Effect of Age on Speech Perception in Noise Abilities Across Different Stimulus

Abstract

Various factors influence speech perception in noise (SPIN): age, hearing loss, cognition, background noise, stimulus redundancy, type of stimulus used, and signal-to-noise ratio. The effect of age on SPIN with different stimuli is yet to be validated in the literature. This study aims to study the effect of age on SPIN results across different stimuli. The study is a cross-sectional study with ninety participants with normal hearing ability. All participants in the study were equally divided into three groups: the children’s group aged 8 to 12 years, the adult group aged 18 to 30, and the older adult group aged 55 to 72 years. Speech perception in the background of Kannada speech babble was assessed across three stimuli: monosyllables (CV), phonemically balanced Kannada words, and Kannada sentences. The stimulus was presented at 60 dBSPL binaurally through a calibrated headphone at 0 dB SNR. The result indicated a significant main effect of age on SPIN across syllables, words, and sentences. Further, Mann-Whitney test results revealed a statistically significant difference between the SPIN scores of children and adults for syllables, words, and sentences. Also, a statistical difference was noted in SPIN scores between adults and older adults for syllables, words, and sentences. However, statistical differences between children and older adults were seen only for syllables. The trend showed that the SPIN scores for syllables, words, and sentences improve from childhood to adulthood, while scores deteriorate from the adult to older adult group. A similar trend was seen for all three stimulus types. However, the reduction in the SPIN score using syllables in the older adult group was more noticeable than words and sentences. It can be concluded that there is an effect of age on SPIN abilities across different stimuli. It highlights the importance of age-appropriate SPIN normative for various stimuli.

Introduction

Speech perception in noise (SPIN) is a complex process that incorporates detecting spectro-temporal cues for speech identification and separating signals from background noise [1]. SPIN is affected by various factors such as age, hearing loss, cognition, background noise, stimulus redundancy, speech rate, the type of stimulus used, signal-to-noise ratio, and language proficiency [2]. SPIN is influenced by age as it is an intrinsic human ability that matures with an increase in age and eventually deteriorates due to aging. The ability to interpret information in the presence of noise involves signal-noise separation, acoustic analysis, sound-to-word mapping, and phoneme mapping which is a distinguishing feature of our auditory system. Aging has been shown to impact these processing abilities even when the peripheral auditory system is intact [3]. People over 60 years experience perceptual difficulties when noise surrounds them and need at least 2 to 3 dB stronger signal-to-noise ratio (SNR) than young adults [4, 5].

In a longitudinal study, Divenyi et al. (2005) examined pure-tone audiometry, speech recognition abilities in quiet and noise, and speech perception of degraded stimuli in older individuals between six and nine decades old. The results showed that the reduction in speech perception ability significantly accelerated between the seventh and ninth decades of life compared to the decline in pure tone audiometric measures [6]. Heidari et al. (2018) also found that elderly individuals had lower SPIN scores than younger ones when comparing signal-to-noise ratios of 0 and − 10 dB. The same study also reported that in speech auditory brainstem responses (ABR), the range of Fundamental frequency (F0) amplitude in older individuals was lower than in adults. The authors reported that F0 enables the person to segregate simultaneous sounds easier.

Studies have also shown that SPIN scores improve with age in childhood [7–9]. SPIN ability matures to adult-like between 13 and 15 years [10]. It has also been reported that no significant differences were noted in word-in-noise tests between adolescents (14–16 years) and adults. However, for the sentence-in-noise test difference was noted between the children and adults [11].

Thus, one of the other factors affecting speech recognition ability is the type of stimulus employed for assessing SPIN. The characteristics of the stimuli and the context in which it is presented can impact perception. Speech perception is influenced by contextual, semantic, and synaptic constraints [12]. Different stimuli to assess SPIN include sentences, words, nonsense syllables, and continuous discourse [13, 14]. Nittrouer and Boothroyd (1990) studied the context effect of words and sentences in identifying phonemes and words using real words and nonsense CVC syllables and sentences in children and older adults. Results indicated that the overall performance of children and older adults was poorer than adults. They observed adult-like responses in children above ten years of age and SPIN scores with phoneme and word stimuli were found to improve with age from 7 to 10 years of age [12].

McArdle et al.‘s (2005) study obtained an SNR-50 in adults using digits in triplet sets, words, and sentences in the presence of multi-talker speech babble. Results reported that SNR-50 was significantly lower for digits than for words and sentences [15]. The authors concluded that digits are presented as triplets and provide a relatively closed set since there are only nine possible items, and a closed set’s lexical access is more limited, which could have improved recognition performance [16].

Thus, studies to assess the effect of the stimulus on SPIN have mainly been done on adults alone or only older adults and children. Studies that have used stimuli with various linguistic contexts are limited, and age’s effect on SPIN with different stimuli is yet to be validated in the literature. There is a need to study the effect of age on speech perception in noise ability across different stimuli. It is hypothesised that the effect of age on SPIN would vary across stimulus. It is also essential to understand whether all the stimuli decline similarly with advancing age. Thus, the present study aims to study the effects of age on speech perception in noise abilities across syllables, words, and sentences in the Kannada language.

Materials and Methods

Purposive sampling was used to recruit participants. A cross-sectional design was used in the study. The employed participants were assigned into three groups: children (group I), adults (group II), and older adults (group III). Speech perception was assessed using syllables, words, and sentences in the presence of speech babble.

Participants

A total of ninety participants were taken for the study. The children’s group (group I) included 30 participants aged 8 to 12 years (mean = 11.13 years, Standard deviation; SD = 1.27). The adult group (group II) had 30 participants aged 18 to 30 (mean = 24.5 years, SD = 1.57). Similarly, the older adult group (group III) included 30 participants aged 55 to 72 (mean = 63.5 years, SD = 3.45). Each group was ensured to have an equal number of male and female participants.

All the participants were native speakers of Kannada. An otoscopic examination revealed that they did not have any noticeable external or middle ear abnormalities. Adults and children had normal hearing sensitivity indicated by pure-tone thresholds of ≤ 15 dB HL for air conduction (0.5 kHz, 1 kHz, 2 kHz, 4 kHz, and 8 kHz) and bone conduction (0.5 kHz, 1 kHz, 2 kHz, and 4 kHz). All the participants of Groups 1 and 2 had an A-type tympanogram with reflexes at or below 100 dB HL at 0.5 kHz, 1 kHz, 2 kHz, and 4 kHz. Older adults had hearing sensitivity of ≤ 15 dBHL for air conduction and bone conduction at 0.5 kHz, 1 kHz, and 2 kHz, and less than 30 dB HL at 4 and 8 kHz. Participants had an A-type tympanogram with reflexes at or below 100 dB HL at 0.5 kHz, 1 kHz, and 2 kHz [17]. Furthermore, the participants had no reported complaints or history of any middle ear pathology, speech and language problems, or psychological and/or neurological problems. All participants had speech identification scores in the quiet of 90% or higher on the “Phonemically balanced word test in Kannada [18].

Written informed consent was obtained from participants and caregivers of children before testing. Participants/caregivers were informed about the aims and objectives of the study and the total duration required to complete the test. All the test procedures followed non-invasive techniques and adhered to the guidelines of the Ethics Approval Committee of the institute [19].

Instrumentation

A calibrated, dual-channel diagnostic audiometer (Grason-Stadler Inc. 61, Eden Prairie, USA) was utilized along with supra-aural earphones (Sennishers HDA 200) and a BC vibrator (B- 71) for assessing pure-tone thresholds. The functioning of the middle ear was determined using an immittance (Inventis Clarinet, Inventis Inc., Padova, Italy).

Stimuli

Speech perception in the presence of speech babble was assessed across three stimuli, namely: monosyllables (CV) [20], phonemically balanced Kannada words [21], and Kannada sentences [22]. Eight-talker Kannada speech babble was used as the competing stimuli [23].

Procedure

The compact disc with the recorded stimulus was played using a computer, the output of which was routed through a Grason-Stadler Inc. 61 audiometer. The participants heard the stimuli through calibrated Sennishers HDA 200 headphones. The SPIN assessment was done binaurally at 60 dB SPL across three sets of stimuli. Binaural testing was opted to reduce test time and eliminate ear effects on speech perception in children [11]. The SPIN abilities were assessed across different stimuli (CV, words, sentences) at 0 dB SNR. The speech stimulus and eight-talker Kannada speech babble [23] were mixed at 0 dB SNR using MATLAB software [24] Version R2010a (The MathWorks, Inc.) using code developed by Gnanateja (2017) [25]. Speech signals can be mixed with noise at different signal-to-noise ratios using this code, and this Matlab function mixes the speech and noise signals in terms of the RMS signal-to-noise ratios.

Perception of speech in the presence of noise was evaluated utilizing a list of twenty monosyllables taken from the “common discrimination test for Indians” test developed by Maydevi (1974). This test contains monosyllables commonly found in Kannada, Hindi, Telugu, and other Indian languages [20]. The test has six lists with 20 syllables in each list, and the material has been validated. Monosyllables and speech babble were presented at 0 dB SNR. Each incorrectly identified monosyllable was given a score of “0.“ The maximum possible score was 20.

For assessing SPIN abilities for words, words from the standardized phonemically balanced Kannada word list were utilized [21]. This material consists of 24 standardized lists in quiet and 21 in noise. For this study, List I was utilized, and the words were presented in the presence of eight-talker speech babble at 0 dB SNR. Each correctly identified word was given a score of “1,“ and an incorrectly identified word was assigned a score of “0.“ The maximum possible score was 20 for words.

The Kannada sentences for assessing SPIN abilities were taken from a standardized Kannada sentences list [22]. The test material comprises 25 homogeneous lists of Kannada sentences. Each list consists of 10 sentences and 40 keywords. For this study, we utilized two lists of sentences comprising 20 sentences and 80 keywords. Sentences were presented in the presence of eight talkers at 0 dB SNR. Scoring for sentences was based on the correct identification of keywords. Every correctly identified keyword was given a score of “1,“ and every incorrectly identified keyword was given a score of “0.“ The maximum possible score was 80 for sentences.

Participants were instructed to repeat the stimulus (monosyllables, words, and sentences) by ignoring the speech babble. An open-set response was obtained from all the participants, and the responses were audio-recorded for scoring. The average test time was approximately 15 min for assessing SPIN abilities across three sets of stimuli. The raw score was converted into a percentage for ease of comparison. Random counterbalancing of all speech material was done to ensure no learning effects across subjects. Also, test retest reliability of 10% of the participants (Nine participants) was done after 1 to 2 days of the first assessment.

Statistical Analyses

For the present study, SPSS (version 25, IBM Corp., NY, USA) was used for statistical analysis. Results obtained in the present study were not normally distributed based on the Shapiro–Wilk test (p < 0.05). As a result, non-parametric tests were done for further analyses. The Kruskal-Wallis test was administered to see the overall effect of age on SPIN abilities for different speech materials. A pair-wise comparison of SPIN abilities between age groups was conducted using the Mann-Whitney test. Friedman test was carried out to check the overall difference in SPIN score for different stimulus materials within an age group.

Results

The within-group comparisons were made to assess the difference in SPIN scores across three different stimuli (syllables, words, and sentences), and between-group comparisons were made to determine the effect of age (children, adults, and older adults).

Effect of Stimulus on SPIN Scores for Three age Groups

Table 1 shows the mean, median, range and SD of the SPIN score across different materials for each group. From Table 1, it can be noted that the SPIN scores are similar across stimuli for all three groups. Further, the Friedman test was done to see if there was any statistical difference between syllable, word, and sentence performance in each group. Results showed that there was no statistical difference in SPIN scores across syllables, words, and sentences among children (χ 2 [2] = 4.624, P = 0.099), adults (χ 2 [2] = 6.077, P = 0.59) and older adults (χ 2 [2] = 3.586, P = 0.166).

Table 1.

Mean, Median, Range, and SD of SPIN score across different materials for each group

Stimulus	Groups	Mean	Median	Range		SD
				Minimum	Maximum
Syllables	Group 1 (Children)	92	92.5	80	100	5.186
	Group 2 (Adults)	96.17	95	85	100	4.292
	Group 3 (Older adults)	85.83	85	70	100	8.103
Words	Group 1 (Children)	88.67	90	65	100	8.298
	Group 2 (Adults)	98	100	90	100	3.107
	Group 3 (Older adults)	87.67	90	70	100	7.038
Sentences	Group 1 (Children)	90.3	90.5	75	100	6.193
	Group 2 (Adults)	98.93	100	91	100	1.893
	Group 3 (Older adults)	89	90	75	99	6.187

Effect of Age on SPIN Scores for Different Stimulus

The box plots of SPIN scores across age groups for syllables, words, and sentences are shown in Fig. 1. From Fig. 1, it can be noted that the adults performed better than the other two groups for all three stimuli. Further, the Kruskal- Wallis test was done to assess the significant main effect of age on SPIN performance for different stimuli. Results indicated that there was a significant main effect of age on SPIN performance for syllables (H [2] = 29.77, p = 0.001), words (H [2] = 37.07, p = 0.001), and sentences (H [2] = 47.43, p = 0.001). Since the main effect of age was observed, the Mann-Whitney U test was performed for pair-wise comparison between the groups.

Fig. 1.

Box plots of percentage of SPIN scores between adults, children, and older adults for (a) syllables, (b) words, (c) sentences

Mann Whitney U test results revealed a statistically significant difference between the SPIN scores of children and adults for syllables (Z =-3.239, p = 0.001), words (Z =-4.923, p = 0.001), and sentences (Z =-5.553, p = 0.0001). Also, a statistical difference was noted in SPIN scores between adults and older adults for syllables (Z =-3.239, p = 0.001), words (Z =-3.239, p = 0.001), and sentences (Z =-3.239, p = 0.001). However, statistical differences between children and older adults are seen only for syllables (Z =-5.914, p = 0.0001) and not for words and sentences. Table 2 shows the pair-wise comparison of SPIN scores across age groups for different stimuli. Figure 1 (a,b &c) represent group comparisons of SPIN performance for syllables, words, and sentences.

Table 2.

Man Whitney result of pair-wise comparison for syllables, words, and sentences across age groups

Stimulus	Sentences
	Group 1 (Children)	Group (Adults)	Group3 (Older adults)
Group 1 (Children)	-	S (Z =-5.553, p = 0.000)	NS
Group 2 (Adults)		-	S (Z =-5.553, p = 0.000)
Group 3 (Older adults)			-
Stimulus	Syllables
	Group 1 (Children)	Group (Adults)	Group3 (Older adults)
Group 1 (Children)	-	S (Z=-3.239,p = 0.001),	S (Z =-5.914, p = 0.0001)
Group 2 (Adults)		-	S (Z =-3.239, p = 0.001)
Group 3 (Older adults)			-
Stimulus	Words
	Group 1 (Children)	Group (Adults)	Group3 (Older adults)
Group 1 (Children)	-	S (Z =-4.923, p = 0.001)	NS
Group 2 (Adults)		-	S (Z =-3.239, p = 0.001)
Group 3 (Older adults)			-

Note- NS- not significant, S – significant

From Fig. 2, we can observe the trend of SPIN performance of syllables, words, and sentences across age groups. SPIN scores for syllables, words, and sentences improve from a child to an adult group, while scores deteriorate from an adult to an older adult group. A similar trend is seen for all three stimulus types, although the reduction in syllable score is more noticeable in the older adult group than in words and sentences.

Fig. 2.

Representation of maturation and deterioration of SPIN score across age for syllables, words, and sentences

Further the test retest reliability of the SPIN scores for syllables, words, and sentences on 10% of the participants was assessed using an interclass correlation test. The ICC values of SPIN scores for syllables (0.866), words (0.929) and sentences (0.913) showed high to excellent reliability.

Discussion

The present study aimed to study the effect of age on SPIN abilities for different stimuli in individuals with normal hearing. SPIN abilities were assessed in children, adults, and older adults using CV monosyllables, words, and sentences. This study’s results showed a significant effect of age on SPIN abilities for different stimuli used. Children and older adults performed poorer in SPIN than adults across syllables, words, and sentences. However, except for syllables, older adults and children had similar SPIN performances in words and sentences. Older adults had poorer performance in syllables compared to words and sentences.

The present study’s outcome agrees with the study by Nittrouer & Boothroyd (1990). They evaluated SNR-50 in the presence of speech babble for phoneme recognition score for real and nonsense words, syllable recognition score for real and nonsense words, and word recognition score in sentences with and without context. The overall performance showed poorer performance in children and older adults than adults.

Poor speech perception abilities in noise in older adults could be due to peripheral and central effects. The peripheral effect refers to the reduction or degeneration of auditory abilities; the central effect is the reduction of cognitive abilities [26]. CHABA (1988) suggested three factors to explain the poor SPIN abilities in older adults: peripheral effects due to cochlear damage, central auditory effects due to damage to the auditory brainstem and cortical structures, and cognitive effects involving age-related degeneration in non-auditory areas responsible for the linguistic and cognitive process. These three factors are not mutually exclusive; all of them may be present in a given individual and interact in a complex fashion [27].

Elliott (1979) suggested that the decrease in SPIN abilities in children could be due to the masking effect of noise (i.e., peripheral effect), because of limited knowledge of linguistic cues, and interference of noise in a child’s ability to utilize linguistic cues (i.e., central effect) [28]. Also, children have underdeveloped sensory skills (like frequency selectivity) for differentiating speech signals from background noise, making it challenging to interpret words in noisy environments (i.e., bottom-up processing) [12, 29]. The inability to use semantic signals may contribute to difficulty in understanding sentences in noisy environments (i.e., top-down processing). These could be the possible reasons for the reduced SPIN performance of children and older adults than adults.

The present study also showed that speech perception in noise matures during childhood, peaking at adult age, and eventually deteriorates with increasing age. Similar results have been reported in the past [1, 8, 11, 14, 30]. In a study by Jain (2016), SPIN abilities were assessed using quick speech perception in noise tests with sentence stimuli. The peak performance was seen at 20 to 29.11 years, and SPIN performance significantly deteriorated after 40 years. The current study shows that the maturation and deterioration of SPIN for CV monosyllables, words, and sentences follow the same pattern.

The current study also analyzed a within-group comparison of SPIN performance across CV monosyllables, words, and sentences. The result indicated that all three groups, i.e., children, adults, and older adults had similar performance in CV monosyllables, words, and sentences. There was no statistical difference in performance between these stimuli. However, children performed similarly to older adults on both the words and sentences, suggesting that children and older adults accessed this information equally well. Also, children exhibited better SPIN performance with monosyllables compared to older adults. This suggests that the deterioration experienced by seniors as they age is greater for the “bottom-up” peripheral processing of the CANS, possibly due to decreased myelination or loss of central auditory nervous system neurons than the decrease in their ability to access or utilise lexical, semantic, or syntactic information in words and sentences; however, the latter is decreased compared to younger adults [29, 31]. However, none of these differences between the stimuli were statistically significant in the study.

Thus, from the present study, we can infer that age significantly affects SPIN abilities across different stimuli; thus, there is a need to develop age-appropriate norms for different stimuli. Utilizing a CV monosyllable helps to overcome the language barrier, and it will be beneficial to assess populations with limited speech and language skills or expressive delay. However, further study on larger populations is needed to validate these findings better. The present study is conducted on a normal population with 0 dB SNR. A similar study with varied SNR levels on both normal and hearing-impaired individuals is necessary to understand their ability to utilize cues for speech perception in the presence of noise.

Conclusion

It can be concluded that there is an effect of age on SPIN abilities across different stimuli. Children and older adults’ SPIN abilities were poorer than adults for all types of stimuli. It highlights the importance of age-appropriate SPIN normative for various stimuli.

Abbreviations

SPIN

Speech Perception in Noise

SNR-50

Signal to Noise ratio where a 50% score is achieved

SNR

Signal-to-Noise Ratio

CV

Consonant Vowel

SD

Standard Deviation

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declarations

Disclosure statement

No potential conflict of interest was reported by the author(s).

Conflict of Interest

Authors declare no conflict of interest.

Informed Consent

Written informed consent was taken from the participants/guardians of all the participants for their willingness to participate in the study.