journal of verbal learning and verbal behavior volume 19 issue 5 1980 [doi...

Upload: venkannareddy-yamnur

Post on 02-Jun-2018

217 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/10/2019 Journal of Verbal Learning and Verbal Behavior Volume 19 Issue 5 1980 [Doi 10.1016%2Fs0022-5371%2880%2990628-3] Maria L. Slowiaczek; Charles Cli

    1/10

    JOURNAL OF VERBAL LEARNING AND VERBAL BEHAVIOR 19, 573--582 (1980)

    Subvocalization and Reading for eaning

    MARIA L SLOWIACZEK AND CHARLES CLIFTON, JR.

    University of assachusetts

    Two experiments demonstrated that subvocalization is of value in reading for certain

    types of meaning. Blocking subvocalization by requiring subjects to count or say cola-

    c o l ac o l a. . . aloud impaired their reading comprehension but generally not their listening

    compre hension. The effect of blocking subvocalization was found to be specific to test s that

    required integration of concepts within or across sentences, as con trasted with tests that

    required only memo ry of individual word concepts . Two hypo these s were offered: first, that

    subvocalization results in a more durable memory representation needed for integration of

    concep ts; and second, that subvocalization enables a prosodic restructuring that makes infor-

    mation needed for sent ence comprehensi on accessible.

    The subjective experience of hearing a

    voice inside one s head while reading seems

    nearly universal. This inner voice is com-

    monly accompanied by act ivi ty of the

    larynx and the articulators (McGuigan,

    1970). Early researchers used innovative

    methods to prevent this covert articulation

    in the hopes of discovering its function.

    Subjects were asked to sing, whistle, count,

    and talk while reading to prevent sub-

    vocalization (Pintner, 1913; Reed, 1916;

    Secor, 1900). Despite these creative efforts,

    the function of subvocal speech in reading

    remains a puzzle.

    One possible function involves lexical

    access. Perhaps strings of printed letters

    are translated to sounds before they are

    recognized as words. Although there is still

    some dispute over the importance of such

    sound translations (cf. Coltheart, 1979), we

    This research followed up experiments presented in

    a Masters thesis by the first author. She would like to

    thank her committee chairman, Arnold Well, and the

    other members o f her committee for their advice and

    assistance. The authors thank Betty Ann Levy for

    providing the materials used in her experimen ts, Dawn

    Piccolomini and Janet Dohert y for preparing materials

    and running subjects, and Rachel Clifton, Alexander

    Pollatsek, and Keith Rayner for their comments on an

    earlier version of this paper. Requests for reprints

    should be sent to Maria L. Slowiaczek, Department of

    Psychology, University of Massachusetts, Amherst,

    Mass. 01003.

    573

    know that direct visual access must be pos-

    sible since we can distinguish between

    homophones and recognize words with ir-

    regular spellings. Also, lexical access

    through sound translation must be possible

    since we can recognize words we have

    heard but never seen before by sounding

    them out. Even though we can use both ac-

    cess routes, the bulk of the evidence indi-

    cates tha t the direct visual route is generally

    faster and therefore relied on more often.

    A second possible function for sub-

    vocalization involves the comprehension of

    entire phrases and sentences. Once words

    are recognized, further syntact ic and

    semantic processing is necessa ry in order to

    understand the meaning of a sentence.

    Subvocalization may play some role in in-

    tegrating words into syntactic relations, or

    in providing a representation tha t facilitates

    semantic integration. One possibility is that

    recoding a sentence by subvocalization

    facilitates keeping it in memory. This hy-

    pothesis is supported by research which in-

    dicates that visually presented materials are

    commonly held in a phonological repre-

    sentation (Conrad, 1972). Kleiman (1975)

    found that a digit shadowing task, assumed

    to block subvocalization, interfered with

    judging the acceptabili ty of sentences

    which subjects read. He suggested that

    sentence acceptability judgments require a

    0022-5371/80/050573-10502.00/0

  • 8/10/2019 Journal of Verbal Learning and Verbal Behavior Volume 19 Issue 5 1980 [Doi 10.1016%2Fs0022-5371%2880%2990628-3] Maria L. Slowiaczek; Charles Cli

    2/10

    574

    SLOWI CZEK ND CLIFTON

    subject to maintain a sentence in memory,

    and that preventing speech recoding im-

    pairs memory for a sentence.

    Further evidence for this memory hy-

    pothesis has been provided in a series of

    experiments on reading stories while sup-

    pressing subvocalization (Levy, 1977,

    1978). Levy (1977) used counting out loud

    as a technique to suppress subvocalization.

    She found that suppressing subvocalization

    interfered when subjects were asked to rec-

    ognize sentences which they had read, but

    had no effect on recognition of sentences

    which they had heard. In a further study,

    Levy (1978) found that subjects could read

    stories and make paraphrase judgments

    without interference from speech suppres-

    sion, but when they were required to rec-

    ognize verbatim sentences, counting sup-

    pression did interfere. Levy concluded that

    reading for meaning was possible without

    subvocalization, and that the purpose of

    subvocalization was to translate visual in-

    formation into a speech code in order to

    maintain it in verbal form.

    Although Levy found accurate under-

    standing of simple sentences even when

    subvocalization was suppressed, her results

    do not provide conclusive evidence that

    subvocalization is unnecessary for com-

    prehension . Hardyck and Pet r inovich

    (1970) found that reading comprehension of

    difficult, but not easy, passages was im-

    paired when they suppressed subvocaliza-

    tion using biofeedback from monitoring of

    laryngeal activity. Levy's reading task may

    be classified as very simple, since para-

    phrase judgments could be made by knowing

    the meanings of the individual words alone.

    Her dist ractor tests introduced new,

    semantically novel, words; for example, the

    story sentence The stronger team beat the

    weak opponents was tested with the dis-

    tractor The stronger team cheered the

    weak amateurs , where neither che ere d

    nor ama teu rs nor any synonym of them

    appeared in the story. Sentence compre-

    hension normally involves more extensive

    processing, where grammatical relations

    in a sentence provide critical information

    for understanding. Perhaps subvocalization

    is necessary when these higher level pro-

    cesses are needed for comprehension.

    In two experiments, Slowiaczek (Note 1)

    replicated Levy's (1978) experiment and

    obtained results consis tent with hers: Sup-

    pressing subvocalization interfered more

    with verbatim recognition judgments than

    with meaning judgment s. In these experi-

    ments, correct test sentences were con-

    structed by substituting synonyms in story

    sentences, while their distractor tests intro-

    duced semantically novel words. In a third

    experiment, Slowiaczek (Note 1) studied

    meaning judgments in a task where some

    distractor items did not introduce novel

    words, but rather replaced one character or

    event in the test sentence with another

    character or event from the story. Suppress-

    ing subvocal speech hurt performance sub-

    stantially in this more difficult reading task.

    The experiments we report here further

    investigate the finding that blocking sub-

    vocalization can interfere with reading for

    meaning. They focus on the hypothesis that

    subvocalization is not needed to identify

    the concepts mentioned in a passage, but

    rather it s used to combine concepts in the

    proper semantic relationships with each

    other. In Experiment ~1, we compared per-

    formance on test items which simply re-

    quired recognition of lexical concepts (noun

    paraphrase and verb paraphrase tests) with

    performance on test items that required

    combining concepts within and across sen-

    tences (crossover and inference tests). We

    introduced a control for nonspecific inter-

    ference by including conditions in which

    subjects heard, rather than read, the pas-

    sages. We assumed that subvocalization is

    not necessary for successful listening com-

    prehension (cf. Levy, 1977), but that other

    sources of comprehension difficulty are

    similar for reading and listening. Unde r this

    assumption, the integration hypothesis pre-

    dicts that suppressing subvocalization will

    disrupt reading comprehension more than

    listening comprehension for inference and

  • 8/10/2019 Journal of Verbal Learning and Verbal Behavior Volume 19 Issue 5 1980 [Doi 10.1016%2Fs0022-5371%2880%2990628-3] Maria L. Slowiaczek; Charles Cli

    3/10

  • 8/10/2019 Journal of Verbal Learning and Verbal Behavior Volume 19 Issue 5 1980 [Doi 10.1016%2Fs0022-5371%2880%2990628-3] Maria L. Slowiaczek; Charles Cli

    4/10

    76 S L O W l C Z E K N D C L IF T O N

    while the sentence Th e older children

    chee red the sad childr en was present ed as

    the crossover distractor. Inference test

    sentences combined information from two

    sentences that had occurred in the story,

    separated by at least one intervening sen-

    tence. To continue our example, the story

    contained a third sentence that established

    two alternative ways of referring to one

    characte r, Th e amateur juggler was a high

    school tea cher . The correct inference test

    sentence combined this sentence with the

    earlier sentence, resulting in Th e amateur

    juggler chee red the sad chi ldre n. Dis-

    tractor inference tests made the wrong

    inference by referring to a different char-

    acter in the story: Th e painted clown

    cheer ed the sad child ren. Notice that in-

    ference distractor tests were not con-

    tradicted by a single story sentence, while

    the crossover distractor tests were. (The

    high school teacher could have been the

    painted clown, but the high school teacher

    could not have been the older children.)

    Thus comprehension of inference test sen-

    tences required combining concepts across

    two sentences, while comprehending cross-

    over tests required combining concepts

    within a single sentence.

    Design and procedure Thirty-two sub-

    jects read the stories and thirty-two sub-

    jects listened to them over headphones.

    Sixteen of the subjects in the reading con-

    dition wore headphones to control for the

    muffling effect of the headphones on the

    sound of counting. Stories were presented

    in the same random order for all subjects in

    both reading and listening conditions, but

    the type of test item was counterbalanced

    across subjects. Half of the stories were

    presen ted while subjects were counting and

    half were present ed while subjects were si-

    lent and free to subvocalize. All test types

    were pres ented equally often for each story

    in both silent and counting conditions, for

    both listening and reading. Each subject

    was presented with an equal number of

    each test type in silent and counting condi-

    tions.

    The stories wer e prese nted in nine blocks

    of eight passages each. One test sentence

    was presented after each story with an

    equal number of each test type in each

    block. The first block was for practice and

    was not included in the data analysis.

    Blocks alternated between silent and

    counting conditions and subjects were told

    at the beginning of each block whether or

    not they should count.

    In the silent reading condition, subjects

    read silently to themselves. In the suppressed

    reading condition, they counted out loud

    from 1 to 10 repe atedly as they read. They

    were encouraged to count at a very rapid

    rate as loudly as they could. The count-

    ing was monitored ove r an intercom system

    to insure that the fast rate was maintained.

    Most subjects were able to count from 1 to

    10 twice between the onset of one sentence

    and the onset of the next.

    The sen tences were presen ted on a

    Hewlett-Packard 2600A video terminal.

    Each sentence remained on the screen for 2

    seconds, with an interval of approximately

    670 milliseconds (including some 170 mil-

    liseconds to write the next sentence on the

    video screen and then move it into a win-

    dow through which the subject could see

    the screen) between sentences. The screen

    was 2 feet away from the subject and sen-

    tences were written in block letters 2.5 by 6

    mm. At the end of each story, the word

    Te s t appeared on the screen for 1 second

    before the test item was presented.

    For the listening condition, stories and

    tests were prerecorded and presented on

    two computer- controlled tape recorde rs at

    the rate of one sentence every 2 seconds.

    The word Te s t appeared visually on the

    video screen for 1 second after each story

    before the auditory test sentence was pre-

    sented.

    For both reading and listening condi-

    tions, subjects were instructed to stop

    counting on the suppressed blocks when

    the word Te s t appeared. Subjects were

    told to respond y es by pushing one but-

    ton if the sentence was true based on what

    they read or heard in the story, and to re-

    spond no by pushing another button.

  • 8/10/2019 Journal of Verbal Learning and Verbal Behavior Volume 19 Issue 5 1980 [Doi 10.1016%2Fs0022-5371%2880%2990628-3] Maria L. Slowiaczek; Charles Cli

    5/10

    S U B V O C A L I Z A T I O N A N D R E A D I N G

    77

    Responses were recorded by the computer,

    and subjects were given feedback on the

    video screen after each response. At the

    end of each block the total number of errors

    for that block was presented on the video

    screen.

    Results andDiscussion

    Accuracy scores were calcu la ted as

    p Hit) - p False Alarm). A preliminary

    analysis showed a larger suppression effect

    for subjects who read without headphones

    mean difference = .29) than for subjects

    who read with headph ones mean differ-

    ence = . 12), F 1,30) = 11.07, p < .002. This

    difference was nearly constant for all types

    of test sentences, which suggests that the

    headphones reduced some general interfer-

    ence caused by hearing oneself count .

    Since none of the interactions involving

    headphones were signif icant , the two

    reading groups were combined for further

    analysis.

    Table 2 presents the accur acy data. They

    were analyzed by combining the two

    paraphrase tests and comparing them in a

    planned contrast to the combined inference

    and crossover tests. As expected, the sup-

    pression effect was greater in the reading

    condition than in the listening condition for

    the inference and cr oss over tests, while this

    difference was less for the paraphrase tests,

    F 1,62) = 3.89, p < .05. The diffe rence

    between the reading effect and the listening

    suppression effect for the inference and

    crossover tests, which required combining

    concepts within and across sentences, was

    .21. The difference was only .02 for the

    paraphrase tests, which required only rec-

    ognition of individual concepts.

    This difference in suppression for reading

    and listening suggests that the suppression

    task interferes with some process which is

    specifically needed in reading. Notice that

    interference from suppression does not

    simply reflect general capacity limitations.

    Performance in the silent conditions for

    reading and listening is comparable, which

    suggests that the reading and listening tasks

    are equally difficult. Also, crossover and

    paraphrase noun tests appear equally dif-

    ficult based on scores in the silent condi-

    tions, but crossover tests show a suppres-

    sion effect for reading and paraphrase noun

    tests do not.

    The suppression effect was assessed sep-

    arately for each type of test item, using an

    error term from an overall analysis of vari-

    ance

    df

    = 186, critical difference forp .05)

    = .13). Bo th the inference and the cross-

    over tests showed significant suppression

    effects in the reading condition, while the

    paraphrase noun tests did not. None of

    these types of tests showed significant sup-

    pression effects in the listening condition.

    Howe ver, the s uppression effect for

    paraphrase verb tests was significant in

    both the reading and the listeninz conditions.

    T A B L E 2

    p HIT ) - p F A ) FOR EXPERIMENT 1 a

    T e s t t y p e

    P a r a p h r a s e P a r a p h r a s e

    n o u n v e r b C r o s s o v e r I n f e re n c e

    R e a d i n g

    Sile nt .59 .68 .63 .44

    Su pp re ss ed .50 .45 .36 .21

    Dif fe ren ce .09 .23 .27 .23

    L i s t e n i n g

    Silen t .66 .71 .67 .47

    Su pp re ss ed .56 .52 .65 .41

    Dif fe ren ce .10 .19 .02 .06

    F A , f a l s e a l a r m .

  • 8/10/2019 Journal of Verbal Learning and Verbal Behavior Volume 19 Issue 5 1980 [Doi 10.1016%2Fs0022-5371%2880%2990628-3] Maria L. Slowiaczek; Charles Cli

    6/10

    7 8 S L O W l A C Z E K A N D C L I F T O N

    While we have no good basis for support-

    ing a particular explanation of this result,

    we acknowledge several possibilities. Com-

    prehending verbs may require some pro-

    cess other than subvocalization that is dis-

    rupted by counting from 1 to 10. Alter-

    natively, subvocalization may be required

    to understand the relations expressed by

    verbs for listening as well as reading. Fi-

    nally, this result might be due to some pe-

    culiar property of the synonym and dis-

    tractor verbs which we used in the tests.

    We can say, though, that the significant

    suppression obtained on verb tests is not

    due to their overall level of difficulty. Per-

    formance on the verb tests in the silent

    condition was comparable to performance

    on the crossover tests, where a suppression

    effect was found only in the reading condi-

    tion. Further, we can note that Experiment

    1 supported the hypothesis that the effects

    of suppressing subvocal speech are specific

    to inference and crossover tests in the

    reading condition, even when the data from

    the verb tests are discarded. The triple in-

    teraction among suppression, modality, and

    type of test remained nearly significant,

    F(1,62) = 3.57, p = .06.

    The second experiment to be reported

    was designed to investigate this critical in-

    teraction further. It differed from the earlier

    experiment in that it used only a few long

    stories, rather than many very short ones,

    in an attempt to engage the reader's or lis-

    tener's interest and to encourage normal

    sentence comprehension strategies. It also

    introduced a new type of inference test

    which required subjects to combine infor-

    mation that they had read with extraex-

    perimental information.

    Finally, the experiment used a different

    speech suppression task, rapid repetition of

    col ac ola col ac ola ... , rather than count-

    ing to 10. We made this change to test the

    generality of the speech suppression effect

    and to eliminate some o f the problems with

    the counting task. An ideal speech suppres-

    sion task would occupy the speech ar-

    ticulators continuously, but introduce no

    memory or decision-making requirements.

    The counting-to-10 task seemed to require

    that a decision to begin counting again be

    made after each cycle through the digits,

    and subjects had to be monitored very

    closely to prevent their pausing at the end

    of each cycle. The col ac ola col a.. . task

    did not require periodic decisions to resume

    the chant, and subjects showed no ten-

    dency to pause at regular intervals.

    EXPERIMENT 2

    M et hod

    Subjects Forty-eight undergraduate vol-

    unteers served as subjects for experimental

    course credit.

    Materials and design

    Four long stories

    were constructed with approximately 70

    sentences in each story. Test sentences

    were constructed to interrupt the stories

    after four to seven sentences had been

    presented.

    An example of a section from one story

    with the test items is presented in Table 3.

    Three types of test sentences were con-

    structed with a correct and distractor sen-

    tence for each type: paraphrase tests,

    transitive inferences, and elaborative infer-

    ences. Paraphrase tests were the same as

    paraphrase noun tests in Experiment 1, and

    the transitive inferences were the same as

    the inference tests in Experiment 1.

    Elaborative inference tests required some

    integration of world knowledge with sen-

    tences in the story in order to determine

    whether or not they were true. In the

    example story the correc t elaborative infer-

    ence requires some understanding of the

    girl's motivation for drinking coffee ( She

    drank coffee to stay out at the cof-

    feeh ouse ). In the s tory, this is not explic-

    itly stated. However, the story does pro-

    vide information which makes this infer-

    ence possible ( She did find a coffeehouse

    with live entert ainment, He r first week-

    day evenings were flooded with coffee ).

    Each subject was tested on all four

    stories, one auditory silent, one auditory

    suppressed, one visual silent, and one vi-

    sual suppressed. Twelve subjects received

    each s tory in each condition. Only one type

  • 8/10/2019 Journal of Verbal Learning and Verbal Behavior Volume 19 Issue 5 1980 [Doi 10.1016%2Fs0022-5371%2880%2990628-3] Maria L. Slowiaczek; Charles Cli

    7/10

    SUBVOCALIZATION AND READING

    TABLE 3

    EXAMPLE OF MATERIALS USED IN EXPERIMENT 2

    579

    The city girl planned to start graduate school

    She moved to the college town in early August

    The town was smaller than she expected

    She met a woman in line at the bank

    The local girl was very outgoing

    She talked about gardening and zucchini pie

    The outgoing woman suggested a rendezvous

    The graduate student appreciated the invitation

    Test

    The local girl suggested a rendezvous. (Transitive infe ren ce- ye s)

    The city girl suggested a rendezvous. (Transitive infere nce -no )

    The new graduate student had trouble adjusting

    She came from New York which buzzed with excitement

    This town se emed to close by 10:00 every evening

    She did find a coffe ehouse with live entertainment

    The first weekday evenings were flooded with coffee

    Test

    She drank coffee to stay out at the coffeehouse. (Elaborative inference-yes)

    She drank coffee to stay awake while studying. (Elaborative inference-no)

    The first month was busy with fixing the apartment

    She bicycled daily to a nearby fabric store

    She needed some prints to brighten the living room

    But making a choice seemed nearly impossible

    Test

    She bicycled daily to a nearby material store. (P ara ph ras e-y es)

    She bicycled daily to a nearby furniture store. (Paraphrase-no)

    o f t e s t i te m a p p e a r e d i n e a c h t e s t p o s i t i o n ,

    t h o u g h i t w a s p r e s e n t e d w i t h b o t h c o r r e c t

    a n d d i s t r a c t o r t e s t s i n al l c o n d i t i o n s a c r o s s

    s u b j e c t s . E a c h s t o r y c o n t a i n e d t h r e e o f

    e a c h o f t h e s ix t y p e s o f t e s t i t e m s .

    Procedure T h e v i s u a l / a u d i t o r y a n d

    s i l e n t / s u p p r e s se d c o n d i t i o n s w e r e b a l a n c e d

    a c r o s s s u b j e c t s . A l l s u b j e c t s w o r e h e a d -

    p h o n e s t h r o u g h o u t t h e e x p e r i m e n t .

    I n t h e r e a d i n g c o n d i t i o n , s e n t e n c e s w e r e

    p r e s e n t e d o n a v id e o t e r m i n a l c o n t ro l l e d b y

    a P D P 8 c o m p u t e r , a t th e r a t e o f o n e s e n -

    t e n c e e v e r y 2 s e c o n d s . T h e s c r e e n w a s 7 0

    c m a w a y f r o m t h e s u b j e c t a n d t h e l e t t e rs

    w e r e w r i t te n i n b l o c k l e t te r s 4 b y 6 m m . T h e

    s t o r y w a s i n t e r r u p t e d b y th e w o r d ' ~ T e s t

    e v e r y f o u r t o s e v e n s e n t e n c e s , a n d a t e s t

    s e n t e n c e a p p e a r e d o n th e s c r e e n .

    I n t h e l i s t e n i n g c o n d i t i o n , t h e s t o r i e s

    w e r e p r e s e n t e d t h e s a m e a s in E x p e r i m e n t

    1. T e s t s e n t e n c e s , h o w e v e r , w e r e p r e s e n t e d

    v i s u a l l y .

    A l l s u b j e c t s w e r e i n s t r u c t e d t o r e p e a t t h e

    w o r d C o l a o u t l o u d o v e r a n d o v e r a s f a s t

    a s t h e y c o u l d in t h e s u p p r e s s e d c o n d i t i o n s .

    C o l a r e p e t i t io n w a s m o n i t o r e d o v e r a n

    i n t e r c o m b y th e e x p e r i m e t e r . A l l s u b j e c t s

    w e r e t o ld t o r e s p o n d Y e s i f t h e s e n t e n c e

    w a s t r u e b a s e d o n w h a t t h e y h e a r d o r r e a d

    in th e s t o ry . F e e d b a c k w a s g i v e n o n t h e

    v i d e o s c r e e n a f t e r e a c h r e s p o n s e .

    Results and Discussion

    A c c u r a c y d a t a a p p e a r in T a b l e 4 . T h e

    s u p p r e s s i o n e f f e c t w a s h i g h ly s i g ni f i c a nt ,

    F ( 1 , 4 7 ) = 3 2 . 7 5, p < . 01 , a n d p e r f o r m a n c e

    w a s b e t t e r w h i l e l i s t e n i n g t h a n w h i l e r e a d -

    i n g , F ( 1 , 4 7 ) = 1 5 .5 8 , p < . 01 . A l t h o u g h t h e

    s u p p r e s s i o n e f f e c t d i d n o t d i f f e r s ig n i fi -

    c a n t l y b e t w e e n r e a d i n g a n d l i s t e n in g ,

  • 8/10/2019 Journal of Verbal Learning and Verbal Behavior Volume 19 Issue 5 1980 [Doi 10.1016%2Fs0022-5371%2880%2990628-3] Maria L. Slowiaczek; Charles Cli

    8/10

    580

    SLOWIACZEK AND CLIFTON

    TABLE 4

    p HIT) -- p FA ) FOR EXPERIMENT 2 ~

    Test type

    Transitive Elaborative Mean

    Paraphrase inference i nference inference

    Reading

    Silent .47 .40 .52 .46

    Suppres sed .30 .10 .19 .15

    Diff erence .17 .30 .33 .31

    Listening

    Silent .66 .60 .56 .58

    Suppres sed .42 .39 .46 .42

    Difference .24 .21 .10 .16

    a FA, false alarm.

    F(1,47) = 1.83, p < . 10, examination of the

    data suggested that the suppression effect

    was greater for inference tests than for

    paraphrase tests after reading, but not after

    listening.

    In order to maximize the power of the

    statistical test, the two types of inference

    tests were combined and compared with the

    paraphrase tests. In this comparison, the

    critical interaction among reading vs lis-

    tening, silent vs suppressed, and type of

    test sentence was significant, F(1,47) =

    4.70, p < .05. Individual cont rast s indicated

    that the difference between performance in

    the silent and the suppressed conditions

    was greater for the inference tests after

    reading than after listening, t(47) = 2.19, p

    < .05, while the suppression effect did not

    differ between reading and listening para-

    phrase tes ts, t(47) = .96. Fur ther, the sup-

    pression effect was greater for inference

    tests than for paraphrase tests after reading,

    t(47) = 2.20,p < .05, while the difference (in

    the opposite direction) was not significant

    after listening, t(47) = 1.15.

    The clear conclusion from Experiment 2

    is that subvocalization helps readers com-

    bine separate pieces of information. Read-

    ers who were prevented from subvocalizing

    had difficulty in making infe rence s,

    whereas listeners did not. Preventing sub-

    vocalization did not specifically interfere

    with reading and comprehending individual

    words; the difference in performance on

    paraphrase tests afer silent vs suppressed

    reading was no greater than the corre-

    sponding difference after listening. In fact,

    the difference of. 17 between performance

    in the silent and suppressed reading condi-

    tions can quite safely be attributed to in-

    terference from the difficult task of repeat-

    ing co la co la co la . . . common to both

    reading and listening, and not due to the

    prevention of subvocalization during reading.

    GENERAL DISCUSSION

    In these experiments, reading for mean-

    ing was severely impaired when subvocali-

    zation was suppressed. Listening com-

    prehension was also reduced by suppres-

    sing subvocalization, but the effect was

    much smaller. We conclude that there was

    a general interference effect from the sec-

    ondary task, but that the significantly

    greater interference effect for reading was

    due to the specific function of subvocaliza-

    tion in reading.

    Earlier research (Levy, 1978) has sug-

    gested that subvocalization was not nec-

    essary in reading for meaning. Our ex-.

    periments demonstrated that although un-

    derstanding individual concepts might be

    possible without subvocalization, combin-

    ing concepts and integrating ideas does re-

    quire subvocalization. We found a large

    suppression effect for recognition of sen-

    tences which required comprehension of

    the relations expressed in a story that was

  • 8/10/2019 Journal of Verbal Learning and Verbal Behavior Volume 19 Issue 5 1980 [Doi 10.1016%2Fs0022-5371%2880%2990628-3] Maria L. Slowiaczek; Charles Cli

    9/10

    S U B V O C A L I Z A T I O N A N D R E A D I N G 5 8 1

    read. However , a s in Levy (1978) , the re

    was an ins ignif icant supp ress ion e ffec t for

    r e c o g n i t io n o f s e n t e n c e s w h i c h o n l y r e -

    q u i r e d r e m e m b e r i n g t h e c o n c e p t s i n t r o -

    duced in the s tory .

    W e o f f er t w o h y p o t h e s e s f o r h o w s u b-

    v o c a l i z a t i o n c a n h e l p i n s e n t e n c e c o m -

    pre he ns ion . The f ir s t, m e m or y , hypo th -

    e s i s h a s b e e n a r g u e d f o r e l s e w h e r e

    (Kleiman, 1975; Baddeley, 1979). This hy-

    pothes is c la ims tha t subvoca l iza t ion t rans-

    l a t e s a v i sua l i npu t i n to a phono log ic a l

    code , and tha t the phonologica l code las t s

    longe r i n w ork ing me mory tha n a v i sua l

    c o d e w o u l d . S u b v o c a l i z a t i o n s u p p r e s s i o n

    shou ld i n t er fe re w i th p roc e ss ing on ly w h e n

    the re p re se n t a t i on o f a se n t e nc e m us t be

    mainta ined in memory for a re la t ive ly long

    t ime . Presumably , th i s i s the case for com-

    plex sentence process ing, such as in tegra t -

    ing concepts and making infe rences . S ince

    spoke n spe e c h p rov ide s t r a nspa re n t c ue s

    tha t eas i ly map onto a phonologica l repre -

    se n t a t ion , subv oc a l i z a t i on shou ld no t be

    ne e d e d fo r l is t en ing c omp re he ns io n . The

    durable phonologica l code i s ava i lable even

    w i thou t subvoc a l i z at i on .

    The m e m ory h ypo the s i s r e l ie s on tw o a s -

    sumpt ions : one , tha t phonologica l informa-

    t ion c a n be ma in t a ine d in me mory longe r

    t h a n v i s u a l i n f o r m a t i o n ; a n d , t w o , t h a t

    m o r e e l a b o r a t e s e n t e n c e p r o c e s s i n g r e -

    qu i re s more t ime , a nd the re fo re , a me mory

    c ode w hic h l a st s l onge r. O ur se c ond , p ro -

    sod ic s t ruc tu re , hypo the s i s doe s no t r e ly

    o n t h e s e a s s u m p t i o n s . T h is h y p o t h e s i s

    sugge s t s t ha t subvoc a l i z a t i on re o rga n iz e s

    the v isua l input in to a representa t ion tha t

    p rov ide s re a dy a c c e ss t o t he i n fo rma t ion

    n e e d e d f o r s e n t e n c e p r o c e s s i n g . S p o k e n

    la ngua ge ha s a n a b unda nc e o f i n fo rma t ion

    ove r a nd a bove the sounds w hic h ma ke up

    the individua l words . In tona t ion contours

    a n d s e n t e n c e r h y t h m s p r o v i d e p a t t e r n s

    which group words in to phrases and high-

    l ig h t n e w a n d i m p o r t a n t i n f o r m a t i o n .

    S t u d i e s i n a u d i t o r y l a n g u a g e p r o c e s s i n g

    have shown tha t l i s teners use th i s informa-

    t ion in se n t e nc e c ompre he ns ion (Le h i s t e ,

    1973; Cutler, 1976).

    In cont ras t wi th the r ich prosodic s t ruc-

    ture provided in spoken language , wri t ten

    l a ngua ge p rov ide s impove r i she d c ue s . I f

    p rosod ic i n fo rma t ion i s use d in se n t e nc e

    proc e ss ing , t he re a de r mus t f i nd some w a y

    to c ompe n sa t e fo r t he l a c k o f p ros ody in

    reading. This m ay be the ro le of subvoca l i -

    za t ion. Through subvoca l iza t ion, the reader

    can reorganize the wri t ten sentence in to a

    prosodic s t ruc ture . Prosodic s t ruc tures en-

    c o d e p a t te r n i n f o r m a t i o n n e c e s s a r y f o r

    com plex process ing and highlight important

    informat ion.

    We do not deny tha t the re i s a c lose con-

    ne c t ion be tw e e n subvoc a l i z a t i on a nd me m-

    ory. Prosodic res t ruc tur ing may make in-

    f o r m a t i o n l a st lo n g e r in m e m o r y , b y

    faci l i ta t ing rehearsal . Just as subvocal iz ing

    permi ts a prosodic s t ruc ture to be c rea ted ,

    a prosodic s t ruc ture supports the rhythmic

    dec is ions tha t must be made in voca l iz ing,

    a nd p re suma bly , subvoc a l i z ing . Subvoc a l

    rehearsa l can re fresh and mainta in infor-

    ma t ion in me mory , a n impor t a n t func t ion

    w he n a n e ndur ing me mory c ode i s ne e de d

    ( c f . t h e a r t i c u l a t o r y l o o p h y p o t h e s i s o f

    Ba d de le y & H i t c h , 1974). H ow e ve r , t he

    prosod ic s t ruc tu re hypo the s i s e mpha s i z e s

    t h a t s u b v o c a l i z a t i o n f a c i l i t a t e s r e a d i n g

    comprehens ion by changing the organiza-

    t ion o f t he me m ory re p re se n ta t i on ra the r

    than i ts durabi l i ty. This hypothesis invi tes

    in t r i gu ing que s t ions a bou t se n t e nc e c om-

    p r e h e n s i o n , a n d p r o v i d e s a c o m m o n

    fra me w ork fo r fu r the r r e se a rc h in re a d ing

    and listening.

    R E F E R E N C E S

    BADDELEY, A. D. W orkin g me mo ry and read ing. I n

    P . A . K o l e rs , M . E . W r o l s t ad , H . B o u m a

    (Eds . ) , Processing of visible language. N e w

    York: Plenum, 1979.

    BADDELEY, A. D., HITCH, G. Wo rking me mo ry. In

    G. A. B ower (Ed. ) , The psychology of learning

    and motivation.

    N e w Y o r k : A c a d e m i c P r e s s ,

    1974. Vol. 8.

    COLTHEART, M. Lexica l access in s imple reading

    t a sks . In G . U nde rw ood (Ed . ) , Strategies of in

    formation processing.

    London : A cademic Pre ss ,

    1979.

    CONRAD, R. Sp eech an d reading. In J. F. Ka van agh

    I. G. Matt ingly (Eds.), Language by eye and by

    ear. Cambridge, Mass.: MIT Press, 1972.

  • 8/10/2019 Journal of Verbal Learning and Verbal Behavior Volume 19 Issue 5 1980 [Doi 10.1016%2Fs0022-5371%2880%2990628-3] Maria L. Slowiaczek; Charles Cli

    10/10

    58 SLOWIACZEK AND CLIFTON

    CUTLER, A. Beyond parsing and lexical look-up. In

    R. J. Wales E. Walker (Eds.), New approaches

    to language mechanisms. Amsterdam: North-

    Holland, 1976.

    HARDYCK, C.D., t~ PETRINOVICH, L.R. Subvocal

    speech and comprehension level as a function of

    the difficulty level of reading material. Journal of

    Verbal Learning and Verbal Behavior 1970, 9,

    647-652.

    KLEIMAN, G. i . Speech recoding in reading.

    Journal

    of Verbal Learning and Verbal Behavior 1975,

    24, 323-339.

    LEHISTE, I. Phonetic disambiguation of syntactic am-

    biguity. Glossa 1973, 7, 107-122.

    LEVY, B. A. Reading: Speech and meaning processes.

    Journal of Verbal Learning and Verbal Behavior

    1977, 16, 623-638.

    LEVY, B.A. Speech processing during reading. In

    A. M. Lesgold, J. W. Pellegrino, J. W. Fokkema,

    R. Glaser (Eds.), Cognitive psychology and in-

    struction.

    New York: Plenum, 1978.

    McGuIGAN, F. J. Covert oral behavior during the si-

    lent performance of language tasks. Psychological

    Bulletin 1970, 74, 309-326.

    PINTNER, R. Inner speech during si lent reading.

    Psychological Review 1913, 20, 129-153.

    REED, H. B. The existence and function of inner

    speech in thought processes.

    Journal of Experi-

    mental Psychology

    1916, 1, 365-392.

    SECOR, W. B. Visual reading. American Journal of

    Psychology 1900, 11, 225-236.

    REFERENCE NOTE

    1. SLOWIACZEK, M. L.

    The importance of speech

    recoding in reading. Unpublished masters thesis,

    University of Massachusetts, 1978.

    (Received December 14, 1979)