline
330 BROWN AND MC NEILL
must be much lower for letters than for syl-
lables because the possibilities are more
nume r ous .
Syllabic Stress
We did not ask S to guess the stress pat-
tern of the target word but the SS words
provide relevant data. The test was limited
to the syllabic location of the primary or
heaviest stress for which The American Col-
lege Dictionary was our authority. The num-
ber of SS words that could be used was
limited by three considerations. (a) Words
of one syllable had to be excluded because
there was no possibility of variation. (b)
Stress locations could only be matched if
the SS word had the same number of syl-
lables as the target, and so only such match-
ing words could be used. (c) Invented words
and foreign words could not be used because
they do not appear in the dictionary. Only
49 SS words remained.
As it happened all of the target words in-
volved (whatever their length) placed the
primary stress on either the first or the sec-
ond syllable. It was possible, therefore, to
make a 2 X 2 table for the 49 pairs of tar-
get and SS words which would reveal the
correspondences and noncorrespondences. As
can be seen in Table 2 the SS words tended
to stress the same syllable as the target
words. The X 2 for this table is 10.96 and that
value is significant with p < .001. However,
the data do not meet the independence re-
quirement, so we cannot be sure that the
matching tendency is significant. There
were not enough data to permit any other
analyses, and so we are left suspecting that S
TABLE 2
SYLLABLES RECEIVING PRIIVfARY STRESS IN TARGET
WORDS AND SS WORDS
Target words
1st syllable 2nd syllable
1st syllable 25 6
u~ 2nd syllable 6 12
in a TOT state has knowledge of the stress
pattern of the target, but we are not sure
of it.
Letters in Various Positions
We did not require explicit guesses for let-
ters in positions other than the first, but the
SS words provide relevant data. The test
was limited to the following positions: first,
second, third, third-last, second-last, and last.
A target word must have at least six letters
in order to provide data on the six positions;
it might have any number of letters larger
than six and still provide data for the six
(relatively defined) positions. Accordingly
we included the data for all target words
having six or more letters.
Figure 1 displays the percentages of let-
ters in each of six positions of SS words
which matched the letters in the same posi-
tions of the corresponding targets. For com-
parison purposes these data are also provided
for SM words. The SS curve is at all points
above the SM curve; the two are closest
together at the third-last position. The values
for the last three positions of the SS curve
quite closely match the values for the first
three positions. The values for the last three
positions of the SM curve, on the other hand,
are well above the values for the first three
0.55 -- o==-,o Words s/m//or in sound {SS)
o- - - o Words similar in meon/no($Ml l
0.50 -
-I
0.45 - -
0. 40 - -
0. 35 --
u_
o 0.30 --
~o.2s -
0.20 - . / ", , , , / -
~o.~5 - I -
n- I --
u.J 0.I0 - -----o,, I
n %
0. 05 - ~,,~I - -
o. oo I I T I I I
1st 2 nd 3 rd 3 rd- 2 nd- Last
Last Last
POSI TI ON IN WORD
FIG. 1. Percentages of letter matches between
target words and SS words for six serial positions.
"TIP OF THE TONGUE" PHENOMENON 331
positions. Consequently the relative superior-
ity of the SS curve is greater in the first
three positions.
The letter-position data were also analyzed
in such a way as to count each target word
just once, assigning each position in the tar-
get a single score representing the proportion
of matches across all Ss for that position in
that word. The order of the SS and SM
points is preserved in this finer analysis. We
did Sign Tests comparing the SS and SM
values for each of the six positions. As Fig. 1
would suggest the SS values for the first three
positions all exceeded the SM values with
p' s less than .01 (one-tailed). The SS" values
for the final two positions exceeded the SM
values with p' s less than .05 (one-tailed).
The SS values for the third-last position were
greater than the SM values but not signifi-
cantly so.
The cause of the upswing in the final
three positions of the SM curve may be
some difference in the distribution of in-
formation in early and late positions of En-
glish words. Probably there is less variety
in the later positions. In any case the fact
that the SS curve lies above the SM curve
for the last three positions indicates that S
in a TOT state has knowledge of the target
in addition to his knowledge of English word
structure.
Chunking of Su~xes
The request to S that he guess the initial
letter of the target occasionally elicited a
response of more than one letter; e.g., ex
in the case of extort and con in the case of
convene. This result suggested that some
letter (or phoneme) sequences are stored as
single entries having been "chunked" by long
experience. We made only one test for chunk-
ing and that involved three-letter suffixes.
I t did not often happen that an S produced
an SS word that matched the target with
respect to all of its three last letters. The
question asked of the data was whether
such three-letter matches occurred more often
when the letters constituted an English suffix
than when they did not. In order to deter-
mine which of the target words terminated
in such a suffix, we entered The American
College Dictionary with final trigrams. If
there was an entry describing a suffix ap-
propriate to the grammatical and semantic
properties of the target we considered the
trigram to be a suffix. There were 20 words
that terminated in a suffix, including fawn-
ing, unctuous, and philatelist.
Of 93 SS words produced in response to
a target terminating in a suffix, 30 matched
the target in their final three letters. Of 130
SS words supplied in response to a target
that did not terminate in a suffix only 5
matched the target in their final three let-
ters. The data were also analyzed in a way
that counts each S just once and uses only Ss
who produced SS words in response to both
kinds of target. A Sign Test was made of the
difference between matches of svffixes and
matches of endings that were not suffixes;
the former were more common with p = .059
(one-tailed). A comparable Sign Test for
SM words was very far from significance.
We conclude that suffix-chunking probably
plays a role in generic recall.
Proximity to the Target and Quality of In-
formation
There were three varieties of positive TOT
states: (1) Cases in which S recognized the
word read by E as the word he had been
seeking; (2) Cases in which S recalled the
intended word before it was read out; (3)
Cases in which S recalled the word he had
been seeking before E read the intended
word and the recalled word was not the same
as the word read. Since S in a TOT state of
either type 2 or type 3 reached the target
before the intended word was read and S in
a TOT state of type 1 did not, the TOTs of
the second and third types may be consid-
ered "nearer" the target than TOTs of the
first type. We have no basis for ordering
types 2 and 3 relative t o one another. We
332 B R OWN AND MC NE I L L
predicted t hat Ss in the two kinds of TOT
st at e t hat ended in recall (t ypes 2 and 3)
would produce more accurat e i nformat i on
about the t arget t han Ss in the TOT st at e
t hat ended in recognition ( t ype 1).
The prediction was tested on the explicit
guesses of initial letters since these were the
most complete and sensitive dat a. Ther e
were 138 guesses from Ss in a t ype 1 st at e
and 58 of these, or 4 2 ~ , were correct. There
were 36 guesses from Ss in a t ype 2 st at e
and, of these, 20, or 56%, were correct.
There were 59 guesses from Ss in a t ype 3
st at e and of these 39, or 66%, were correct.
We also anal yzed the results in such a way
as to count each word onl y once. The per-
centages correct were: for t ype i , 50%;
t ype 2, 62%; t ype 3, 63%. Finally, we per-
formed an analysis counting each S j ust once
but averagi ng together t ype 2 and t ype 3
results in order to bri ng a maxi mum number
of Ss into the comparison. The combining
action is justified since bot h t ype 2 and t ype
3 were st at es ending in recall. A Sign Test
of the differences showed t hat guesses were
more accurat e in the st at es t hat ended in
recall t han in the st at es t hat ended in recog-
nition; one-tailed p < .01. Suppl ement ary
analyses with SS and SM words confirmed
these results. We conclude t hat when S is
nearer his t arget his generic recall is more
accurat e t han when he is fart her from the
target.
Special interest at t aches to the results from
t ype 2 TOTs. I n the met hod of our experi-
ment there is nothing to guarant ee t hat when
S said he recognized a word he had real l y
done so. Perhaps when E read out a word, S
could not help thinking t hat t hat was the
word he had in mind. We ourselves do not
believe anyt hi ng of the sort happened. The
single fact t hat most Ss claimed fewer t han
five positive TOTs in a 2-hr period argues
agai nst any such effect. Still it is reassuring
to have the 36 t ype 2 cases in which S re-
called the intended word be]ore it was read.
The fact t hat 56% of the guesses of initial
letters made in t ype 2 st at es were correct is
hard-core evidence of generic recall. I t may
be wort h adding t hat 65% of the guesses
of the number of syllables for t ype 2 cases
were correct .
Judgments o] the Proximity of SS Words
The several comparisons we have made of
SS and SM words demonst rat e t hat when
recall is i mmi nent S can distinguish among
t he words t hat come to mind those t hat re-
semble the t arget in form from those t hat
do not resemble the t arget in form. Ther e
is a second ki nd of evidence which shows
t hat S can tell when he is getting close (or
" wa r m" ) .
I n 15 instances Ss rat ed two or more SS
words for comparat i ve si mi l ari t y to the t ar-
get. Our analysis cont rast s those rat ed "most
si mi l ar" (1) with those rat ed next most
similar (2). Since there were very few words
rat ed (3) we at t empt ed no analysis of t hem.
Similarity poi nt s were given for all the fea-
tures of a word t hat have now been demon-
st rat ed to pl ay a par t in generic r ecal l - - wi t h
the single exception of stress. Stress had to
be disregarded because some of the words
were i nvent ed and their stress pat t er ns were
unknown.
The probl em was to compare pai rs of SS
words, rat ed 1 and 2, for overall si mi l ari t y
to the target. We determined whether each
member mat ched the t arget in number of
syllables. I f one did and the other did not,
then a single si mi l ari t y poi nt was assigned
the word t hat mat ched. For each word, we
counted, beginning wi t h the initial letter,
the number of consecutive letters in common
with the target. The word havi ng the longer
sequence t hat mat ched the t arget earned one
similarity point. An exactly comparabl e pro-
cedure was followed for sequences st art i ng
from the final letter. I n sum, each word in
a pai r could receive from zero to three sim-
i l ari t y points.
We made Sign Test s compari ng the t ot al
scores for words rat ed most like the t arget
"TIP OF THE TONGUE" PHENOMENON 333
(1) and words rated next most like the tar-
get (2). This test was onl y slightly inappro-
priate since only two target words occurred
twice in the set of 15 and only one S re-
peated in the set. Ten of 12 differences were
in the predicted direction and the one-tailed
p = .019. I t is of some interest t hat sim-
ilarity points awarded on the basis of letters
in the middle of the words did not even go
in the right direction. Figure 1 has already
indicated t hat t hey also do not figure in Ss'
j udgment s of the comparat i ve similarity to
the target of pairs of SS words. Our conclu-
sion is t hat S at a given distance from the
target can accurat el y judge which of two
words t hat come to mind is more like the
target and t hat he does so in terms of the
features of words t hat appear in generic
recall.
Conclusions
When complete recall of a word has not
occurred but is felt to be imminent there is
likely to be accurate generic recall. Generic
recall of the abstract ]orm vari et y is evi-
denced by S' s knowledge of the number of
syllables in the target and of the location
of the pri mary stress. Generic recall of the
partial vari et y is evidenced by S' s knowledge
of letters in the t arget word. This knowledge
shows a bowed serial-position effect since
it is better for the ends of a word t han for
the middle and somewhat better for begin-
ning positions t han for final positions. The
accuracy of generic recall is greater when S
is near the target (complete recall is im-
minent) t han when S is far from the target.
A person experiencing generic recall is able
to judge the relative similarity to the target
of words t hat occur to hi m and these judg-
ments are based on the features of words
t hat figure in partial and abst ract form re-
call.
DISCUSSION
The facts of generic recall are relevant to
theories of speech perception, reading, the
understanding of sentences, and the organiza-
tion of memory. We have not worked out alI
the implications. I n this section we first at-
t empt a model of the TOT process and then
t ry to account for the existence of generic
memory.
A Model of the Process
Let us suppose (with Kat z and Fodor,
1963, and many others) t hat our long-term
memory for words and definitions is organ-
ized into the functional equivalent of a dic-
tionary. I n real dictionaries, those t hat are
books, entries are ordered alphabetically
and bound in place. Such an arrangement is
too simple and too inflexible to serve as a
model for a mental dictionary. We will sup-
pose t hat words are entered on keysort cards
instead of pages and t hat the car ds are
punched for various features of ".he words
entered. Wi t h real cards, paper ones, it is
possible to retrieve from the total deck any
subset punched for a common feature by
put t i ng a metal rod t hrough the proper hole.
We will suppose t hat there is in the mi nd
some speedier equivalent of this retrieval
technique.
The model will be described in terms of a single
example. When the target word was sextant, Ss
heard the definition: "A navigational instrument
used in measuring angular distances, especially the
altitude of sun, moon, and stars at sea." This deft-
nifion precipitated a TOT state in 9 Ss of the total
56. The SM words included: astrolabe, compass,
dividers, and protractor. The SS words included:
secant, sextet, and sexton.
The problem begins with a definition rather than
a word and so S must enter his dictionary back-
wards, or in a way that would be backwards and
quite impossible for the dictionary that is a book.
It is not impossible with keysort cards, providing
we suppose that the cards are punched for some
set of semantic features. Perhaps these are the
semantic "markers" that Katz and Fodor (1963)
postulate in their account of the comprehensioh of
sentences. We will imagine that it is somehow pos-
sible to extract from the definition a set of markers
and that these are, in the present case: "navigation,
instrument, having to do with geometry." Metal
rods thrust into the holes for each of these features
might fish up such a collection of entries as: astro-
labe, compass, dividers, and protractor. This first
334 BROWN AND MC NEI LL
~et ri eval , whi c h is i n r es pons e t o t he def i ni t i on,
mu s t be s emant i cal l y ba s e d a n d i t will not , t her ef or e,
a c c ount / or t he a ppe a r a nc e of s uc h SS wor ds as
sextet a nd sexton.
The r e ar e f our ma j o r ki nds of out c ome of t he
first r et r i eval a n d t hes e out c ome s c or r e s pond wi t h
t he f our ma i n t hi ngs t h a t h a p p e n to Ss i n t he "TOT
exper i ment . We will a s s u me t h a t a def i ni t i on of each
wor d r et r i eved is ent er ed on i t s car d a n d t h a t i t i s
possi bl e to check t he i n p u t def i ni t i on a ga i ns t t hos e
on t he car ds. Th e fi rst possi bl e out c ome i s t h a t
sextant is r et r i eved al ong wi t h compass a n d astro-
labe a n d t he ot he r s a n d t h a t t he def i ni t i ons ar e
specific e nough so t h a t t he one ent er ed f or sextant
r egi st er s as ma t c h i n g t he i n p u t a n d al l t he ot he r s
as n o t - ma t c h i n g . Thi s is t he case of cor r ect r ecal l ;
S h a s f o u n d a wo r d t h a t ma t c he s t he def i ni t i on a n d
i t is t he i nt e nde d wor d. The s econd possi bi l i t y is
t h a t sextant is n o t a mo n g t he wor ds r et r i eved a nd,
i n addi t i on, t he def i ni t i ons ent er ed f or t hos e r e-
t r i eved ar e so i mpr eci se t h a t one of t h e m ( t he
def i ni t i on f or compass, f or exampl e) r egi st er s as
ma t c h i n g t he i nput . I n t hi s case S t hi nks he ha s
f o u n d t he t ar get t h o u g h he real l y has not . Th e
t hi r d possi bi l i t y is t h a t sextant is n o t a mo n g t he
wor ds r et r i eved, b u t t he def i ni t i ons ent er ed f or t hos e
r et r i eved ar e specific e nough so t h a t none of t h e m
will r e ~s t e r a ma t c h wi t h t he i nput . I n t hi s case, S
does n o t k n o w t he wo r d a n d real i zes t he f act . Th e
a bove t hr ee out c ome s ar e t he c o mmo n ones a n d
none of t h e m r epr es ent s a T OT s t at e.
I n t he TOT case t he f i r st r et r i eval mu s t i ncl ude a
car d wi t h t he def i ni t i on of sextant ent er ed on i t
b u t wi t h t he wo r d i t sel f i ncompl et el y ent er ed. Th e
car d mi ght , for i ns t ance, ha ve t he f ol l owi ng i n-
f or ma t i on a b o u t t he wor d: t wo- s yl l abl es , i ni t i al s,
fi nal t. The e n t r y wo u l d be a p u n c h c a r d e qui va l e nt
of S _ _ _ _ T . Pe r ha ps a n i nc ompl e t e e n t r y of t hi s
s or t is J a me s ' s " s i ngul ar l y def i ni t e ga p" a n d t he
basi s f or gener i c recall.
The S wi t h a cor r ect def i ni t i on, ma t c h i n g t he i n-
put , a n d a n i ncompl et e wo r d e nt r y wi l l k n o w t h a t
he k n o ws t he wor d, will feel t h a t he a l mos t h a s i t ,
t h a t i t is on t he t i p of hi s t ongue. I f he is a s ke d
t o gues s t he n u mb e r of syl l abl es a n d t he i ni t i al
l et t er he s houl d, in t he case we h a v e i magi ned, be
abl e to do so. He s houl d al so be abl e to pr oduce SS
wor ds. The f eat ur es t h a t a ppe a r i n t he i nc ompl e t e
e n t r y ( t wo- s yl l abl es , i ni t i al s, a n d fi nal t ) can be
us ed as t he basi s f or a s econd r et r i eval . Th e s ubs e t
of car ds def i ned by t he i nt er s ect i on of al l t hr ee
f eat ur es wo u l d i ncl ude car ds f or secant a n d sextet.
I f one f e a t ur e wer e n o t us e d t h e n sexton woul d be
a dde d t o t he set .
Whi c h of t he f act s a b o u t t he TOT s t at e can n o w
be a c c ount e d f or ? We k n o w t h a t Ss wer e abl e, wh e n
t he y h a d not r ecal l ed a t ar get , t o di s t i ngui s h be-
t ween wor ds r es embl i ng t he t a r ge t i n s o u n d (SS
wor ds ) a n d wor ds r es embl i ng t he t a r ge t i n me a n i n g
onl y ( SM wor ds ) . Th e basi s f or t hi s di s t i nct i on i n
t he model woul d seem to be t he di s t i nct i on bet ween
t he fi rst a n d s econd r et r i eval s. Me mb e r s h i p i n t he
fi rst s ubs e t r et r i eved defi nes SM wor ds a n d me m-
ber s hi p i n t he s econd s ubs e t def i nes SS wor ds .
We k n o w t h a t wh e n S h a d pr oduc e d sever al SS
wor ds b u t h a d n o t recal l ed t he t a r ge t he coul d
s ome t i me s accur at el y r a nk- or de r t he SS wor ds f or
s i mi l ar i t y to t he t ar get . Th e model offers a n a c c ount
of t hi s r a nki ng pe r f or ma nc e . I f t he i ncompl et e e n t r y
f or sextant i ncl udes t hr ee f eat ur es of t he wo r d t he n
SS wor ds h a v i n g onl y one or t wo of t hes e f e a t ur e s
(e.g., sexton) s houl d be j udge d less si mi l ar t o t he
t ar get t h a n SS wor ds h a v i n g al l t hr ee of t h e m
(e.g., secant).
Wh e n a n SS wo r d ha s al l of t he f eat ur es of t he
i ncompl et e e n t r y (as do secant a n d sextet i n our
exampl e) wh a t pr e ve nt s i t s bei ng mi s t a ke n f or t he
t a r ge t ? Wh y di d n o t t he S who pr oduc e d sextet
t h i n k t h a t t he wo r d wa s " r i g h t ? " Becaus e of t he
def i ni t i ons. Th e f or ms me e t al l t he r e qui r e me nt s of
t he i ncompl et e e n t r y b u t t he def i ni t i ons do not
ma t c h.
Th e TOT s t a t e of t en ended i n r e c ogni t i on; i.e.,
S fai l ed to recall t he wo r d b u t wh e n E r e a d o u t
sextant S r ecogni zed i t as t he wo r d he h a d been
seeki ng. The mode l a c c ount s f or t hi s out c ome as
f ol l ows. Suppos e t h a t t her e is onl y t he i nc ompl e t e
e n t r y S____. T i n me mo r y , pl us t he def i ni t i on. Th e
E n o w s a ys ( i n effect ) t h a t t her e exi st s a wo r d
sextant whi c h has t he def i ni t i on i n ques t i on. Th e
wo r d sextant t he n sat i sfi es al l t he da t a poi nt s avai l -
abl e to S; i t ha s t he f i ght n u mb e r of syl l abl es, t he
r i ght i ni t i al l et t er , t he r i ght fi nal l et t er , a n d i t is
sai d to h a v e t he r i ght def i ni t i on. The r es ul t is r ecog-
ni t i on.
Th e pr opos e d a c c ount ha s s ome t es t abl e i mpl i ca-
t i ons. Suppos e t h a t F. wer e t o r e a d out , wh e n r e-
call fai l ed, n o t t he cor r ect wo r d sextant b u t a n
i n v e n t e d wo r d like sekrant or saktint whi c h s at i s -
fies t he i ncompl et e e n t r y as wel l as does sextant
i t sel f. I f S h a d n o t h i n g b u t t he i ncompl et e e n t r y a n d
E' s t e s t i mony to gui de h i m t he n he s houl d " r ecog-
ni ze" t he i nve nt e d wo r d s j u s t as he r ecogni zes
sextant.
Th e a c c ount we h a v e gi ven does n o t accor d wi t h
i nt ui t i on. Our i nt ui t i ve not i on of r ecogni t i on i s
t h a t t he f eat ur es whi c h coul d not be cal l ed wer e
act ual l y i n s t or age b u t less accessi bl e t h a n t he f ea-
t ur es t h a t wer e recal l ed. To s t a y wi t h our exampl e,
i nt ui t i on s ugges t s t h a t t he f eat ur es of sextant t h a t
coul d n o t be recal l ed, t he l et t er s be t we e n t he f i r st
a n d t he l ast , wer e ent er ed on t he c a r d b u t wer e
~TIP OF THE TONGUE '~ PHENOMENON 335
less "legible" than the recalled features. We might
imagine them printed in small letters and faintly.
When, however, the g reads out the word sextant,
then S can make out the less legible parts of his
entry and, since the total entry matches E' s word, S
recognizes it. This sort of recognition should be
"t i ght er" than the one described previously. Sekrant
and saktint would he rejected.
We did not try the effect of invented words and
we do not know how they would have been re-
ceived hut among the outcomes of the actual ex-
periment there is one t hat strongly favors the
faint-entry theory. Subjects in a TOT state, after
all, sometimes recalled the target word wi t hout any
prompting. The incomplete entry theory does not
admit of such a possibility. I f we suppose t hat the
ent ry is not S_____T but something more like
Sex tanT (with the italicized lower-case letters rep-
resenting the faint-entry section) we must still
explain how i t happens t hat the faintly entered,
and at first inaccessible, middle letters are made
accessible in the case of recall.
Perhaps it works something like this. The fea-
tures that are first recalled operate as we have
suggested, to retrieve a set of SS wo.rds. Whenever
an SS word (such as secant) includes middle letters
that are matched in the faintly entered section of
the target then those faintly entered letters become
accessible. The match brings out the missing parts
the way heat brings out anything written in lemon
juice. In other words, when secant is retrieved the
target entry grows from Sex tanT to SEx rANT.
The retrieval of sextet brings out the remaining
letters and S recalls the complete word--sextant.
I t is now possible to explain the one as yet un-
explained outcome of the TOT experiment. Subjects
whose state ended in recall had, before they found
the target, more correct information about i t than
did Ss whose state ended in recognition. More cor-
rect information means fewer features to be brought
out by duplication in SS words and so should mean
a greater likelihood that all essential features will
be brought out in a short period of time.
All of the above assumes that each word is en-
tered in memory just once, on a single card. There
is another possibility. Suppose that there are entries
for sextant on several different cards. They might
all be incomplete, but at different points, or, some
might be incomplete and one or more of them com-
plete. The several cards woul d be punched for
different semantic markers and perhaps for different
associations so t hat the entry recovered would
vary with the rule of retrieval. With this concep-
tion we do not require the notion of faint entry.
The difference between features commonly recalled,
such as the first and last letters, and features that
are recalled with difficulty or perhaps only recog-
nized, can be rendered in another way. The more
accessible features are entered on more cards or
else the cards on which they appear are punched
for more markers; in effect, they are wired into
a more extended associative net.
The Reason ]or Generic Recall
I n a d u l t mi n d s wo r d s a r e s t o r e d i n b o t h
vi s ua l a nd a u d i t o r y t e r ms a n d b e t we e n t he
t wo t h e r e a r e c o mp l i c a t e d r ul es of t r a ns l a -
t i on. Ge n e r i c r e c a l l i n v o l v e s l e t t e r s ( or pho-
n e me s ) , affi xes, s yl l abl es , a n d s t r es s l oc a t i on.
I n t hi s s e c t i on we wi l l di s cus s o n l y l e t t e r s
( l e gi bl e f o r ms ) a nd wi l l a t t e mp t t o e xpl a i n
a s i ngl e e f f e c t - - t h e s er i al pos i t i on ef f ect i n
t he r e c a l l of l et t er s . I t is n o t c l e a r h o w f a r
t he e x p l a n a t i o n c a n be e xt e nde d.
I n br i e f o v e r v i e w t hi s i s t he a r g u me n t .
T h e de s i gn of t he En g l i s h l a n g u a g e is s uc h
t h a t one wo r d is us ua l l y d i s t i n g u i s h e d f r o m
al l o t h e r s i n a mo r e - t h a n - mi n i ma l wa y, i . e. ,
b y mo r e t h a n a s i ngl e l e t t e r i n a s i ngl e pos i -
t i on. I t is c o n s e q u e n t l y possible t o r e c ogni z e
wo r d s wh e n one ha s n o t s t or e d t he c o mp l e t e
l e t t e r s e que nc e . T h e e v i d e n c e i s t h a t we do
n o t s t or e t he c o mp l e t e s e q u e n c e i f we do n o t
h a v e t o. We be gi n b y a t t e n d i n g c hi e f l y t o
i ni t i a l a nd f i nal l e t t e r s a n d s t or i ng t hes e.
T h e o r d e r of a t t e n t i o n a n d of s t or a ge f a vor s
t he ends of wo r d s be c a us e t he e nds c a r r y
mor e i n f o r ma t i o n t h a n t h e mi ddl e s . An i n-
c o mp l e t e e n t r y wi l l s e r ve f or r e c ogni t i on, b u t
i f wor ds a r e t o be p r o d u c e d ( or r e c a l l e d)
t h e y mu s t be s t or e d i n ful l . F o r mo s t wor ds ,
t hen, i t is e v e n t u a l l y n e c e s s a r y t o a t t e n d t o
t he mi d d l e l et t er s . Si nce e nd l e t t e r s h a v e
be e n a t t e n d e d t o f r om t he f i r st t h e y s houl d
a l wa ys be mor e c l e a r l y e n t e r e d or mo r e
e l a b o r a t e l y c o n n e c t e d t h a n mi d d l e l et t er s .
Wh e n r ecal l is r e qui r e d, of wor ds t h a t a r e
n o t v e r y f a mi l i a r t o S, as i t wa s i n our ex-
p e r i me n t , t h e e nd l e t t e r s s houl d o f t e n be
acces s i bl e wh e n t he mi d d l e a r e not .
I n b u i l d i n g p r o n o u n c e a b l e s e que nc e s t h e
En g l i s h l a ngua ge , l i ke al l o t h e r l a ngua ge s ,
ut i l i zes o n l y a s ma l l f r a c t i o n of i t s c o m-
b i n a t o r i a l pos s i bi l i t i e s ( Ho c k e t t , 1958) . I f
336 BROWN AND MC NEILL
a l anguage used all possi bl e sequences of
phonemes (or l et t er s) its wor ds coul d be
short er, but t hey woul d be much mor e vul -
ner abl e to mi sconst r uct i on. A change of any
si ngl e l et t er woul d r esul t in r ecept i on of a
di fferent word. As mat t er s ar e act ual l y ar-
r anged, most changes r esul t in no wor d at
al l ; for exampl e: textant, sixtant, sektant.
Our wor ds ar e hi ghl y r edundant and fai rl y
i ndest r uct i bl e.
Under wood ( 1963) has made a di st i nct i on
for t he l ear ni ng of nonsense syl l abl es be-
t ween t he " nomi na l " st i mul us whi ch is t he
syl l abl e pr esent ed and t he " f unct i onal " st i m-
ul us whi ch is t he set of char act er i st i cs of t he
syl l abl e act ual l y used t o cue t he response.
Under wood r evi ews evi dence showi ng t hat
col l ege st udent s l ear ni ng pai r ed- associ at es do
not l ear n any mor e of a st i mul us t r i gr am
t han t hey have to. If, for i nst ance, each of
a set of st i mul us t r i gr ams has a di fferent
i ni t i al l et t er, t hen Ss ar e not l i kel y to l ear n
l et t ers ot her t han t he first, si nce t hey do
not need t hem.
Fei genbaum ( 1963) has wr i t t en a com-
put er pr ogr am ( EPAM) whi ch si mul at es
t he sel ect i ve- at t ent i on aspect of ver bal l ear n-
i ng as wel l as many ot her aspect s. " . . .
EPAM has a noticing order ]or letters of
syllables, whi ch prescri bes at any moment a
l et t er - scanni ng sequence for t he mat chi ng
process. Because i t is obser ved t hat subj ect s
gener al l y consi der end l et t er s bef or e mi ddl e
l et t ers, t he not i ci ng or der is i ni t i al i zed as
fol l ows: first l et t er, t hi r d l et t er , second l et -
t er " (p. 304) . We bel i eve t hat t he di fferen-
t i al recal l of l et t er s in var i ous posi t i ons, re-
veal ed in Fi g. 1 of t hi s paper , is t o be
expl ai ned by t he oper at i on in t he per cept i on
of real wor ds of a rul e ver y much l i ke Fei gen-
baum' s.
Feigenbaum's EPAM is so written as to make
it possible for the noticing rule to be changed by
experience. If the middle position Were consistently
the position that differentiated syllables, the com-
puter would learn to look there first. We suggest
that the human tendency to look first at the begin-
ning of a word, then at the end and finally the
middle has "grown" in response to the distribu-
tion of information in words. Miller and Friedman
(1957) asked English speakers to guess letters for
various open positions in segments of English text
that were 5, 7, or 11 characters long: The percent-
ages of correct first guesses show a very clear serial
position effect for segments of all three lengths.
Success was lowest in the early positions, next
lowest in the final positions, and at a maximum in
the middle positions. Therefore, information was
greatest at the start of a word, next greatest at the
end, and least in the middle. Attention needs to be
turned where information is, to the parts of the
word that cannot be guessed. The Miller and Fried-
man segments did not necessarily break at word
boundaries but their discovery that the middle
positions of continuous text are more easily guessed
than the ends applies to words.
Is there any evidence that speakers of English do
attend first to the ends of English words? There is
no evidence that the eye fixations of adult readers
consistently favor particular parts of words (Wood-
worth and Schlosberg, 1954). However, it is not
eye fixation that we have in mind. A considerable
stretch of text can be taken in from a single fixa-
tion point. We are suggesting that there is selection
within this stretch, selection accomplished centrally;
perhaps by a mechanism like Broadbent's (1958)
"biased filter."
Bruner and O' Dowd (1958) studied word per-
ception with tachistoscopic exposures too brief to
permit more than one fixation. In each word pre-
sented there was a single reversal of two letters and
the S knew this. His task was to identify the
actual English word responding as quickly as pos-
sible. When the actual word was AVIATION, Ss
were presented with one of the following: VAIA-
TION, AVITAION, AVIATINO. Identification of
the actual word as AVIATION was best when S
saw AVITAION, next best when he saw AVIA-
TINO, and most difficult when he saw VAIATION.
In general, a reversal of the two initial letters made
identification most difficult, reversal of the last two
letters made it somewhat less difficult, reversal in
the middle made least difficulty. This is what should
happen if words are first scanned initially, then
finally, then medially. But the scanning cannot be
a matter of eye movements; it must be more central.
Selective attention to the ends of words should
lead to the entry of these parts into the mental
dictionary, in advance of the middle parts. However,
we ordinarily need to know more than the ends
of words. Underwood has pointed out (1963), in
connection with paired-associate learning, that
while partial knowledge may be enough for a stim-
ulus syllable which need only be recognized it will
~TIP OF THE TONGUE" PHENOMENON 337
not suffice for a response item which must be
produced. The case is simi!ar for natural language.
In order to speak one must know all of a word.
However, the words of the present study were low-
frequency words, words likely to be in the passive
or recognition vocabularies of the college-student Ss
but not in their active vocabularies; stimulus items,
in effect, rather than response items. If knowledge
of the parts of new words begins at the ends and
moves toward the middle we might expect a word
like numismatics, which was on our list, to be still
registered as NUM____I CS. Reduced entries of
this sort would in many contexts serve to retrieve
the definition.
The argument is reinforced by a well-known
effect in spelling. Jensen (1962) has analyzed
thousands of spelling errors for words of 7, 9,
or 11 letters made by children in the eighth and
tenth grades and by junior college freshmen. A
striking serial position effect appears in all his sets
of data such that errors are most common in the
middle of the word, next most common at the end,
and least common at the start. These results are as
they should be if the order of attention and entry
of information is first, last, and then, middle. Jen-
sen's results show us what happens when children
are forced to produce words that are still on the
recognition level. His results remind us of those
bluebooks in which students who are uncertain of
the spelling of a word write the first and last let-
ters with great clarity and fill in the middle with
indecipherable squiggles. That is what should hap-
pen when a word that can be only partially re-
called must be produced in its entirety. End letters
and a stretch of squiggles may, however, be quite
adequate for recognition purposes. In the TOT
experiment we have perhaps p!aced adult Ss in a
situation comparable to that created for children by
Jensen' s spelling tests.
T h e r e a r e t wo p o i n t s t o c l a r i f y a n d t he
a r g u me n t i s f i ni s hed. T h e Ss i n our e xpe r i -
me n t we r e col l ege s t ude nt s , a n d so i n o r d e r
t o o b t a i n wo r d s on t he ma r g i n of k n o wl e d g e
we h a d t o us e wo r d s t h a t a r e v e r y i n f r e q u e n t
i n En g l i s h as a whol e. I t is n o t our t h o u g h t ,
h o we v e r , t h a t t he T OT p h e n o me n o n oc c ur s
o n l y wi t h r a r e wor ds . T h e a b s o l u t e l o c a t i o n
of t h e ma r g i n of wo r d k n o wl e d g e is a f unc -
t i on of S' s a ge a n d e duc a t i on, a n d so wi t h
o t h e r Ss we wo u l d e x p e c t t o o b t a i n T OT
s t a t e s f or wo r d s mo r e f r e q u e n t i n Engl i s h.
F i n a l l y t h e ne e d t o p r o d u c e ( or r e c a l l ) a
wo r d is n o t t he o n l y f a c t or t h a t is ] i kel y t o
e n c o u r a g e r e g i s t r a t i o n of i t s mi d d l e l et t er s .
T h e a mo u n t of de t a i l ne e de d t o s pe c i f y a
wor d u n i q u e l y mu s t i nc r e a s e wi t h t h e t ot a l
n u mb e r of wo r d s k n o wn , t he n u mb e r f r o m
whi c h a n y one i s t o be di s t i ngui s he d. Cons e -
q u e n t l y t he g r o wt h of v o c a b u l a r y , as wel l as
t h e ne e d t o r ecal l , s houl d h a v e s ome p o we r
t o f or ce a t t e n t i o n i nt o t he mi d d l e of a wor d.
R E F E R E N C E S
BARNIIART, C. L. (Ed.) The American college dic-
tionary. New York: Harper, 1948.
BROADBENT, D. E. Perception and communication.
New York: Macmillan, 1958.
BRUNER, J. S., AND O'DowD, D. A note on the in-
formativeness of words. Language and Speech,
1958, 1, 98-101.
FEICENBAU~t, E. A. The simulation of verbal learn-
ing behavior. In E. A. Feigenbaum and J. Feld-
man (Eds.) Computers and thought. New York:
McGraw-Hill, 1963. Pp. 297-309.
HOCI~ETT, C. F. A course in modern linguistics. New
York: Macmillan, 1958.
J A~S, W. The principles of psychology, Vol. I.
New York: Holt, 1893.
JENSE~, A. R. Spelling errors and the serial-position
effect. J. educ. Psychol., 1962, 53, 105-109.
KATZ, J. J., AND FODOR, J. A. The structure of a
semantic theory. Language, 1963, 39, 170-210.
M I L L E R , G. A., A N D FRIEDI~AN, E L I Z A B E T H A. T h e
reconstruction of mutilated English texts. In-
lOTto. Control, 1957, 1, 38-55.
THORN'DIKE, E. L., AND LORGE, I. The teacher's word
book o] 30,000 words. New York: Columbia
Univer., 1952.
UNDERWOOD, B. J. Stimulus selection in verbal
learning. In C. N. Corer and B. S. Musgrave
(Eds.) Verbal behavior and learning: problems
and processes. New York: McGraw-Hill, 1963.
Pp. 33-48.
WENZL, A. Empirische und theoretische Beitriige
zur Erinnerungsarbeit bei erschwerter Wort -
findung. Arch. ges. Psychol., 1932, 85, 181-218.
WENZL, A. Empirische und theoretische Beitr~ge
zur Erinnerungsarbeit bei erschwerter Wort -
findung. Arch. ges. Psychol., 1936, 97, 294-318.
WOODWORTI~, R. S. Psychology. (3rd ed.). New York:
Holt, 1934.
WOODWORTICI, R. S., A-~'D SCIILOSBERG, H. Experi-
mental psychology. (Rev. ed.). New York:
Holt, 1954.
(Received January 15, 1965)