Speech with Maxillary Implant Prostheses: Ratings of Articulation

Abstract

Speech is often perturbed after placement of maxillary implant-retained prostheses. We tested the hypothesis that the rate of speech errors varies with prosthetic design. Thirty edentulous subjects with mandibular implant prostheses entered two within-subject crossover trials. Subjects wore maxillary fixed prostheses and removable long-bar overdentures (Trial 1), or overdentures with and without palates (Trial 2). Test words from a French language speech battery were recorded after each prosthesis had been worn for two months. The percentages of stops, fricatives, and vowels correctly perceived by lay judges were calculated. Subjects produced a significantly higher percentage of sounds correctly with overdentures than with fixed prostheses. Between-treatment differences were significant for stops and fricatives (p < 0.01), but not for vowels. There were no significant differences in error rates between the two overdentures. In conclusion, maxillary implant overdentures with and without palates enable patients to produce more intelligible speech than fixed prostheses.

INTRODUCTION

Dentists know that dental prostheses sometimes alter speech, and patients identify speaking ability as an important factor in their satisfaction with implant prostheses (Awad and Feine, 1998; Zitzmann and Marinello, 2000).

Speech problems with maxillary fixed prostheses are frequently reported, mostly during the first weeks after delivery (Haraldson and Carlsson, 1977; Lundqvist et al., 1992a). Lundqvist et al.(1992a) reported that 60% of the patients in a clinical trial had distorted speech soon after treatment, and 3 yrs later the rate was still 30%. However, it is hard to extrapolate these findings to the general population, because 67% of the subjects had hearing deficits, which themselves have a negative impact on speech (Lundqvist et al., 1992a,b). Another study showed that approximately 9 yrs after the placement of implant prostheses in the mandible and/or maxilla, 82% of patients still made articulatory errors, compared with only 52% of subjects with natural teeth (Jacobs et al., 2001).

The gap between mucosa and fixed prostheses is thought to be a major cause of speech errors (Lundqvist et al., 1992a). It can be closed if removable appliances are used, but these usually cover the palate, which may also interfere with speech (Petrovic, 1985). Indeed, when the palate of a dentate subject is covered experimentally, the articulation of consonants is often abnormal, even after prolonged periods of adaptation (McFarland et al., 1996; Baum and McFarland, 1997).

We have carried out a series of within-subject crossover comparisons of maxillary implant-supported prostheses, and have reported on patients’ satisfaction with and ability to chew with fixed prostheses, removable overdentures without palates, and removable overdentures with palatal coverage (de Albuquerque et al., 2000; Heydecke et al., 2003).

In this paper, we report on the quality of speech produced by the study subjects. Sounds were selected to cover the major categories used in spoken French and English: vowels, stops, and fricatives. Speech sounds are classified as voiced or unvoiced, depending on the presence or absence of vocal fold vibration. All vowels are voiced; consonants are either voiced or voiceless. Fricative consonants are produced when oral airflow is restricted, creating turbulence, while stops are produced when air flow is blocked, and there is a rapid release of air (Shriberg and Kent, 1982; McFarland and Lund, 1995).

MATERIALS & METHODS

Study Design

Thirty completely edentulous French-speaking subjects aged 30–60 yrs initially enrolled in two within-subject crossover trials. The 15 chosen for Trial 1 had enough residual bone in the maxilla to receive 6 root-form implants. The subjects entering Trial 2 had sufficient bone for 4 implants. Subjects had normal hearing, as assessed by pure-tone audiological testing, were non-smokers, and had no known history of disorders of the respiratory, laryngeal, or speech articulatory systems. A speech pathologist assessed them, and only those with no speech pathology were admitted to the trial. Details of the recruiting process and study design have been described (de Grandmont et al., 1994; de Albuquerque et al., 2000; Heydecke et al., 2003). The protocol was reviewed and approved by the Institutional Review Board for the protection of human subjects. Written informed consent was obtained.

In Trial 1, the 15 subjects were given maxillary implant-retained long-bar overdentures without palate (LBO₁) and fixed maxillary prostheses (FP). They were already wearing mandibular removable implant overdentures. The 15 subjects in Trial 2 received long-bar overdentures without (LBO₂) and with palatal coverage (LBOP). They wore mandibular fixed implant prostheses (Fig.). In each trial, we used a quasi-random process to determine which maxillary prosthesis was to be worn first (Cox, 1958).

Treatment

Six (Trial 1) or 4 (Trial 2) implants (Brånemark^®, Nobel Biocare, Göteborg, Sweden) were placed in the maxilla. Six mos later, healing abutments were attached. The FP was built on a framework with a distal cantilever of 10–15 mm. A parallel-sided gold bar (Cendres et Métaux SA, Montreal, Canada) was fabricated for the long-bar overdentures. LBO₁ and LBO₂ had no palate; the LBOP had a palate about 2 mm thick. Contour, vertical dimension, and occlusion were replicated to match the two different types of prostheses for each patient as closely as possible, and the same types of teeth were used (de Albuquerque et al., 2000; Heydecke et al., 2003). Both prostheses for each patient were designed by the same prosthodontist.

Analysis of Speech

Three sessions were held at two-week intervals following a two-month adaptation period with the prostheses. Phrases were shown on a computer screen, and subjects were asked to repeat them. Sample words from a French articulation test battery (Centre Hospitalier Côte-des-Neiges, 1986; for examples, see Table 1) containing 12 consonants were embedded in a carrier phrase ("vous dites____encore") to mimic speech articulation in normal conversation. They were also asked to pronounce 3 vowels. Practice trials were given prior to each session. Speech was recorded on digital audio-tape by a microphone placed 10 cm from the mouth of each subject. The phrases were presented in random order.

To assess intelligibility of speech to average listeners, we recruited two native French-speaking lay judges (Sereno et al., 1987; Flege et al., 1988). They listened to isolated consonants and vowels edited from the carrier phrase by using the Brown Lab Interactive Speech System (BLISS) software (Mertus, 1988). For stop and fricative consonants, only the sounds from the onset of the release burst to the onset of voicing for the following vowel were included. Vowels were edited from the final [t] or [ts] of “dites” through the end of voicing.

For each treatment, we digitally isolated and stored 1170 isolated stops, 1170 fricatives, and 585 vowels. To reduce the burden on judges, we randomly selected 20% of the total sounds (Sereno et al., 1987; Flege et al., 1988). Test files comprised of 234 stops, 234 fricatives, and 117 vowels were created and presented in random order to the judges by means of headphones.

Judges then chose the sound that most closely matched the one they heard from a list. Later, the percentage of correctly identified sounds was calculated from the 5 repetitions for each consonant and vowel. The percentages of correctly perceived sounds by each judge were designated as the “responses” for each sound.

Patient Self-assessment

After 2 mos of adaptation, subjects rated their ability to speak with each prosthesis on 100-mm Visual Analog Scales (VAS). The anchor words were “totally dissatisfied” and “completely satisfied”.

Statistical Analyses

Agreement between judge responses was assessed according to Pearson’s correlation for continuous data. Comparisons of the two treatments within Trial 1 and Trial 2 were carried out by t tests. Correlations between patients’ VAS ratings of ability to speak and judge ratings for all sounds were also calculated (Pearson’s r).

RESULTS

Thirteen subjects completed the study in each of the trials (Fig.). Details of the course of study, drop-outs, and patient characteristics have been reported elsewhere (de Albuquerque et al., 2000; Heydecke et al., 2003).

Agreement between the two judges was very high and significant. In Table 2, we give the Pearson’s correlation coefficient for each of the comparisons that were made between treatments. This shows that r was > 0.9 for each. Because of this, ratings from the two judges were pooled for the other analysis.

Between-treatment Comparisons

No significant treatment, period, or treatment/period interaction effects were found in either trial for the judge ratings of speech. Therefore, direct comparisons were carried out with the full dataset from both treatment phases (Hills and Armitage, 1979). In Trial 1, judge ratings indicated that a significantly higher percentage of consonants was correctly produced with the LBO₁ than with the FP, and between-treatment differences were significant for all classes of stops and fricatives (Table 3). Voiced and voiceless consonants were equally affected, but the rate of error was slightly higher for fricatives than for stops. When comparisons were made on the basis of tongue and lip position, it was seen that the frequency of errors was greatest in the production of linguo-palatal fricatives (j, ch), and these were also the sounds that showed the greatest effect of treatment (▵ = 38.7%). There were no errors recorded for velar stops (g, k) in the LBO₁ group, but even for these sounds, which are made with the tongue tip at the back of the mouth, the error rate in the FP group was almost 25%. There were only 4 errors in the production of vowels in the FP group and none in the LBO₁ group, and the difference was not significant. In Trial 2, there were no significant differences in judge ratings of sounds correctly produced between the LBOP and LBO₂ groups (Table 3 ).

The mean rating given by subjects for their ability to speak with the LBO₁ was significantly higher (▵ 20.2 mm VAS) than the rating for the FP (p = 0.023). However, the difference between overdentures with and without palates was small (4.7-mm VAS) and not significant.

Correlations between patient VAS scores and mean error scores were moderate but significant for the FP-LBO₁-treatment comparison (r = 0.48, p = 0.03) but low and non-significant for the LBOP-LBO₂-comparison (0.24; p = 0.29).

DISCUSSION

Our results indicate that more speech errors were associated with wearing implant-supported maxillary fixed bridges than with removable prostheses. However, the presence or absence of palatal coverage seems to make no difference to intelligibility. The subjects also rated their speaking ability to be lower with the fixed prosthesis.

Vowels were not significantly affected by the prostheses. This seems to fit with the fact that compensation for changes in the form of the oral cavity is more immediate and complete for vowels than for consonants (e.g., Hamlet and Stone, 1976, 1978; Petrovic, 1985; McFarland and Baum, 1995; McFarland et al., 1996; Baum and McFarland, 1997). Vowels are produced with a relatively non-constricted vocal tract; therefore, the location and form of the prosthesis may have little impact on vowel articulation.

We also found no between-group differences in the articulation of consonants in the LBOP-LBO₂ comparison. The only difference between these prostheses was the 2-mm-thick palate of the LBOP. Although an artificial palate disturbs the pronunciation of linguo-palatal fricatives in dentate subjects (McFarland et al., 1996), all our subjects had previously worn complete upper dentures. They had probably developed compensatory strategies for palatal coverage that aided adaptation.

There were no significant differences in the sound intelligibility between the overdentures tested in Trial 2 (LBO_1,2) and in Trial 2 (LBOP). This suggests that the two types of mandibular prostheses (removable overdenture, Trial 2; and fixed prosthesis, Trial 1) had no influence on speech articulation. This is consistent with subjects’ self-assessment of their speaking ability when wearing fixed and long-bar mandibular overdentures (Heydecke et al., 2003), and with the fact that many speech sounds are produced with the tongue approximating the maxilla, not the mandible.

The percentages of correctly produced consonant sounds were significantly lower for the FP. This was true for stops and fricatives regardless of whether the sound was voiced or voiceless. Consonants, and fricatives in particular, are very susceptible to changes in oral form, perhaps because they require high articulatory precision (Stoel-Gammon and Dunn, 1985; McFarland and Baum, 1995). Earlier research had shown that the [s] sound, a linguo-alveolar fricative, is influenced by the shapes of complete dentures and by palatal appliances (Pound, 1977; Petrovic, 1985; Lundqvist et al., 1992a,b; McFarland et al., 1996; Baum and McFarland, 1997). Stops, also significantly affected by the wearing of a FP, fall between fricatives and other consonants in their susceptibility to changes in oral form (Garber et al., 1980; McFarland and Baum, 1995). We tested speech after a two-month adaptation period. Consequently, our results are likely to predict long-lasting effects of the FP on speech and sound production.

Space is left between the alveolar ridge and a fixed maxillary prosthesis, and air passing through it may be the cause of the higher error rate for linguo-alveolar and linguo-palatal stops and fricatives in the FP group. However, we were surprised to find that the error rates for bilabial stops and labio-dental fricatives were also significantly higher in the FP group. It is probable that the space also has an impact on the build-up and release of intra-oral pressure for all stops, and on turbulent air generation during formation of fricatives, regardless of tongue position. Stops and fricatives are particularly sensitive to increases in the distance between the jaws (Flege et al., 1988; McFarland and Baum, 1995), but this was not a factor in our study, because the two types of dentures were of the same height.

The subjects perceived that their ability to speak was significantly worse with the FP than with the LBO, but, like the judges, they perceived no difference in their speech with and without palate. However, the degree of correlation between patient VAS scores and judge ratings was moderate for the FP-LBO₁ comparison (r = 0.48). The lack of a significant correlation in the LBOP-LBO₂ comparison is not surprising, because the variability in speech errors and VAS scores was very low.

Our findings are in contrast to those of Zitzmann and Marinello (2000), who found no significant differences between LBO and FP prostheses for patient satisfaction with speech. However, this was a between-group study, and subjects were not randomly assigned. The fixed prosthesis was given to the subjects with the most residual bone, and the removable appliance to those with the least (Zitzmann and Marinello, 2000). This may have minimized residual air space in the fixed group.

The sample size for this trial was calculated for ratings of general satisfaction and not speech intelligibility (de Albuquerque et al., 2000; Heydecke et al., 2003). Nevertheless, it was large enough to show significant differences between FP and LBO. Although it is possible that a difference in speech between LBOs with and without palates could be shown with a large population, the differences in speech errors and VAS scores between groups were very small and probably of no clinical importance.

In conclusion, maxillary implant overdentures with and without palates enable patients to produce more intelligible speech than do fixed prostheses. Together with previous findings, this leads us to conclude that the long-bar overdenture is the treatment of choice for patients with an edentulous maxilla.

Table 1.

Classification of Sounds Used in the Judge Evaluation of Speech

		Bilabial		Linguo-alveolar		Velar
Stops	Articulation	Sound	Sample Word	Sound	Sample Word	Sound	Sample Word
	Voiced	b	banc	d	dent	g	guerre
	Voiceless	p	port	t	temps	k	code
		Labio-dental		Linguo-alveolar		Linguo-palatal
Fricatives	Articulation	Sound	Sample Word	Sound	Sample Word	Sound	Sample Word
	Voiced	v	vin	z	zone	j (3)	jour
	Voiceless	f	fer	s	sept	ch (∞)	champs
Vowels			a		i		u

Table 2.

Agreement Between the Ratings of the Two Judges (J1, J2) for Each Treatment within Each of the Two Treatment Arms (Pearson’s r)

	J1-J2LBOP	J1-J2LBO 1	J1-J2FIXED	J1-J2LBO 2
Pearson correlation	0.910	0.941	0.942	0.905
p (two-tailed)	0.0001	0.0001	0.0001	0.0001
N	585	584	584	584

Table 3.

Judge Ratings of Correctly Produced Speech Sounds (%) within Each of the Two Trials and Between-group Comparisons (t tests)

			Trial 1—Treatment								Trial 2—Treatment
			Fixed (n = 13)		LBO 1 (n = 13)			Mean	95% CI		LBOP (n = 13)		LBO 2 (n = 13)			Mean	95% CI
Classification of Sounds^c			Mean	SD	Mean	SD	p	Difference	Lower	Upper	Mean	SD	Mean	SD	p	Difference	Lower	Upper
^a Number of observations per treatment group.
^b Cannot be computed because the standard deviations of both groups are 0.
^c The sounds were classified by articulation and by voicing.
Stops (n = 234^a)
	Bilabial	(n = 78^a)	68.3	22.5	88.3	7.8	0.005	−20.1	−33.2	−7	92.7	7.2	92	3.5	0.775	0.7	−4.5	6
	Linguo-alveolar	(n = 78)	69.7	24.8	97.5	4	0.001	−27.8	−41.7	−13.9	97.7	3.4	95	4.1	0.113	2.7	−0.7	6.2
	Velar	(n = 78)	75.7	26.3	100	0	0.003	−24.3	−38.9	−9.8	98.2	6	100	0	0.353	−1.8	−5.8	2.2
Fricatives (n = 234)
	Labio-dental	(n = 78)	76.7	25.8	99.2	2.9	0.005	−22.5	−36.9	−8.1	98.2	3.4	98	3.5	0.905	0.2	−3	3.3
	Linguo-palatal	(n = 78)	67.3	23.1	90.8	21.1	0.011	−23.5	−41.1	−5.9	78.6	26.2	83	13.6	0.642	−4.4	−23.7	15
	Linguo-alveolar	(n = 78)	43.8	21	82.5	31.9	0.002	−38.7	−61.5	−15.8	86.4	14.8	88	11.6	0.783	−1.6	−13.9	10.6
Stops (n = 234)
	Voiceless	(n = 117)	70.4	14	95.3	4.6	0.000	−24.8	−32.9	−16.7	96.4	4.3	95.7	4.5	0.721	0.7	−3.3	4.7
	Voiced	(n = 117)	72.2	15.6	95.3	3.6	0.000	−23.1	−31.9	−14.3	96.1	2.5	95.7	2.7	0.735	0.4	−2	2.8
Fricatives (n = 234)
	Voiceless	(n = 117)	58	16.8	86.4	13.9	0.000	−28.4	−40.6	−16.2	87	10.5	86.7	7.4	0.940	0.3	−8.1	8.7
	Voiced	(n = 117)	62.2	17.2	90.6	16.8	0.000	−28.3	−41.9	−14.8	88.5	13.3	92.7	6.6	0.381	−4.2	−13.9	5.6
Vowels (n = 117)		98.8	4	100	0	0.341	−1.2	−0.039	0.015	1	0	1	0	^b
Ability to speak (n = 13)		72.9	27.3	93.1	7.6	0.023	−20.2	−37.0	−3.3	88.6	14.8	83.9	18.3	0.479	4.7	−8.8	18.2

Figure.

This Fig. shows the distribution into the two treatment arms of the two cross-over clinical trials and the sequence of data-gathering sessions.

Footnotes

Notes

A supplemental appendix to this article is published electronically only at .

Acknowledgements

The authors are deeply indebted to Dr. J. Larivée for her help during the surgical part of the study, to Drs. P. Boudrias, P. deGrandmont, and G. Gauthier for fabrication of the maxillary prostheses, to Dr. Ling Tang for gathering patient based data, and to Stephanie Wollin for editorial assistance. This study was supported by a Canadian MRC University-Industry Grant. Industrial sponsors were NobelBiocare Canada and Laboratoire Dentachrome Inc. G Heydecke was supported by the Deutsche Forschungsgemeinschaft (DFG; HE 3441/1-1).

References

Awad MA, Feine JS -1998-. Measuring patient satisfaction with mandibular prostheses. Community Dent Oral Epidemiol 26:400–405.

Baum SR, McFarland DH -1997-. The development of speech adaptation to an artificial palate. J Acoust Soc Am 102:2353–2359.

Centre Hospitalier Côte-des-Neiges -1986-. Protocole Montréal-Toulouse de l’examen linguistique de l’aphasie Montréal. Montréal: Centre Hospitalier Côte-des-Neiges.

Cox DR -1958-. Planning of experiments. New York: Wiley.

de Albuquerque RF Junior, Lund JP, Tang L, Larivée J, de Grandmont P, Gauthier G, et al. -2000-. Within-subject comparison of maxillary long-bar implant-retained prostheses with and without palatal coverage: patient-based outcomes. Clin Oral Implants Res 11:555–565.

de Grandmont P, Feine JS, Tache R, Boudrias P, Donohue WB, Tanguay R, et al. -1994-. Within-subject comparisons of implant-supported mandibular prostheses: psychometric evaluation. J Dent Res 73:1096–1104.

Flege JE, Fletcher SG, Homiedan A -1988-. Compensating for a bite block in /s/ and /t/ production: palatographic, acoustic, and perceptual data. J Acoust Soc Am 83:212–228.

Garber SR, Speidel TM, Siegel GM, Miller E, Glass L -1980-. The effects of presentation of noise and dental appliances on speech. J Speech Hear Res 23:838–852.

Hamlet SL, Stone M -1976-. Compensatory vowel characteristics resulting from the presence of different types of experimental dental prostheses. J Phonetics 4:199–218.

10.

Hamlet SL, Stone M -1978-. Compensatory alveolar consonant production induced by wearing a dental prosthesis. J Phonetics 6:227–248.

11.

Haraldson T, Carlsson GE -1977-. Bite force and oral function in patients with osseointegrated oral implants. Scand J Dent Res 85:200–208.

12.

Heydecke G, Boudrias P, Awad MA, Albuquerque RF, Lund JP, Feine JS -2003-. Within-subject comparisons of maxillary fixed and removable implant prostheses: patient satisfaction and choice of prosthesis. Clin Oral Implants Res 14:125–130.

13.

Hills M, Armitage P -1979-. The two-period cross-over clinical trial. Br J Clin Pharmacol 8:7–20.

14.

Jacobs R, Manders E, Van Looy C, Lembrechts D, Naert I, van Steenberghe D -2001-. Evaluation of speech in patients rehabilitated with various oral implant-supported prostheses. Clin Oral Implants Res 12:167–173.

15.

Lundqvist S, Lohmander-Agerskov A, Haraldson T -1992a-. Speech before and after treatment with bridges on osseointegrated implants in the upper jaw. Clin Oral Implants Res 3:57–62.

16.

Lundqvist S, Haraldson T, Lindblad P -1992b-. Speech in connection with maxillary fixed prostheses on osseointegrated implants: a three-year follow-up study. Clin Oral Implants Res 3:176–180.

17.

McFarland DH, Baum SR -1995-. Incomplete compensation to articulatory perturbation. J Acoust Soc Am 97:1865–1873.

18.

McFarland D, Lund JP -1995-. The control of speech. In: Studies in physiology. Vol. 3. Neural control of skilled human movement. Cody FWJ, editor. London: Portland, pp. 61–74.

19.

McFarland DH, Baum SR, Chabot C -1996-. Speech compensation to structural modifications of the oral cavity. J Acoust Soc Am 100:1093–1104.

20.

Mertus J -1988-. Brown Lab Interactive Speech System -BLISS- user’s manual. Providence, RI: Brown University.

21.

Petrovic A -1985-. Speech sound distortions caused by changes in complete denture morphology. J Oral Rehabil 12:69–79.

22.

Pound E -1977-. Let /S/ be your guide. J Prosthet Dent 38:482–489.

23.

Sereno JA, Baum SR, Marean GC, Lieberman P -1987-. Acoustic analyses and perceptual data on anticipatory labial coarticulation in adults and children. J Acoust Soc Am 81:512–519.

24.

Shriberg LD, Kent RD -1982-. Clinical phonetics. New York: John Wiley & Sons.

25.

Stoel-Gammon C, Dunn C -1985-. Normal phonological development. In: Normal and disordered phonology in children. Stoel-Gammon C, Dunn C, editors. Baltimore: University Park, pp. 15–46.

26.

Zitzmann NU, Marinello CP -2000-. Treatment outcomes of fixed or removable implant-supported prostheses in the edentulous maxilla. Part I: Patients’ assessments. J Prosthet Dent 83:424–433.