ASA Lay Language Papers
161st Acoustical Society of America Meeting

Adaptive learning of voice quality behavior as a possible cause of muscle tension dysphonia

Kari Urberg Carlson,
Benjamin Munson,
Peter Watson,
Department of Speech-Language-Hearing Sciences
University of Minnesota-Twin Cities
164 Pillsbury Dr.
Minneapolis, MN 55455

Popular version of paper 5aSC14
Presented Friday morning, May 27, 2011
161st ASA Meeting, Seattle, Wash.

Muscle tension dysphonia (MTD) is a common voice disorder, in which a person is hoarse, but there is not any physical damage to their larynx (voice box). The cause of MTD is currently unknown. One influential theory is that it is a psychogenic disorder that is caused by psychological conflict (Aronson, 1990). This theory is supported by indirect evidence. For example, Roy, Bless and Heisey (2000) showed that patients with MTD on average had higher scores for “negative emotionality” on the Multidimensional Personality Questionnaire (Tellegen, 1982) than patients with a different voice disorder (vocal nodules) or non-patients. However, this theory does not account for the fact that MTD often occurs during or after other causes of hoarseness such as laryngitis (Verdolini, Rosen & Branski, 2006). A second theory is that MTD is a behavior that compensates for something that has gone wrong with the larynx. Think about limping. If a person injures their knee, they start to limp. They are not doing it on purpose. Rather, their body is responding to something that has gone wrong. The limp is the easiest way to walk given the injury that the person has sustained. Sometimes the knee gets better, but the person still limps. It's as if the body has forgotten what normal walking feels like, and the limp has become the normal way of walking.

A theory that can explain this phenomenon is called adaptive learning. Each time someone makes a movement, the adaptive learning system compares what it expects to happen to what actually happened. If there is a mismatch, it makes changes to the instructions that are sent to the muscles the next time that movement is made. For example, if a person puts on heavy roller skates, their brain notices that their feet are moving slower than expected, and uses more force to lift the legs. When the roller skates are removed, the feet move too fast until the brain adjusts again to use less force.

Adaptive learning has been shown to occur in a number of different types of speech behaviors. For example, when the pitch of a person’s voice is artificially changed so that it sounds slightly higher or lower than it actually is, they change the pitch they are producing to compensate (Larson, 1998). Other studies have shown that people's articulation of vowels (Houde and Jordan, 2002) and consonants like /s/ (Schiller, Sato, Gracco, & Baum, 2009) changes when their feedback is altered.  One aspect of speech production that hasn't yet been studied is the production of voice quality, which refers to whether the voice is clear, hoarse, breathy, or strained.  We conducted the current study to investigate whether adaptive learning processes explain how MTD occurs. The inflammation associated with laryngitis makes it difficult or impossible for a person to produce a normal voice quality.  The inflamed vocal folds don't always close properly, making the voice sound breathier than usual.  Some people respond to laryngitis by resting their voices, some respond by simply producing a breathy voice, while others respond by trying to counter the breathiness by contracting the muscles in the larynx so that the vocal folds close more completely.  The resulting voice quality is especially strained and hoarse. Our research proposes that when changes to the larynx make the voice sound different—like those that occur during laryngitis— the brain changes the instructions that it sends to the muscles, resulting in a hoarse-sounding voice.  In some speakers, this adaptive process becomes habituated, resulting in the maintenance of a strained, hoarse voice after laryngitis has resolved. 

To begin to test this theory, the current experiment looks at what happens to people’s voice quality when their perception of their own voice is disrupted. Participants’ voices were recorded, then modified to sound breathier than they originally were by mixing in a special kind of speech-shaped noise (Kreiman & Gerratt, 2005). The artificially breathy voice that they heard is similar to the kinds of breathy voices that people hear themselves produce when they have laryngitis.  This breathy voice was played to them over headphones while they spoke.  We measured how their voices changed in response to hearing their own voice changed so that it was breathier. The phases of the experiment are shown in figure 1, click on the links to hear the sound files.

Baseline phase - no noise | Control phase - speech babblenoise | Training phase - breathiness noise | Adaptation phase - no noise

Figure 1: experimental conditions with types of noise

As of this writing, the analysis of the full data set is not completed.  However, preliminary analyses do support our hypothesis. The attached sound files show the behavior of one subject, whose voice was breathier during the training and adaptation phases than during the baseline and control phases.

This preliminary experiment suggests that hoarseness is a behavior that shows adaptive learning just like pitch does, and it may have a role in the development of MTD. It also suggests that voice rest during episodes of hoarseness may be helpful to avoid practicing behaviors that may lead to longer-term hoarseness.                                                                                           


Aronson A.E. (1990). Clinical Voice Disorders, an Interdisciplinary Approach. New York: Brian C. Decker.
Houde, J. F., & Jordan, M. I. (2002). Sensorimotor adaptation of speech I: Compensation and adaptation. Journal of Speech, Language, and Hearing Research: JSLHR, 45(2), 295-310.  
Kreiman, J., & Gerratt, B. R. (2005). Perception of aperiodicity in pathological voice. J Acoust Soc Am, 117(4 Pt 1), 2201-2211.
Larson, C. R. (1998). Cross-modality influences in speech motor control: the use of pitch shifting for the study of F0 control. J Commun Disord, 31(6), 489-502; quiz 502-483; 553.
Roy. N., Bless, D.M. & Heisey, D. (2000). Personality and voice disorders: a multitrait-multidisorder analysis. J Voice, 14, 521–548.
Shiller, D. M., Sato, M., Gracco, V. L., & Baum, S. R. (2009). Perceptual recalibration of speech sounds following speech motor learning. The Journal of the Acoustical Society of America, 125(2), 1103-1113.
Tellegen A., (1982). Brief Manual for the Multidimensional Personality Questionnaire. Unpublished manuscript, University of Minnesota, Minneapolis, MN.Verdolini K, Rosen C, Branski R (Eds). Classification m