You are on page 1of 12

FREQUENCY OF OCCURRENCE OF PHONEMES IN MALAYALAM N.

Sreedevi1 & Irfana, M 2 Abstract Malayalam is a Dravidian language primarily spoken in the southwest of India. The frequency of occurrence of phoneme data in a particular language is essential to understand the language structure and also has wide applications particularly in the area of rehabilitation of communication disorders such as audiology and speech language pathology. The frequency of phonemes obtained will aid audiologists and speech language pathologists in developing and updating the existing test material for evaluating various communication disorders and also for selection of treatment targets in such population. The study has implications in the area of linguistics, speech synthesis tasks and automatic speech recognition systems also. Phoneme frequency in written material is available in a few Indian languages. However details on spoken language are preferred for most of the applications. Hence the purpose of the present study was to obtain the frequency of occurrence of various phonemes in Malayalam using conversation samples. Participants included fluent native speakers of Calicut dialect of Malayalam in the age range of 30 to 55 years. Seven separate conversation samples were recorded for 25-30 minutes. The samples obtained were transcribed using IPA transcription. Inter judge and intra judge reliability of phonetic transcription was evaluated for 10% of the recorded samples. Further it was analyzed using the Systematic Analysis of Language Transcripts (SALT) software for obtaining frequency of phonemes. Mean and standard deviation of frequency of phonemes of all the seven samples were obtained. The results show that /a/ was the most frequently occurring vowel and /k/ was the most frequently occurring consonant in Calicut dialect of Malayalam. The other most frequently occurring vowels were /I, , a:, e/ and consonants were /n , , t, l/. Overall, vowels constituted 42.54% and consonants 57.26% of the conversational data. The results obtained will be made available to communication professionals to develop and update several test materials in Malayalam for evaluation and speech and language therapy purposes. Key words: Frequency of phonemes, Dravidian languages, Malayalam, SALT software, Place of articulation, Manner of articulation, Vowels, Consonants. 1. Reader in Speech science, AIISH, Mysore, E mail: srij_01@yahoo.co.in; 2. JRF, AIISH, Mysore, . E mail: fanairfana@gmail.com. In a language, a phoneme is the functional unit of speech that is used to distinguish meaning of a word. Different languages have different phonological systems and the same sound or the same groups of sounds do not necessarily have the same function in one language as in another. Although the dialects are the variation of a single language, it differs in terms of vocabulary, grammar, pragmatics and pronunciation (Siegel, 2010). The frequency can vary from language to language and dialect to dialect. There are about 600 consonants used in different languages around the world and some occur more frequently than others. The most common are the voiceless stops and about 98 percent of the worlds languages have the three voiceless stops /p/, /t/, /k/. Like consonants, an account of the number of vowels across the world may be difficult as one vowel glide into another easily. The

well known forms of English have 14 vowels (Californian English) to 20 vowels (BBC English). Ladefoged (2000) report /a/ to be the most common vowel followed by /e/, /i/, /o/, /u/ in English. The frequency of occurrence of phonemes is in research since 1930s. As the phonological structure varies with the language use and the dialects, several studies were carried out in different languages. Up to 1970s, the studies of phonemes were mainly on the written corpus where the frequencies were mostly calculated from sources like newspapers, journals and scripts of plays. Mines, Hanson and Shoup (1978) used conversation in an interview and database of about 1, 03,887 phoneme occurrences were obtained in English. The frequency of occurrences of phonemes were listed in descending order as / a, n, t, i, s, r, l, d, /, which accounted for 47% of the total data. The three nasals /m/, /n/ and // account for 18.45% of all consonants produced in initial, medial and final positions. /l/ and /r/ were occurring with nearly the same frequency in American English just over 6% of all consonant sounds in adult speech. Stops showed 29.21% of occurrence. Delattre (as cited in Edwards, 2003) considered 2000 syllables in English from different sources like narration and dramatization and found that the most occurring vowels were //, /I/, // and the consonants were /t/, /n/, /r/ and /l/. There exist observable differences in frequency of occurrence of phonemes between written and spoken data in the same language. Sandoval, Toledano, de la Torre, Garrote and Guirao (2008) compared the syllabic and phonemic frequency in spoken and written context in Castilian Spanish and reported that /s/ occurred for 8% and 7% in spoken and written contexts respectively. On similar lines, initial studies on Indian

languages are by Bhagwat (1961) where he calculated the phonemic and morphemic frequencies in Marathi. Ghatage and Madhav (1964) calculated the phonemic and morphemic frequencies in Hindi using written source of materials. Results showed that vowels occurred more frequently than consonants. Followed by Hindi, Ghatage and Madhav (1994) studied phonemic and morphemic frequencies in Malayalam using 1, 00,000 words from various written materials. The results indicated /a/ and /I/ were the most frequent vowels. Among consonants, palatal nasal / / was most occurring followed by /k/ and /m/ respectively. Kelkar (1994) studied phonemic and morphemic frequencies in Oriya and indicated that // was the most occurring vowel followed by /a/ and /I/. Among consonants /r/, /k/ and /t/ were the most found phonemes in origin. Frequency of occurrence of phonemes in Kannada was first studied by Ramakrishna (1962) in written form. The results revealed that long vowels and aspirated phonemes are used relatively less frequently and vowel /a/ is the highest occurring vowel and consonants like /r/, dentals / d/ and /t/ are the highly used consonants in Kannada. Jayaram (1985) also studied phonemes in Kannada from news paper sources and listed out a series of vowels and consonants by rank order of their frequency of occurrence and the order was /a, I, n, r, u, d, e, t, l/. Recently, Sreedevi, Smitha and Vikas (2012) found that in conversational or spoken Kannada, vowel /a/ was the most frequently occurring phoneme followed by /n/, /I/. /e/, /r/, /a:/, /d/, /l/, /u/, /g/ and /k/. Phonemes /h/, /s/, /p/, //, //, // were less frequent and vowels constituted 44.3% and consonants 55.3% of the conversaional data in Kannada.

In Telugu, Kalyani and Sunitha (2009) found that consonants (51.21%) were predominant in spoken data than vowels (44.98%). Open vowels were higher in frequency and among consonants, alveolars showed significant occurrence followed by bilabials and velars. Kumar and Mohanty (2012) also found similar results and reported that aspirated stops were not significant in Telugu spoken mode of communication. There is a wide application using the statistics of phoneme occurrence in a language. These phoneme frequency is data of use in automatic generation of speech for reading machine for the blind. For the same reason the knowledge about the frequency of phoneme provides a database for phoneme identification process of an automatic speech recognition system. The frequency of occurrence of phonemes plays a crucial role in the development of linguistic theories in a number of areas including the grammatical relations, semantic structure etc. Audiologists use several speech materials in the form of word lists, involving phonetically balanced phonemes in a language for hearing assessment and intervention. As these test materials are language specific, they require most frequently occurring phonemes in that particular language (Egan, 1948; Campbell, 1965). For speech language pathologists, these
studies would provide a database for developing speech materials for assessment and selecting treatment targets for various communication disorders, knowledge about the most frequent phonemes aids in targeting those phonemes in speech therapy for the hearing impaired which leads to better intelligibility of their speech.

and there are different dialects under each language system. A database of different dialects with its phoneme frequency will help us to know a language better. The present study investigates the frequency of occurrence of phonemes in Calicut dialect of Malayalam, a Dravidian language spoken in the state of Kerala in India. The earlier study on phoneme frequency in Malayalam by Ghatage and Madhav (1994) was from various written materials. Also their findings may not be apt at present as there are a lot of new words, modified and borrowed words used in day to day conversation. Hence the present study is planned to obtain the frequency of occurrence of phonemes in conversational speech samples in Malayalam. Method Participants: Fluent adult native speakers of Calicut dialect of Malayalam in the age range of 30 to 55 years with a minimum of 10 to 12 years of schooling in Malayalam medium of instruction were selected for the study. The participants did not have any clinical background of speech, language, hearing or any neurological problems. Instrumentation: A digital recorder (Olympus WS 100) was used for recording the conversation samples. Procedure: The data was collected through conversations in controlled natural environments for about 25 to 30 minutes of duration. There were seven conversational groups including 4 to 5 participants in one group. The digital recorder was kept at equidistance from all the speakers. Participants themselves initiated the conversation without any specific topic being assigned to them. The participants were instructed to avoid words from other languages and to speak naturally in Malayalam only. They were not restricted

The phoneme statistics are also applicable to the fields of linguistics and speech, to teach the language as a foreign language. India is a country with diverse languages. And these languages have different phoneme system

from using commonly used loan English words (E.g.: Bus, ticket, phone, car etc). Each of the seven recording sessions involved different participants. Data Analysis: The conversation samples were transcribed using International Phonetic Alphabet for Regional languages (Malayalam) by Asher and Kumari (1997) which is provided in Appendix 1. The transcribed data was analyzed using the software SALT RV version 9 for the frequency count. A database of Malayalam phonemes was prepared and saved in the editable standard wordlists of SALT software. The SALT software compared the database and the loaded phoneme file and provided the phoneme count.
1. Malayalam is a Dravidian language primarily spoken in the southwest of India. According to Summer (2009), it is the official language of Kerala state and Lakshadweep union territory. Within India alone there were over 35 million speakers of Malayalam, not including the other nearly 500,000 speakers outside India. Malayalam has 11 monothongs and 2 diphthongs and 52 consonant phonemes, encompassing 9 places of articulation which are bilabial, labiodentals, dental, alveolar, alveolopalatal, retroflex, palatal, velar and glottal and 8 manners of articulation which include plosive, nasal, trill, tap/flap, fricative, affricate, central approximant and lateral approximant (Jian, 2010). Based on reports of Asher and Kumari (1997), places of articulations in Malayalam are labials, dentals, alveolars, retroflex, palatals, velars, glottal and manner of articulation include stops, nasals, fricatives, taps/trills, lateral and glides. 2. Calicut is a northern district of Kerala where more of Muslim community is accommodated. The dialect of Malayalam spoken by mappila Muslims shows deep influence of Arabic language. And the dialect is partially different from mainstream Malayalam.

Inte r judge and intra judge reliability 10% sample of each conversation recording was subjected to inter judge and intra judge reliability measures. Three graduate speech language pathologists served as judges for inter judge reliability measures. For intra judge reliability, the 10% of each conversational recording were transcribed and analyzed by one of the authors. Cronbach alpha test was applied and a reliability index () of 0.89 was obtained for inter judge and 0.92 was obtained for intra judge reliability. Results The aim of the present study was to obtain the frequency of occurrence of phonemes in conversational speech in Calicut dialect of Malayalam. Following the SALT analysis, as per Fig 1, maximum number of phonemes were recorded in the sixth session (R6), which amounted to 14,358 phonemes and minimum phonemes were recorded from fifth (R5-10,199 phonemes). Each recording session elicited more than ten thousand phonemes and the grand total was 83,561 phonemes from the 7 recordings.

3. SALT, the Systematic Analysis of Language Transcripts, is a computer program designed to help in analyzing and interpreting language samples from one or more speakers during a communicative interaction. It can be used to analyze samples from everyday speech like conversation and narration. The SALT program provides clinicians and researchers with the means to transcribe language samples into a common format and to compute a series of general analyses of lexical, syntactic, semantic, and pragmatic, rate, fluency, and error categories.

Number of phonemes in recording sessions


13527 12453 12893 14358 10199 10324

10662

The mean percentage of occurrence of each phoneme in the corpus is depicted in Table 1. Vowel /a/ occurred most frequently in all the recordings. The mean percentage of occurrence of /a/ was more than 10% in all seven groups. Some vowels showed highly reduced frequency of occurrence. Such as long vowels and diphthongs.

Table 1:
R1 R2 R3 R4 R5 R6 R7

Figure 1. Number of phonemes obtained in each recording session

Mean percentage and standard deviation of occurrence of vowels and consonants in spoken Malayalam Vowels /a/ /a:/ /i/ /i:/ U /u:/ /e/ Mean % (SD) 13.83 (1.3) 4.31 (1.74) 6.07 (0.24) 0.71 (0.17) 3.46 (0.6) 0.64 (0.2) 3.75 (0.45) 0.88 (0.42) 0.15 (0.07) 1.23 (0.49) 1.67 (0.39) 0.027 (0.02) 5.58 (1.24) Consonants /k/
h

The transcribed data consisted a total of 83, 561 phonemes including both consonants and vowels. Based on the descriptive statistical analysis the mean percentage of occurrence of consonants was more than vowels. Mean occurrences of consonants were 57.26% and vowels were 42.54% of the total corpus as shown in Fig 2. Occurrence of diphthongs was fewer in all the seven recording sessions.
70

/k/ /g/
h

/g/ /c/
h

/c/ // // //
h

60
50 40 42.54%

57.26%

/e:/ /ai/ /o/ /o:/ /au/


Vowels Consonants

30
20 10

// // // /r/

// Figure 2. Mean percentage of occurrence of consonants and vowels.

Mean % (SD) 7.06 (0.73) 0.02 (0.02) 0.21 (0.08) 0.01 (0.01) 1.59 (0.35) 0.002 (0.00) 0.19 (0.11) 0 (0.00) 4.44 (0.5) 0.10 (0.10) 0.09 (0.06) 0.007 (0.00) 2.01 (0.20)

Consonants /t / /t / /d / /d / /p/ /p/ /b/ /b/ // // // /n/ /rrr/

Mean % (SD) 4.20 (0.44) 0.12 (0.06) 0.37 (0.20) 0.38 (0.18) 3.24 (0.39) 0.024 (0.01) 0.46 (0.10) 0.17 (0.08) 1.48 (0.32) 0.78 (0.31) 2.49 (0.2) 5.32 (1.07) 0.53 (0.13)

Consonant s /n / /m/ /j/ // /l/ /v/ /s/ // // /h/ // // /f/

Mean % (SD) 1.53 (0.7) 3.73 (0.5) 2.68 (0.31) 2.65 (0.34) 4.04 (0.39) 2.19 (0.14) 1.35 (0.21) 0.30 (0.06) 0.44 (0.14) 0.15 (0.05) 1.89 (0.25) 0.24 (0.12) 0.035 (0.06)

Mean percentage of occurrence


30 20 28.34%

10
0

8.21%

Long vowel

Short vowel

Figure 3. Mean percentage of occurrence of long and short vowels. Considering consonants, all the places of articulation except glottal showed significant occurrence in the conversational corpus where as voiceless velar /k/ showed highest percentage of occurrence which was followed by alveolar nasal /n/ and retroflex // respectively. Glottal /h/ occurred least frequently compared to all other places of articulation. Dentals showed predominant occurrence followed by labials, retroflex and velars. Figure 4 depicts the mean percentage of occurrence of consonants based on places of articulation.

Occurrences of short vowels were considerably more than long vowels. Open vowel /a/ (13.83%) was highly predominant than closed front vowels (6.78%) and closed back vowels (4.00%). Figure 3 shows the mean percentage of occurrence of long and short vowels.

Frequency of occurrence of consonantsplaces of articulation


15.3

9.66

7.64

9.31

5.42

8.78
0.15

Figure 4. Mean percentage of occurrence of consonants based on places of articulation.

Based on manner of articulation, stops showed predominant occurrence followed by nasals, laterals, glides and trills. Fricatives were the lowest occurring manner of articulation. Figure 5 shows the mean percentage of occurrence of consonants based on manner of articulation.

mere 1.32% in the conversational corpus. Voiced aspirated stops (0.58%) showed relatively higher occurrence than unvoiced aspirated stops (0.3%). Figure 4 indicates the mean percentage of occurrence of stops in the total corpus obtained.
Mean percentage of occurrence of stops 25 20.53

Percentage of occurrence of consonantsManner of articulation


22.68% 13.80%
2.27%

20 15 10 5 0 Unvoiced

1.32

0.301

0.567

Voiced

Unvoiced Voiced aspirated aspirated

4.66%

5.30%

5.11%

Figure 6. Mean percentage of occurrence of stops. All nasal phonemes showed their presence in the corpus with significant percentage of occurrence. Alveolar /n/ occurred for 5.32% followed by bilabial /m/ with 3.73% of occurrence. Palatal // was less found than other nasal sounds. Malayalam is known to have the maximum places of articulation among nasals and Calicut dialect showed all these six places of nasals. Among fricatives /s, , , h/, unvoiced alveolar fricative /s/ occurred for 1.35% of the data. Glottal fricative /h/ occurred for 0.15% of the total corpus and it was least seen among fricatives. Liquids including both tap/trill and laterals occurred in the conversational corpus with considerable percentage of occurrence. Lateral retroflex / / occurred for 1.89% and lateral alveolar /l/ occurred for 4.04%. Tap /r/ and trill / / had similar percentage of occurrences which was 2.01% and 2.65% respectively. In the category of approximants, // showed 0.24% of occurrence. Labial /v/ and palatal /j/

Figure 5. Mean percentage of occurrence of consonants based on manner of articulation. Among stops, unvoiced stops were significantly higher in occurrence compared to their voiced counterparts. Occurrence of unvoiced velar consonant /k/ was highest and occurred for 7.06% of the total phonemes. Occurrence of unvoiced retroflex stop // was 4.44% followed by unvoiced dental stop / t /. Unvoiced palatal affricate /c/ was comparatively reduced in the sample. (Affricates are classified under stops according to Asher & Kumari, 1997). Respective voiced counterparts of the above mentioned consonants were highly reduced in all the seven recordings. Occurrence of aspirated phonemes were also significantly reduced compared to their unaspirated counterparts. Unvoiced stops accounted for 20.53% and voiced stops occurred for only a

occurred more than 2% of the total phoneme occurrences. As per the results, the mean percentages of frequency of occurrence of the ten prominent phonemes are shown in Figure 5. The most frequently occurring vowels were /a, I, , a:, e/ and consonants were /k, n , , t, l/. As discussed above, low-central vowel /a/ occurred more frequently followed by velar stop /k/. Velar /k/ was produced for 7.06% of the total corpus of all seven recordings. Followed by /a/ and /k/, highfront vowel /i/, mid- central vowel // and dental nasal / n / showed prominent occurrence. Appendix 2 presents the occurrence of all phonemes in descending order.
16 13.83% 14 12 10 7.06% 8 6.07% 5.58% 5.32% 4.44% 6 4.31% 4.2% 4.04% 3.75% 4 2 0 a k i n a: t l e

predominant and is concomitant with several earlier studies (Ghatage & Madhav, 1964 in Malayalam; Ramakrishna, 1962; Jayaram, 1985; Sreedevi et al, 2012 in Kannada, Mines et al, 1978; Sandoval et al, 2008; Ladefoged, 2000 in English). Hence predominance of vowel /a/ was seen across languages. In the present study, among various manners of articulation, stops occurred the highest which is in concurrence with several earlier studies in English (Mines et al, 1978) and Indian languages (Jayaram, 1985; Sreedevi et al, 2012; Kalyani & Sunitha, 2009; Kumar & Mahanty, 2012). Considering the place, present study evidenced high occurrence of voiceless velar /k/. However, other Indian studies have reported different places of articulation, alveolars in Telugu (Kalyani & Sunitha, 2009; Kumar & Mahanty, 2012) and dentals in Kannada (Jayaram, 1985; Sreedevi et al, 2012). The present study observed significantly high occurrence of unvoiced stops compared to their voiced counterparts. There are no earlier reports on such disparity between unvoiced and voiced stops. Ghatage and Madhav have not commented on the voiced/unvoiced distinctions in written sources of Malayalam. The aspirated sounds were negligibly present, which is in agreement with all the other Indian language studies. This is possibly because the aspiration feature is rarely used in colloquial conversations. Conclusions The frequency of occurrence of different phonemes in Calicut dialect of Malayalam was determined using conversational samples collected from seven different recording sessions and the results showed that consonants (57.26%) constituted larger part of the total corpus than vowels (42.54%). The most frequently occurring

Figure 5. Mean percentage of frequency of occurrence of phonemes in descending order. Discussion This study aimed to investigate the frequency of occurrence of phonemes in the Calicut dialect of Malayalam. Results showed some salient findings. Based on descriptive statistical analysis, the mean frequency of occurrence of consonants were predominant than vowels. Some of the studies in Kannada (Jayaram, 1985; Sreedevi et al, 2012) and Telugu (Kalyani & Sunitha, 2009) also showed similar findings. Within the vowel category, /a/ was highly

vowels are /a, I, , a:, e/ and consonants are /k, n , , t, l/. Aspirated sounds and long vowels were rarely present in the data. The findings of the present study are applicable for both diagnostic and therapeutic aspects of communication disordered population and also for use in automatic generation of speech reading machines for the blind. It is also applicable in the field of linguistics and will serve as a data base for computerized speaker identification softwares. References Asher, R. E., and Kumari, T. C. (1997). Malayalam (pp. 405-435). London: Routledge. Bhagwat, S. V. (1961). Phonemic frequencies in Marathi and their relation to devising a speed-script. Pune: Deccan College Campbell, R. A. (1965). Discrimination and word test difficulty. Journal of Speech and Hearing Research, 8, 1322. Edwards, H. T. (2003). Applied Phonetics: The sounds of American English (3rd edition). Canada: Delmar Learning. Egan, J. P. (1948). Articulation testing methods. Laryngoscope, 58, 955 991. Ghatage, and Madhav, A. (1964). Phonemic and Morphemic frequencies in Hindi. Poona: Deccan College Postgraduate and Research Institute. Ghatage, A. M. (1994). Phonemic and morphemic frequencies in Malayalam . Mysore: Central Institute of Indian Languages. Jayaram, M. (1985). Sound and Syllable distribution in written Kannada and

their application to Speech and Hearing. Journal of All India Institute of Speech and Hearing, 16, 19-30. Jiang, H. (2010). Malayalam: a Grammatical Sketch and a Text. Rice University. Online version: http://www.owlnet.rice.edu. KalyanI, N and Sunitha, K.V.N ( 2009). Syllable analysis to build a dictation system in Telugu Language. International Journal of Computer Science and Information Security, Vol. 6. Kelkar, A. R. (1994). Phonemic and Morphophonemic frequency in Oriya. Mysore: CIIL. Kumar, R. S. B., and Mohanty, P (2012). Speech recognition performance of adults: a proposal for a battery for Telugu. Theory and practice in language studies, vol. 2, pp: 193204. Ladefoged, P. A. (2000). Vowels and Consonants (2nd ed.).Oxford, UK: Blackwell. Mines, M. A., Hanson, B. F., & Shoup, J. E. (1978). Frequency of occurrence of phonemes in conversational English. Language and Speech, 21, 221-41. Ramakrishna, B. S. (1962). Some aspects of the relative efficiencies of Indian languages: a study from information theory point of view . Bangalore: Indian Institute of Science. Sandoval, A. M., Toledano, D. T., de la Torre, R., Garrote, M., & Guirao, J. M. (2008). Developing a Phonemic and Syllabic Frequency Inventory for spontaneous spoken Castalian

Spanish and their comparison to Text-based Inventories. Proceedings of the Language Resources and Evaluation Conference. Available at: www.lrecconf.org/proceedings/lrec2008/pdf/2 83_paper.pdf (accessed: 14 March 2012). Siegel, P (2010). Second language acquisition (1 st ed.). Cambridge University Press. US.

Yegerlehner and Voegelin (1957). Frequencies and Inventories of Phonemes fromNine Languages. International Journal of American Linguistics, Vol. 23, No. 2 pp. 85-93 ACKNOWLEDGEMENTS We extend our sincere gratitude to Prof. S.R. Savithri, Director, and Prof. Y V Geetha, Head, Department of Speech Language Sciences, All India Institute of Speech and Hearing for permitting us to carry out this study. We also thank all the participants of our study for their kind cooperation. This study is a part of the ARF project titled Frequency of occurrence of phoneme s in Malayalam

Sreedevi, N., Smitha, N and Vikas (2012). Frequency of occurrence of phonemes in Kannada. ARF project. AIISH.

Appendix I CONSONANT SEGMENTS International Phonetic Alphabet for Regional languages (Malayalam) Asher & Kumari (1997) Manner of Articulation Stops Voiceless Voiceless Aspirated Voiced Voiceless Aspirated Fricatives Nasal Liquid Tap /Trill Lateral Approximants Glide Place of Articulation Labial Dental Alveolar Retroflex Palatal Velar Glott al

p p B b f m

t t d d

c ch h

K k G g H

s, n N

r, L v j

Appendix II Percentage of occurrence of phonemes in descending order Vowels /a/ /i/ // /a:/ /e/ /u/ /o:/ /o/ /e:/ /i:/ /u:/ /ai/ /au/ Mean % 13.83 6.07 5.58 4.31 3.75 3.46 1.67 1.23 0.88 0.71 0.64 0.15 0.027 Consonants /k/ /n/ // /t / /l/ /m/ /p/ /j/ // // /v/ /r/ // Mean % 7.06 5.32 4.44 4.20 4.04 3.73 3.24 2.68 2.65 2.49 2.19 2.01 1.89 Consonants /c/ /n / // /s/ // /rrr/ /b/ // /d / /d / // // /g/ Mean % 1.59 1.53 1.48 1.35 0.78 0.53 0.46 0.44 0.38 0.37 0.30 0.24 0.21 Consonants // /b/ /h/ /t / // // /f/ /p/
h

Mean % 0.19 0.17 0.15 0.12 0.10 0.09 0.035 0.024 0.02 0.01 0.007 0.002 0

/k/
h

/g/ //
h

/c/ //
h

You might also like