We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
After finishing this chapter, you will be able to:
} Know the phonetical foundations in Sanskrit Language
> Understand the basic structure of Sanskrit Language Grammar
| » Get introduced to some computational aspects in Astadhyayi
> Appreciate the importance of verbs in Sanskrit Language
> Appreciate the potent
Sanskrit offers in Natural Language Processing,
Panini’s work on Sanskrit grammar has
been acclaimed to be one of the best
in the field of linguistics and phonetics.
It has attracted wide attention. The
accompanying figure is a Birch bark
manuscript from Kashmir of the
Rupavatara, a grammatical textbook
based on the Sanskrit grammar of
Panini. It was composed by Dharmakirti,
a Buddhist monk from Ceylon. The
‘manuscript was transcribed in 1663.
Source: hetps://[Link]!wiki/File:Birch_bark_MS_from_Kashmir_of_the_Rupavatra_ Wellcome.
0032691 jpg = 114 _Introdveton to Indian Knowledge SystemC
‘An Ecosystem for Sans!
nt languages (Sanskrit
Sanskrit is one of three an
Greek, and Latin). While ancient Greek and Latin
have attained a state of being classical, the Sanskrit
language continues to be in use, albeit in specific
domains by researchers. Hi in recent years,
there is a growing recognitio
of Sanskrit and several in ;
taken globally to integrate technology aids with
nguage. One of them is building
n of the importance
jatives have been
the Sanskrit
an ecosystem of language processing tools for
Sanskrit.
The efforts began in the late 1980s with several
projects digitizing Sanskrit texts. Over a decade
2 central registry of digitized Indic texts was
created in some Universities in Germany. Another
progress in this direction was the creation of
digital dictionaries of Sanskrit. The Cologne Digital
Sanskrit Lexicon project digitized the Monier
Willams’ Sanskrit-English Dictionary. Furthermore,
the Digital Dictionaries of South Asia project at
the University of Chicago digitized Apte’s and
MacDonell's Sanskrit-English dictionaries. The next
step in tis direction was the processing of Sanskrit
text using machine coding. Academy of Sanskrit
Research, Melkote developed a verbal cognition
generator for Bhandarkar's Sanskrit primer in the
early 1990s.
By the turn of the century, the Ministry of
Communication and Information Technology,
Government of india set up the Technology
Languages program (TOIL).
has been a spurt in the efforts
international
Develo
Further to this, t
evident from multiple
collaborative efforts in this direction
developed Anusdraka’,
The Akshar Bharati gro
that employed techniques
ed on Pénini's Astadhyayi. Beginning in 20
f Sanskrit computational tools was developed
rough several research projects. These activities
y (INU), Delhi
as been the most effective
ugh some basic comm 9S can
elab d compl
concepts and Applic
for our commun
andled v
% ideas. Scientific discovery, the advancement of
Jkrit Language Processing
University of Hyderabed, IT Kenpo
and Rashtriya Sanskrit Vidyapeeth, Tirgpatt
Global, other Sanskrit linguistic researchers et
allaborting on these
Some of the fallout of these are developin
Sensi including the following
«+ Web-based Sanskrit reader program at Brown
University called Kramapstha
2 digital edition of Whitney's roots that
served as the source of verbal stems forthe
ilectional generation sofware
The international Digital Sanskrit Library
Integration projectin the Classics Department
at Brown University
Sanskrit Library Phonetic encoding that
allows all sounds represented in Vedic texts
to be represented digitally
Sanskrit Heritage platform (SH), centred
around an electronic version of the Sanskrit-
French Sanskrit Heritage dictionary
Sanskrit Reader, allowing segmentation
(andhi analysis), tagging, and parsing
In 2008, the Indian Government funded a
major consortium project to develop various tools
for analysis of Sanskrit text and a Sanskrit-Hindi
machine translation system. Sanskrit scholars and
computational linguists collaborated to develop
several tools for enhancing the Sanskrit language
processing ecosystem. This includes Sandhi and
Samasa analyser, morphological analyser, generator,
sandhi joiner and splitter, and fullledged parser.
These tools are now being deployed in some
machine translation systems for testing and further
refinements
Source: Adopted from, Goyal, P., Huet, G., Kulkami,
A,, Scharf, P, and Bunker, R. (2012). "A Distributed
tform for Sanskrit Processing’, Proceedings of
Papers, pp. 1011-1028,
COLING
Mumbai, December
ication since time immemorial.
th gestures, language becomes inevitable sowledge, and collaborative working require a common method of commun\ca¥er ine
tenga this roe ina eviized society. As technology and computing prowess improves, we aFe
ys o develop better applications using Artificial Intelligence (AN) techniaues, This requires
able Nevelop efficient Natural Language Processing (NLP) capabilities. A systematic OT
eguages and their capabilities for the evolving requirements has become important for the
langice and technology community. In this chapter we shal see the developments in the fie
sf yanguage in ancient India,
5.1 COMPONENTS OF A LANGUAGE
ow many of us can read and comprehend Shakespeare's works or native Indian {ete
Heth as Ramayana? Even if we know the language in which these texts are written, it IS difficult
icause the way the language is used undergoes changes.§
pecmtuage 1s a tool used by everyone ina community and # One of the Vedangas Kt
i karana focuses on linguistics
itis very difficult to maintain it unchanged. We also notice 4 a
tide there are differences in the same language spoken by and Phonetics aspects of San
people from different regions. For these reasons, the, itaghya
EO a ee Tees neompreheraiis fr te Weta a ra
people in the future though they may be speaking the same and the best avalable descriptive
fanguage. Studying the structure of the language helps us model of a languages
not lose the underlying principles that govern a language
and ensures that the received wisdom from the ancestors is not Jost. Italso maintains continuity
in language processing.
‘Communication is key to trade, science and technology, and societal progress. It hinges on
our ability to effectively process language as itis central to all human transactions and pursuits
Language processing has two dimensions: receptive
and productive. The receptive part of a language deals
With the ability of an individual to receive language
inputs from multiple sources and process them to
decipher the intended message and comprehend them.
On the other hand, the productive part of the language
is to transmit back to others for their consumption.
The focus in the former is on listening and reading,
whereas it is on speaking and writing in the latter.
Viewed from another perspective, sound (listening
and speaking) and script (reading and writing) are the
essential elements of a language. Therefore, language
processing can be represented in a 2x2 framework as agua
ge
shown in Figure 5.1. Linguistics addresses all these
aspects of a language. Phonetics will cater to the receptive and productive aspects of the sound
and a syntactical structure will cater to the scriptural aspects of a language.
Linguistics is a branch of language research that provides a scientific study of a language.
It iS a systematic study of language to understand speech sounds, grammatical structures, and
mea
ing It helps us analyse the language form and meaning and identify systematic methods
“integral to the Tanguage to derive the word forms a
nd their meaning using structured rules and
syntax. The earliest approach to a systematic treatment of linguistics is attributed to the Indian
FIGURE 5.1 Components of a ‘oncepts and Applications
to Indian Knowledge Syster
ho lve inthe 6th Century BCE, His work on Sanslert grammar
grammarian, Panint tum opus on linguistics. In this chapter, we shall see more dets ae
jum opus on lin
as Astadhydyris a magn!
5.2. PANINI’S WORK ON SANSKRIT GRAMMAR
in the Indian cont e the pi /edas and their meanings was given
ext, since the preservation of the Vedas anc ae
ae ae a wy Sas an important discipline to be studied by everyone. Therena ames
importance, linguistics was
ref
siigas, Siksa, Nirukta and Vyakarana focused on phonetical and ae
a vogee hve Brel disused these ln Chapter 2 of the Book Deeg
aspects of ARS. erin Sanskrit can comprehend ancien Ween the epics,
eee aie ee ie carpas ina lossless fashion. Several fa] farig aaa borrowed the
Partha sed ghonedi structure, language organization, and vocabalaes aa Panini’s work
aaa gettimar Therefore, iis important to knove this in somalia aman woe
aera raat, kno as Agiidkyi sl process of refiteuaet and syntactical suc
Se ence a eae aa
eation of human intelligence. - ae
athe china is “a culmination of a_long grammatical tradition. Panini brought hig
originality into action and proposed a structure to the grammar. Te is the best available
descriptive model of a language. The greatness of this work is evident from the fact that fe
has overshadowed all previous attempts on Sanskrit grammar, Moreover, several commentaries —
written on its another proof of its prominence in linguistics. In the 4th century BCE, Katyayana
composed a commentary (vartika) on the Paninan work, which served to: provide further
explanation of the work and clarified certain aspects of Astadhyayi. Another great work on
Astadhyayi is the Mahabhasya, a commentary on the Astadhyayi, authored by Patafijali in
the 2nd century BCE. These three are considered the three great sages (tri-muni) of Sanskrit
grammar.
: While framing a syntax and linguistic framework, the challenge is to find an efficient set
of rules to accommodate all the observed patterns and variations in the use of a language by
the society, If there are robust rules that govern a language
¢ Pinini composed 3,983 rules to leaving several exceptions, it is not useful, At the other
accommodate all the patterns and extreme, to accommodate all the possible variations, the
Te spoanikt language: | number of rules framed may become very large, again
* Te dotinguching teen wend making it less useful. The greatness of the Sansleit gramme
Sat ae laid out by Papin bes In fe asiienec i ee
eternal in its appeal framework for linguistics that nicely balanced both these
— aspects, Panini did not write a set of rules and insisted that
s to use the language. This is possible when we develop a
for example. Panini composed 3,983 rules (known as
Patterns and variations in the Sanskrit language and
jm in eight chapters (therefore, the name Astédhyayi). Each chapter is
| into 4 quarters, thereby making it 32 quarters in all ck ye
f PSnini and its distinguishing features make Sanskrit a po
d eternal in its appeal. These are summarized below:
everyone must follow these
computer language such as C or Python,
) to mostly accommodate the
cd th
basic approach of
he entire vocabulary of the Sans
rit language could be created using the 3,983 rules.
“wv exceptions need special handling. The rules are aphorisms (known as siitras),
share easy to commit to memory. What it implies is that ifsomeone is familiar with Linguistics 117
__ ee er et
these rules and how it needs to be applied, then it amounts to gaining unambjgvov
mastery of Sanskrit language
ie. The educational systems in India until the introduction
of the Macaulayan system of
ff education in the 19th century CE ensured this mastery
for the students.
‘+ Language processing and word generation are strictly rule-based and derivative
ature. Proper application of the rules must result in a valid word (a form of the
Taguage), There is no need to make any additional assumptions to derive any word
+ The derivation of words using the rul
les could be done using a step-by-step process:
What it implies is that technically speaking, st
tarting from a base (a verbal root or
wfominal Toot), the logic can search if any of the 3,983 rules could be applied to
the current transformation of the root. If the rule can modify, the current structure
it will perfo
rm the operation. The procedure stops when none of the rules could be
further applic
.d. The result is the final word. This rule-based recursive structure to
language is highly amenable to computer-based processing. The entire process !s
logical, unambiguous, and rational.
'¢ The entire scheme for word generation follows a highly modular approach. Two basic
components form a word. Each word is formed out of a base (verbal or nominal root);
‘To this, one or more suffixes are added to generate the word. On account of this method
of word generation, the Sanskrit language inherently has a high degree of ‘patterns’ of
words and word structures that makes it very efficient to develop vocabulary.
+ Since the entire language is rule-based, the vocabulary is not fixed or static for the
language. The rules can be used for generating new words, as long as the rules are
not violated. Therefore, the language is dynamic, can construct new words as demand
‘rises, and can maintain its relevance. This has implications for lexicographic studies
in the language.
+ Astidhyayi deploys several interesting data structures and computational elements
that make it unique among linguistic studies.
‘These distinguishing aspects of Panini's grammar are discussed details in Sections 5.5,
5.6 and 5.8 of this chapter. Since language has the phonetic (sound) and syntactical (script)
dimensions, we shall see Panini’s approach to these in the Sanskrit language.
5.3. PHONETICS IN SANSKRIT
Phonetics is the study of sounds in a language, particularly the production of sound in a
language and how it communicates the language corresponding to the scripts of the language.
It also addresses the issue of how the sound is perceived in the language. Phonetics in the
Sanskrit language has been addressed in some detail, since this is vital because the ancient
Indian knowledge tradition is oral. The entire transmission of the Vedas from time immemorial
has survived several thousand years on account of scientific methods of oral rendering, This
has been possible on account of a well-developed science of phonetics. This also ensured an
impeccable textual transmission superior to the classical texts of other cultures. This is perhaps
the reason for UNESCO recognizing Vedas in the form of oral knowledge as a heritage for
preservation.
The science of the study of sound, known as Siksa, forms one of the six of the Vedaigas. In
Chapter 2, we have a brief introduction to Siksa. In the earliest Indian traditions, Pratisakhyas sand Applications
eton to Indian Krowdve =
isare produced. Reveda-Pratisakhya and the Taittirtyaq
e issue of how sounds are Pro o these texts, the sound pri
enon ‘on the subject. According to Primarily.
are the earliest works on one se nest due to the movement of the breath in the jael
the junetion of the Tey a diferent locations resulting in various sound pattem
sin the oral cavity As technology and computing prowess » According to Sanskrit grammar, two pho
improves, we are able to develop better are considered to be homogenous ff thet
ications using Artificial intelligence (Al) produced with the same articulatory
chniques. This requires us to develop and at the same place of pronunciation,
* Natural Language Processing (NLP) > The entire scheme for word generat
follows a highly modular approach. Two |
reation of components form a wor
st available suffix (pratyaya).
Each word is formed out of a base (verbal 6
tominal root). To this, one or more si
are added to generate the word.
One of the distinguishing aspects of #
Paninian approach to linguistics and Sanskti
fadhyayi is considered a fir
human intelig
criptive model of a language.
3,983 rules (known as
mmodate most of the patterns
skrit language and
apters,
According to earliest Indian
primarily arises at # the sound grammar is the use of several featur
sri tw cher a Junction of the throat which in modern parlance map to certal
e chest due to the movement of the om,
breath in the body and manifests -omputational concepts.
Cavity at diferent locations resuting in various” 20INIS system of applying grammatiel
sound panes. ue conditions to derive words works exactly like
2 rule-based engine, Every conceivable word in the Sanskrit language can be derived using
a set of rules in a strict algorithmic fashion.
Astadhyayl employs recursive logic to process
ammatical applications. One such example
the logic of samasa, a method of creating
‘compound wards from a group of nouns
On account of the karaka concepts and linguistic
arrangements, the sentence formation has a
robust structure and leaves very little scope
for ambiguity in the Sanskrit language.
Most words (both the verb forms and noun
forms) originate from the dhatus. In Sanskrit,
we often find several synonyms for a word.
REVIEW QUESTIONS
1. Briefly sketch the organization of Astadhyayi. How many rules are there in As
Zz
3
4
10.
1.
12.
13,
Linguistics 135
Each synonym for 2 word is often derived from
a dhatu,
Prefixes, known as upa-sargas, are appended
to the verb forms in order to create additional
words.
There are 22 upa-sargas and these could be
prefixed to a ver’ form. By adding the upa:
sargas, it is possible to express the meaning
in many ways.
Studies show that certain inherent advantages
in the structure of the Sanskrit language and
the grammar makes it an attractive candidate
for NLP and artificial intelligence-related work.
\dhyayi?
What are the unique aspects of the Sanskrit grammar propounded by Panini?
How Is the issue of phonetics addressed in the Sanskrit language? What are the phonetical aspects,
pertaining to vowels specified in Sanskrit grammar?
‘What do you understand by the terms ‘Prakrti' and ‘Pratyaya’? What is the relevance of these in
Sanskrit grammar?
Write short notes on the following:
(@)_ The Sanskrit language is derivational in nature
(b) Sanskrit grammar is rule-driven
(Q) Sanskrit grammar is modular in structure
‘What are the ways by which one can generate noun forms in Sanskrit? Similarly,
forms can be generated,
explain how verb
What do you understand by the terms ‘subanta’ and ‘titanta’?
What do you understand by the term ‘mnemonics’? How is it used in Astadhyayi?
Briefly explain how compound words are generated in the Sanskrit language,
What is the relevance of the ‘Karaka concept? Using an example identify the Karakas in your
example. Can you also relate it to the vibhaktis?
Write a one-page note explaining why knowledge of dhatus in Sanskrit is very valuable.
What happens when an upa-sarga is prefixed to a verb form? Give some exam
your argument.
ples in support of
Comment on the statement, “The Sanskrit language has potential for use in NLP and Al
applications",