The effects of word length and visual complexity on verbal reaction times.

THE

EFFECTS OF WORD LENGTH AND VISUAL COMPLEXITY ON VERBAL REACTION TIMES

DONALD G. BRENNAN and WALTER L. CULLINAN

University of Oklahoma, Oklahoma City

Thirty adult subiects learned to associate nonsense names varying in word length with nonsense visual stimuli varying in visual complexity. Simple reaction times ( SRTs), visual duration thresholds (VDTs), matching response latencies (MRLs), and naming response latencies (NRLs) were then obtained from these subjects. The data indicate that SRTs, VDTs, and NRLs are significantly related to word length and that VDTs and NRLs are significantly related to visual complexity. There is also a tendency for MRLs to increase with increases in word length, particularly for "no" responses. However, the effects of word length on VDTs, MRLs, and NRLs may be confounded with the number of trials needed to learn the paired associates or with the number of overlearning trials. The data are consistent with an interpretation that motor planning or some form of implicit speech process may be a part of the total time required for the naming response. The existence of an inverse relationship between the time it takes to respond with the name of an obiect and the frequency of occurrence of that word in the language has been demonstrated frequently (Boysen and Cullinan, 1971; Newcombe, Oldfield, and Wingfield, 1965; Oldfield and Wingfield, 1964, 1965; Oldfield, 1966; Wingfield, 1968). After investigating the nature of the relationship of word frequency with naming response latency, with visual duration threshold, and with a "yes"/"no" matching response latency using pictured stimuli, Wingfield (1968) concluded that the major source of variance in naming latencies for common and rare objects might be attributed to differences in the time needed to search for the objects' names following completion of perceptual identification. Oldfield and Wingfield (1965) speculated that words may be stored in the brain in such a way that access time for frequently occurring words would be shorter than for less frequently occurring words. The frequency of occurrence of words in the language, however, is confounded with word length in that common words tend to be shorter than rare words. The question arises then as to whether the naming response latency is affected by word length, as appears to be the case for the reading response latency (Eriksen, Pollack, and Montague, 1970), independently of word frequency. It may be that implicit speech occurs prior to the spoken utterance in 141

Downloaded From: https://jslhr.pubs.asha.org/ by a Universite Laval User on 03/25/2018 Terms of Use: https://pubs.asha.org/ss/rights_and_permissions.aspx

naming as well as in reading (Eriksen et al., 1970). Or it could be that longer utterances require longer motor planning responses prior to initiating speech. Brennan and Cullinan (1974) for example, have suggested that the naming latency should include, along with the time needed for perceptual identification and the search for the name, the time it takes to initiate the verbal response once the name has been located. The degree of visual complexity of stimuli used in verbal reaction time tasks would appear to be another important factor affecting response latency (Bisiach, 1966; Milianti and Cullinan, 1974). Fraisse and Elkin (1963), for example, reported that detailed drawings resulted in lower recognition thresholds than did outline drawings. Temporal differences in threshold would be expected to be reflected in differences in naming latency. The purpose of the present experiment was to investigate the effects of word length and degree of visual complexity of stimuli on some of the processing times in the naming response. The specific measures obtained were simple reaction time (SRT), visual duration threshold (VDT), matching response latency (MRL), and naming response latency (NRL).

METHOD Sub/ects Thirty graduate students in the Department of Communication Disorders, University of Oklahoma Health Sciences Center, were randomly assigned into three subject groups of 10 subjects each. All subjects were experienced in phonetic transcription and had normal speech, normal hearing, and normal or corrected to normal vision. Verbal Stimuli A preliminary group of 63 nonsense words was presented to eight observers, all of whom were members of the faculty or staff of the Department of Communication Disorders. Nine words were selected from the larger group for inclusion in this study according to the following criteria: (1) 75~ or more of the observers rated the word as having average ease of pronunciation; (2) the word reminded 25~ or less of the observers of a real word; and (3) each of the one-syllable words was the first syllable of one of the two-syllable words, and each two-syllable word comprised the first two syllables of one of the three-syllable words. To insure that syllable stress patterns were constant across stimuli, the unstressed schwa vowel was substituted for the vowel in the second syllable of the two- and three-syllable words. Thus, the first and third syllables were always stressed and the second syllable was always unstressed. The nine nonsense words selected were: /zov/, /zov~9/, /zov~9id/, /duv/,/duv~d/,/duwdib/,/mob/, [mobon/, and/mobonez/. 142 1ournal o[ Speech and Hearing, Research


19 141-155 1976

Visual Stimuli Forty-five visual stimuli were chosen from the Abstract Reasoning Subtest of the Differential Aptitude Tests, Form L, and were presented to six observers, graduate students in the Department of Communication Disorders who were instructed to rate each form on a three-point scale of visual complexity. Each observer was then asked to describe each form as completely as possible so that a person who had not seen the form would be able to draw it from the verbal description. These descriptions were tape recorded and the number of words used to describe each form was counted. Three forms representing each of the three levels of visual complexity were selected according to the following criteria: (1) Level 1 stimuli were judged as less complex than most by at least four of the six observers and required the least number of words to describe them; (2) Level 2 stimuli were judged as presenting average complexity by at least four of the six observers; and (3) Level 3 stimuli were judged as more complex than most by at least four of the six observers and required the most number of words to describe them. Visual stimuli for all three levels were selected by an experimenter so as not to be similar in form to the other two stimuli selected for each level. The test stimuli were prepared in the form of black line-drawn tracings of the nine visual forms with the overall size of the forms kept relatively uniform.

Apparatus The experimental situation and equipment were similar to those described by Milianti and Cullinan (1974) and Brennan and Cullinan (1974). A Harvard four-channel digital timer (Model 300-4T), lamp driver (Model 402), and an experimenter were in the control room of a two-room sound-treated suite. The subject, experimental assistant, and the exposure cabinet of a twofield Harvard tachistoscope (Model T-2B) were in the experimental room. A two-way intercom system allowed the experimenter to communicate with the subject.

Preexperimental Learning Task A flash card technique was used to teach the nine nonsense words as names for the nine nonsense visual forms. Subject Group I learned one-syllable names paired with Level 1 visual stimuli, two-syllable names with Level 2 stimuli, and three-syllable names with Level 3 stimuli. Subject Group II learned two-syllable names paired with Level 1 visual stimuli, three-syllable names with Level 2 stimuli, and one-syllable names with Level 3 stimuli. Subject Group III learned three-syllable names for Level 1 visual stimuli, onesyllable names for Level 2 visual stimuli, and two-syllable names for Level 3 stimuli. This arrangement of the stimulus pairs allowed for the evaluation of the main effects of word length and visual complexity using the Lindquist Type II Mixed Design ANOVA ( Lindquist, 1956). BRENNAN, CULLINAN:Verbal Reaction Times 143


The phonetic symbols for the sounds in the names were presented to each subject to insure that each subiect was familiar with the pronunciation associated with each symbol. The subiects were instructed to read the phonetic symbols on each flash card until all of the sounds were pronounced correctly on two consecutive trials. Each subject was then presented a set of flash cards containing the nine nonsense names and was instructed to pronounce each name aloud until he had pronounced all nine names correctly on three consecutive trials. Nine flash cards, each containing only a visual form on the side facing upward, were presented in random order to the subject. The subject turned each card over, exposing the visual form paired with its nonsense name, pronounced the name of the form aloud and proceeded to the next card. This procedure was repeated until all nine forms and names were presented. The stimulus pairs were reordered randomly and the subject attempted to name each form. The subject turned the card over to see if his naming response was correct and again pronounced the name of the visual stimulus aloud. This procedure was repeated until the subject had correctly named all nine visual forms on three consecutive trials. Overlearning trials were then presented until the subject named all stimuli correctly for five additional trials. A second series of overlearning trials followed for which subjects were instructed to name the visual stimuli as rapidly as possible. Trials were repeated until each subiect had three trials in which all stimuli were named correctly and each stimulus was named within three seconds. This completed the first day's learning session. On the following day, the subjects again were presented the stimulus pairs using the flash card technique and were encouraged to name the visual forms as rapidly as possible until they had correctly named the forms within three seconds per stimulus for five consecutive trials. After reaching this criterion, each subiect participated in the four experimental tasks, which were presented in random order to each subject. Simple Reaction Time Task The experimenter presented one of the nonsense names aloud and the subject was instructed to repeat the nonsense name to be sure that he heard the experimenter correctly. A "ready" signal was then given followed by a two- to three-second interval before a two-second presentation of a stimulus light. The subject was requested to produce the nonsense name as rapidly as possible upon seeing the signal light. Each subject received three practice trials with other nonsense syllables before the presentation of the nine experimental nonsense names in random order. Visual Duration Threshold Task Each subject was instructed to name the nonsense forms which would 144 Iournal o[ Speech and Hearing Research


19 141-155 1976

appear briefly on the viewing screen. He was instructed that if he was unable to see the pictured form he was to say "no" and the form would be shown again. A ready signal was given followed by a two- to three-second interval before the presentation of each stimulus picture. Each form was initially presented for 5 msec and the time of exposure was increased in 5-msec steps until the form was correctly named twice at the same exposure time. Each presentation of a visual form was immediately followed by a 1-sec masking stimulus to reduce the effects of visual afterimage. Four additional stimuli (copies of four of the nine experimental stimuli) were included bringing the number of stimuli to 13. These four were added to reduce the possibility that the subject might correctly guess the stimuli presented late in the sequence due to the process of elimination, but only the responses to the first appearance of each of these four were used in subsequent analyses. Before randomly presenting the test stimuli, three pictures were presented to familiarize the subject with the task. Visual duration thresholds (VDTs) were obtained for the nine test stimuli by recording exposure time directly from the digital timer.

Matching Task Before the presentation of each visual stimulus, the name of one of the visual forms was presented aloud by the experimenter, and the subiect was instructed to repeat the name to be sure that he had heard the experimenter correctly. The subiect was told that he was to say "yes" as quickly as possible if the name spoken was appropriate for the visual form which would appear on the viewing screen. The subiect was instructed to say "no" as rapidly as possible if the name was not appropriate for the form. The experimenter gave a ready signal followed by a two- to three-second interval prior to the presentation of each visual form. Three practice stimulus pairs were presented before the random presentation of the same 13 experimental stimuli used in the VDT task. Each stimulus was exposed for a duration of three seconds and was immediately followed by a dark poststimulus field. The stimuli were arranged so that there were five "yes" and five "no" responses for each of the nine forms for each subiect group. For the stimulus pairs which required a "no" response, the incorrect name which was given was always the same word length as the correct name for the visual form.

Naming Task Each subiect was instructed to name the visual forms as rapidly as possible. Three practice pictures were presented prior to the random presentation of the 13 visual forms used in the VDT task. The experimenter gave a ready signal followed by a two- to three-second interval before the three-second stimulus presentation.

Response Recording During the simple reaction time, matching, and naming tasks, subiects' Bru~NAN, Ctrta~lNXN: Verbal Reaction Times 145


verbal responses were picked up by an Electro-Voice cardioid microphone (Model 664) and recorded on Channel 1 of an Ampex two-channel tape recorder (Model 440). The start control of the digital timer which initiated the tachistoscopic presentation was wired in such a way as to produce simultaneously a stimulus voltage on Channel 2 of the same tape recorder. The tape-recorded samples were transferred to a Sanborn oscillographic chart recorder (Model 7702A) for the measurement of the simple reaction times (SRTs), matching response latencies (MRLs), and naming response latencies (NRLs). The criteria of measurement and the experimenter's reliability in making this type of measurement have been established and reported elsewhere ( Brennan and Cullinan, 1974). RESULTS

SRT, VDT, MRL, and NRL Measurements The mean SRT, VDT, MRL, and NRL measurements are displayed graphically in Figures 1 through 4, respectively, for each level of visual complexity and for each word length. Shown also are the corresponding standard errors

370

40

(a)

(a)

35 "10 tO

"~'.~=~o"350

oM

30 25

E

E

t'-

330

I Level I

.~_

I Level 2

1 Level 3

E

I

I

Level 1 Level 2 Level 3 Visual Complexity

t--

.~_ 40

(b)

~ ~

20

O c-

e,

Visual Complexity

~' 370

-o

(b)

a5

350 25 330

~

I

20

i

One Syllable TWOSyllable Three Syllable

I ~ I One Syllable Two Syllable Three Syllable

Word Length

FIctraE 1. (a) Mean simple reaction time and standard error of the mean for each level of visual complexity. (b) Mean simple reaction time and standard error of the mean for each word length. 146

Word Length

Ficurus 2. (a) Mean visual duration threshold and standard error of the mean for each level of visual complexity. (b) Mean visual duration threshold and standard error of the mean for each word length.

Journal of Speech and Hearing Research


19

141-155

1976

750

-E .~_ v

(a)

1300

700

1200

650

llO0

6O0 E 550

a Level 1

,t-.

Visual Complexity

._J

~" 750 o $ a=

1000

I Level 3

Level 2

(a)

90C

I

I

I

Level 1

Level 2

Level 3

3

Visual Complexity

N 1300

(b)

(b)

~ 1200

700

.~_ E 1100

650

Z

=E

I000

6OO 550

i

900

l

,, I

I

,

I

One Syllable Two Syllable Three Syllable

One Syllable Two Syllable Three Syllable

Word Length

Word Length

FICURE 3. (a) Mean matching response latency and standard error of the mean for each level of visual complexity. (b) Mean matching response latency and standard error of the mean for each word length.

Fxcum~ 4. (a) Mean naming response latency and standard error of the mean for each level of visual complexity. (b) Mean naming response latency and standard error of the mean for each word length.

of the mean. A Lindquist T y p e II A N O V A (Lindquist, 1956) was performed for each of the four sets of measures. A s u m m a r y of the ANOVAs is presented in Table 1. An inspection of the means displayed in Figures 1-4 indicates an increase in all four measurements as word length increases. An increase in SRT and TAbLe. 1. F values for tests of significance of word length, visual complexity, and interaction factors for SRT, VDT, MRL, NRL, and number of learning trials. Source

Between subjects Word length X visual complexity Within subjects Word length Visual complexity Word length • visual complexity

df

SRT

2, 27

1.60

2, 54 2, 54 2, 54

VDT

3.09* < 1 1.12

MRL

NRL

Trials

< 1

< 1

< 1

< 1

15.15"** 5.03*** < 1

1.37 < 1 < 1

12.43"** 3.59** < 1

10.36"** < 1 < 1

*p < 0.10. **p < 0.05. **.p < 0.01.

BRENNAN, CULLINAN: Verbal Reaction Times


147

decreases in VDT, MRL, and NRL accompany an increase in visual complexity level. The effects of visual complexity and word length are statistically significant (p < 0.05) for the VDT and NRL measurements but nonsignificant for the SRT and MRL measurement (see Table 1). All word length by visual complexity interactions are nonsignificant. Since the effect of word length on SRT could be considered significant at the 0.10 level, the appropriate sums of squares were pooled and the Duncan's New Multiple Range Test was used to test the differences among the treatment means for the three word lengths. Significant differences (p < 0.05) were found between the treatment means for the one- and three-syllable names and between the means for the two- and three-syllable names, but not between the means for the one- and two-syllable names. The failure of the main effect of word length on MRL measurements to reach statistical significance should be interpreted cautiously. An examination of the mean MRLs for "yes"/"no" responses depicted in Figure 5 for each group of subjects suggest a "yes"/"no" response by word length by subject group interaction. For all three groups, "no" response latencies increase monotonically as stimulus word length increases from one to three syllables. On the other hand, the form of the relationship of "yes" response latencies to

~.

800[

(a)

700

"no".-- .......

" " ""

C

~u 600 ~- "yes" ,---_'~ 550 L E 800 (b)

,

i

/. i f

~

700

--

"no"

,',

550

"" 1:0 .c:

800

~

700 600 550

--''

I

(c)

..--

"yes"~ ';no"

i i L One Syllable Two Syllable Three Syllable Word Length

FictraE 5. Mean "yes" and "no" matching response latencies for (a) Subject Group I, ( b ) Subject Group II, and (e) Subject Group III. 148 ]ournal of Speech and Hearing Research


19 141-155 1976

word length varies among subject groups. This interaction may have influenced the results of the ANOVA.

Duration of SRT Responses The duration of each of the 270 verbal responses obtained in the SRT task was measured. The mean duration for all subjects was 361 msec for onesyllable names, 419 msec for two-syllable names, and 592 msec for threesyllable names. A two-way ANOVA with repeated measures of one factor (Winer, 1971) showed the effect of word length to be significant (F = 389.4; d[ = 2, 54; p < 0.01).

Number of Learning Trials and Overlearning The differences among subject groups in the mean number of trials in each of the four paired-associate learning and practice sessions did not exceed one trial, and, in three of the four tasks, did not exceed one-half trial. These data suggest that the paired associates for any one group were not appreciably more difficult to learn, on the average, than those for any other group. While this experiment was not designed to explore the relationships between the various processing times studied and the order in which "names" were learned, some information bearing on this question has been obtained from the data. The criterion for learning of an association was defined as the third consecutive trial in the first learning session in which the subject correctly named the visual form. Pearson product-moment correlation coefflcients, obtained as measures of the relationships between mean number of trials to criterion and the various processing times, were 0.33, 0.75, 0.57, and 0.89 for mean SRTs, VDTs, MRLs, and NRLs, respectively. Although all four processing times tend to be positively related to the number of trials required to learn the names, only the coefllcients for VDTs, MRLs, and NRLs, are significantly different from zero (p < 0.05). It was previously noted that "yes" and "no" response latencies in the MRL task may not be similarly related to word length. For this reason, correlation coefllcients were obtained for the mean latency for the "yes" responses (r = 0.35, t -- 1.87, p > 0.05) with the mean number of trials to learn based on the data for each stimulus pair from those five subjects who gave "yes" responses and for the mean latency for the "no" responses (r = 0.49, t -- 2.81, p < 0.05) with the mean number of trials to learn for each stimulus pair based on the data from the five subjects giving the "no" responses. Thus, the correlation coefficient for "no" latencies with the learning data is significantly different from zero but when the "yes" latencies are paired with the learning data the coefficient is not significant. It appears, then, that some of the processing times obtained in this experiment are related both to word length and to level of visual complexity. It appears also, however, that the times may be related to the order in which BRENNAN,CULLINAN"Verbal Reaction Times 149


the names were learned. Thus, the number of trials to criterion also may be related to word length and to level of visual complexity. The mean number of trials required to learn the one-syllable paired associates was 7.1 trials, for two-syllable paired associates, 8.9 trials, and for three-syllable paired associates, 9.1 trials. A Lindquist Type II ANOVA (see Table 1), showed that the effect of word length on the number of trials to criterion was significant, but the main effect of visual complexity and the interactions were not significant. The appropriate sums of squares were pooled and the Duncan's New Multiple Range Test was used to test the differences among the means of word length. Significant differences (p < 0.01) were found between the means of one- and three-syllable responses and between one- and two-syllable responses. The difference between the means of the two- and three-syllable responses was not significant. Since in the first learning session, all nine pairs were practiced until all visual stimuli were named correctly on three consecutive trials, those pairs which were learned first underwent more overlearning trials than those learned in later trials. Thus, the relationships noted between the various processing times and the number of trials to criterion might actually be reflecting relationships between the times and the amount of practice or overlearning for the pairs. The number of trials to learn is confounded with the number of overlearning trials in that the subiects in the three groups participated in different numbers of trials in the learning session. Are the various processing times actually related to word length or do they just appear to be because both the times and the word lengths are related to the number of trials to criterion and/or the number of overlearning trials? Some information bearing on this issue was obtained by computing mean SRTs, VDTs, and NRLs for responses for subiects who showed the same number of trials to criterion, and therefore the same number of overlearning trials for all three words beginning with the same syllable. Whereas there were only five instances where a subject learned all three names beginning with the same syllable in the same number of trials, there were 18 instances where the number of trials for the three words for a subject differed by no more than one trial. The latter criterion was used to obtain the means presented in Table 2. Even though in some comparisons all names, subjects, and groups do TABLE2. Mean response times (in msec) for SRTs, VDTs, and NRLs and mean number of trials to criterion for various word lengths with the number of learning trials controlled. Word Length Experimental Tasks SPtT VDT NRL

Number o[ Responses

1 Syllable

2 Syllables

3 Syllables

18 18 16"

337 23 856

337 28 952

361 28 1096

*Some NRLs missingfor 2 subjects. 150 lournal o[ Speech and Hearing Research


19 141-155 1976

not occur equally often and word length and visual complexity level are confounded, the relationship of the obtained means and word length are of interest. Response latencies for the MRL task were not computed because of the confounding effect of "yes" and "no" latencies. When the number of learning trials is controlled thusly, SRTs did not increase from one to two syllables but did increase from one to three and from two to three syllables. VDTs increased from one to two syllables and from one to three syllables, but did not vary from two to three syllables. NRLs increased from one to two to three syllables. DISCUSSION Word Length

Within the limitation imposed by the use of newly learned nonsense names for nonsense stimuli rather than familiar names of real objects, the results of this study support the point of view that SRT, VDT, and NRL are to some extent related to word length independently of word frequency. In addition, there is a tendency for MRL to increase with increases in word length, particularly for "no" responses. Carroll and White (1973) have suggested that objects whose names are learned early in life are named faster, that is, they have shorter NRLs. They suggested further, that word retrieval may be a one-stage process that depends upon the age at which a word is learned. Since nonsense stimuli were used in the present study, the findings cannot be explained in terms of the age at which tile names were learned. The finding of a significant correlation between NRL and number of learning trials might suggest, however, that the naming latency is related to the order in which names are learned. The results of the analysis in which the number of learning trials was controlled supports the hypothesis that NRLs are related to word length independent of learning order. Thus, it appears that long words as opposed to short words may be both more difficult to learn as names and more difficult to retrieve from storage and/or to initiate in overt speech. The finding of a significant increase in SRT when word length increased from one- or two-syllable names to three-syllable names is unlike the findings of other studies. Eriksen et al. (1970) and Klapp (1971) failed to find significant syllable effects for SRTs. Eriksen et al. (1970) suggested that when subjects knew the verbal response ahead of time in their SRT task, any implicit speaking of the word would have been completed during the two- to three- second interval between experimenter's presentation of the word and the signal to respond. When the number of learning trials was controlled statistically, the increases in SRT with increases in word length were similar in magnitude to those reported in the original analysis. It would appear then, that the increase in SRTs from one to three and from two to three syllables is due, at least in part, BRENNAN,CULLINAN'Verbal Reaction Times 151


to the effects of word length and not entirely, if at all, to the number of learning or overlearning trials. In addition, a small nonsignificant correlation was found between mean SRT and mean number of learning trials. Since the visual forms were not used in the SRT task, one would not expect paired-associate learning involving the visual forms to be a factor in SRT. The significantly longer mean SRT for three-syllable names than for one- and two-syllable names suggests that some form of implicit speech might occur, but not until after the presentation of the signal light. If implicit speech occurs and if syllabic duration variations in implicit speech parallel those in spoken words, a greater difference in SRT between two- and three-syllable names than between one- and two-syllable names would be anticipated since stressed vowels in vocalized speech tend to be associated with longer durations than unstressed vowels (Fry, 1955). The first and third syllables of the names were stressed and the second syllables were unstressed. The obtained difference between the mean durations of one- and two-syllable responses was 58 msec while the difference between the means of two- and three-syllable responses was 173 msec. Thus, these results are consistent with the contention that implicit speech occurs before the spoken response in the SRT task but following the presentation of the signal light. The differences in response duration are much greater than the differences in SRT. If implicit speech does occur, then, as Colegate and Eriksen (1970) suggest, there is not a one-to-one temporal relationship between overt speech and implicit speech. Landauer (1962), however, has stated that implicit speech requires roughly the same duration as overt speech. It would seem that either (1) Landauer is right and implicit speech is not occurring following the stimulus presentation in the SRT task; (2) Landauer is in error and implicit speech and overt speech do not require the same duration; or (3) the term implicit speech as used in the present study and by Colegate and Eriksen does not refer to the same behavior as it did when used by Landauer. If the word length or syllable effect observed in this study for SRT is not due to implicit speech as defined by Landauer, then how is this effect to be explained? The subjects in Landauer's study most likely performed some of the neuromuscular responses associated with overt speech whereas implicit speech in the SRT task involved only a more central process of covert speech. One might speculate further that this process is one of motor planning for speech and that the time involved in motor planning increases as the length of the verbal response increases. The difference between the results of this study and of the Eriksen et al. and the Klapp studies may be related to the nature of the verbal stimuli used. Possibly motor planning occurs before the signal when familiar, real words are used, but when unfamiliar or newly learned words are used some brief rehearsal of the motor planning may be needed following the signal to respond and immediately preceding the initiation of the overt speech attempt. Further research seems necessary for understanding the reasons for the apparent word length effect for the SRT task and the nature of the motor planning for speech. 152 1ournal of Speech and Hearing Research


19 141-155 1976

The increase in NRL as a function of word length is consistent only in part with an interpretation that an implicit speech (as defined by Landauer) process may be a portion of the total time required for the naming response to be made. The average durations of the spoken one- and three-syllable words differed by 231 msec and the corresponding difference in NRL was 254 msec. The difference, however, between response durations for one- and two-syllable words was only 58 msec, whereas the corresponding differences for NRL was 146 msec. Thus, it appears that the differences in SRT with word length parallel more closely the differences in response duration, though with much smaller magnitudes, than do the differences in NRL. Wingfield (1968) suggested that information derived from a visual display must be coded in some more durable form than a visual image ff it is to be retained through a masking pattern. Although Wingfield did not speculate as to the nature of the coding process, Colegate and Eriksen (1970) suggested that the subiect , through a scanning or noting process, encodes information from the visual afterimage using an implicit speech response. Our finding that word length is a factor in VDTs add support to the hypothesis of Colegate and Eriksen. Klapp (1971), using visually presented digits, reported MR'Ls comparable in magnitude and in direction of change with increases in number of syllables to those obtained in the present study. The mean matching latencies reported by Klapp were 645 msec and 662 msec for two- and three-syllable numbers, respectively, compared to mean MRLs of 644 and 669 msec found in the present experiment. The mean "yes" latency reported by Klapp was 649 msec while the mean "no" latency was 675 msec. In the present experiment, the mean "yes" latency was 610 msec and the mean "no" latency was 675 msec. It should be noted also that the difference between the means of "yes" and "no" latencies and the differences among the means of word length across "yes" and "no" responses in the Klapp experiment were significant (p < 0.001). It may be that "yes" or "same" responses are not related to word length in the same way as "no" or "different" responses. Bamber (1969) has proposed that subiects employ two distinct stimulus comparison processes simultaneously, with a "fast identity reporter" subserving "same" decisions and a "slower serial processor" subserving "different" decisions. Wingfield (1968) has suggested that the subiect has an initial set for a "same" decision and the need to reiect this set for a "different" decision results in longer matching latencies. Klapp and Bischoff (1972), who failed to find a syllable effect for matching latencies, have suggested that subjects do not use an implicit speech process to mediate "same"/"different" decisions with printed names and nonsense forms. Klapp and Bischoff, however, used only two word lengths, paired "real" words to nonsense forms, and two nonsense forms were simultaneously presented to subjects for matching. It is questionable how much implicit speech involving the associated names would be used in a picture-picture matching task. Posner and Mitchell (1967) propose that the "same"/"different'" iudgment in a case such as this is based on "physical identity." Since both forms BRENNAN,CULLINAN:Verbal Reaction Times 153


would be present in the perceptual field at the same time, it is possible for a match of this type to be made even if the stimuli had never been seen before. In addition, if verbal responses are used by the subject in the picture-picture matching task, questions may be raised as to the influence on the syllable effect of the subject's prior experience with the real words. Another possible explanation for the lack of syllable effect in the Klapp and Bischoff experiment may involve the nature of the response words themselves. Their two-syllable words are comparable in stress pattern to the two-syllable words in the present investigation. It has already been suggested that the addition of an unstressed syllable would not increase the time required for implicit speech as much as adding a stressed syllable. Response latency differences also may be more apparent in comparing three-phoneme one-syllable responses to five-phoneme two-syllable responses, as in the present study, than when comparing one- and two-syllable responses which differ, on the average, by only one phoneme as in the Klapp and Bischoff study. The results of the analysis in which the number of learning trials was controlled, along with the significant correlation between VDT and the number of learning trials, can be taken to suggest that number of learning trials may be a more important factor than word length in the perceptual identification of nonsense forms.

Visual Complexity An inverse relationship was found between VDTs and visual complexity levels in this experiment supporting a similar finding by Fraisse and Elkin (1963). The association of low VDTs with visual stimuli from complexity Level 3 may be due to the greater detail in the drawings of that level. Conversely, the comparative lack of detail in complexity Level 1 drawings may have created more ambiguities at below threshold exposures and thereby resulted in increased thresholds. The data indicate also an inverse relationship between NRL and visual complexity. Stimuli associated with Complexity Level 3 may have been named faster because they have lower thresholds of recognition. However, the difference in mean VDTs of Complexity Level 1 stimuli and Complexity Level 3 stimuli was only 5 msec while the difference in mean NRLs between Level 1 and Level 3 stimuli was 97 msec. The small differences in recognition thresholds would not seem to account for the larger differences in the NRLs. Further study is needed to provide more and better information about the factors which affect the processing times associated with the naming response. ACKNOWLEDGMENT This paper is based on a doctoral dissertation completed by the first author at the University of Oklahoma Health Sciences Center. Requests for reprints should be sent to Donald G. Brennan, Department of Communication Disorders, St. Louis University, 15 North Grand Boulevard, St. Louis, Missouri 63103. 154 1ournal o[ Speech and Hearing Research


19 141-155 1976

REFERENCES BAMBER, D., Reaction times and error rates for "same"-"different" judgments of multidimensional stimuli. Percept. Psychophys., 6, 169-174 (1969). BISIACH, E., Perce[~tual factors in the pathogenesis of anomia. Cortex, 2, 90-95 (1966). BOYSEN, A. E., ana CULLINAN, W. L., Object-naming latency in stuttering and nonstuttering children. J. Speech Hearing Res., 14, 728-738 (1971). Br~,~NN^N, D. G., and CULLINAN, W. L., Object identification and naming in cleft palate children. Cleft Palate .l., 11, 188-195 (1974). CAnr~OLL, J. B., and WHITE, M. N., Word frequency and age of acquisition as determiners of picture-naming latency. Q. 1I. exp. Psychol., 25, 85-95 (1973). COt.ECATE, R., and ERIKSEN, C. W., Implicit speech as an encoding mechanism in visual perception. Am. 1. Psychol., 83, 208-215 (1970). EmKSEN, C. W., POLLACK, M. D., and Mo~VrAGUE, W. E., Implicit speech: Mechanism in perceptual encoding? I. exp. Psychol., 84, 502-507 (1970). Fr~AISSE, P., and ELrdN, E., Etude genetique de l'influence des modes de presentation sur le seuil de reconnaissance d'objects familiers. Annee psychol., 63, 1-12 (1963). FRY, D. B., Duration and intensity as physical correlates of linguistic stress. I. acoust. Soc. Am., 27, 765-768 (1955). KLAPP, S. T., Implicit speech inferred from response latencies in same-different decisions. J. exp. Psychol., 91,262-267 (1971). KLAPP, S. T., and BISCHOFF, D. M., Does implicit speech in same-different decisions extend to nonsense forms? Percept. Psychophys., 11, 363-364 (1972). LANDAUER, T. K., Rate of implicit speech. Percept. Mot. Skills, 15, 646 (1962). LINDQUIST, E. F., Design and Analysis of Experiments in Psychology and Education. Boston: Houghton Mifflin (1956). MmIANTI, F. J., and CULLINAN, W. L., Effects of age and word frequency on object recognition and naming in children. 1. Speech Hearing Res., 17, 373-385 (1974). NEWCOMBE, T., OLDr~mLO, R., and WINGFmLD, A., Object-naming by dysphasic patients. Nature, 207, 1217-1218 (1965). OLDFI~.LD, R., Things, words, and the brain. Q. II. exp. Psychol., 18, 340-353 (1966). OLDFIELD, R., and WINCFIELD, A., The time it takes to name an object. Nature, 202, 10311032 (1964). OLDFIELD, R., and WINCFtELD, A., Response ]atencies in naming obiects. Q. Jl. exp. Psychol., 17, 273-281 (1965). POSNER, M. I., and MITCHELL, R. F., Chronometric analysis of classification. Psychol. Rev., 74, 392-409 (1967). WINmn, B. J., Statistical Principles in Experimental Design. (2nd ed.) New York: McGrawHill ( 1971 ). WINGFmLD, A., Effects of frequency on identification and naming of obiects. Am. J. Psychol., 81, 226-234 (1968). Received April 9, 1975. Accepted October 25, 1975.

BRENNAN, CULLINAN: Verbal Reaction Times


155

Effects of walking on reaction and movement times among elders.

Effects of task difficulty on subsequent preference for visual complexity.

Word recognition cues used in matching verbal stimuli within and between auditory and visual modalities.

The effects of word length and emotionality on hemispheric contribution to lexical decision.

Effects of Word Width and Word Length on Optimal Character Size for Reading of Horizontally Scrolling Japanese Words.

The effects of recall-concurrent visual-motor distraction on picture and word recall.

Effects of word length on eye movement control: The evidence from Arabic.

Visual field effects and short-term memory for verbal material.

Effects of the series length on Lempel-Ziv Complexity during sleep.

Word-length effects in immediate memory: Overwriting trace decay theory.

Effects of stimulus-response correspondence and verbal expectancy on choice reaction time.

The role of elicited verbal imitation in toddlers' word learning.

Word length and frequency effects on children's eye movements during silent reading.

Computational Modeling of Morphological Effects in Bangla Visual Word Recognition.

Grip on complexity in chemical reaction networks.

Visual reaction time as a function of locus, area, and complexity of stimulus.

Effects of alcohol, practice, and task complexity on reaction time distributions.

Performance of children on the Turkish Nonword Repetition Test: Effect of word similarity, word length, and scoring.

Emerging Object Representations in the Visual System Predict Reaction Times for Categorization.

The Effects of Repeated Testing, Simulated Malingering, and Traumatic Brain Injury on Visual Choice Reaction Time.

On the functional nature of the N400: Contrasting effects related to visual word recognition and contextual semantic integration.

The effect of word frequency and parafoveal preview on saccade length during the reading of Chinese.

The effects of incidentally learned temporal and spatial predictability on response times and visual fixations during target detection and discrimination.

The effects of reward and punishment on reaction times and autonomic activity in hyperactive and normal children.