ACOUSTIC AND PERCEPTUAL ASPECTS OF ALARYNGEAL SPEECH

PRESENTED BY: RITHU M FACULTY: Dr M PUSHPAVATHI Mr GOPIKISHORE P

Laryngectomy??

Types of laryngectomy ??

Laryngectomy

Partial

Total

Hemilaryngec tomy

Supraglottic

supracricoid

Subtotal

Cordectomy

Rehabilitation

Artificial/el ectromech anical

Esophageal

Prosthesis

Artificial/electromechanical

Transcervical

Transoral Intraoral

Esophageal

Consonant Injection Injection Method Inhalation Method Swallowing

Prosthesis

Non-Indwelling Prosthesis:
Must be removed every 3-4 days

Indwelling Prosthesis:
Can stay in place for 3-6 months

Patient can change prosthesis independently More education is required for removal, cleaning, etc. Must have 2cm or greater tracheostoma Must pass esophageal insufflation test

Requires SLP to remove/replace

Less maintenance required

Must have 2cm or greater tracheostoma Must pass esophageal insufflation test

TEP

Overview:
F0 in phonation, speech
Intensity Perturbations Range Temporal aspects - VOT, Rise Time, Fall Time in phonation, MPD, Vowel duration, Rate of speech, Pause time and Total duration Spectral aspects Format structures, LTAS Prosody in alaryngeal speech

Fundamental Frequency (Fo)

Electrolarynx/ Artificial Larynx:

Most of have the mechanical speech aids are Some a variable frequency electronic and have a manually adjustment. adjustable fundamental frequency. Because Fo is determined by the electronic design of the These are typically set to specific a low pitch for a instrument, little data have been reported male voice (about 100 Hz) and, where on the Fo speech possible, tocharacteristics a higher valueof for a female produced with the electro larynx. voice (about 200 Hz).

Esophageal Speech:
The F0 of the esophageal voice is typically about 1 octave lower than the average laryngeal Fo of a male voice, whereas the female esophageal voice is about 2 octaves lower than normal. Better esophageal speakers tend to produce somewhat higher Fos whereas poorer speakers may produce somewhat lower Fos.

Slavin and Ferrand (1995) grouped according to their average Fo and variability characteristics

26 esophageal speakers

Most of them had difficulty controlling their Fo during dynamic speech.

esophageal speakers exhibit greater variability than normal speakers.

Some authors believe the Fo of esophageal voice depends on the exact location of the vibrating segment, but there is little evidence to support this hypothesis.

 Weinberg (1980) normal pattern of high Fo with high vowels higher Fo in females than males. This is due to the morphology of PE segment in females which is smaller and thinner.

Ranges from 29.37Hz (Perry & Tikofsky, 1965) to 86.50 (Horri, 1982).

F0 characteristics of Esophageal Speech (Reading Task):

Study Damste, 1958 Snidecor & Curry, 1959) N 20 6 Sex M M Mean 67.50 62.80 S.D 4.80 Range -

Shipp, 1967
Wienberg & Bennet, 1972 Robbins et al. 1984

16
15

M
F

64.74
86.65

4.98
3.94

16.00
21.25

15

77.10

4.43

34.23

pharyngoesophageal segment

The total laryngectomy procedure produces a defect in the hypopharynx that must be reconstructed to form the pharyngoesophageal segment (PES)

This tubular shaped region, composed largely of skeletal muscle tissue, serves as the neoglottis and enables production of esophageal voice

Extends over C4, C5 and C6

Spasticity or hypertonicity results in poor speech

Air flows through the PE segment causing it to vibrate

Morphology at Rest
Length The excellent TEP speakers- the shortest visible vibratory segment, followed in increasing order by the good, fair and poor speaking groups
These differences are generally very small Length of the PES is a significant contributor to TEP speech proficiency

Thickness
Multilayered and/or mucosally redundant structure Excellent, good, and fair groups- very thin and very thick extremes. Poor speaking group exhibit mildly to moderately thickness

 Subjects with thicker vibratory segments generally produce more hoarse-harshstrained vocal quality, greater dysfluency, and pitch and loudness dyscontrol.

PES thickness voice and speech proficiency.

lack of synchronous mucosal vibratory activity caused by thickened PES mucosa

Biomechanics During Phonation

Vibratory rhythmicity
Dyssynchronous PES vibratory patterns Positive correlation between the synchrony of PES vibrations and the associated level of TEP speech proficiency

Vibratory stiffness
Exhibits at least some degree of vibratory stiffness

Excellent

good

fair

poor

PES spasmodic hypertonic vibratory activity

speech proficiency.

As the pathophysiologic signs increase in severity, communication efficiency decreases

Mucosal waves
excellent, good and fair groupsmoderately retarded mucosal waves more severe disturbances observed in the poor group

strong positive correlation

degree of perceived PES spasms or hypertonicity mucosal wave abnormalities.

A relatively strong interrelationships

TEP speech proficiency PES mucosal wave integrity.

Muscular control
The speakers in the excellent, good, and fair groups- moderate degrees of PES muscular control poor group- mild degree o f PES muscular control

Tracheoesophageal Speech:
Tracheoesophageal speakers tend to produce Fos that are closer to normal laryngeal speakers, at least for male speakers. The variability of Fo is also somewhat less than esophageal speakers, but individual speakers may show considerable variation.

 Juarbe et.al (1989) collected data from 10 subjects with flap reconstruction. For these 10 subjects, the range in Fo was the most limited. F0 Ranges from 50.40 (Kyatta, 1964) to 100 (Zanoff et al., 1990).

Weinberg (1980)
Higher Fo in TEP compared to esophageal speech due to pulmonary air supply.

As noted above the Fo of TEP is commonly aperiodic. Damste (1958) quoted reasons for this aperiodicity
Due to variation in subneoglottic pressure. Length and elasticity of the PE segment is not constant and adjustable as in normals.

Fo characteristics of TE Speech

(Reading Task):
Study Robbins, et al, 1984 Trudeau & Qi, 1990 N 15 Sex M Mean 101.70 S.D 3.56 Range 37.46

10

108.6

2.68

Moon & 16 Wienberg, 1987 Wienberg & Bennet, 1972

Merwin et al, 1985

M M

64.74 72.73

4.98 .91

16.00 22.44

83.80

Comparison of Fundamental Frequency characteristics in normal, TEP and EP individuals: (Robbins, et al, 1984).

Vocal Intensity
Electrolarynx/ Artificial Larynx:
Users of an electro This level is typical larynx can produce of normal laryngeal average intensity speakers during levels during speech ordinary ranging between 75 conversation or and 85 dB reading.

There is some evidence for a reduced intensity range for users of electro larynges.

As was the case for Fo, the intensity of the electronic vibrator is largely determined by the design of the instrument.

Intensity characteristics of individuals with Electrolarynx

Study N Sex Mean S.D Range

Hymen, 1955

83.00
7.00

Weiss & 5 Komshian, 1979

74.00

1.87

5.00

Esophageal Speech:
The intensity of esophageal speech is more variable and somewhat lower in overall loudness than normal.

The range of voice intensity that esophageal speakers are able to produce is much less than the intensity range of normal laryngeal speakers (about 10 dB vs. 30 dB).

Intensity characteristics of individuals with Esophageal Speech:

Study Hymen, 1955 N 7 Sex M Mean 73.00 S.D 11.00 Range

Snidecor & Isshiki, 1965 Hoops & Noll, 1969 Baggs & Pine, 1983

85.00

20.00

22

62.40

3.60

10.55

8.96 (Recorded in mm from a graphic level recording. Not converted to dB). 59.30

1.58

4.33

Robbins et al, 1984

15

10.09

Tracheoesophageal Speech:
The intensity of tracheoesophageal speech appears to be only slightly less than the levels produced by laryngeal speakers. Variation of intensity may be somewhat greater than normal speakers. Some tracheoesophageal speakers habitually produce greater than normal intensity levels.

 Robbins et al (1984) compared TE, esophageal and normal speech under identical sets of conditions.

In terms of vocal intensity laryngeal speech occupied the middle ground, being on the average 10 dB more intense than the esophageal speech and 10 dB less intense than the TE speech in oral reading and sustained vowel phonation.

Intensity characteristics of individuals with TEP:

Study N Sex Mean S.D Range

Robbins, et al, 1984

15

79.40

2.10

13.8

Trudeau & Qi, 1990

10

70.80

8.50

29.00

Baggs & Pine, 1983

19.56

3.22

15.69

Author Baggs and Pine (1983)

Method Comparison intensity Esophageal speakers.

Results of vocal Larger intensity in TEP between speakers. Due to greater and TE intraoral pressure.

Singer (1983)

Esophageal speaker and Considerable lower TEP speaker. intensity with TE speaker.

Blood (1984)

Laryngeal and TEP

Higher intensity with TEP speakers.

Robbins et al (1984)

15 normals, esophageal, Sustained vowels: TEP N: 76.9 dBSPL sustained vowels, Eso: 74 dBSPL Paragraph reading. TE: 88 dBSPL Paragraph reading: N: 69.3 dBSPL Esophageal: 59.3 dBSPL

Debruyne (1994)

12 TE, 12 Esophageal

TEP: 79.4dBSPL Vowel

Esophageal: 79.7 dBSPL

Veena.K.D (1998)

TE: 65 dBSPL 5 each normals, N: 72.3 dBSPL Esophageal and TE Esophageal: 35.5 dBSPL TE: 32.6 dBSPL

Comparison of Intensity characteristics in normal, TEP and EP individuals: (Robbins, et al, 1984).

Perturbation Measures

Frequency perturbation (jitter)

reflects the frequency stability of the vocal folds. mean period difference, jitter ratio, jitter factor, relative average perturbation (RAP), and directional perturbation.

 Jitter ratio - ratio of the average period directional Jitter ratio difference and the average period. jitter Directional jitter - number of sign changes of the period differences divided by the total frequency perturbation number of periods. alaryngeal This ratio is then in multiplied by 100 to yield a sp percentage measurement.

Electrolarynx/ Artificial Larynx:

No reported studies of frequency perturbation in speakers using an electro larynx. However, jitter expected to be directly related to the stability of the electronic circuit producing the tone would not reflect the speech characteristics of the speaker.

Esophageal Speech:
Esophageal speech more unstable than normal laryngeal speech - as reflected in much larger jitter ratios. Directional jitter is about the same magnitude as normal speakers.

Author
Hoops and Noll (1969)

Method
22 esophageal rainbow passage

Results
Jitter(%): 41.1%

Smith et al (1978)

9 esophageal phonation /a/

Jitter: 0.62 to 5.13 msec Jitter ratio: 95:47

Tracheoesophageal Speech:
The data on jitter characteristics of tracheoesophageal speakers are unclear. One study reports a jitter ratio very similar to normal speakers, whereas another reports a much higher than normal value.

Jitter values to be similar to those of esophageal speakers as both groups of speakers use the same anatomical system as the vibrator, that is, the PE segment.

Author Robbins et. al (1982) Kinshi and Amatsu (1986) Trudeau and Qi (1990) Pindzola and Cain (1989)

Measure % jitter Mean jitter Jitter ratio Mean jitter Jitter ratio Directional jitter Jitter %

Laryngeal 0.77 0.07 10

TE 5.14 0.47 30 1.78 msec 134.8 63.7 4.59

Esophageal 18.25 0.82 60

2.03

7.65

Rajashekar (1990) Single case

Rajashekar (1991)

20 TE and Esophageal speakers

Bertino et al (1996)

Extent of fluctuation Speed of fluctuation extent of fluctuation speed of fluctuation

19 Hz 36 Hz

9.2 Hz 14 Hz

13.3 Hz 14.6 Hz

10.4 Hz 16.5 Hz

Jitter and shimmer of TE is more similar to normal speakers than esophageal

In TE speech
more regular pattern in jitter values due to expiratory airflow which is more efficient driving force than the small ejections of air out of esophagus.

Larger jitter in females for TE speakers

attributed to their higher Fo and small VC.

jitter ratio

elapsed time between laryngectomy and voice recording

Trudeau and Qi (1990)

These combined findings seem to indicate the type of surgery, particularly as the surgery transplants other tissue into the area of the PE segment, affects the acoustical nature of speech produced by the puncture.

Amplitude perturbation (Shimmer)

index of the stability of a sound source The average difference in amplitude between adjacent cycles of vibration (dB) Directional shimmer, like directional jitter, is the number of changes of sign between adjacent periods divided by the total number of period differences, again multiplied by 100.

Electro larynx

reflect the electronic design and construction of the instrument and not the inherent anatomical or physiological capabilities of the speaker.
Esophageal speakers Shimmer of is greater than normal whereas directional shimmer is very similar to normal speakers Tracheoesophageal Both shimmer and directional shimmer are greater than normal speakers.

Author Robbins (1982) Robbins (1984) Rajashekar (1991)

Method Shimmer ratio

Task /a/

Laryngeal 0.43 0.3 dB

TE 10.55 0.80

Esophageal 24.15 1.90

Mean shimmer /a/ 20 TE, 20 Esophageal

6.8 dB 28.4 dB

3.8 dB 3.3 dB

Extent of fluctuation
Speed of fluctuation

Pauloski et Lower shimmer in TE speakers who wore low pressure al (1989) prosthesis and spoke by digital occlusion.

Temporal Characteristics
Temporal measurements reported on alaryngeal speech

words per minute (wpm)

syllables per second

total duration of reading

words or syllables per air charge

wpm as a measure of speech rate.

total vowel duration, or the maximum time a speaker can sustain a vowel.

percentage of silence during reading aloud, used as a measure of pause time.

To a large extent, all of these measures reflect the speakers ability to control the regressive air stream. For the esophageal speaker, they also reflect the ability to quickly recharge the esophagus with sufficient air.

 For users of an electro larynx, phonation time is dependent on the vibrator Esophageal speaker on the speakers Silence is dependent Small air facility with the on/off button. volumes present

in the esophagus

TE speakers

full pulmonary air supply

 The reading rate of normal adults speakers (between 40 and 70 years of age; ages most appropriate for comparison with laryngectomies) is about 173 wpm. Rates much less than 140 wpm are usually perceived as slow and rates above 185 wpm are perceived as fast (Franke, 1939). Normal speakers can produce about 13 words per breath of air, which averages to about 4 seconds in duration (Snidecor & Curry, 1959).

 Reading rates are slower when using an electro larynx compared to normal phonation or to tracheoesophageal speech (Merwin et al. 1985; Weiss & Yeni-Komshian, 1979).

We might expect longer reading times for electro larynx users because of the need to produce more precise articulation to maintain an acceptable level of intelligibility.

Esophageal speakers read slower than normal laryngeal speakers.

Rates between 100-115 wpm appear typical for these speakers, which is about 60-70% of the rate of normal speakers.

Esophageal speakers generally spend about 30-45% of their reading time in silence.
These abnormally long silent periods reflect the more frequent need to recharge air supply.

Better esophageal speakers have much shorter periods of silence

more rapid air intake with less interruption of speech flow.

A much shorter sustained duration of phonation than normal speakers, typically less than 6 seconds (vs. 15-20 seconds for normal speakers).

small volume of air in the esophagus.

Tracheoesophageal speakers read at a slower rate than normal speakers but faster than esophageal speakers. difficulty in controlling the PE segment and the need to articulate precisely. These speakers spend bout 10-30% of their time in silence The ability to use full pulmonary air supply to drive the PE segment.

Tracheoesophageal speakers also can produce long phonation durations (about 12 seconds) for the same reason

TE speakers (97-136 wpm)

esophageal speakers (110-115 wpm)

laryngeal speakers (166 wpm)

Studies on Esophageal speech:

Author Snidecor and Curry (1960) Filter and Hyman (1975) Sanyogeetha (1993) Results Eso: group average of 113 wpm

2.5 syllables per second for good Esophageal speaker Rate of speech was less in Esophageal compared to normals

Studies done on TE speakers:

Author
Singer (1983). Pauloski et al. (1989)

Method
4 TE TE Duck-bill Vs Lowpressure TE

Results
97-136 wpm. High rate of speech with low pressure prosthesis 2.86 syllables/seconds

Sedory et al (1989)

Robbins (1984); Sedory (1989)

TE

Fast rate of speech ranging from 2.6 to 3.6 syllables per second in TE speakers

Rate of speech across groups:

Author Method Laryngeal Esophageal TE

Baggs and Sentences Pine (1983) Robbins et Rainbow al (1984) passage Veena K.D 5 each (1998). normals, Eso and TE

182.5 wpm. 117.7 wpm.

132.4 wpm

172.8 wpm. 99.1 wpm.

127.5 wpm

5.43 1.85 syllables syllables per second. per second

3.44 syllable per second

Comparison of WPM in normal, TEP and EP individuals: (Robbins, et al, 1984)

Other temporal characteristics: RT-FT in phonation

Pause time

VOT

MPD

Total duration

VOT
physical characteristics of neoglottis

myoelastic

motor control properties

responsible for VOT in alaryngeal speech

Author Klor and Milanti (1980)

Method VOT for pre-vocalic stop consonants Laryngeal, Esophageal speakers Esophageal and TE speakers

Results Reduced VOT in alaryngeal speakers

Weinberg (1982)

Esophageal speakers are far less consistent than normals in effective variations in timing of voicing onset Longer VOT Laryngeal>TE>Esophage al

Robbins, Chrinstensen and Kempstar (1986)

VOT in voiceless consosnants Normals, Esophageal and TE speakers Normals and TE speakers

Santhosh Kumar (1993)

Greater VOT in TE than normals (contrasts with Robbins et al)

Author
Venkatraj Ajthal (1997)

Method
Normals & TE

Results
VOT for /p/ /t/ /k/ and /th/ was longer in TE than normals in both initial and final positions. Slightly shorter VOT for TE for/b/ /d/ /g/ and /dh/ compared to normals in both initial and medial positions.

Sacco, Mann and Schultz (1967); Marshall (1974)

Esophageal

Listeners misidentified consonant voicing contrasts in Esophageal. He attributed this as a cause for reduced intelligibility.

Chrinstensen, Weinberg and Alfonso (1978)

VOT in a large number of consonants

Average VOT associated with prevocalic voiceless stops of Esophageal was significantly shorter than normal

2. Rising time; Falling time in phonation

Author Rajashekar et al. (1990). Method TE Results Greater RT and FT in TE. Attributed to more pressure required to initiate and sustain phonation in TE speakers RT shorter than normals. TE showed longer FT than normals on/i/ and /u/ whereas normals showed longer FT in /a/.

Santhosh Kumar (1993)

Normals and TE speakers

3. MPD
Author Baggs and Pine (1983) Robbins (1984). Results Longer PD in TE compared to Esophageal, however, MPD in TE was shorter than normals Attributed reduced MPD in TE to High airflow rates Poor digital occlusion of the stoma Poor MPD in Esophageal to limited air supply MPD: Laryngeal: 22 secs. TE: 12 secs. Esophageal: 6 secs Lower mean MPD in TE compared to normals.

Robbins, Fisher, Blom and Singer (1984)

Santhosh Kumar (1993).

Comparison of MPD in normal, TEP and EP individuals (Robbins, et al, 1984).

4. Vowel duration:
Author Christensen and Weinberg (1976) Robbins, Chrinstensen and Kempstar (1986). Hariprasad (1992). Method VD Results Longer VD in voiced for Esophageal as against the voiceless in normals Normals had shorter VD, Esophageal intermediate and the TE longest. Alryngeal speaker uses longer VD as a compensatory strategy to increase intelligibility of speech Esophageal had longer VD than normals for /a/ /o/ and /u/. shorter VD for /u/ /a/

normals and Esophageal

15 each normals, Esophageal and TE.

10 vowels Esophageal.

Sanyogeetha (1993

Normal and Esophageal

Longer VD in TE speakers attributed to:

Pulmonary air as a driving source. Greater air pressure and sustained flow rates driving the neoglottis, producing slower decay in PE segment vibration.

5. Word duration:
Author Venkataraj Aithal (1997) Method Laryngeal and TE speakers. Word reading task. Results TE used longer WD compared to normals.

This is attributed to lack of efficient timing control in initiation and termination of voice in Te speakers and also changes in articulatory behavior secondary to laryngectomy.

Pause time:
Esophageal: 30-40% in silence. Better Esophageal speakers-shorter PT. TE: 10-30%
Author Method Laryngeal
0.62

Esophageal TE
0.65 0.89

Robbins et al Rainbow (1984) passage

Spectral aspects:
Esophageal:

Weinberg (1982): elevated formant frequency.

Sindecore (1968): irregular striations.

Author
Sanyogeetha (1993)

Method Normals, Esophageal

Results

Higher except /o/, /u/ in Mean F1, F2, and F3 for Esophageal vowels /a/, /i/, /u/, /o/, /e/

Hariprasad (1992).

Normals and Esophageal

Space between formants increase, speech intelligibility increases

TE
Author Method Results Wider space between formants reduced F3 Christensen and vowels Weinberg (1976) Santhosh (1993) Kumar /a/ /i/ /u/ /e/ /o/

VenkatrajAithal (1997) Hammberg and Nord (1989)

10 vowels

Higher Fo, F2 and F3

Normals and TE

Alaryngeal voice had weaker Fo than F1

Prosody in alaryngeal speech

Intonation and stress: Weinberg (1980): TE were able to control Fo duration. Intonation and stress as like normals but change in frequency is discontinous.

TE and Eso-produce stress syllable but not on the same syllable. Intonation contrasts were seen in laryngeal, TE and Eso but Electro-larynx-not able to achieve these intonation distinctions.