Week 4 PDF

Created with Coconote - https://coconote.app Understanding Speech Audiometry Components This I need the same person. Yeah. Because that that person was engaged in, like, 2 things. They Alright. Thank you everyone for making it here. I'm glad you're all here safely. I'm gonna oh, it just stopped....

Created with Coconote - https://coconote.app Understanding Speech Audiometry Components This I need the same person. Yeah. Because that that person was engaged in, like, 2 things. They Alright. Thank you everyone for making it here. I'm glad you're all here safely. I'm gonna oh, it just stopped. That's nice. You. I I think we'll be done early today, which is the goal for real this time. The laugh. Have you talked to students for 4 more years? Yeah. I want to be done early today also because I think it's gonna get worse from what I've heard, so we'll wrap this up. But, yeah. Thank you guys for for being here. I'm glad you're safe. I hope you all take your time going home and have snow tires or good boots or whatever. One of my daughters now commutes from Halifax to Truro everyday for work, which is temporary, but it's yeah. I always I'm, like, nervous everyday because she gets old around these days. Alright. Alright. Let's get started. So today is a lot lighter than last week. Masking is a heavy topic. So, you know, if if anyone the rest of the class won't be all like that. It's all different topics. So if you were panicked after last week, it's just one week of masking. So if you figure that out, the rest is mostly downhill. There are a few other concepts that are tricky, but nothing. I think masking is the trickiest. And hopefully by now most of it makes sense. Again, if you're still, like, if you, you know, done the readings and kind of work through it, you're still struggling with stuff, you know, or whatever, if you're stuck on a question, feel free to reach out to me or Madeline. It's probably an easy answer. It's just it's easy to get stuck on something. So, yeah, don't beat yourself up or, you know. Anyway, alright. Any questions before we start on that note? Can you hear me okay, slow? Alright. So today we're gonna get into speech audiometry, which you may have guessed is doing audiometry with speech. Class over. It is it is a lighter it is a easier class than last week's old. Alright. Speech audiometry. Before we dive into this, because this is an important part of pretty much every audiological assessment is speech audiometry. One of the reasons is obvious. Right? Most people when they come to see us have trouble hearing speech. That's the most important thing they have to hear. It's not, you know, I'm just struggling with the trombone, and maybe, but that's rare. Right? I've never heard it. It's always speech. So there are some reasons we want to make measurements to speech, and we'll talk about that in a second. The specific reasons we do it and the different tests. Yeah. Oh, is it? Okay. We'll be we'll be done by then actually. That's good. Yeah. That gives me a nice hard deadline. Yeah. Yeah. I'm not surprised. It's alright. So we won't waste any time. I mean, if we do if we have to get cut off, we'll just pick up next week. It's no stress. Okay. So there's at least four levels of speech perception. There's if we're talking about how we, hear and understand speech. The most basic of these would be awareness. That you know someone's talking, you hear that there is speech, but you can't make any of it out. I like to think of, Charlie Brown's teacher, if you've ever watched that. Right? You know, or if you're listening if you've traveled and you're listening to someone speak or could be here in a different language you don't know at all. It's just, you know, if it's if it's French or something you probably know some words even if you don't speak French because there's some forward words or German. Right? But if they're speaking a language you have no connection to, you know they're talking, you have no idea what they're saying. That's speech awareness, Which and it is it's important at some level, and there are some reasons we might measure that in the clinic. Right? Because it's looking at can you hear the sounds of speech. The next level up is discrim, discrimination. That means you can tell 2 apart. So I know that those two sounds are different. It's a little artificial. It's something that is often looked at in, studies. You know, could sing different, you know, brain responses. Can a child tell the difference between, a p and a b? Right? Can they pick up voicing? That those that's discrimination. But it's identification is one level above that. Not only can you tell it apart from other sounds, but you know what that sound is. I know that that is the p sound, that is the buh sound. I know that that word, is, you know, Beetlejuice, whatever. It doesn't mean you understand it, you just can identify it. Understanding would be that last level, which is the goal generally. We mostly speak to be understood most of the time, right? We want we want the person to understand what we're saying. Okay. So understanding is the ultimate goal that but in audiology, we very rarely look at understanding. And the reason is this, that as we go through from awareness to understanding, linguistic experience plays a bigger role. It plays obviously none at the level of awareness, but at the level of understanding, you know, it's very important. And we're very often interested in whether someone can hear or whether the sounds are clear. We're not interested in those language issues. So almost everything we do is actually at this level of identification. Even a test we'll talk about today that is typically called word discrim or word discrimination. No one ever says the -ation, no one calls it word discrim. It's an identification task. They're all identification. And the way you would test identification is, you know, can you say the word, repeat back the word to me. So it's a very simple way to do it. It's very rare that we look at understanding. A few exceptions, but So why would why would we do speech audiometry? Like, what are a few reasons you might want to measure? That'll be really simple. Like I'm doing a hearing test, doing thresholds with pure tones, so I know what their thresholds are. Why would I wanna also look at speech? Because, like, difference in, like, frequencies, like, to see, like, where their range is? Yeah. I mean, you get some of that with you get some of that with the thresholds. Right? You're looking across frequencies, but you don't get, like, whether people can distinguish them. Right? Like, there's no clarity in that. Thresholds just I hear something. Yeah. Yeah. Right. Speech is much more complex. And it's you know, if you're thinking of vision and you could test whether someone can test whether someone can see us a dim light or you can see whether they can read clearly. Can they make out the shapes of the letters. Right? It's more complex. You need to be able to pick up differences over time and over frequency. And so part of why we look at speech is that, to look at that issue of clarity. Because sometimes people will have very very poor clarity even though their thresholds are good. You can have normal hearing thresholds and not be able to make up speech sounds because of a complete lack of clarity. You may not have any, like, really poor temporal processing or really poor frequency discrimination of auditory neuropathy would be condition that could lead to that. How can cause that? What other reasons? There's 2 other ones that are at least Actually, I got a general standpoint. Like, people complain about not feeling understand other people, so why not test, like, speech itself? Yeah. It's it's direct, like you're getting a picture of their problem more directly. And actually that's a good one. That also means it's probably worth measuring speech perception in noise, because that's what most people are complaining about. And it is. It's generally better. You can only predict speech perception in noise if you measure it in noise. It has a very loose relationship with speech and quiet. Yeah? Yeah. So that's I sort of put that under differential diagnosis. You're trying to figure out, like, what's going on. And the speech measures can sometimes give you some really good information about that. Is this conductive, sensory neural, which the audiogram also, you know, already tells you. But is it sensory or neural? Is there a neural problem or is this just a regular cochlear hearing loss? That's a good reason. One more, might be a little tricky. Okay. The other reason, which is actually probably the most important reason, is as a check to our pure tone test. Because it's very hard to screw up a test that just says repeat back what I say. It's very simple. So we use it as what we call a cross check. All of those are important. I don't know if I wanna say that's the most, but it's an it's an integral component of why we do it. So cross check of thresholds. It's very easy to misunderstand a test where we say, I want you to, you know, raise your hand or push the button to the sound even if it's very soft. Right? Some people will wait till it's more comfortable or, you know, there's there's that criterion as to when they respond. And particularly if there are language issues or cognitive issues, age issues, right, it's maybe hard for young kids to understand, maybe hard to convey if if there's limited linguistic, connection. You know, if someone is, doesn't speak your language well, might be hard. It might think, oh, okay. You want me to say when it's too loud or you might want me to say when it's comfortable. Right? Whereas if I say repeat back the word, it's hard to mess up. It's hard to misconstrue that or misunderstand that. The other is it's a person trying to fake the test. It's very, very difficult to fake a speech test to the same level that you speak fake a pure tone test. So someone's trying to falsify during loss. We're gonna talk more about that next week. That is a legitimate issue depending on your work setting. Some people will run into that frequently. My work settings, I ran into that infrequently. Most or it's more infrequent, but you will run into it. Someone will come in for some reason, want to appear as if they have a hearing loss. The speech results often are the tip-off that it's it's not right. Differential diagnosis, can we distinguish, particularly cochlear from neural or retro cochlear behind the cochlea? Losses. Talk about that more. And this other sort of thing that I think we've hit on in a from a few different angles, it's a more complete diagnosis of your speech perception ability. It's more realistic. You're looking at issues of clarity, maybe speech and noise, all the complexity of speech, you know, all of those things. Are the deficits beyond what the audiogram would predict? Because we can't predict what your speech perception will be based on the audiogram. This is going to the next level. That was my my GIF of someone confused. Sorry. I don't think that's what he's doing. Anyway, so it's kind of this absolute versus differential sensitivity. Right? Absolute, you know, can I see that something's there? Differential is, can I resolve it? Can I is it clear? Can I differentiate this from something else? That frequency from this frequency, that moment from this moment. Right? Pitch from this pitch, so on. Alright, so let's talk about this first one first. I guess you always talk about the one you do first first. One of those days. Okay. The speech reception threshold, which no one ever says in full. Everyone calls it the SRT. So you only hear that probably in class and then every audiologist you ever talk to will say SRT, because we like to make everything three letters. It's just it's an audiology thing. So the speech reception threshold, this is a threshold. So it's not a score or anything like that. It's the threshold for identifying a speech sound. So it's gonna be in DBHL, the threshold. Now we want it to match your pure tone average, which is what? The average of 5,000,000, and 2,000. Right? Your pure tone average, we want it to match that, and I'll come back to that in a minute. I don't think that was on the slides. Oh, no. It is. Did is that on your slides? Okay. Good. Good. I updated them. So the premise of the SRT is this cross check principle. We do the SRT to cross check to know that our thresholds are legit, are right. It's a great it'll often be the flag that something's wrong with the hearing test, is the SRT. A lot of audiologists do this even before thresholds, so they know right away if the thresholds aren't there's something being misunderstood. I know at hearing and speech, a lot of people do it first. So did they understand pseudohypocuses, like, pseudo like hypolowhearing. Right? So it's a falsified hearing test for some reason. It's not real. Person might not be lying. Right? They could be it could be a psychological issue. Right? It's not always, but it's it's not a real hearing loss. It also establishes a baseline for other speech tests. So you get some kind of base measure of speech perception. Okay. So this should be within 10 dB of the pure tone average of that 5, 1, and 2, 5,000, and 1,000, and 2,000 should always be within 10 dB. What would explain it if it doesn't? This is why would it might why might it not be within 10? Yeah. If they're faking it or if there's something wrong, the headphones stopped working, they misunderstood. There's one other thing, which I'll talk about a bit later too. If they have 2 thresholds, like, if let's say, 501,000 are pretty good. Am I ringing a bit? I think I am. No? Okay. If, let's say, 501,000 are good, and 2,000, there's a precipitous drop to 2,000, it might actually match the 2 best. Right? Sometimes it'll match. You can hear enough sometimes if you hear, like, let's say, 501,000 to match your SRT, even if 2,000 drops off. And and so we'll we'll come back. You can sometimes guess, and it doesn't you don't have to match all 3. So why should it match? Why would speech, your threshold match your speech perception? That's weird. Right? Like, threshold if you have you guys done the lab yet? Has anyone done it? A few guys? Okay. Because it's it's hard. You're bare like, it's at the very edge of perception. I I think I heard a beep. Why would we be able to understand speech at that level? You think it would be harder. You'd have to hear the speech sounds to be able to repeat them back. And the answer is yes. It is much harder to understand speech than to just hear a tone. The reason they match is because we sort of fake it with calibration, and I'll talk about that right here. Speech is not one level. Right? A tone, a beep can be one level. It can be like that is 10 dB or 20 dB. But speech is always a bunch of levels. This is a graphic from, from the web, from the ListenLab channel. It shows the time waveform of speech, and it shows how much it varies over time. Right? You've got loud and soft parts, you know, even in the sentence I'm saying right now that that's the s's are a little more intense. If I say an f sound that's very soft, the vowels tend you know, the nucleus of the syllable tend to be more intense. You have constantly fluctuating level. Right? And it's constantly different across the frequencies. So there there are all these different, what we would call, short term levels in speech. It's never just one level. Never. Even if we do a vowel that's continuous, like, that's one level at the formant peaks, which, you know, at 812100 Hertz, whatever. It's a different level between the formant peaks. Right? It's different levels of different frequencies. So speech is never one level. So when we want to describe hearing aid fittings and things like this. We measure speech almost always in 1 eighth of a second windows. Why would we do that? Why 1 eighth of a second? Think down deep. Think to audition 1. It has to do with how we perceive level. Temporal integration. Do you guys remember that from audition 1 a little bit? Fuzzy in the mists of time. Okay. So when we perceive sound, we kind of integrate the levels over about an eighth of a second. Right? So if I play something shorter than that, it's gonna seem softer just because it's shorter. Right? We sort of average time. Our our perceptual system averages perceived level over about an eighth of a second, a 125 milliseconds. Not exactly, for some frequencies it might be a little more, might be up to 200, but it's roughly an eighth of a second. Okay. So this matches human perception. Right? That's about how our auditory system, averages things. It doesn't update our level perception, like, every millisecond. Right? It's, you know, it updates it, but it's a smoothing window, I should say. And we typically also do it in narrow frequency bands about a third of an octave because that approximates a critical band, which I think you've talked about last term. Right? Critical band sort of is is how we perceive level across frequency. Right? We divide it into these kind of bands, these critical bands. It's about a third of an octave. So when we do short term levels of speech, we typically do it every eighth of a second in time and every third of an octave in frequency. We're dividing it up like a grid into the kind of the perceptual units of the auditory system, how it perceives level across frequency and across time. Because we care it's humans that we care about hearing speech. Right? So it should follow human perception. Right? It might be, you know, maybe you want to see if your dog can hear you. I don't know. You'd have to calibrate these parameters for a dog. Right? We we usually want to see how humans hear. So when we do that, we can look at the distribution of short term levels, like where the highest the peaky the most intense ones are, where the softest ones are across the frequencies. Right? So if I measure the response, let's say, to a sentence with a bunch of levels, high and low, in all the different frequency ranges, what we find is that there's a distribution of about 30 dB from top to bottom across all the frequencies for human speech. So this is showing ignore the y axis for a minute, it's level, but it's it's a different it's not what we're used to looking at, but this is level on this side, frequency on this side, and this is speech measured over time. Each line is a percentile of the distribution. So the top line is that is the top percentile, the 1%, the 1st percentile. Okay. So those are the most intense portions of speech. The bottom line here is the 80th percentile. So 80% of the short term one eighth of a second levels in the speech exceed that or higher than that. Now we don't usually go below the 80th percentile because there are silences in speech. There's gaps sometimes between words and so on, so we ignore the very very softest bits, the the silent parts. Okay? So this is showing you the levels, the distribution of levels across time. These were the softest little eighth of a second chunks at, let's say, a 1000 Hertz, and the most intense chunks were way up there. And this is about a 30 dB range. So when you think about speech in terms of hearing, the sounds of it, it's about 30 dB wide, from the softest bits to the top, to the highest bits over time. Right? At any given moment, it's one of those things, but over time, it's up and down. Right? Because I could be saying, you know, the balloon is is giant. Right? And as I say ba, we suddenly have peaks at the performance of the ah. But then when I'm saying giant, I'm gonna have a dip at that point. Right? Because it's a different vowel. So those levels are always going up and down. You guys with me? Any questions from there? Usually, you wanna measure for at least, like, 10 seconds or more to get a good distribution of speech, so it's usually a sentence that's ongoing. If you've seen someone measure a hearing aid, you there's typically a sentence that's put through it. The the classic one, if you've ever been in an audiology clinic to listen is, a carrot is a red and yellow vegetable that belongs to the parsley family. Has anyone heard that? I see a few hints. Yeah. So it's like an ongoing sentence and the levels of that sentence are measured so you can see how what proportion someone can hear of the speech sounds. K. So there's a short term levels. This is how it looks if we're in DB SPL, in the ear canal. This is I have the web tool, it does this. The the top red line is are the peaks. The bottom red line are the softest ones, the 80th percentile. The you know, not including the silences. Right? The gray line in the middle would be the average. And the term for that sorry. I'll put that up here. The red lines are the peaks and valleys. The center line is the average, what we call the long term average speech spectrum. It's the average of all of those levels. Audiologists often call it the LTAS. You might hear the LTAS. That's like the the average of speech, kind of the middle. Okay. But it isn't, interestingly enough, right in the middle. It turns out that the peaks for speech are about 12 dB above the average, its average level. And the troughs or the valleys, the softest bits are about 18 dB below. So you have the average, the peaks are about 12 dB above, valleys are about 18 dB below. For music, it's different. For music, your peaks are about 20 dB above the average. Right? So it depends on the kind of thing you're looking to. But mostly, we care about speech for this. Okay. So if the hearing threshold matched the average speech level so let's I put hearing threshold there, the black line, right at the l tasks, at the long term average speech spectrum. How much of the speech would you hear? Would you hear all of it? The top 12 dB, the peaks, or the what would you be hearing? You're just gonna get this bit that's above threshold. Right? If this is the softest sound you can hear, you're only gonna hear what's above that. Right? So this is dB SPL. If that's your threshold in dB SPL, you're only gonna get those top 12 dB. You're gonna be missing half of it. So why would you be able to understand it? Probably wouldn't. Right? So why does the SRT match the pure tone average? It's because of this. On an audiometer, speech is calibrated to produce a level that's usually 12 and a half dB higher. Okay? So it matches it. In other words, the speech is just boosted up. That's not perfect across the frequencies, but it's boosted up by 12 and a half dB, so that most of it will be above threshold. In other words, if I have my if I am set to speech in my audiometer and it says 0 dB, that's actually like 12.5 dB. It's actually higher. K? We do that funky calibration so that our pure tone average will match our speech threshold. We do it to make them we force them to match. Yep. It's all done it's all We do it to make them we force them to match. Yep. It's all done it's all calibrated in the system. So yeah. So if I if I let's say measure a threshold average of 10 dB HL for that person, and then I measure them with speech, I should also get 10 dB HL for speech. Really? That's, like, 22 dB. So how would it work if you do the speech one first, like you said something to do? Still, it's always calibrated that way. Yeah. So speech on the speech when you talk about speech in DBHL, like, the way it's calibrated on on an audiometer, it's always, like, 12 and a half dB louder and dB SPL is boosted. So, for an example, when you you'll often see if you're doing audiometry that you want to set the speech level to about 40 dB HL, because that's a normal average speech level, 40 dB. And you think, well, no. Speech in conversational speech is higher than that. It's more like, you know, 60 some odd dB SPL, right, depending on how loud someone is speaking. The 40 is really in the 50 range, so it's a soft speech. Okay. So if I said 40 dB HL is really like 52 dB, or 52.5 dB SPL. K. So it's you don't see that as an audiologist. You just see the level on the dial, but the speech level has is already sort of adjusted behind the scenes so that it will match your pure tones. We wanna make it match because we wanna make life easy for the clinician. So measures in DBHL for speech are very different than measures in DB SPL, like for fitting into a hearing aid or something like that. K. I sense confusion a little bit. Yeah. K. So, we put it this way. If I had calibrated my audiometer for the pure tones in the speech using the same ret spools, right, like using the same exact calibration values, you would have to turn the audiometer up to get for people to understand speech. So let's say their threshold was 10, I'd probably have to turn the dial up to, like, 22 for them to hear speech, right, if I had calibrated them the same. But since that's complicated, we wanna make your clinical life easier, we just adjust it so that when you set the audiometer to 10 and you choose speech, it's really 22 and a half dB. Right? But it says 10. It says 10 on the dial. So speech, when you're saying dbh gel, like the audiometer level, it's always a lower number. So, like, speech that would be 50 dB SPL, in HL, that's 38 37.5. Right? It reads on the dial much lower. Right? That's just so that on the dial, pure tones match speech. Okay? So you don't have to do anything with that really. Just to understand, if we're talking about speech levels in DVHL, they're different than DV SPL. They'll seem like lower levels, and that's because they they are. Right? We're just we're calling them lower levels so they match your pure tones. K. Alright. Okay. So we want to make it easy. We force it to be the case that your pure tones match your speech thresholds by lying, essentially, by using a lower number for what the speech level really is. Okay. Because you want to make it easy. You want to make it easy, you want those numbers to match. Okay. So you want it to match your pure tone oop. Average. By leaning ahead, it moves. It's so weird. Yeah. Sorry. Just one more question. Sure. Sure. Yeah. So the SPL is higher than the HL? SPL is higher than the HL. Yeah. Yeah. Yeah. So, like, speech that's, like, 40 dB HL will be, like, 52 and a half dB SPL. It's actually higher. Yeah. That's and it's just so the HL matches for the 2. Yeah. So we use spondees when we're doing the SRT. If you've done poetry or English, you probably know, like, spondee is a it's a word with equal emphasis on both syllables. Two syllable word with equal emphasis. Right? There's spondees and trumpees and all these things. Right? So words like Batman, hot dog, toothbrush, you know, where there's you've got equal emphasis on both. Why do we use spondees? We use them because they're really, really easy. I think I why do we use them? Because they're really really easy to hear. And if if you're dealing with a really easy set of words, in fact, we we present them all to you first before we do a hearing test with them, so you know what the words are gonna be. We present them all at a high level. So we're dealing with a very small easy set of words to to listen to, like baseball, hot dog, bathtub, because we're not we don't want to test your discrimination, your clarity. We're just testing your threshold. We want to see this is the cross check for thresholds. We wanna make them easy to hear. If you have a familiarize, like a simple set of spondees to listen to, your ability to hear them almost entirely depends on the vowels. Right? If if I let's say it wasn't very clear and I said, like, rnf all or a hog, you know, like, just hearing that it's an a and an you know it's bathtub. You know it's not baseball. Right? Or That's a. You know, you know I said hot dog. Right? You didn't have to hear the tuh and the duh. Right? You just you just need the vowels. This is why you should match your pure tone average, because your pure tone average is 500, 1,000, 2,000. That's the range at which your formants happen. Right? Think of going back to speech perception or speech, with the course. Speech science. Right? Your formants are basically a little below 500 to a little above 2,000. Right? So 5, 1, and 2, those 3 frequencies in the pure tone average, we we want that average because it matches where the formants will be. And to separate out spondees, you just have to hear the vowels. To hear vowels, you need to hear formants. Right? So it's those two things should match really nicely and cleanly. Okay? Familiarize because we want it easy. So we read the list to them first or we present it from the audiometer first, because we don't want any clarity issues. And, also, if a person misses a word when it's at a nice high level, we just drop it. You don't want any language issues or any confusion. So if you say baseball and they don't get it, you don't use baseball on the test anymore. You just drop it. If you're working with little kids, we'll often use some really easy kid words. And if they don't know one of them, they don't get it the first time, we just drop them. We want easy words they get really easily to do thresholds with. Okay. And they match peritone average because 512, those three frequencies match the formants and speech. They match really nicely. K. Monitored live voice was most common until recently, until we were able to present them from the audiometer, like individual words. In the old days when we had to have a CD player or, a cassette tape recorder hooked up to the audiometer, it was it was a hassle. So people usually just did them through the microphone, which gave you a lot of flexibility. If you're working with an older audiometer in your clinic, you might see people still do this. And it's fine to do. You have to make sure when you're doing this that your voice, your speech level hits 0 on the VU meter for each syllable. So when you speak into the microphone, you'll notice on screen a little VU meter. That's basically the input level, and one of those levels is 0. And you have to hit 0 with your voice for it to match the level on the audiometer that says on on the dial. So you if you're saying bathtub, you watch the v u meter v u meter in bathtub, and you have to get used to that. We used to practice in school, speaking at a totally monotone level and hitting 0 with their speech level over and over again. It's fun. But now we just use the machine mostly. If you work with an older audio, you might end up doing this. But it also makes it customizable. They don't get a word that's very easy to drop it. And if you're doing it there, usually let them see your face when you familiarize it, but then you cover your face when testing. You don't have to cover your face if you're using the audiometer to make the sounds. Okay, and then this is what it looks like. Right, because they don't want the the other Bluetooth LEDs. So okay, here. This is there's the VU meter. It's an old school one. Now they're all of course digital, because it's computer screens. But I want it. Audiologist. K. That's Matt Mosse. They did a recording of, I don't know why my thing isn't working, Of a audiologist doing the SRT in a song. It's the first song I bought on iTunes. I'm like, I gotta get that. Stop it. Stop. It's awesome, buddy. We'll go to the next slide. Here here we go. Damn. I hope you wanted to mess with me today. Anyway, I encourage you all to get that song if you're in audiology. Anyway. Okay. The procedure is very simple though. You're doing those words, wants to familiarize, and it's just like a hearing test. Down 10, up 5. You want them to hear it several times on an increase just like when you're doing the thresholds. That's the Martin and Dowdy procedure. It's the most common. So you're just testing the spondes like you would with the tones, but you're getting the person to say it back. So you say bathtub or you push bathtub, and you want them to say bathtub, and so on. Okay? And that should be Oh, here we go. Should be within 10 dB of the pure tone average, and then in parentheses I say, or the Fletcher average. The Fletcher average is the best 2 of 3. It's the average of 2 of the 3. So when I said before, let's say their hearing's 20 dB at 501,040 dB, 40 dB HL at 2,000. Right? They're probably gonna guess what the vowel is just by hearing 501,000. So it'll actually match just those 2 upper frequencies. You can ignore one that's way worse, essentially. And that's called the Fletcher average. It's the best 2 out of 3 of the pure tone average. Okay? Should match the pure tone average, or if you have some kind of precipitous thing going on, you should at least match the Fletcher average. Should match those 2. Okay? So here's an example of it. It's a super easy test to do. You guys will do this in the in the lab. It's very easy. This one is showing what kind of hearing loss. Let's just do that first before we go to the SRT. I think I heard it. Mild. Yes. It's mild loss. Which configuration? It's well, configuration is flat. Right? It's conductive. If that's masked, right, this doesn't have masking symbols, but let's assume they're masked symbols, then that would be a conductive flat mild hearing loss. The SRT is 40. Would you trust those thresholds? Yeah. Right? Because the pure tone average, 5, 1, and 2 are gonna be, like, 40 plus 40 plus 35. That's gonna be, like, what, 38 or something. And so that's within 10 dB of 40. Right? So when it makes sure that SRT 40 is within 10 dB of the average of those 3 frequencies, those middle frequencies, right, because that's what that's what you would need to hear those formants to make up to to hear that out. Okay. Here's another one. We're looking at what there's a sloping mild to moderately severe sensory neural hearing loss in the left ear. SRT is 35. That makes sense? Yeah, right? It's, I mean, I don't think we need to use a Fletcher average here, but it's pretty close to the upper two frequencies. If we take the average, that's gonna be within 10 dB, right, the SRT. So you're you're doing that at a glance. You're doing that in your head, thinking what's the pure tone average? Is my SRT in line with that? Or is there something wrong? Like this case, here's SRT is still 35, but what's the pure tone average there? It's 55 plus 60 plus 70, so that average is going to be a little above 60. Right? Do a quick sort of average in my head. What's it doesn't match at all. What might have happened there? Any guesses? So the bone conduction bone thresholds, actually have a pure tone average pretty close to 35. My air thresholds are far worse, but my SRT is matching my bone thresholds. Yeah. Yeah. This is actually not that uncommon if you're using TBH or supraoral collapsing canals, right? So it's it's like a temporary conductive loss. Because clearly when you did the SRT, they were hearing where the bone thresholds were. For some reason, I'm getting this conductive hearing loss that makes it worse, Right? So something seems to have changed in the test. It's likely collapsing canals, or headphone wire is broken or something like that, or headphone moved off or something. Some reason why you're picking up a transient conductive loss. Sorry. You might have already said this. Yes. Could you go over how you're doing, like, the SRT of 35? So that's just doing going back to doing, you know, baseball, hot dog, you're doing the hearing test with that. Right? So down 10 up 5, you start at, like, 40 or something. So is that with the average? So that is that is your final threshold. The SRT, like, would be where they can repeat it twice after an increase. So it's down 10, up 5. Yeah. Yeah. Just like thresholds, exactly the same. That's the threshold. So that's in dB HL. SRT is that'd be 35 dB HL. Yeah. Yeah. Sorry it's what happened when my matmos was going crazy. But yeah, it's just like exactly the same as thresholds except with the word instead of the tone. Did you have someone else have a hand? No? Okay. Alright. Does that make sense? Yeah. So I would do let's say, at 40, I would have I'd say bathtub, and the person says bathtub. I drop it down by 10. At 30, I say hot dog. They say hot dog. I drop it down by 10. Then I say, peanut, and then they don't hear peanuts, I go up by 5. Right. And then so it's just the same as audiometry, except instead of them hearing a tone, they're repeating a word back. That's the yeah. Yep? Is it do you, like, get, like, all the words, or just, like, you change it each time? It's it's a new word each they have a list. So they hear all the words first at a high level to make sure they can repeat all those back. They like, so you run through it once first, and then you start doing the down 10 up 5 with it. It's, so it's a new word each time, but you can go back, like, if you, you know, if you have, you know, 30 words, it doesn't matter if you're cycling back to in a list again. Right? It's yeah. If you go by down by 10 with the word, they don't hear it. Yeah. Go up by 5, do you keep the word? No. You do a new word. So you go down by 10, they don't hear it. When you go up by 5, it's a new word. It's always a new word. Yeah. Yeah. Always a new word. And you wouldn't want to you have to be careful if you're doing it, let's say, with someone who's had a lot of hearing tests, especially a kid who's grown up and done lots of hearing tests. If you always use the same word list, they'll just know the words, right? After a while, all the ones they've heard. So you have you want to go in a different order, you know, that's easy, right? Or use a different word list. Yeah. So if we look at the same of this one here, so you're always doing a quick average calculation in your head, and you have to be within 10 so you don't have to do the math usually exactly. But in this case, fresh the pure tone average would be what? 35 plus 40 plus was that 55? 55. Right? So what is that's coming out to 35 plus 55 is 90 plus 40 is a 110 divided by 3 is what? 30, 36 roughly, right? That's like 36 dB, which actually in that case matches the SRT perfectly, right? But let's say that this one was like way down here. Let's say it was precipitous. So in that case that pure tone average would be off my SRT. And likely they would have the SRT will probably match just the best too. Because like even hearing from 500 to 1000 probably gives you enough to figure out what that word was. So you just do an average of the best tube. That's all it is. Yeah, so when the pure tone average, like when you have 3 frequencies, one of the frequencies is very different. It's usually the 2,000. Like, in the pure tone average, you just take the, the 2 good ones. The 2 the 2 good ones of the 3 is the Fletcher average. Yeah. The official pure tone average is always 500, 1000, 2000. Fletcher is, that's 2 out of 3. Yeah, usually 501,000, because you don't get precipitous losses, low frequency precipitous losses. For a complex reason, it's kind of interesting that would probably take too long, and we want to be out of here by 1, so we won't go through that. Okay. So so your test is just like pure tones and exactly the same, except you're always doing a new spondee, and they have to say the word back. And it's hard and let's say they were trying to fake a mild loss. It's very hard to know what words they would be able to hear and not hear, especially considering that, you know, speech isn't even calibrated the same as pure tones, right? Like it's very hard to know what they should be hearing in speech for a given hearing loss. So typically the SRT won't match if they're faking the loss. Any more questions before we go on? You guys will go through this from practice practice this in the lab, so you'll have it'll yeah. It'll all be really straightforward. Yeah. Can you just go over again why you think that's, like, a collapsing now? Oh, yeah. Yeah. In this case, I mean, it's possible it's something else. It's possible it's a fake tearing loss or something like that. The reason I would think so is the bone thresholds, 35, 40, 50, are very close to the SRT. SRT isn't done with bone conduction. It's done through the same earphones that were used doing the the thresholds, the air conduction with. So the fact that it matches the bone thresholds so well makes me think that conductive hearing loss thanks, Siri. It does this thing if I raise it to my makes me think that, the conductive loss isn't real because they were hearing it suggests they were hearing through air conduction, through the speech test, the same way they were hearing with bone conduction. Right? So it depend a bit on order of the tests, like in actual practice. So if I did, SRT first and then I did air conduction thresholds, and then bone, which is often done. It might be that it was fine when I first did the SRT, and then the canal fully collapsed, and then there's a conductive loss. Right? It would just be strange it's strange that the SRT through air conduction matches the, bone conduction threshold so well. If it was one where they misunderstood the test, they'd probably misunderstand bone the same way they misunderstand air, so I wouldn't think it would be that. This could still be faked or something. Or it could be, just for some reason these air thresholds seem off because the the SRT, you trust it in a certain way more than the other thresholds because it's just so hard to get that wrong. You know, say bathtub bathtub, like, it's hard to mess up. So you kind of use that as your guideline to where the air threshold should be. It suggests that the air thresholds should be closer to 35, which is the bone threshold. So then you think, why they're not? Something happens with my earphone in the middle? You know, so it would depend on, like, test order and all those other things. And, Yeah. Jig jiggle your cords, you would check for collapsing canals, you take it off and push against you push here and see if it does collapse, make sure that the placement isn't off, all those kinds of things. Yeah. Anything else? Other thing would be something really variable that would be a conductive loss, like a eustachian tube problem, and they're now plugged up when they weren't in the beginning of the test. That would affect their hearing. That's also can happen. Clotzing canals is just a common one. So I've actually seen that not this exact audiogram, but I've seen that come in audiograms that have been sent to the school where it's clear there's something wrong with the air thresholds, that there's a fake conductive overlay. And so it's so easy to get a fake conductive loss. You just have to have your earplunged for some reason for a minute to get that. Does that make sense? Okay. Alright. So if we will use the speech awareness or speech detection threshold, those mean the same thing, Speech awareness, speech detection, which is really just knowing that there was speech there, in cases where someone can't repeat the word back. So the obvious one is for young kids. If they're pre linguistic or very very small, they might not be able to repeat the word back. So you just use speech awareness threshold, or if there's a complete linguistic barrier, it's not optimal, but obviously they're not going to be able to repeat the word back. You can do other things. Let's say, you talk to the parent and they say, well, they know their body parts, so you say, oh, you know, point to your head. Even if not everything is a spondee, close enough, right? You can get them to you can do color sometimes. So they'll know certain things that you can do with kids. But if you can't do any of those things, you do speech awareness threshold. It's not quite the same as your SRT, so you tend to get this, like, 2 to 6 or 5 dB, because we usually go in 5 dB steps, better than the SRT. Because the person doesn't have to repeat it back. They just have to hear that there are sounds there. Right? So typically, the SAT or SDT, speech awareness or speech detection threshold, is a little better. So you have to do that. You you would expect it to be a little higher than the pure tones. Okay? Alright. Alright. SRT. We're rolling. It's 12:30. Hey. Should we just go through? Because we're gonna be done soon. Yeah. Okay. Alright. Suprathreshold speech testing. So that's the SRT. Remember, just like pure tones, you'll have the words in the audiometer. So, Madeline will show you that in the clinic. Old school ways to read them off a list with your face covered. We used to have lists printed out. The other tests we do are above threshold. So these are not threshold tests. These are the tests, to look at whether speech is clear, or maybe to do differential diagnosis. Is there something you can get out of this? And here's an example. So we have an audiogram there. And we do we have, normal thresholds on the left ear. Right? And we've got a flat sensory neural loss on the right ear that's mild to mild mild to moderate. Moderate, right, on that ear. So it goes a little past 40. It's sensory neural because the bone thresholds are pretty close. So there's an asymmetry there. Now if I did this, I would expect that's better. In the left ear, my SRT is 10, just as I would expect, looks perfect. In the right ear, I get an SRT of 35, looks just about perfect. It's within 10 dB of that average, of average of 45, 40, and 35. But let's say I get I do a word test. I do a word recognition score, so I read a list of words, and they get 4%. Is that what I would expect? What if they got 75%? That seems better. Right? If they got 75%, you think they've got a mild to moderate loss, probably gonna get most of the words. If I these are typically done at an audible level. It's turned high enough for you to hear. It's above threshold. So the question is, how do we use these numbers, these percentages? What's abnormal? What's normal? What would I'd be doing this to see is your speech clarity how is it? Like, can I fit a hearing aid to that ear? If one speech is audible, you're only getting 4% of the words. A hearing aid's not gonna help that much. Right? If they're getting 75% of the words, much better. Right? Does 4% tell me something? Does it tell me that maybe it's a neural issue, it's not just a cochlear loss? So we have to know how to interpret these super threshold numbers. Okay. It's getting at this differential sensitivity. Like what should their score be for a given hearing loss? You know that people come to us because they're having trouble hearing speech. Remember this from the 1st week. Do speech problems overestimate or underestimate physiologic damage? Like, is speech easy to hear in quiet? It is really in quiet, right? Like, you don't need to have hear a lot of fine details is what I'm getting at. Right? Like, you're it's a big, broad, easy signal. Like it's as long as it's audible, you should be able to make it out. Like you can mess with speech a lot. This the stuff that people can hear is amazing. Right? Like speech isn't shouldn't be a hard signal to detect. So, speech problems underestimate damage. Right? You can have a very damaged ear and still understand speech and quiet. Right? Speech and noise is a little more sensitive. Speech and noise is harder. And where you hit but speech in quiet, you don't really need details at all. Remember, it's this big, broad, easy to hear thing. K. So speech problems underestimate damage. Speech should be easy. Yep. Yeah. It's a good question. So phonological like, language issues will cause problems with it. Right? Depending on the type of speech test you're doing. That's a good question. Like, I I don't know how much of a phonological disability you would need to interfere with, we typically use for for most of our basic speech tests, just consonant vowel consonant. But, yeah, it probably would interfere with that. Yeah. Versus, like, yeah. Yes. Yeah. So if you if you you know, if the speech was much worse than you would expect given the loss, like, language issues could be a factor. Right? It's yeah. So one of the reasons why you'd wanna have connections to local SLPs. I haven't seen that in particular, but you might see it more if you're working with certain, like, young kids or with certain populations. Yeah. Yeah. It's a good question. So the but the question is what's normal? What should we expect? If there's no extra disorder, if it's just a hearing loss, 75% is closer to what we would expect in that case, which I think is intuitive. Right? It's all to moderate loss. If I make the words loud enough, you're gonna get most of them, probably. Getting 4% would be weird. Right? So what is normal has been a question that has been answered in various studies over the years. One of the references that had been used years ago, Judy Debno, who's still working back in 95, had published lowest maximum word score, relative to various pure tone averages. And so she's called your your word score, PB Max, and the reason that term is stands for phonetically or phonemically balanced. Some of the lists are balanced, have all the the phones. Most of them are balanced, have all of the phonemes, so we call them phonetically or phonemically balanced. And your max score, we do you test multiple times what's the best one you get. That's called PB Max. So if I do a word list with a phonemically balanced word list and you get you do it a couple times and you get 80% on one of them, your PB Max is 80%. Sounds like a peanut butter or something, but that's what PB Max stands for. It's your best word score. And WRS is word recognition score. I should say that we use word recognition score interchangeably with word discrimination score. Those two words are used interchangeably in audiology. So people will often write w r s and then say that's descrim. Right? Because descrim is word descrim, word discrimination score. Sometimes it'll be w d s as well. So just as a warning, you'll see those both. Okay. So, so in that previous loss we looked at, right, that person, their pure tone average was, well their SRT was about 35, the pure tone average was 40. 45 plus 40 plus 35 comes up to 40. So for a 40 dB pure tone average, they should be getting at least 64% of the words on the test, and they were getting 75%. So 64 is the lowest score that you should get. So we say that's normal. Okay? So the standard way to do this, to do word recognition or word discrimination, again totally interchangeable in audiology, is to use open set monosyllables. So we want here we're trying to measure clarity. So we're not gonna tell them what the words are first on a closed set. It's words that they just have to hear. Right? And they're monosyllables, consonant vowel consonant. So, you know, pop, top, cat, you know, words that are are difficult to hear, so we don't familiarize it first. They haven't heard it before. They're embedded in a carrier phrase. Say the word cat. Say the word top. Like that, because it makes it more realistic when they're in a carrier phrase. Almost always, the standard approach is due to to do it at one constant level. I'll talk about what level we choose in a second, but we we just do them all at one level. So you say the word cat, say the word top, say the word hat, and so on, all at one level. Okay. That's word recognition or word discriminate. What you get out of it is a percent correct. How many did they get correct? On your audiometer, you'll have buttons to say correct, incorrect, correct, incorrect. So you don't even have to keep track. Okay. So, like, tap, nut, job, and so on. You don't want a new monitored live voice because the way it's enunciated matters a lot. I was actually tested by a friend, before I got into audiology. She was, she was in her 1st year and she did this monitored live voice with me, but she had a strong accent. This is me at Western. I was an undergrad. It's probably why I got into audiology because it was, like, this seems fun. But I scored really poorly on this because I could not understand her accent. And after the test, she said, your speech perception is really bad. And I just wanted to say, well, you can't talk right. It's funny. But so people used to always do this because it was cumbersome to hook up your cassette recorder or your CD player and play the word and stop and play the word and stop. And the person who had this job before me, Brad Stack, said, if time precludes the use of taped, it's old school, right, taped speech materials, time should be preserved by not administering speech tests at all. Like, it's a waste of your time. Nowadays this is not as much of an issue because the audiometers have the words built in. Right? Digital, easy, present one word at a time. So, but you shouldn't do this with recorded, like, with your own voice because it's the way you unless if you're super clear maybe, but then with the next clinic have the same speaker, you want to use a a recording, standard recording. Standardized with, like, exact speech noises or music as well? Yeah. So there's standard recordings of the speech test. Yeah. So your score will vary a bit if you have a different recording of the speech test. Yeah. It changed. I've gotten them done for years years. But it's changed over. The last one I got is the guy had, like, a southern draw. He pronounced like white. It's a bad speech test to use. He's using a speech test, or she or he the the clinician is using a speech test that is not the same dialect, which is not ideal. Yeah. Yeah. You you should which raises an important point. You should be using the language they speak and the same dialect that they speak, the same accent, ideally. Like, you if you don't, your scores will be different because of it. Alright? Yep. Would you have to put the same Yeah. Ideally so if people come to your clinic with different languages, you should have recordings corresponding to their language, right? Ideally, a clinician that speaks the language is the best as well, because it's hard for you to score it if you can't, you know, understand it. It's maybe less of an issue if it's English French and you're not super fluent in French but you speak it enough and you can, you know, you can still make it out. But if you really don't know, it's tough. But you'd ideally have different recordings, yeah, for these. I mean, it has been done in a lot of different languages. Ideally, not in a southern drawl if you're a cat you know, either. Yeah. Where you do this what time are we at? Okay. Where you do this varies a little bit. There are different rules of thumb, and different clinicians use different approaches. A standard one one is 40 dB above the pure tone average. Right? So you take the pure tone average, let's say, it was 10, you hit 50. That kind of breaks down, though, if there's a lot of hearing loss. Let's say their pure tone average was 70. You're not gonna do this at 110. Right? So, of limited usefulness. Another approach is to do it at the most comfortable level. Say, is that comfortable? And they say, yes, and you do it there. A more used approach, one that I think is defensible, actually it's the way I used to do it, so I was kind of biased, is to do the upper limit of most comfortable level. So you say, Is this a comfortable level? And they say, Yes, you go up by 5 dB. How is that? Is this still comfortable? And if they say, Yes, you go up by 5 dB. And you keep doing that until they say, No, no, that's too loud. Then you go back to the highest level that was comfortable for them. That's a good way to get their PB max, their best speech score, is at the upper end of what they find comfortable to listen to, because they will do better at that level than they will at a comfortable soft level. So it's, not an uncomfortable, but the highest level they say is comfortable. You just keep doing that until they say it's uncomfortable. Some people will also do it at a standard high level. I think in our clinic here we just do it at 80 dB for a certain range of hearing losses. You just pick one level, which is also reasonable, because it for a lot of people that's gonna be okay. And some people will do it at multiple levels, like as an example is most comfortable level and a high level. And we'll talk about why you might want a couple in a second. But you you do want to get you don't wanna be soft. You wanna be at a point where they're at their PB max, where they're getting their best percent correct. That's the goal of this test, to get their best percent correct. So you don't wanna be too soft. I would say be careful with MCL. You kind of want to make sure it's the upper end of MCL, that it's comfortable, but not soft. K. Remember, this is an easy thing to make a mistake with. PB max is the best score they get. PB max is in percent. Right? PB max means, like, their best score was 80% or was 83%. Right? It's not the level where they get it. So maybe they get 80% when the speech is at 80 dB, but the PBmax is not 80 dB. It's 80%. K. PBmax is a score, not a level. It's an easy thing to miss. Do you test the word recognition score for profound hearing loss? Yeah, you can. You obviously can't like go 40 dB above their you'd probably that's where you'd want to do MCL, upper end of MCL. You probably only have a range of 5 dB. It's it's it's actually really good to know, right, because some people with profound are still doing okay with that loss. It's typically not they're not going to be getting super high scores, whereas some are getting almost no intelligibility. Right? That's the territory where you're definitely gonna be looking at cochlear implants. Yeah. Is it is the word recognition score used as, like, an eligibility criteria for certain experience? Yeah. It's often used. Yeah. It's it's your ability to, like, to work with the hearing, what intelligibility you're getting through it. It's not the perfect measure because a hearing aid will shape the low and high frequencies and fit to your loss. This is just a level change, but this it does factor into those decisions, like what's an appropriate rehab strategy. This person's getting 5% when it's audible hearing aid's probably not gonna do it. Yeah. Yeah. It's a good question. Alright. Word recognition scores. Sorry. I love that picture. In a nutshell, before you go for a couple of these details, you get a better, more stable word recognition score if you use a lot of words. If I get you to repeat back a 100 words, and you're getting 26% of them, that's a pretty good estimate, like I've done a 100 words. A lot of clinicians will, to save time, use 10 words. Right? Ten words, which, you know, and I I get it sometimes it's justifiable. Right? Let's say this person is doing really well, you don't think there are any speech difficulties, and I do 10 words and I get them all right. Maybe it makes sense stop there, right? Or if I do 10 words and I get them all wrong, right? Maybe that makes sense to stop. It turns out that if you do short lists, they're pretty stable if people get, like, near one end to the other, if they get them all right or get them all wrong. If they're getting in the middle, let's say they're getting 5 out of 10 words, it's not giving you that information. You want to be doing more words. So you can actually look, this is from 1978, it's really old, looking at different lengths of word lists. So the top line here is 10 items, like you're just reading 10 words. Next line down is 25 items. Next one is 15, which is kind of a standard list. And what they're plotting here is your speech recognition score, so what they got on the test, and what the standard deviation is in terms of predicting a proper score if you measure with hundreds of words. So your true score should fall within plus or minus 2 standard deviations. In other words, let's say you did 10 words and they got 5 of them right, that's 50%. Right? The standard deviation is about 15 there, which means 2 standard deviations and is 30. So that's plus or minus 30%. So that means 50 is probably not different from 80 or for, from 20. Okay, so if they get 5 in other words, if they got 5 out of 10 right, and you only did 10 words, that's not really different from getting 8 out of 10 right or 3 out of 10 right. Right? Like, it's just maybe they got that word, and they would have got it the 2nd time with a different word. Like, there's a lot of randomness in that if I'm only doing 10 words. If I'm doing lots of items, those standard deviations are much smaller. Right? The other thing to pick up from this is if I do 10 words and they get them all right, the standard deviations are pretty tight. Okay. It's 10 words is probably enough if they get them all right, and 10 words is probably enough if they get them all wrong. But if that's not the case, you probably want to go up to 25 words. Right? Or 50 words, if they're somewhere in between. A lot of audiologists do do that. They'll start with 10 words. If you get them all right, they'll stop. But if you don't, they're gonna go on and continue to do 25 words. Alright. There's there's actually a really nice little, article that Marshall Chasen did in our CAA's old magazine with his daughter when she was in high school to kind of do the stats for that. Like, when is when should you jump to a 25 word list? When should you jump to a 50? And it's cute. Anyway, I can pull that up for you guys if you're interested. Alright, but more words are better, they just take long. And it's also boring for the person to go through a lot of words. So we usually do start with 10. These are the actual ranges. So if we do 10 items and they get, let's say, 3 of the 10 right, that's not different from getting 1 out of the 10 right or 7 out of 10 right. That's the range. Right? So, it doesn't tell me that much about how they're doing. Okay? So I might want to go to 25 items, and that has a tighter range for predicting what the speech score is. Where does that matter is, let's say I do their speech test and they come back next time, I do the speech test, they say it sounds clear now, and I do the speech test again and I get a different score. Is that score different? Right? I need to know how different does it have to be to count as actually different. Or let's say I put a hearing aid on and you do a speech score. Are they doing better? Right. I need to use more than 10 words in order to really make those comparisons. Alright. So that's the range at which you you'd say there's no significant difference in this range. Alright. This has been formalized a little bit in a nicer way with what's called the sprint charts. Linda Thibodeaux did these in 1999, and these are kind of what time are we at? Kind of neat in that if they let you do 2 different things. So I'll walk there's 2 things you can do with this chart. The first is, let's say you're doing their speech score twice to compare before and after. Let's say you do a speech score, you put the hearing aid on, and you do a snub you do the speech score again, and you want to say if they're doing better. Right? You're comparing 1 speech test to another speech test. Or let's say you do one as a baseline, and then they come back next year, and they say they're they want you to see if they're doing any better or worse, so you do another speech score. So you wanna compare the 2. Anytime you want to compare 2 speech scores, what you do is you would compare the percent from the first score here with the percent on the second score there. And if it falls within the arrows, they're not significantly different. If it's outside the arrows, they're significantly different. Okay. It's doing a significant different test for you simply in a graph form. This is the sprint 25. So this is if you're doing a 25 word list. That's why it goes from 0%, 4%, 8%, because you can't get anything in between, right? It's 25 words. Okay. So if they got 40% on 1, and let's say you do something and they got 80% after you do it, well 40 is significantly different from 80, the 25 words. But if they got 40, and then you did something and they got 44, those aren't different. It's within the arrows. Okay? That's one way you can use it. The other way is to say, is this loss within the range of what we would expect for that hearing loss? If it was a normal sensory loss, there was no sort of extra neural issue going on, like neuropathy or something. It's just a normal, run of the mill, noise induced, age related kind of hearing loss, cochlear loss. What would I expect? That's where I just drew this red line on. It's not really on there. It's the shaded versus unshaded on the actual chart. That's where you want to see, am I in the expected range or the unexpected? So you can pair compare their speech score to their pure tone average. Average. Okay, so it's another way you can use the same chart. So if their pure tone average was 0, which is the very top row, the normal score for them is a 100%. Anything below a 100% is abnormal. 96, in other words, if they got 24 out of 25, it's anything below getting all the words is abnormal. It's gray, it's shaded gray. If they have a pure tone average of 20, right, you can see 20, you'd go over and say, well anything from 88% to a 100% is normal, anything below that is abnormal. So anything in the gray shaded shaded area is abnormal for that loss. Okay. So it's normal if you have, you know, let's say, an 84% pure tone average, it's normal to only hear 8% of the words. Alright. You're probably gonna miss a lot of them, even if it's made to be audible. It'd be abnormal to get 0, you should actually get more than 0 for that. If you're in the profound territory, it's not abnormal to not hear any of them. Okay? Does this make sense? Okay. What abnormal means, like falling into the gray area, means this is not just, there's something else, this is not just a cochlear loss, there's probably a neural or retro cochlear problem. Could be a tumor, could just be neuropathy, pathology of the nerve, something else going on. That's what we're on the lookout for. Yeah. So for the first one, when you're comparing to each score, was it the bottom versus this Yeah. Yeah. And then, like, you just do the, like, that to see where it falls? You just see where it falls. You look at the intersection. Yeah. Same same with this. You'd so if you're then seeing whether it's a test is normal or abnormal, you take the pure tone average and the percent correct and just look for where they meet. If it's gray, it's abnormal. If it's white, it's normal. Okay. Alright. So that's these are done for different word list lengths. So this is the sprint 25, this is when you've done 25 words. The next one here is the sprint 100. If you do 100 word lists, you, you know, your your arrows are more narrow and so on, you get a more precise measure when you do this. Okay. We don't do these word lists as much, but, you know, you've got Sprint 10, Sprint 25, Sprint 100, Sprint 50, and so on. Right? So you can use them in 2 ways. This is good if you want to compare 2 speech scores. Using the pure tone average as your point of deciding whether something's normal or abnormal isn't the best approach. So we don't tend to do this as much. You might see this in the clinic, it's not a terrible thing to do, but the pure tone average is the average of 500, 1,000, and 2,000. Your percent correct on a consonant vowel consonant test isn't gonna match that so well. Right? Because a lot of consonant information is higher than that. 500, 1,000, and 2,000 gives you what part of speech? Vowels. Right? Those are your vowel formants. Right? So it makes sense for a competitive PTA makes sense when you're looking at spondes. It's not so good for predicting speech perception in general. So we'll we'll talk about we'll finish it up next week, but we'll talk about a better approach to this. But you might still see this. You should know how this works. One last thing before we finish up is some clinicians will do speech tests at multiple levels, and obviously this would be harder if you're doing a 100 words at each level. So, you know, typically you're doing 10 words at each level, maybe 25 at most. Okay? And if you do that, so you present speech at different levels, let's say, this is your your client, and at 0 they're not getting any. You might not bother doing 0 because they're probably not gonna get any. At, you know, 30, they got 10 words. When you raised it to 20, they got 80 words. You can trace out this function. We often call it a performance intensity function, a PI function. The one to the left here, a, is what you would expect if someone has normal hearing. Right? Their maximum score they're getting a 100% by the time you're at about 30 dB HL, which is about 42 and a half dB SPL. It's like soft speech. They're gonna get a 100% of the words. If they get something like b, that would be a conductive hearing loss. It's just like a, but everything has to be louder because their ear is plugged for some reason. Right? That's a conductive pattern, beam tracing. C, D, E are what you would expect with sensory neural loss because then there's some kind of degradation as well. Right? So C is kind of is a common pattern. So as you raise up the level, they get up to 80%, which is their their best score, and their best score is called PB Max. Right. Pbmax is 80% for that, the best score. And then at higher levels, they're not really doing any better, that's their top score. Okay. You also will see some people that have a pattern like e. Their pbmax is 64%, but as you go to higher levels, their performance drops. It gets worse. So they were doing pretty good at 70. When you turn it up to 90 dB, their score got worse. That's a sign of a neural hearing loss. If you have something like a tumor or some kind of degradation of the nerve, your score will often get worse above PB max, at the highest levels. It starts to drop, and there's a term for that drop it's called rollover. Where the curve rolls over. So it you can only know this if you do speech at a few different levels, which is why some clinicians will do like kind of a high comfortable level, and also a very high level. They'll do at least they might not do the whole curve, but they might do a few levels. Because if you get rollover, that's a pretty good indication that there's a neural problem. It has very good specificity. It's there's very few things that would lead to that apart from having a neural issue. Alright. So it gives you some more information than the audiogram does. The audiogram gives you sensory sensory neural versus conductive. This gives you a bit of sensory versus neural, to to see rollover. Okay? You you probably wouldn't do that many levels, but there is a way to calculate, a rollover index. So this is just repeating what I said. For sensory loss, you expect performance to be a little worse, right, like curve c. For retrocochlear or neural loss, you often get rollover where it gets worse at high levels. So when you're doing your sprint chart, what you're really comparing is someone's PB max, like 80%, 76, 64, with their pure tone average to say, is this what you'd expect with a normal sensory neural loss? Okay. With a neural loss, it might be too low, and or you might get rollover. Both of those are possibilities. Okay. With neural loss, the PB max might be lower than you'd expect for the loss, and also at high levels it might get worse again. Okay. Rollover index, if you want to calculate it, is just this is the formula, the ri. It's pdmax minus pdmin. Pdmin is your percent score at the highest level. So it's a difference of those two points, divided by PD max. So for this one, let's say it's 64 and 38%. So it'd be 64 minus 38 divided by 64, so it's a rollover index of 0.41. You can get a single number for that. And for different word lists, different rollover indices are like a cutoff for that's when you'd say there's probably a neural issue. Not everyone would use those exact cutoffs. You might just you might be a little more, loose with it. If it looks like there's some rollover, we'll make the referral, which is typically the approach. If in doubt in the clinic, if you think there's a possibility of something going on, you're usually gonna follow-up on it. K. Alright. Let's stop there. We'll look at some cases, and we'll look at one more thing with speech next week. And be safe, guys. Be safe. I will put the recording up too.