ஏதேனும் ஒரு 
தலைப்பைச் 
சொடுக்குக.

                               18.The Association between Sound and Meaning - A Statistical study


	 {Paper presented at the Fourth Linguistics Conference conducted on behalf of PILC and AITLA in Pondichery on May 20-22, 1994}


1. INTRODUCTION

1.1 PURPOSE


	The object of this paper is to analyse and bring out the association, if any, between the sound and meaning of words as found 
in literary texts in Tamil. Such an analysis may be very useful in the better understanding and appreciation of the feelings of the author 
or the character as found in the text. This analysis would also pave the way for similar studies in future on current spoken language data 
also and this may lead to interesting studies on the speech habits of different speech communities. However, the main purpose of this 
analysis is to attempt to study the existence of association between sound and meaning as found in ancient Sangam or Post Sangam literature.

1.2 DATA

	It is very important that in any statistical study, frequencies of linguistic elements in study should be stable in the sense that the 
outcome of the study should not vary from sample to sample. It is an established fact that about 5000 units of the linguistic element involved 
in the study taken from a running text would provide the stability required. Since the basic unit involved in this study is ‘sound’, the phoneme 
is taken as one unit. Since separate study has to be carried out for vowels and consonants, running texts on minimum of 6000 phonemes were 
chosen for this analysis.

	The following are the different materials used for this analysis:

	1. All complete texts of the Ten Idylls (பத்துப்பாட்டு முழுதும்)
	2.Selected passages from akanAnURu (akam - அகநானூறு)
	3. Selected passages from puRa nAnURu (puRam - புறநானூறு)
	4. Complete text of tirukkuRaL (திருக்குறள்)

1.3 METHODOLOGY

	The selected passages were entered into a digital computer after transcription into English on one phoneme – one key basis and 
stored in diskettes. Programs were written in TURBO PASCAL to analyse the stored material and produce the desired outputs. In getting 
the frequency counts of various phonemes, utmost care was taken in identifying the phonemes sounding as allophones depending upon the places 
of utterances. Even though how these phonemes were uttered exactly during the time of writing of these texts was not known, it was assumed 
that it would have been mostly close to the modern way of uttering it. The ‘hard’ consonants becoming voiced as in the case of 
nungku (நுங்கு), anjcu (அஞ்சு), ANTu (ஆண்டு), enpu (என்பு), atu (அது), aTu (அடு), apu (அபு) etc., and softened as in the case of akam (அகம்), 
acal (அசல்). etc., were taken into account in the counting process.

2. ANALYSIS

2.1 STUDIES IN AKAM

2.1.1 CONSONANTS

	For purposes of this study, I have restricted my data only to certain selections from akanAnURu (அகநானூறு) and 
puRanAnURu(புறநானூறு). In akam, songs were selected from each of the five tiNai’s (திணைகள்), namely, kuRinjci, mullai, marutham, neythal and 
pAlai. It was believed that as these five tiNais depict different facets of the inner life of a person, some of joy, some of sorrow, some of patience, 
some of impatience etc., there would definitely be a significant variation in the usage of sounds used to picturise these feelings. To my surprise, 
I found no such significant difference in the distribution of majority of sounds. Table I describes the distribution of these sounds grouped in the 
usual way and expressed in percentage of occurrence.

			

	A quick look at these results reveals the fact that these are almost uniformly distributed in all the five tiNais. 
Especially in the case of vallinam consonants, we see that approximately 43% of the consonants are vallinam. About 30% to 32% of the consonants 
are iTaiyinam and as far as mellinam is concerned we see a slightly significant variation in their distribution, with pAlai having the least 
occurrence of these consonants. A close look can be had with the help of the occurrence of the individual phonemes as described in TABLE II.

			

	In the case of N (ண) and n (ந) there is a clear variation in their use in the different tiNais and between the least (in pAlai) and the 
most (in neytal) there is a significant difference at 90% level as tested by the test for difference of proportions between two large samples. 
This particular result is also confirmed by a similar analysis done in tirukkuRaL as described in TABLE III

			

	Here we find a 4% increase in the use of voiced consonants in the third division in kuRaL which deals with the concept of ‘love’. 
Among these mellinam consonants, a clear trend in variation of use can be seen in the case of N (ண) and n (ந). 
From the above we can conclude that, there appears to be no definite choice of the vallinam or iTaiyinam consonants in the expression of soft 
feelings while there clearly seems to be a preference to more use of N (ண) and n (ந) in the expression of soft feelings as we see in the sangam 
or post sangam literature.

2.1.1 VOWELS

	In the case of vowels, there is a striking similarity in the distribution of frequencies, among the various tiNais. 
This is illustrated in TABLE IV

			

	We find that mostly 77% or more are short vowels in all the tiNais. But when a similar study was conducted on selected works on 
puRam, the results were surprisingly different. Some poetic works relating to moments of great sorrow (கையறுநிலை) and feelings of great pride 
(பாடாண் திணை) were selected for the study and their results were compared.

2.2 STUDIES IN PURAM

			

	While in the case of pATAN (பாடாண் திணை) songs, we see that the distribution of short and long vowels agrees with the general 
pattern of distribution of those sounds as found in the akam works, in the case of kaiyaRu nilai songs, we see a significant increase in the use of 
long vowels. These songs are the wailing songs sung to bemoan the death of a great king or waarior and hence they include a large number of 
long vowels. A closer look at the exact distribution of these sounds will reveal a picture of their greater usage to depict moments of sorrow.

			

	While we find a general increase in the use of long vowels like A (a: ஆ), U (u: ஊ), E (e: ஏ) and O (o: ஓ) in moments of sorrow, 
we see a particularly very significant increase in the use of E (e:) in the songs. This observation has become all the more significant when the 
author has observed the occurrence of the E sound in all the ten idylls to be around 2.3%, the maximum occurring 2.6% in tirumurukARRuppaTai. 
Hence it is evident that the E sound is more frequently used for the depiction of deep sorrow. It may also be observed that frequent repetition of 
the sounds A(a: ஆ), I (i: ஈ)and O(o: ஓ)  may serve to portray high or deep feelings of pride or sorrow, splendor or fear as they are found equally 
high in both kaiyaRu and pATAN but significantly differ from the other passages.

3. CONCLUSION

	There appears to be a definite preference over the use of certain consonants and vowels when the poets try to depict feelings in the form 
of sounds. New statistical tools may be developed to measure these to bring out their intensity, depth etc.,

Reference:

1. Marie Ludrikova ., On the Occurrence of Syllables in different word positions., Prague Studies of Mathematical Lunguistics., 
Vol. 6., North-Holland Pub.Co., 1978.

2. Gustav Herdan., The Advanced Theory of Language as Choice and Chance., Springer Verlac., 1966.