---

This paper studies the phonetic aspects of vowel reduction in Persian. Vowel reduction was examined in unstressed syllables using duration and F1 and F2 spectral values across different consonantal contexts. Results showed a strong effect of stress on duration across all consonantal contexts, but the results for F1 and F2 frequencies varied with vowel type as well as consonantal context. Most vowels had non-significant formant changes in unstressed condition as compared to stressed condition; however, the vowel /a/ had significant spatial reduction in the direction of the center of the vowel space. The results support neither of the two known articulatory hypotheses concerning vowel reduction, namely, centralization (in unstressed position) by target undershoot, and reduction due to co-articulation with the consonant context. Instead, the results lend support to the dispersion theory of speech perception which argues that vowel systems cross languages tend to utilize the available acoustic vowel space so that maximal auditory contrast is maintained. It is further argued that the position of vowels in the acoustic vowel space is influenced by this dynamical repelling force which has come to be called adaptive dispersion.   

---

Down syndrome (hereafter DS) is a chromosomal condition caused by the presence of all or part of a third copy of chromosome 21. People with DS may have a variety of intellectual disabilities, a characteristic facial appearance and weak muscle tone. In addition, a global delay that affects their motor development, cognition, communication and language is often reported. This study aims to investigate and compare vowel spaces of healthy and DS children. This research is quantitative and to investigate vowel spaces of participants, twelve subjects (6 healthy children with the average age of 8 and 6 DS children with the average age of 10) attended the research project. The data were 36 monosyllabic (CVC) words and for each of them, the codas and onsets were occupied by similar voiced and voiceless Persian stops. The acoustic data were recorded in a soundproof chamber and the subjects’ production was recorded using SHURE® microphone (SM81 Condenser) placed at the mouth level. Each speech sample was segmented and analyzed acoustically using Praat® speech analysis software (Version 5.2.24). Making a text grid, it was possible to analyze both sound waves and spectrograms simultaneously, securing reliable analyses. Therefore, using PRAAT script, first formant frequency (F1) and second formant frequency (F2) of all the Persian vowels were measured for all the participants. In the next step, repeated measure ANOVA and SPSS (version 17) were performed to investigate differences between vowels of all subjects. The level of significance was α: 0.05. Thus, the confidence interval would be 95%. If the p-value of the repeated measure ANOVA was less than 0.05, post-hoc Bonferroni test would be employed to analyze the dependent variables. Statistical information on mean F1 of vowels shows that F1 of [ɑ, æ, o] in healthy children is more than that in children with DS. This result indicates that these vowels are articulated in a higher position in subjects with DS. From the other hand, mean F1 of [u, i, e] in healthy children is more than that of children with DS. It can be concluded that in children with DS, these vowels are articulated in a lower position. Analyzing statistical information on mean F2 of vowels in subjects with DS indicates that F2 of [æ, i, e] is less than those in healthy children. In other words, these vowels are articulated in a more back position in these participants. Also, analyzing F2 of [ɑ, u, o] in DS children shows that these vowels are articulated in a more front position. Another part of this research is dedicated to analytic statistics of vowels. It was hypothesized that the effect of DS on F1 and F2 of vowels is significant. This hypothesis is partially supported by the findings. The results indicate that there is no significant difference between F1 of all Persian vowels [æ, e, o, ɑ, u, i] in the two groups of subjects. Nevertheless, it is worth noting that the difference between F2 of all Persian vowels except [ɑ] is significant between healthy and DS children. All in all, these results support the fact that formant values are more centralized for children with DS, which means a reduced vowel space area. Consequently, it can be concluded that, in DS children, structural features of oral cavity, like smaller oral orifice, narrower but not higher palatal curve and shorter vocal cords have an effect on formant structure of vowels.
 

---

We live in a world of language and we communicate with each other using language (Fromkin et al., 2007). Deafness occurs when a person cannot understand speech through hearing, even when the sound is amplified. The degree of hearing loss is measured according to the sensitivity of deaf people’s ears to the sounds (Ganji, 2017) and then those people are asked to use hearing aids or do cochlear implantation surgery (Ebrahimi, 2006). Children have the opportunity to hear others’ speech during normal speech development while children with hearing impairment lose much of the auditory input, which may hinder speech development (Tseng et al., 2011 as cited in Serry and Blamey, 1999; Blamey et al., 2001; Peng et al., 2004; Bouchard et al., 2007). Different factors will affect speech development. One of these factors is DISABILITY, any disorder in hearing will affect speech and subsequently will cause problems in communication (Amiri et al., 2014). Some speech characteristics of children with hearing loss are different from each other and different from normally hearing children (Kord et al., 2012). GENDER is another factor affecting speech. While analyzing vowels produced by Azari speakers, Mirahadi et al. (2018) proved that men produce some vowels different from women.
Vowel space is a vowel quadrilateral which is used to shows the first and the second formants. Phoneticians put vowels in the vowel space according to their position in the oral cavity, which was first introduced by Essner (1947) and Joos (1948) (as cited in Harrington & Cassidy, 1999). The first formant will increase as we go down the vowel space, it shows highness of tongue while producing vowels in the oral cavity. The second formant shows anterior or posterior position of tongue in the process of production of vowels in the oral cavity and will increase as we go to the left side of the vowel space (Hayward, 2013: 227, 228 & 502).
The present study aims to compare the vowel space of hearing aided, cochlear implanted, and normally hearing children, with DISABILITY and GENDER as factors, in order to investigate the difference between the vowels produced by these children (Their average age was 10 years and 4 months and their participation in the study was optional, after getting permission from their parents.) and find out more effective hearing aid tools among two groups of children with DISABILITY (They did not have any mental or physical problems except hearing impairment and took part in language therapy classes for about 450 hours.), the group which makes vowels similar to normally hearing children will prove the effectivity of the hearing aid tools. So, the results will be useful for language therapists while helping children learn how to pronounce vowels.
Therefore, the participants, 5 girls and 5 boys in each group, were asked to repeat thirty-six CVC framed words. Each word consists of one of the stops [p, b, t, d, k, g] in the onset and offset of the pattern and one of the Persian vowels [æ, e, o, a, i, u] in the center. Then the words were recorded using a Shure microphone in PRAAT software. After that, the first and the second formants of each vowel were measured and by using SPSS software the data was analyzed.
The results showed that the effect of DISABILITY on the first formant of vowels [æ, a, i, u] and the second formant of vowels [a, e, o, u] was significantly different (p < 0.050). The results of a Post Hoc Bonferroni test showed that children with DISABILITY produced vowels [u, i, e, ᴂ] in a lower place in the vowel space comparing to normally hearings. Cochlear implanted children produced vowel [a] in a lower place and vowel [o] in a higher place in the vowel space comparing to normally hearings. Hearing aided children produced vowel [a] in a higher place in the vowel space comparing to normally hearings. All vowels in cochlear implanted children and vowels [ᴂ, a, e, i, u] in hearing aided children were produced in an anterior place in the vowel space comparing to normally hearings.
The results also proved that the effect of GENDER on the first formant of vowel [a] and the second formant of vowel [i] was significantly different (p < 0.050). The results of a Post Hoc Bonferroni test revealed that the mean of the first formant of vowel [a] in girls was more than boys and the mean of the second formant of vowel [i] in girls was less than boys.
It is also important to mention that hearing aided children had the smallest vowel space among three groups of participants while the vowel space of cochlear implanted children was not so different from the vowel space of normally hearings. So, language therapists have to try more to teach vowels to the children who use hearing aids comparing to the children who are cochlear implanted.

---

One of the dysarthria features is inaccurate articulation of consonants and vowels, which is the main cause of speech unintelligibility. However, it is not clear what acoustic measures are more related to speech intelligibility in dysarthria. On the other hand, the number of vowels, their articulation and acoustic characteristics are various in different languages. As far as we know, no study investigated vowels articulation in developmental dysarthria in Persian, so there is not enough information about movement details which make their inaccuracies articulation. Hence, the present study aimed to investigate some acoustic measures including vowel formants, vowel space area (VSA), formant centralization ratio (FCR) and vowel articulation index (VAI) and aim to study their correlation with speech intelligibility in Persian speaking children with dysarthria due to spastic cerebral palsy. In this cross ـ sectional study, 11 monolingual children with dysarthria aged 3 ـ 9 years and their typically developing matched peers were selected through convenience sampling. For each of the six Persian vowels, two monosyllable words were acoustically analyzed using PRAAT software. To determine intelligibility of words and vowels, the percentage of correct detected ones was calculated. Results showed a significant difference between the two groups in the third formant of the vowels /a/ and /u/ and the second formant of the vowel /o/. The vowel space area and intelligibility were significantly reduced in dysarthria. VAS was more correlated with intelligibility and is a good acoustic measurement to assess speech intelligibility in dysarthria. However all three studied acoustic indices had moderate significant correlation with intelligibility.

1. Introduction
Abnormalities in power, speed, range, muscle tone, stability, and accuracy of movements necessary for respiration, phonation, production, articulation, or prosodic aspects of speech lead to dysarthria (Duffy, 2019). Depending on the location of the damage in the nervous system, there are different types of dysarthria and each creates different speech characteristics (Duffy, 2019). Cerebral palsy is one of the main causes of dysarthria in children. One of the main and distinctive features of dysarthria is imprecise production, which is the main factor for low speech intelligibility (De Bodt et al., 2002). However, there is limited information on the motor features that cause this inaccuracy in children’s production, and awareness of these motor events is a prerequisite for providing appropriate therapeutic approaches (Allison et al., 2017). If we consider syllables as speech motor units (Yunusova et al., 2008) and vowels as the core components of syllables, vowel distortion can affect the structure of syllables and consequently speech. Describing the vowels of a language has more auditory aspects than articulatory ones. However, gradually linguistics realized that if the vowels of a language are displayed based on F1 and F2 features, the result will be a two-dimensional space that they named vowel space. Since speech intelligibility is the primary goal of clinical interventions, and since vowels are a good predictor of speech intelligibility in dysarthria, it is necessary to accurately determine the correctness and accuracy of vowels with the help of appropriate diagnostic or perceptual tools to assess the effects of direct and indirect therapies on the proper production of vowels and the improvement of speech intelligibility. This diagnosis guides therapists in selecting appropriate treatment methods or continuing therapy.
Research Question(s)
Is the formant of each of the 6 vowels in Persian-speaking children with spastic dysarthria different from that of typically developing children? Are speech intelligibility, acoustic parameters of vowel space, vowel production indices, and formant centralization ratio different in children with dysarthria based on age? Is there a correlation between some acoustic parameters of vowels and speech intelligibility?


2. Literature Review
In 2019, a study was conducted in China on the Mandarin language with the aim of determining acoustic parameters for speech evaluation in children with dysarthria due to cerebral palsy. Thirty children aged 5 to 15 with cerebral palsy and 13 typically developing children participated in the study. Formants F1 and F2 were calculated for vowels /a, i, u/. The VSA, VAI, and FCR indices were also calculated and analyzed. In children with cerebral palsy, there were significant differences in F1 and F2 for the vowel /a/ and F2 for the vowel /i/, while there were no significant differences in F1 and F2 for the vowel /u/ and F1 for the vowel /i/. There were significant differences between the two groups in the FCR, VAI, and VSA indices (Mou et al., 2019). In another study in 2017 aimed at improving the understanding of speech characteristics related to dysarthria in English-speaking children with cerebral palsy in the United States, formants were analyzed at the word and sentence levels. The speech of 10 children with dysarthria due to cerebral palsy was compared to 10 typically developing children. The results showed that children with dysarthria had a smaller vowel space compared to typically developing children at both the word and sentence levels (Allison et al., 2017). The same researchers conducted another study in 2018 with the aim of determining acoustic characteristics of connected speech in children with dysarthria due to cerebral palsy and typically developing children. Twenty children with dysarthria were matched with typically developing children in terms of age and gender. The results showed that children with dysarthria differ from typically developing children in terms of speech-segmental and suprasegmental features (Allison & Hustad, 2018). In 2016, a study titled “Vowel Intelligibility in Children with or without Dysarthria” was conducted. Researchers in this exploratory study examined the vowel intelligibility of American English vowels in American children. Generally, children with dysarthria had lower vowel intelligibility compared to typically developing children, but there was no statistically significant difference in intelligibility among different vowels. These studies highlight the significance of focusing on the vowel system during therapy to enhance intelligibility (Levy et al., 2016).

3. Methodology
In the present cross-sectional study, 11 children with spastic dysarthria due to cerebral palsy were selected from rehabilitation clinics in Semnan city, and 11 typically developing age- and gender-matched children from preschools and primary schools in the same city. The inclusion criteria for children with cerebral palsy in the study were mild to moderate dysarthria due to spastic cerebral palsy, aged 3 to 9 years, and the ability to produce single words. Exclusion criteria were the presence of sensory or motor problems, bilingualism, dysphagia, and difficulties in producing consonants present in test words. Initially, 12 monosyllabic words with a CVC structure including 2 words for each of the 6 vowels in the Persian language were selected by a linguist and a speech-language pathologist, which were simple to produce in terms of consonants and did not affect the central vowel. Each word was uttered twice, so 4 words were spoken for each vowel, with 3 of them selected for formant analysis using PRAAT software. Subsequently, the spoken words by each child were presented to adults unfamiliar with the child, who were free of hearing or language issues, to write down the words after hearing them. The researchers determined the number of correct words and the vowels identified by adults to establish the percentage of vowel and speech intelligibility for each child. Finally, the two groups of children were compared in terms of vowel space and speech intelligibility.

4. Results
In this study, each group of children with normal speech and dysarthria included 4 girls and 7 boys with a minimum age of 3.5 and a maximum age of 9 years. The mean age in children with dysarthria was 1.9 ± 5.6 years and in the group with normal speech was 1.8 ± 5.7 years, with no significant difference between them (p > 0.05). Since results in speech production change with age due to vocal tract lengthening and children’s motor skill development, the children in this study were categorized into 2 subgroups of 3.6 to 6 years and 6 to 9 years. Comparison of vowels in children with dysarthria and their age-matched typically developing peers using the Mann-Whitney test showed a significant difference between the two groups only in 6 to 9-year-old children in the first formant of /i/ and /æ/ vowels, the second formant of /æ/ vowel, and also the third formant of /e/ and /a/ vowels. Speech intelligibility at the single-word and vowel level, acoustic parameters of vowel space, vowel production indices, and formant centralization ratio were compared between the two groups of children with dysarthria at different ages using the Mann-Whitney test, which showed a significant difference in word and vowel intelligibility between the two groups. All acoustic indices also showed a significant difference between the two groups. Vowel space, calculated based on the /a/, /i/, and /u/ side vowels in this study, was smaller in the group of children with dysarthria compared to the normal group. In both age groups, the dispersion of vowel production in the dysarthric group was higher compared to the normal children. The results of the Spearman correlation coefficient test showed a significant moderate relationship between all indices of vowels and speech intelligibility. This relationship was direct for the Vowel Space Area (VSA) and Vowel Articulatory Index (VAI) and inverse for the Formant Centralization Ratio (FCR).