Patrick C.M. Wong - pcmwong@ccwf.cc.utexas.edu
Randy L. Diehl
Dept. of Psychology
University of Texas at Austin
Austin, TX 78712
 
Popular version of paper 4aSCb1
Presented Thursday morning, March 18, 1999
ASA/EAA/DAGA '99 Meeting, Berlin, Germany
 
 
According to the estimation by the World Health Organization (WHO), there were 4 million people worldwide suffering from Parkinsons Disease (PD) in 1990 (WHO Information Fact Sheet N152 - Parkinson's Disease (April 1997)). Incidence is higher in the elderly population (the average age of onset is in the mid 50s). As the population ages, the prevalence of PD will increase.

Currently incurable, PD is a progressive neuro-degenerative disorder resulting from a deficiency of the chemical messenger dopamine, which is produced by the brain and influences the initiation, planning, and execution of movements. While many people may recognize the tremoring hands of a PD patient, they may be less likely to recognize the patient's speech problems. Having movement problems means there will potentially be speech problems because the production of even the simplest speech sound requires the effort of different muscles in the body including the respiratory muscles, laryngeal muscles (muscles in the "voice box" or "Adam's apple" responsible for controlling the vocal cords), and articulatory muscles (e.g. tongue and lips). While a normal healthy person may take their speech capability for granted, a person with PD may have great difficulty producing speech. Furthermore, as the disorder progresses, the ability to produce speech may become increasingly difficult. Without this fundamental communication ability, social and emotional life may also be affected.

Research has been conducted to investigate characteristics of PD patients speech. In the early 60's, Canter found that PD patients were unable to produce rapid movement of the tongue, lips, and vocal chords. Their production of some plosive consonants (consonants requiring the build-up of sufficient oral pressure, e.g. /p/ as in "Pot") were imprecise. They had problem sustaining phonation (e.g. holding the vowel /a/). Furthermore, the pitch range and pitch variability used in their speech was more restricted than that of normal speakers. The problem of speaking with a restricted pitch range is that it affects the prosody of speech. Prosody is a combination of the pitch, stress, timing, intensity, as well as emotional aspects of speech. For example, to show anger, people speak louder and with a higher pitch. Caekebeke and colleagues (1991) found the loudness, pitch and duration variations necessary to signal different moods, including angry and hesitation, are absent in PD speech.

Almost all studies on PD speech have studied PD patients who speak nontone languages like English. What will happen if the PD patients are speakers of a tone language? A tone language is a language that uses pitch to signal meaning at the word level. For example, Mandarin is a tone language with four different tones. The syllable /ma/, if spoken with a high pitch, means 'mother'. However, the same syllable, if spoken with a falling pitch, will mean 'to scold.' Reduced pitch variability may impair the ability to signal moods for speakers of a nontone language. But for speakers of a tone language, the ability to produce words correctly may be impaired, and thus, 'mother' may become 'to scold' or vice versa.

It is probably correct to say that most languages in the world are tone languages (Fromkin & Rodman, 1995). In Africa alone, there are more than 1000 tone languages. Almost all languages of the Far East are tone languages. Many of the Native American languages are also tonal. Statistics from WHO showed that PD can affect people of all ethnicity, not just Europeans and Americans who speak nontone languages. Hence, to solely investigate the speech problem of nontone PD speakers is ignoring the speech abilities of the majority of PD patients in the world.
 
In the present study, we investigated the impact of Cantonese PD speakers inability to vary pitch on speech communication. Cantonese is a tone language spoken primarily in Hong Kong and the nearby southeastern Chinese cities. In this study, two speakers, a female PD speaker, and a female normal speaker, were asked to produced the Cantonese syllable /si/ with the six different tones: */si1/ 'teacher,' /si2/ 'history,' /si3/ 'to try,' /si4/ 'time,' /si5/ 'market,' and /si6/ 'yes.' They were also asked to produce these six words in a semantically-neutral context: /ha6 yat1 go3 zi6 hai6 ___/ 'The next word is ___.' The results showed that the Cantonese PD patient spoke with a more restricted pitch range compared to the normal speaker. When only the pitches of the target words (all the six /si/s) were measured, we found that the pitch range (or the "tonal space") where the six tones fell was also more restricted in PD speech. We called this pitch range the "tonal space" because it was the space where all the six tones were contained. These finding were not surprising given what we have learned from PD English speakers. In a subsequent perceptual experiment, we presented the speech of both speakers to normal native Cantonese listeners. We found that listeners had more trouble identifying what the PD speaker said. In an experimental condition where listeners were asked to identify tones spoken in a context ("/ha6 yat1 go3 zi6 hai6 ___/"), listeners could only identify 69% of the PD tones correctly while they could identify 97% of the normal tones. We also found that correct identification is related to the concept of tonal space (the pitch range where the tones fall). In an experimental condition where the tonal space of the PD tones equals the tonal space of the normal tones (we chose three tones as stimuli in this condition, and the three tones fell within a pitch range that was equal for both speakers), listeners identified tones spoken by the PD speaker and the normal speaker equally well.

The results of this study are consistent with early findings regarding to the perception of tones spoken by normal speakers. In an earlier study, we found that to successfully identify tones, listeners relied on a context (i.e. what they previously heard from the same speaker). For example, the Cantonese /si3/, when spoken in isolation, cannot be identified correctly. However when it is followed by a precursor sentence (e.g. /ha6 yat1 go3 zi6 hai6 si3/), listeners can easily identify the word. We found that successful identification of the target word relies on listeners estimation of the pitch distance between the context (the precursor sentence: "/ha6 yat1 go3 zi6 hai6/") and the target (/si3/). This "context-target" pitch distance has to be sufficiently large for listeners to correctly identify tones. In the present study, we found that the context-target pitch distances in PD speech was smaller than normal speech which might explain listeners poor performance in identifying PD tones.

This study is the first study investigating the speech characteristics of PD patient speaking a tone language. Unlike earlier studies which focused on the production of PD speech, this study investigates the comprehension of PD speech. Findings from this study not only help us to better understand how tones are disrupted in PD speech, they help us to better understand how normal speech is processed. From this study, we know that the size of the tonal space (the pitch range where the tones fall) and the context-target pitch distance (the pitch distance between the precursor sentence and the tone in question) are two factors related to tone language speakers ability to understand tone.
 
*The six Cantonese tones.