diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with...

27
Dept. for Speech, Music and Hearing Quarterly Progress and Status Report Diagnostic rhyme test for speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR volume: 17 number: 2-3 year: 1976 pages: 040-058 http://www.speech.kth.se/qpsr

Upload: others

Post on 02-Mar-2021

13 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

Dept. for Speech, Music and Hearing

Quarterly Progress andStatus Report

Diagnostic rhyme test forspeech audiometry with

severely hard of hearing andprofoundly deaf children

Risberg, A.

journal: STL-QPSRvolume: 17number: 2-3year: 1976pages: 040-058

http://www.speech.kth.se/qpsr

Page 2: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR
Page 3: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR
Page 4: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

The explanation of the difference in tes t r e su l t s must therefore be

that subjects with the s&me pure-tone audiogram can differ corisideraMy

in their ability to extract the information-bearing elements in the acoust-

i c speech signal. This can be due to the hearing impairment itself but

can also be a resu l t of differences in training. The great effoft made to

t ra in hearing impaired children to perceive o ra l language through the

combined use of visual and auditory information makes i t important to

develop tes t methods that can, a s ear ly as possible, give a detailed diag-

nosis bf a n individual chi ld 's ability to l ea rn to use acoustic cues for the

perception of speech, This tes t Iliethod must consist of a battery of

t e s t s thatmeasures differ tnt aspects of the difficulties that can be en- I

countered. The following can be mentioned as examples of tes t s in

this battery: pure-tone audiometry, t e s t s that measure frequency and

ampliiude d isc~i tn ina t ion ability, objective dudiametric tes t s , t es t s

with syhthetic speech stimuli, speech t e s t s with syiiables, wards, and

sentences, audiovisual t e s t s with different speech mater ial , t e s t s that

diagnose central language d isorders , etc. By combining Pesults f rom

different t e s t s i t might be possible to predict the difficulties an individual

child will meet in learning to use acoustic information and to integrate

auditory and visual information. In this paper a speech tes t is described

using rhyming words. This tes t i s intended a s one pa r t of a tes t battery

of the above type.

General description of rhyme t e s t s

The perception of speech i s based on the recognition of a number of

acoustic cues in the speech signal (Liberman et a1 1967). In a situation

where noise is present o r the signal i s f i l tered to suppress some par t of

the frequency spectrum, i t will not be possible for the l is tener to hear

all acoustic cues that descr ibe the speech sounds. He must then base h is

answer on the remaining cues. By analyzing the e r r o r s made it will be

possible to find out the character is t ics of the distortion introduced by

the technical system on the speech signal. If the sys tem i s a hearing im-

pairment, a n analysis of the confusion will give important information

about the nature of the impairment.

A rhyme tes t i s a closed response tes t where only one element i s

different between the t e s t i t ems i n a set . In speech r e sea rch rhyme t e s t s

consisting of CV, CVC, or VC- syllables a r e often used. By varying the

Page 5: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

STL-QFSR 2-3/1976 42.

consonants i n the syllables i t is possible to study how a specific tes t situa-

tion influences the ability to perceive the different acoustic cues that a r e

important for consonant perception. An example of this type of experi-

ment i s the study of Miller and Nicely (1955) of how low- and high-pass

filtering influences consonant confusion.

Tes t s with nonsense syllables can be made with older hearing impaired

children but i t is very important to make su re that the subjects have under - stood the task. A s an example, E r b e r (1972) used nonsense syllables to

study auditory, visual, and auditory-visual perception of consonants by

normal-hearing, severely hard of hearing, and profoundly deaf children.

Walden e t a1 (1975) has made the same type of study with adult subjects.

A rhyme tes t consisting of se t s of words has been constructed by

Fairbanks (1958). This tes t h a s been modified by House et a1 (1965).

There a r e s ix response alternatives in this l a s t test. The words a r e of

the CVC-type, where the initial and final consonant is varied. The sub-

jects answer by underlining the word they thought they heard on a response

sheet. This tes t has been adopted for use in audiology by Kruel et a1 ( 1968).

Picket t e t a1 (1972) used a modification of the tes t to measure the speech

perception ability of severely hard of hearing and profoundly deaf students

a t Gallaudet College, The responses were analyzed based on three features:

place of articulation, voicing, and low continuant.

Griffiths(l967) modified the rhyme t e s t of House et a1 ( 1965) to improve

i t s diagnostic power. In t h e tes t the l i s teners ' ability to discr iminate

minimal contrasts in pl ace and manner of articulation of the consonants

was measured. The analysis was however complicated by the fact that not

a l l consonants had the same frequency of occurrence in the test .

Simple, m o r e or l e s s systematic, rhyme t e s t s have been used by sev-

e r a l investigators to measure speech perception ability of severely hard of

hearing and profoundly deaf children (Boothroyd 1972, Hudgins 19 53, and

Watson 1967).

Rhyme t e s t s have several advantages. By giving the subjects a l imited

set of response alternatives i t has been shown that the differences in ling-

uistic background a r e minimized (Pollack e t a1 1959). This i s very im-

portant in testing children with a congenital severe hearing l o s s a s they al-

ways suffer f r o m a pronounced but often unknown language retardation.

By using rhyme t e s t s i t is a l so easy to construct a number of different tes t

Page 6: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

l i s t s with the same difficulty. I t has a l so been shown that the learning

effect on repeated t r i a l s i s small (House et a1 1965). This effect i s es -

pecially useful in for example hearing aid fitting. By choosing the r e -

sponse alternatives in a systematic manner i t is possible to design t e s t s

that measure specific perceptual abilities. One problem with rhyme t e s t s

is that i t might be difficult to predict the ability to perceive everyday

speech f rom the tes t scores .

The t e s t s developed by Fairbanks (1958), House e t a1 (1965), Kreul l

et a1 (1968), and Griffiths(1967) a r e not suited for use by severely hard

of hearing and profoundly deaf children with a congenital hearing l o s s a s

they often a r e too difficult and contain many words that a r e not known to

the children. I t is a l so difficult to use this type of t e s t s i f a detailed

picture of the perceptual difficulties of an individual child i s wanted.

Description of diagnostic rhyme t e s t s

In designing the diagnostic rhyme tes t described in the following i t

was decided that the test should meet the following cr i ter ia :

(1) Only words well known to the subjects should be used.

(2) The tes t should consist of a number of sub-tests where each sub-test measured the subject ' s ability to use acoustic in- f ~ r m a t i o n mainly along one signal dimension.

(3) The tes t should be easy to use by the teacher in his daily work.

Several experiments with different rhyme tes t s of the above type have

been done a t the Dept. of Speech Communication during the l a s t ten years .

The a i m of the experiments has been to t r y this type of tes t for hearing

aid fitting, to study the relation between speech perception and speech

production, to study the perception of speech rhythm, etc. Four tes t

rounds have been made with l a rge r groups of subjects in order to collect

statist ical data about the speech perception problems of the severely hard

of hearing and profoundly deaf. Two of these tes t rounds were made a s

thesis studies a t the teacher training college for teachers of the deaf in

Stockholm (Edlund 1967, Adolvsson and Forsbn 1968). The two other

studies were done by MQrtony. Some resu l t s f rom these l a s t studies

have been published ear l ie r ( ~ Q r t o n y e t a1 1972, M&rtony 1974). In this

paper the r e su l t s f rom a l l four tes t rounds a r e summarized.

Table IV-A-I shows the contrasting phonemes that have been used in

different sub-tests in the four tes t rounds. All sub-tests a r e intended to

Page 7: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

TABLE IV-A-I. Contras t ing phonemes i n d i f ferent sub- tes t s .

Sub- tes t Te s t round Contrast ing phoneme Posi t ion no.

I one and two syl lables

2 /u:/-lo:/-/a:/ cv c 3 /u:/-/i:/, /a : / - /&: / CVC, CV

I one and two syl lables

2 /u: /- 10: /* 10: / - / a : / cv C

3 0 - /a/-/:/, u - : cvc* CVC/?/'

5 /n/-/l/ Initial & media l

6 /p/-/b/, /t/- /d/, /k/-/g/ Initial

I11 & IV 1 one and two syl lables

2 /u:/-/o:/, / o : / - / a : / cvc 3 /i:/-/y;/, /i:/-/ti:/ CVC, CV

4 /s/- /s t / , /s/- /t/, /st/-/t/ Initial

TABLE IV-A-11. 'Formant f requencies of long Swedish vowels. Male ta lker . Fan t (1973), p. 96.

Vowel F l F2 F3 F4

tu:3 290 595 2330 3260

Lo: 1 390 690 2415 3160

[a: 1 600 925 , 2540 3320

E m : 1 625 1720 2500 3440

r e :I 505 1935 2540 3370

[e:] 345 2250 2850 3540

[i: 3 255 2190 3150 3730

Page 8: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

STL-QPSR 2-3/1976 45.

measure one specific ability of the subjects to use acoustic speech in-

formation. The selection of contrasting phonemes i s based on this a s -

sumption but i t i s , however, not possible to avoid the use of secondary

acoustic cues by some subjects to select the correct response alternative.

As the f i rs t sub-test a test that measures the ability of the subjects

to tell i f a word has one or two syllables has been used i n al l tes t rounds.

Words a r e chosen that i n their one- syllable form a r e nouns, e. g. l/ma:t/

(food) and that form verbs by adding an /a/-sound: /ma:ta/ (feed). The

syllable in the two- syllable word has about equal s t ress . Word pai rs of

this type where the final consonant in the one-syllable word i s a plosive o r

an unvoiced fricative a r e easier than words, where the final consonant i s

a voiced continuous consonant, a s e. g. /ta:l/-/ta :la/ (speech, s peak)

(Mhrtony 1974). The test can therefore be divided in two sub-tests if

more detailed information about the subject' s ability to detect time-inten-

sity variations is needed. It i s possible to increase the number of r e -

sponse alternatives to three by selecting tr iads of the type /ma:t/ (food),

/ma:ta/ (feed), /ma :tada/ (fed). It i s of course also possible to select

pa i rs or triads with very different phoneme structure. In this case i t i s ,

however, not possible to tell i f a correct answer i s based on the percep-

tion of the number of syllables in the word or on the difference in pho-

neme structure. This might, however, be possible to determine f rom

the resul ts of other sub-tests. A test of this type has been described by

Erber and Alencewicz (1976). This test i s especially useful for testing

young children a s words can be selected that a r e wellknown to the sub-

jects and which also can be illustrated.

More difficult tests of the ability to perceive speech rhythm can be

constructed by selecting response alternatives that differ with respect to

the s t r e s s pattern in words or sentences. Results from this type of tests

have been reported by ~ h r t o n y (1974). I

As the second sub-test a test has been used that measured the ability

to use formant frequency differences in a frequency range below 1500 Hz.

In the test word pairs containing the back vowels /u:/, /o:/, and /a : / a r e

used. The formant frequencies for these vowels for a male talker a r e

shown in Table IV-A-11, Fant ( 1973).

Page 9: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

STL-QPSR 2-3/1976 46.

The intensity of t h t third and fourth formants in the back vowels /u:/,

/o:/, and / a : / is about 30 dB below the intensity of the f i r s t formant. The

higher formant can however influence the discrimination between the vowels

for subjects with better hearing i n the higher than in the lower frequencies.

This type of hearing los s is very uncommon.

I t i s possible to find t r iads of rhyming words with these vowels and

this was used i n tes t round I. In Swedish, however, this often has the

resul t that one of the words is ve ry uncommon o r is a v e r b fo rm that I

might not be known by the subjects. 1 i I The total intensity of the three phonemes i s not the same and the inten-

sity difference between /u:/ and /a :/ i s especially large. To avoid the

use of the intensity difference a s a cue only the p a i r s /u:/-/o:/ and

lo:/-/a:/ a r e used in tes t rounds 11, 111, and IV.

The third sub-test is a high-frequency vowel test . Words a r e select-

ed that differ with respec t to a vowel phoneme where the frequency of the

f i r s t formant is about the same but where the difference in the frequency

of the second formant in the frequency range above 1500 Hz is large. In

the four t e s t rounds different contrasting phonemes have been used, see

Table IV-A-I. The formant frequencies of the vowels a r e shown in Table

IV-A-11.

Based on the same principle a l a rge number of different sub-tests that

measures the ability to use different acoustic cues that distinguish the

consonants can be constructed. I n the four t e s t rounds reported on he re

three consenant t e s t s have been included, see Table IV-A-I. In tes t

round I1 a t e s t that measured the ability to identify i f a word contained

/n/ or /1/ in initial and medial position was included. The acoustic dif-

ference between these sounds l i e s in the spectrum of the /n/ and /I/-

sounds, in the transit ions to the following vowel and i n the presence of

nasalization nf the vowel followed by /n/ ( ~ a r l s o n et a1 1972. Dalston

1975, Ohman 1962). In the same tes t round a sub-test was included that

measured the ability to detect if an initial plosive sound was voiced o r

unvoiced. This distinction i s signaled by a number of acoustic cues.

The most important cues a r e the presence o r absence of a voiced occlu-

sion, the t ime lag between the explosion and the onset of voicing and the

length of the formant transit ions in the vowel (Slis and Cohen 1969).

In t e s t rounds 111 and IV a consonant tes t was included that measured

the ability to identify /s/, /st / , and /t/ in initial position. This sub-test

Page 10: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

STL-QPSR 2 - 3 / 1 9 7 6

measures the ability to detect the presence and timing of noise energy in

a frequency range above 2000 Hz. As a secondary cue the length of for-

mant transition in the following vowel can be used. These transitions a r e

long for /s/ and /st/ but short for /t/.

Test procedure

Test l i s t s were cqnstructed by choosing 4- 10 word pairs that contained

the contrasting phinemes that were to be tested. A l i s t of 15-30 pai rs was

constructed by repeating the selected pairs .

Test l i s t s were recorded in a silent room using a good tape recorder

and microphone. In test rounds I and I1 a female talker was used and in

test rounds I11 and I V a male talker. The words were preceded by a car-

r i e r phrase "Now you will hear . . . ". In test rounds I and I1 the test word

after the ca r r i e r was spoken once and in test rounds I11 and IV twice.

Response sheets were prepared where al l pa i rs were listed in the order

they were presented. The subjects heard one of the words in the pair and

underlined the word he thought he heard.

Before testing i t was made sure that the subject knew the words and

under stood the task. Short training was given on each list.

The subjects listened through the group amplifier they used daily and

they adjusted the output during the training part of the test to what they

considered optimum.

Subjects

As the subjects must be able to read, only subjects from the age of

10 years were used. In Table IV-A-I11 and IV-A-IV the number of sub-

jects in different iO-dB intervals i s shown. InTable IV-A-111 the mean

hearing loss at 500, 1000, and 2000 Hz i s used a s a measure of the degree

of hearing loss and in Table IV-A-IV the hearing loss a t 2000 Hz. The sub-

jects a r e f rom the school for the deaf i n Stockholm (I and 11). from the

school for the deaf in tjrebro (JII), and f rom the school for partially hear-

ing in Stockholm (IV). All subjects have been auditory trained for a t least

four years using group amplif iers and individual hearing aids.

Results and discussion - In calculating the resul ts on the rhyme test the per cent correct rela-

tive to guessing i s used. If the number of correct answers i s below what

Page 11: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR
Page 12: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

TABLE IV-A-IV. Number of subject6 in 10 dB intervals. Hearing loss at 2000 Ha.

N u m b e r of s u b j e c t s Test Age Total

50-59 dB 60-69 d B 70-79 d B 80-89 dB 90-99 dB 100-109 d B 110- dB

I 13- 17 I 2 3 6 4 7 21 44

Total 10 2 3 3 3 32 19 33 84 234

Page 13: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR
Page 14: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

100

50

NUMBER OF . SYLLABLES

X ALLTESTS t " " 50- 60- 70- 80- 90- 100- 110- 59 69 79 89 99 109

M E A N HEARING LOSS. ~ B M 500,1000,2000 Hz

Fig. IV-A- I . Median and interquar t i le range on the sub- t e s t that m e a s u r e s the abil i ty to detect i f a word h a s one o r two syllables.

MEAN HEARING LOSS. ~ B M 500.1000,2000 Hz

Fig . IV-A-2 . Median and interquar t i le range on the sub- tes t that m e a s - u r e s the abil i ty t o u se formant frequency information below 1500 Hz. Contrasting phonemes /u:/, /o:/, / a : / ,

Page 15: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

- - - - - - - -

50 - - - - v m - X ALLTESTS

0 - I I I t 1 I I

- b

50- 60- 70- 80- 90- 100- 110- 59 69 79 89 99 109

HEARING LOSS AT 2000 Hz, dB,

Fig. IV-A-3. Median and interquartile range on sub-test that measures the ability to use formant frequency information above 1500 Hz. Contrasting phonemes, see Table IV-A-I.

HEARING LOSS AT 2000 Hz,dB

Fig. IV-A-4. Median and interquartile range on the sub-test that measures the ability to detect timing of noise energy. Contrasting phonemes /s/, /st/, and /t/.

Page 16: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

STL-QPSR 2-3/1976 5 1.

111 and IV only words of the type /&:dl-/ba:da/ were used. In these

words the intensity variations in the two-sylliible word a r e more pro-

nounced than in words of the other type. See also the resul ts published

by ~ S r t o n y (1974). In test rounds 111 and IV training on the test material

was also given and the resul ts shown a r e those obtained after four test

se s sions.

For sub-test 2, the-.difference-between-/u-:/, /o:/, an d / n , : / , the me-

dian value i s about the same for all test rounds. 50 QJo of the- children with a

mean hearing loss in the range 90-99 dB manage this test but only about

25 70 in the group with a mean hearing loss in the range 100- 109 dB. For

a mean hearing loss of more than 100 dB only about l o - % of the- children got

a score of more than 50 OJo over guessing. I The main acoustic difference between the phonemes /u:/, /o:/, and 1

/a :/ i s the frequency of the combined formant one and formant two group,

see Table IV-A-I. There i s , however, also a difference in total intensity

and this intensity difference can be enhanced by the shape of the audiogram

and the frequency response of the amplifying system. I t i s possible that

some children with a severe loss use this intensity difference a s the prim-

a ry cue and not the difference in formant frequency. The possibility that

secondary cues a r e used by some children must always be taken in ac-

count and i t i s therefore necessary to interpret the resul ts from this kind

of simple rhyme tests with caution. " By.. comparing the resul ts from dif-

ferent sub-tests i t might often be possible to determine if pr imary o r

secondary cues have been used. Another possibility is to use synthetic

speech sounds instead of r ea l speech. In this case secondary cues can

more easily be controlled (Pickett and MBrtony 1970).

The third sub-test, the high-frequency vowel test, i s plotted against

the hearing loss a t 2000 Hz in 10-dB groups a s the important acoustic dif-

ferences in this case lie in a frequency region between 1500 and 3000 Hz.

This test i s quite difficult and the degree of hearing loss, where about 50

% of the subjects get a score better than 50 70 over guessing, l ies around

75 dB. Even in the rahge 50- 59 dB hearing loss about 25 % of the children

do not get a score better than 50 70 over guessing.

The contrasting phonemes in sub-test 3 were not the same in the dif-

ferent test rounds, see Table IV-A-I. In the f i rs t two test rounds the dif-

ference was between front and back vowels and in the las t two test rounds

Page 17: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR
Page 18: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

50- 60- 70- 80- 90- 100- 110- 59 69 79 89 99 109

MEAN HEARING LOSS, ~ B M 500,1000,2000 Hz

Fig. IV-A-5 . Median and interquartile range on the sub-test that meas- ures the ability to identify the difference between /n/ and /1/ and between voiced and unvoiced plosive sounds.

Page 19: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

STL-QPSR 2-3/ f976 53.

only included in test found 11. The number of subjects in each 10 dB-

interval i s small and the resul ts must therefore be seen a s preliminary.

Recent studies with the same sub-test, however, confirms the general

resul ts shown in Fig. IV-A-5. Both tests a r e very difficult for subjects

with a mean hearing loss of more than about 70 dB.

The pr imary acoustic difference between /n/ and /1/ i s in spectral

differences, formant transitions in the adjacent vowel, and the presence

of nasalization in the vowel followed by /n/. The main difference in the

frequency spectrum of /n/ and /1/ l i e s around 2000 Hz. This information

i s difficult to use for sevepely hard of hearing subjects, especially if low-

frequency information i s simultaneously present. These subjects have

also difficulties in using information in formant transitions, a s was shown

by Danaher, Osberger, and Pickett (1973)~ These difficulties seem to ex-

plain the poor resul t on this sub-tedt.

For the difference between voiced and unvoiced plosive sounds many

different acoustic cues can be used. The most important cue i s probably

the presence of a voiced occlusion in voiced plosives, the length of the 1 formant transition, and the time lag between the explosion and onset of the

vowel ( ~ l i s and Cohen 1969). Other cues a r e changes in the fundamental

frequency, the ra te of change of energy in the low frequencies etc. It is

difficult to say which of these cues that a r e pr imary cues for the severely

hard of hearing subjects used in the study. In a study by Bennet and Ling

(1973) it was, however, shown that profoundly deaf subjects cannot use

voice onset time a s a cue for voiced/voiceless identification. A s the dis-

tinction between voiced and unvoiced plosive sounds cannot be seen on the

talker' s face (woodward and Barber 19 60), i t i s very important to study

these problems to find methods for training this distinction. Synthetic 1

speech with enhanced contrast between the different cues might be used in

this training . The use of the rhyme tes ts in diagnosis

The resul ts obtained in the four test rounds a r e a t present used a s a

preliminary standard against which the resul ts f rom individual subjects a r e

compared. As an example, Fig. IV-A-6 shows the resul ts obtained f rom

two groups of subjects. The f i r s t group consists of s ix subjects with a

hearing loss acquired at the age of 4- 6 years , and the second group of s ix

congenitally deaf subjects that have been intensively auditory trained. The

Page 20: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR
Page 21: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

STL-QPSR 2- 3/1976 54.

age of the subjects in the f i r s t group is 14 years and in the second 16

yea r s a t the t ime of testing.

The tes t resu l t s f rom the two groups do not differ on the low-frbquency

vowel t e s t s but on the high-frequency vo\kel tes t one of the subjects with

a congenital deafness get a very high score. On the consonant tes t the

grsup with acquired hearing los s a p p e l r s to be better than the group with

congenital hearing lass. Thik might then show the effect of.very ear ly

auditory kraining. A s far a. is known the intensive auditory training of

the group with congenital hearing s tar ted ra ther late. I

In Fig. IV-A-7 some resu l t s a r e shown for dlder subjects with differ-

ent etiology. Especially interesting is subject A F , where the cause of

deafness is the t reatment with streptomycine. This subject gets a poor

score on the two vowel t e s t s but reasonably good sco res on the consonant

tests. The difficulties with the vowel tes t seem to be explained by a very

poor frequency discrimination ability. Evans ( 197 5) has recently shown

that canamycin and other drugs of the same type can resul t in a hearing

impairment characterized by a reduced frequency discrimination ability.

In Fig, IV-A-7 the frequency discrimination ability for sinusoidal signals

is a lso shown (Risberg and Agelfors 1975) .

Conclusion

The rhyme test of the described type can be easily administered to

severely hard of hearing and profoundly deaf subjects. The limitation i s

that the subjects must be able to read and that the test r e su l t s might be

influenced by variations of the subject ' s ability to combine an orthographic

symbol with a sound. These difficulties can be avoided by selecting the

words in the t e s t ve ry carefully and in some cases selecting words that

can be i l lustrated. The subjects must a lso be carefully instructed. The

t e s t can be used in a learning situation where different scramblings of the

tes t list a r e used and the subjects a r e trained to a crude asymptote. In

the training teaching machines can be used (Doehring and Ling 197 1). In tes t rounds I11 and IV this technique was used and the resu l t s shown a r e

the final r e su l t s a f te r 4-5 training sessions. A training effect was ob-

tained on al l t e s t s but was statistically significant only on sub-test 2;

low-frequency vowel discrimination. In F ig . IV -A-8 the relation be-

tween resu l t s obtained on the f i r s t and the fourth test sessions is shown.

Page 22: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

I s / - / s t / - I t / OIo I 8 I I I I I +

1 0 0 - \ - -

50 -

- 0 - &-- -

t I I I I I I

So- 70- 90- 110- 59 79 99

Of0 loo

5 0 1 0

SUBJECT AGE ar % CAUSE O F DEAFNESS 250 500 lk Hz

0 AF , 31 7 .- - STREPTOMYCIN, 15 YEARS

$ AL 35 3 2 3 MENINGITIS, 2 YEARS I I

U KR 28 7 - 15 HEREDETARY, PROGRESSIVE

V B M 24 3 3 2 CONGENITAL, UNKNOWN

A IB 35 20 20 5 - 1' - - 1' -

Fig. IV-A-7. Results from measurements with rhyme tests on older subjects with different etiology. Frequency discrim- ination ability for sinusoidal signals is also shown.

Page 23: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

TEST SESSION 1.

Fig . I V - A - 8 . Effect of training on r e s u l t s l r o m a rhyrne tcs t . The f i g u r e shows the r e s u l t s f r o m the f i r s t a n d the fourth t c s t s e s s i o n with the tcs t that m e a s u r e s the abi l i ty to use fo rmant f requency informat ion i n the f re quency r a n g e below 1500 H z i n tcs t rounds I11 and IV. The t r a in ing w a s g iven by m e a n s of a teaching ~ n a c h i n e and the sub jec t s w e r e told if t h e i r r e s p o n s e s w e r e cor rec t o r not.

Page 24: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

STL-QPSR 2-3/1976 5 5.

The effect of training i s significant in the two groups with the mean

hearing l o s s between 80-89 and 90-99. In the groups with a mean hearing

l o s s of 70-79 dB and in the groups with a mean hearing los s above 100 dB

no significant effect of training was obtained. If training is used in the

test , i t i s possible that the subject l ea rns to use secondary acoustic cues

in the words. I t is therefore probably necessary to t ra in the subjects on

one set of words o r the recording f rom one speaker and then tes t them on

another set of words o r a recording f rom another speaker.

By establishing the median and quarti le values for subjects in differ-

ent hearing los s and age groups, i t i s possible to establish a t e s t score

against which different sub groups o r individuals can be compared, a s

shown in Figs . IV-A-6 and IV-A-7. By means of this technique i t might

be possible to study some of the factors that influence the resu l t s of

auditory training.

The tes t method i s based on acoustic differences between the contrast-

ing phonemes in the t e s t words. This means that the same type of t e s t i can be constructed for a l l languages and a standardization made in one

country can be used in another country, where a different training method I

o r technical equipment i s used. The tes t r e su l t s shown in this paper

might be used as a prel iminary standardization but to get a more reliable

standard the different factors influencing the r e su l t s must be analyzed and

a standard tes t procedure established.

Acknowledgments

This r e sea rch has been supported by the Swedish Boaxdfor- Tech-

nical Development. The measurements have been made a t the School for

the Deaf in Stockholm (Manillaskolan), the School for the partially hearing

in Stockholm (Alviksskolan), and the School for the Deaf in Orebro. I

wish to thank the staff of these schools for their cooperation.

The measurements summarized in this repor t have been made by

Edlund, Adolvsson, and F c c s b a s a thesis work a t the teacher training

college in Stockholm and by Jsnos Msrtony a t the Department of Speech

Communication of the Royal Institute of Technology in Stockholm.

Page 25: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

STL-QPSR 2-3/1976 56.

References:

ADOLVSSON, K . and FORSEN, H. ( 19 68) : I f Samband mellan talfor s t l e l s e och talkvalitet" (Relation between speech perception and speech quality), Thesis study a t the Teacher Training College in Stockholm.

BENNETT, C. W. and LING, D. (197.3) : "Discrimination of the voiced- voiceless distinction by severely hearing-impaired children", J. Aud. Res. - 13, pp. 271-279.

BOOTHROYD, A. (1972): "Audiological evaluation of severely and pro- foundly deaf children", in P r o c . of Int. Symp. on Speech Communication Ability and Profound Deafness, Stockholm 1970 (Ed. G. Fant), A. G. Bell Association for the Deaf, Washington, D. C . , pp. 47-56.

CARLSON, R . , GRANSTROM, B. , and PAULI, S. (1972): "Perceptive evaluation of segmental cues", STL-QPSR I/ 1972, pp. 18- 24.

CRAMER, K. D. and ERBER, N. P. (1974): "A spondee recognition tes t for young hearing-impaired childr ent' , J . Speech and Hearing Dis. - 39, pp. 304-31 1.

DALSTON, R. M. (1975): "Acoustic character is t ics of English /w, r , 1/ spoken correct ly by young children and adults", J. Acoust. Soc. Am. - 57, pp. 462-469.

DANAHER, E. M., OSBERGER, M. J. , and PICKETT, J . M. (1973): "Discrimination of formant frequency transit ions i n synthetic vowelsf1, J. Speech and Hearing Res . - 16, pp. 439-451.

DANAHER, E. M. and PICKETT, 3. M. (1975): "Some masking effects produced by low-frequency vowel formants in persons with sensorineural hearing loss", J . Speech and Hearing Res. - 18, pp. 261-271.

DOEHRING, D. G. and LING, D. (1971): "Programmed instruction of hearing impaired children. in the auditory discrimination of vowels", J. Speech and Hearing Res. - 14, pp. 746-753.

EDLUND, A-L. ( 1967): "Under stikning av s v l r t h6r selskadade barns for- mHga till frekvensdiskrimination och rytmuppfattning" (Study of the f r e - quency discrimination ability rhythm perception ability of severely ha rd of hearing children), Thesis study a t the Teacher Training College in Stockholm.

ERBER, N.P. (1972): "Auditory, visual, and auditory-visual recogni- tion of consonants by children with normal and impaired hearing", J. Speech and Hearing Res . 15, pp. 41 3-422.

ERBER, N.P. (1974): "Pure-tone thresholds and word-recognition abili t ies of hearing-impair ed children", J. Speech and ~ e a r i ; ~ Res. -* 17 pp. 194-202.

ERBER, N.P. and ALENCEWICZ, C. M. (1976): "Audiologic evaluation of deaf children", J. Speech and Hearing Dis. 41, pp. 256-267. . ! EVANS, E. F. (1975): { 'The sharpening of cochlear frequency selectivity in the normal and abnormal cochleau, Audiology 14, pp. 419-422.

FAIRBANKS, G. (19 58) : "Test of phonemic differentiation: The rhyme test", J.Acoust. Soc.Am. 30, pp. 596-600. - FANT, G. (1973): Speech Sounds and Fea tures , The MIT P r e s s , Cambridge, Mass . , and London, pp. 94-99.

GRIFFITHS, J. D. (1967): "Rhyming minimal contrasts: A simplified diagnostic articulation t e s t t f , J. Acoust. Soc. Am. 42, pp. 236-241. -

Page 26: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

STL-QPSR 2-3/1976 57.

HOUSE, A. S. , WILLIAMS, C. E. , HECKER, M. H. L. , and KRYTER, K. D. ( 1965): "Articulation- te sting methods: Consonant differentiation with a closed-response set", J. Acoust. Soc. Am. - 37, pp. 158- 166.

HUDGINS, C. V. (moderator) (19 53) : "Speech and speech perception!' A summer meeting Pane l Discussion, June 18, 1952. In Volta Review 5 5 -, pp. 20-35.

HUDGINS, C. V. (1954): "Auditory training: I t s possibil i t ies and l imita- tions", Volta Review - 56, pp. 339-349.

HUDGINS, C.V. (1954): "A method of appraising the speech of the deaf", Volta Review - 51, pp. 597-

KREUL, E. J. , NIXON, J. C . , KRYTER, K.D., BELL, D. W . , and LANG, J. S. ( 1968) : "A proposed clinical t es t of speech discrimination", J. Speech and Hearing Res . fi, pp. 536-552.

LIBERMAN, A. M. , COOPER, F. S. , SHANKWEILER, D. P. , and STUDDERT-KENNEDY, M. ( 19 67) : "Perception of the speech codeI1, Psychol. Rev. - 74, pp. 431-461.

MARTIN, E . S. and PICKETT, J. M. (1970): "Sensorineural hearing l o s s and upward spread of masking", 3. Speech and Hearing Res . - 13, pp. 426-437.

MARTONY, J. ,RISBERG, A . , SPENS, K-E. , and AGELFORS, E . (1972): "Results of a rhyme-test for speech audiometry", in P r o c . of Int. Syrnp. on Speech Communication Ability and Profound Deafness, Stockholm 1970 (Ed. G. Fant), A. G. Bell Association for the Deaf, Washington, D. C., pp. 75-80.

MARTONY, J. (1974): "On a rhyme test", STL-CPSR 2-3/1974, pp. 57-7 1. I

MILLER, G.A. and NICELY, P , E . (1955): "An analysis of perceptual confusion among some English consonants", J. Acoust. Soc. Am. - 27, pp. 338-352.

NUMBERS, M. E. and HUDGINS, C. V. (1948): "Speech perception i n presen t day education for deaf children", Volta Review 50, pp. 449-456. - PICKETT, J. M. and ARTONY, J. (1970): "Low-frequency vowel formant discrimination in hearing-impair ed l isteners", J. Speech and Hearing Res . - 13, pp. 347-359.

PICKETT, J. M. , MARTIN, E. S., JOHNSON, D. , BRAND SMITH, S. , DANIEL, Z., WILLIS, D. and OTIS, W. (1972): "On pat terns of speech feature reception by deaf l is teners" , in P r o c . 1nt. symk. Speech corn- munication Ability and Profound Deafness, Stockholm 1970 (Ed. G. Fant), A. G. Bell Association for the Deaf, Ip?ashington, D. C. , pp. 119- 133.

POLLACK, I . , RUBENSTEIN, H. , and DECKER, L . (1959): "Intelli- gibility of kaown and unknown message sets", J. Acoust. Soc. Am. 31, - pp. 273-279.

RISBERG, A. and I ~ R T O N Y , J. ( 1972): "A method for the classification of audiograms", i n P r o c . Int. Symp. on Speech Communication Ability and Profound Deafness, Stockholm 1970 (Ed. G. Fant), A. G. Bell Associ- ation for the Deaf, Washington, D. C. , pp. 135- 139.

RISBERG, A., AGELFORS, E , , and BOBERG, G. (1975): "Measurements of frequency-discrimination ability of severely and profoundly hearing- impaired children", STL -QPSR 2- 3/1975, pp. 40- 48.

Page 27: Diagnostic rhyme test for speech audiometry with severely hard of … · speech audiometry with severely hard of hearing and profoundly deaf children Risberg, A. journal: STL-QPSR

STL -0PSR 2- 3/1976 58.

SLIS, I. H. and COHEN, A. (1969): "On the complex regulating the voiced-voiceles s distinction II", Language and Speech - 12, pp. 137 - 155.

WALDEN, B. E. , PROSEK, R. A. , and VJORTHINGTON, D. W. (1975): "Auditory and audiovisual feature t ransmission in hearing-impair ed adults", J. Speech and Hearing Res. Is, pp. 272-280.

VAN UDEN, A. (1970): "New realizations i n the light of the pure ora l method", Volta Review 72, pp. 524-537.

WEDENBERG, E. (1954): "Auditory training of severely hard of hearing preschool children", Acta Oto-Laryng. , Suppl. 110.

WOODWARD, M. F. and BARBER, C. G. (1960): "Phoneme perception in lipreading", J. Speech and Hearing Res. - 3, pp. 21 2- 222.

WATSON, T. J. (1967) : The Education of Hearing Handicapped Children, pp. 60-63, University of London P r e s s , Ltd. , London.

OHMAN, S. (1962): "On the perception of Swedish consonants in inter- vocalic position1', Speech Transmission Laboratory, KTH, Report No. 25.