COMD 3700 – Lesson 10
Speech Audiometry II
COMD 3700 for Distance Education. This is lesson 10 on speech
audiometry. This is the second lesson on speech audiometry. This lesson
will cover pages 139-142, 147 (recording speech-recognition test results)
and 149-160 in Chapter 5 of your textbook.
As we discussed in lesson 9, there are 5 Factors in Speech Audiometry:
Test Equipment, Test Environment, The Patient, The Clinician and The
Test Procedure. We will continue to discuss the testing procedure in this
lesson.
The speech testing we will discuss in this lesson are the MCL, UCL and
WRS or speech recognition testing, not to be confused with the speech
recognition threshold.
Let's discuss most comfortable loudness level. The purpose of the MCL is
to determine the best level for the patient to hear and understand speech.
The MCL gives important information both with regard to the patient and to
the fitting of a hearing instrument. In a patient with normal hearing, most
comfortable loudness level is somewhere from 40 to 55 dB HL. The
stimulus for finding this is cold running speech. As discussed in lesson 9,
cold running speech is an informative but non-emotional series of
sentences, recorded at a constant level. Cold running speech carries a lot
of redundant information. The patient hears and understands the subject
without realizing that each individual word may not be clear. Because
presentation of this speech may be comfortable at any sensation level
above SRT, a bracketing method determines a more accurate MCL. The
reason for this is that the MCL is a little hard for a patient to identify. It may
be hard to define. A person could be 5 or 10 dB on the high or low side.
We'd like to narrow that a bit and try to get a finite dB level that represents
MCL. To accommodate this, we use the bracketing method.
First, we instruct the client how to behave and respond. The instructions to
the client may be something like this, “Now you are going to hear
continuous speech. You do not have to repeat any of the words. I want to
find the most comfortable level for you to listen. I will increase the volume
slowly, so you can listen at different levels. You can help me by pointing
your thumb up if you want it louder, or down if you want it quieter. Please
let me know when we find the volume you prefer. Do you understand?”
So we start our bracketing level by first presenting a stimulus to them that
we know will be below their comfortable listening level or too soft. This will
be at a hearing level slightly above the SRT. Some audiologists start at a
40 dB sensation level or 40dB above SRT. Once you start speaking, the
patient will raise their thumb that the sound needs to be increased. We're
not going to increase it by five dB or ten dB and then say a word like we did
in finding SRT. Instead we'll continue with cold running speech and quickly
increase in 5 dB steps until the speech is being presented above their
threshold but just below their uncomfortable loudness level. They will put
their thumb down for softer. Then decrease the speech in 10 dB steps until
they say it is comfortable. You can go up 5 or down 10 until you find the
comfortable level for them. The forced choice method will help determine
an accurate MCL. To do this you always advance the attenuator dial to a
point a little above where the patient reports speech as comfortable. While
really listening quietly, realization may come to the patient that speech can
be louder and perhaps more comfortable. Continue to increase slowly in 5
dB steps until the patient decides that speech was better lower, then return
to that level. This level is noted on the audiogram as the MCL. Have
patient think about how they’d listen to the TV. Do they want the volume
louder or softer?
Next we will determine the uncomfortable listening level or UCL of the patient.
The UCL is not the threshold of pain; it is a level where speech becomes
uncomfortably loud. The purpose is to determine the point where no further
amplification will be permitted in a hearing instrument or accepted by the
patient. For patient with normal hearing, UCL is usually 100-110 dB. 110 dB is
usually the maximum level we deliver to any client. Levels exceeding 110 dB
have the power to damage hearing. Usually cold running speech is used to
find UCL. We talked about running speech as being something along "Mary
had a little lamb. Its fleece was white as snow. And everywhere that Mary
went, her lamb was sure to go." An important point that I want you to
remember is that hearing losses do not necessarily extend the UCL. That
means you'd assume if a person had normal hearing and the UCL is 100-110
dB, you'd assume someone with a hearing loss would have a UCL of 120 to
130 dB. This is not necessarily true. People with sensorineural hearing losses
in fact have many times the opposite condition where the uncomfortable
loudness level is lowered. So, instead of being able to tolerate 100-110 dB,
they may be able to only tolerate 80-90 dB so their ceiling is lowered. The
UCL is also known as: TD: Threshold of Discomfort, Tolerance Level and
LDL: Loudness Discomfort Level.
So an example of patient instructions would be, “Now I want to find a level
where sound starts to bother you, not where it is just a little too loud. I want to
find where it is uncomfortable for you, but not painful. I don’t want you to hold
your ears in pain. Please raise you hand or stay ‘stop’ whenever the sound
starts to bother you to the point where you couldn’t listen to it or when
exceeding the loudness level would be intolerable.” We begin by delivering
the speech at a comfortable loudness level. We'd then begin extending the dB
or the intensity, making the cold running speech more and more intense.
They'd raise their hand eventually. You don't just stop and turn it off when they
say that's as loud as they tolerate. You would then take the hearing level dial
back down and increase it again back to an uncomfortable level, noting the
highest level they could tolerate and record that as their uncomfortable
loudness level.
Another use for the UCL is to help define the dynamic range of the patient.
This is also known as the range of comfortable loudness (RCL). The
dynamic range is the range of usable hearing, between the SRT and UCL.
This is calculated by subtracting the SRT from the UCL. So, for example for
patients with normal hearing, their dynamic range may be 100 dB, or
120dB (UCL) minus 20 dB (SRT). Usable hearing takes place within this
dynamic range, or area of comfortable loudness. If the frequency and
intensity of a sound are inside this area, we readily hear it. Any sound that
falls below or outside this area remains unheard. So if you present the
speech at a stimulus lower than the SRT, you're not going to function very
well. If sounds are more intense than the UCL you will be uncomfortable.
So, a person with a dynamic range of 50 dB has no problem listening to
recorded music, since most albums are produced at about a 35 dB range
from the faintest to loudest sound. The volume can be set at a comfortable
level where the soft sounds can be heard and not bothered by the louder
ones. This person has more difficulty with a live orchestra, where the
dynamic range would be 70 dB or greater. Sitting in the back, soft sounds
would be missed, whereas by moving closer, the person would be bothered
by the loud sounds.
Now we will discuss speech recognition testing. There's a distinction
between a threshold for speech and ability to understand the speech that is
heard. Like I mentioned before there's a lot of complaints by patients with
hearing impairments saying, “I can hear speech but I can't understand it” or
“Words are not clear”, “speech sounds muffled or distorted” or “I mistake
one word for another”. The common theme here is that the speech heard
by the patient is lacking in intelligibility. This problem is experienced in
terms of inaccurately receive messages or reduced clarity. So that's why
we have speech recognition testing.
Speech recognition testing is also referred to as Word Recognition Testing.
This to me is easier to understand. But the book refers to it as speech
recognition testing. We have discussed the speech recognition threshold
testing so I want you to realize that there's a difference between the SRT
and what is referred to as speech recognition testing. This type of testing
can also referred to as Word discrimination testing or Speech
Discrimination (SD) testing. The term “discrimination testing” is older, but
still popular. You will often hear an audiologist refer to this testing as
“discrim testing”, as in “I did SRT and discrim”. The measurement of
speech discrimination is referred to as: Speech recognition score (SRS),
Word recognition score (WRS) or the Speech discrimination score.
The purpose of speech recognition testing is the quantitative determination
of a patient’s ability to discriminate speech. We need to determine how well
the patient hears and understands speech. This helps the audiologist in
several ways: It determine the extent of the speech-recognition difficulty, it
aids in diagnosis of the site of the disorder in the auditory system, it assists
in the determination of the need for and proper selection of hearing
instruments and it helps the clinician make a prognosis for the outcome of
treatment efforts.
So although there are many terms to define this testing, the speech intelligibility testing can be tested clinically in a very
straightforward manner. The most common approach is to present the patient with a list of phonetically balanced (PB) test words.
Phonetically balanced just means that the percent of time any given sound appears on the list is equal to its occurrence in the
English language. There are several different lists that are available for testing. Most commonly used are the CID Auditory Test
W-22 and the NU-6 lists. CID stands for Central Institute for the Deaf. Hirsch and others brought these lists together in 1952. They
put fifty words in a list and had four lists. There are six versions of each list with the words in a different order. These lists
provided an adequate number of words for finding word recognition score for a person. So each list contain 50 phonetically
balanced words. The NU-6 lists were developed at Northwestern University. There are 4 50-word lists scrambled into 4
randomizations. Either list is acceptable. For example, one Speech Discrimination list includes words such as: "are," "bar," "clove,"
"fraud," "hive," "nook," "pants," "rat," "slip," "there," "use," "wheat" and 38 others. There are other lists that are used for specific
purposes, such as testing children, but we will not be discussing them in this course. All of the words are one-syllable words. They
are designed to be common words that should be within the vocabulary of the average person. There are a variety of speech
recognition test materials, including CID Auditory Test W-22 and NU-6 provided in the Professional Resources section on the CD
included with your textbook. Although it is recommended that the full 50-word list is used in testing. Sometimes only a ½ list is
used. It is permissible to use a half list if an individual's score is greater than 90% on the first 25 words (misses no more than 2 out
of 25 words). Sometimes it may be necessary to administer more than one list. The audiologist does NOT familiarize the person
with the words. A new list of words has to be used each time the test is conducted (on the same day). The use of a carrier phrase
is advisable. The goal is for the words to have natural inflection. The test item should actually be slightly lower in intensity than the
“say the word” carrier phrase portion. The inflection should drop slightly at the end. For instance, say out loud “This is a book.”
Your pitch and loudness probably dropped slightly when saying “book”. This same intonation pattern should be used when you
say, “Say the word ‘carve’.” You don’t want to use a rising, singsong pitch when you present the test item. This is unnatural and
will create poorer scores if the patient has a high-frequency hearing loss. It also sounds unprofessional. Because “live voice”
testing by individual audiologists creates variability in the test presentation, it is preferable to use recorded materials. When
recorded materials are used, if patient performance differences from one test date to another, it’s easier to interpret that those
difference are because of the patient rather than because of the person administering the test.
The test presentation level can vary based on what the audiologist is trying to determine. The standard method is to set
the presentation level at a 40 dB sensation level (SL) above the SRT. Some clinics may also use 30 dB SL. Remember
that the sensation level will vary from patient to patient depending on the degree of their hearing loss. There was some
confusion about the sensation level after the last exam, so let me explain it again. Let's take three patients. Patient #1
has an SRT of 10 dB. So the softest speech they could hear was 10 dB. Patient #2 has an SRT of 20 dB. Patient #3 has
an SRT of 30 dB. So we have three patients here with three different speech reception threshold levels. We want to
deliver a word recognition score to each of these patients that will be perceived at the same level. We use 40 dB SL as
the presentation level for our clinic. So when a person comes in to our clinic, we'll administer our word recognition level
at 40 dB SL. So when we administer the test to a patient #1 (with an SRT of 10 dB) the test presentation level would be
50 dB HL. Then patient #2 (with SRT of 20 dB), we'd deliver the test to them at 60 dB. We would present the words to
patient #3 (SRT of 30 dB) at 70 dB HL. We are adding 40 dB to the SRT of 30 dB. That is why we administer the test at
70 dB. So although it is measured at 70 dB HL, to the patient it is perceived as 40 dB or 40 dB sensation level. Once the
level has been established for each patient, the intensity is not changed during the test. The intensity is left at that SL
level. What happens is we now have each patient receiving the test at the same 40 dB SL so each patient has no
advantage over any other. So hopefully you now understand how we found the 40 dB SL for each patient. Some patients
with moderate sensorineural losses or worse probably have significant recruitment (can’t tolerate loud sounds), and
would find even 30 dB SL too loud. If these levels are uncomfortably loud for the individual then testing should be
performed at the most comfortable listening level (MCL) of the patient. The textbook also discusses performing the
testing at a minimum of two levels. The first level is presented at 5 to 10 dB above the patient’s MCL. The second level
is presented at a higher intensity (90 dB). You can read more about this on page 151. If the audiologist wishes to know
how well a person understands words when they are presented at a conversational loudness, they may present the test
items at a level of 45 to 50 dB HL. (All words are presented at one intensity.) If the audiologist wants to know how well
the person can do when speech is made comfortably loud, the testing can be conducted at MCL.
The patient instructions are very straightforward and may be something
like, “You are going to hear a series of words. Please repeat each of the
words you are asked to say. If you have difficulty understanding the words,
it helps me if you guess. I have a better idea what part of the word was
difficult for you. If the volume is not comfortable for you, or you would like
the words louder or quieter, please let me know. Do you understand?”
So far we have been expressing the speech scores and speech test results
as finite dB levels such as: SRT is 30 dB, the UCL is 90 dB, the dynamic
range is 55 dB, the MCL is 35 dB. With word recognition we'll talk in terms
of percentage. So we'll report an individual's word recognition score as
48% or 92%. The percentage of the test words correctly repeated by the
patients is recorded. Each correct response is worth 2% if using a 50-word
list and 4% if using a 25-word list. If live voice testing is used then you have
a form where you record next to each word if they got it correct or if they
missed it. When using recorded material on a computerized audiometer,
then the list of words is on the computer and after each word you indicate
whether or not the patient responded correctly. This is the picture on the
slide. So, If 50 words are given, and the patient misses 11 items, what is
the percent correct score? 78% because each error is worth 2 points when
50 words are given. If 25 words are given, and the patient misses 6 of the
words, what is the percent correct score? 76% because 25 words are
given, each error counts as 4% off (since 4 X 25 = 100). The list used and
the level at which the testing was performed is also noted on the
audiogram.
Once the word recognition testing is complete, the results provide a relative
indicator of the difficulty the person has understanding speech. The WRS
suggests the approximate degree of difficulty the individual will experience
understanding speech for each ear in a quiet listening condition. So, what is a
"good" score? How do you interpret a word recognition score? The chart shows
the general word recognition ability based on the percentage of words that the
patient repeated correctly:
92-100% Excellent
82-90% Good
72-80% Fair
52-70% Poor
22-50% Very Poor
0-20% Extremely Poor
However, there are some things to keep in mind. Normal scores are near 100%,
if the stimulus is presented at 30 dB SL or above, at MCL or above. The hearing
impaired probably will not have a normal score, but will miss some items. When
interpreting the word recognition score, it is important to consider the
presentation intensity. A normal hearer would score poorly if tested at 5 dB SL. A
score of 80% would not mean the person hears 80% of words correctly in
everyday life. If the testing had been conducted at the patient’s MCL of 80 dB HL
(which is much louder than conversational speech), they may have done much
better. So, the intensity of presentation must always accompany the report of the
percent correct. It is simply not true that a score of X% means that a person
hears that percentage of speech. Even if the testing was conducted at a
conversational loudness, speech understanding is different from understanding
single words presented in quiet in a test room. Speech has linguistic redundancy
– the listener does not need to hear each word in order to understand the
message. Word recognition testing uses single words, not contextual speech.
Also, note that testing is conducted in a quiet test booth. The everyday world is
usually rather noisy. Speech is usually only 8 dB louder than the background
noise! So speech recognition tests are helpful in diagnosis, but are far from
perfect for predicting real-world communication.
The speech recognition test results also alert the audiologist to patient who
has unusually poor (or unusually good) scores. Very poor scores are a sign
of possible retrocochlear involvement (which we haven’t discussed yet),
and further testing is needed. Low scores when testing is conducted at a
comfortable level indicate that the person may have limited benefit from
hearing aids. For a patient with normal hearing the score should be 90% or
better. If they have a cochlear disorder the score is usually consistent with
the degree of hearing loss; the greater the loss, the poorer the WRS. If they
have a retrocochlear disorders the score is usually considerably poorer
than expected based on the degree of hearing loss.
Another use for speech recognition testing is to assess the PI function; also
known as the performance intensity function and the PB max. When
assessing the PI function, word recognition testing is repeated after
increasing the presentation level to a higher intensity level. Normally, the
word understanding would remain essentially the same, if not improve. In
some cases, the word understanding will decrease and be abnormal. This
is a sign of possible retrocochlear hearing loss. This means that there may
be some damage beyond the cochlea, such as lesions in the higher
auditory centers. When the ear with a damaged nervous system is
stimulated with a very loud sound, the nerves may be unable to handle the
increased signal load. Consider this analogy. If you injured your arm, you
might be able to lift a small weight, but would be unable to lift a heavy
weight. Similarly, the damaged VIII nerve may be able to transmit a
conversational level speech signal, but “tire out” when faced with the
demand of sending strong, sustained messages. So, we need to review the
PI function chart, which is also located on page 133 of your textbook. On
the chart are the dB levels, the sound pressure level. The side is the word
recognition scores or your performance. If you get 100 up the side along
the ordinate, you'd have 100 percent word recognition or sixty percent word
recognition score. Along the bottom is the intensity you received the
speech material. You could have received it at 30, 40, or 60 dB. This
performance, that's the ordinate, intensity is along the abscissa. So look at
the results for #4. This is word recognition performance. This person has
an obvious hearing loss because they're only getting about 15% of the
words correct at 30 dB. As we increase the intensity of the presentation
level, as we go from 40 dB to 50 dB to 60 dB, the performance is
improving. But when we get to 75 dB, all of a sudden, we reach a
performance of 60%. If we increase the intensity to 80 dB orr90 dB, and on,
the performance decreases. The peak we look at is the PB max or the
maximum phonetically balanced score using PB word list. If you see PB
max written, you know we've done a word recognition score on an
individual, and they've reached a maximum level. If we continued to
increase intensity, word recognition score diminishes. This person on curve
four has a PB max of about 75 dB. So the point where the scores begin to
decrease is known as the rollover effect. This occurs in some ears when
there are lesions in the higher auditory centers. We won’t go into that any
further in this course. But I just want you to be aware of another function of
word recognition testing.
It is important to remember that each of the speech tests help the
audiologist to verify the accuracy of the audiogram. For example, the SRT
verifies the PTA. So when testing, you need to ensure that each of your
tests are supporting the audiogram, the type of loss, the case history, the
tympanogram and the visual inspection of the ear. All of these tools are
used to ensure that the correct results have been found. I would
recommend that you review the audiograms with the speech results found
on page 157-159 of your textbook to ensure that you understand the
speech testing results. The summary of tests used in speech audiometry
found on page 160 is also a helpful review.