By - Jayashree Aanand Gajjam, Diptesh Kanojia,Malhar Kulkarni
IIT Bombay, India
IITB-Monash Research Academy, India
AbstractThe Sanskrit grammatical tradition which has commenced with Pāṇini’s Aṣṭādhyāyī mostly as a Padaśāstra has culminated as a Vākyaśāstra, at the hands of Bhartṛhari. The grammarian-philosopher Bhartṛhari and his authoritative work ‘Vākyapadīya’ have been a matter of study for modern scholars, at least for more than 50 years. The notions of a sentence and a word as a meaningful linguistic unit in the language have been a subject matter for the discussion in many works that followed later on. While some scholars have applied philological techniques to critically establish the text of the works of Bhartṛhari, some others have devoted themselves to exploring philosophical insights from them. Some others have studied his works from the point of view of modern linguistics, and psychology. Few others have tried to justify the views by logical discussions.
This paper presents a fresh view to study Bhartṛhari, and his works, especially the ‘Vākyapadīya’. This view is from the field of Natural Language Processing (NLP), more specifically, what is called as Cognitive NLP. We have studied the definitions of a sentence given by Bhartṛhari at the beginning of the second chapter of ‘Vākyapadīya’. We have researched one of these definitions by conducting an experiment and following the methodology of silent-reading of Sanskrit paragraphs. Team collected the Gaze-behavior data of participants and analyze it to understand the underlying comprehension procedure in the human mind and present our results. We evaluate the statistical significance of our results using T-test, and discuss the caveats of our work. We also present some general remarks on this experiment and usefulness of this method for gaining more insights in the work of Bhartṛhari.
Bhartrihari (c. 450—510 C.E.)
Bhartrihari may be considered one of the most original philosophers of language and religion in ancient India. He is known primarily as a grammarian, but his works have great philosophical significance, especially with regard to the connections they posit between grammar, logic, semantics, and ontology. for detailed works visit link
Bhartṛhari, for the first time, deals with the semantic issues in the second Kāṇḍa i.e Vākyakāṇḍa of Vākyapadīya (VP). We can observe a comprehensive treatment on various theories of sentence and their meanings along with their philosophical discussions. He enumerates eight views on the notion of a sentence which are held by earlier theorists in India.
The verse is:
Ākhyātaśabdaḥ saṅghāto jātiḥ saṅghātavartinī
Eko’navayaḥ śabdaḥ kramo buddhyanusaṃhṛtiḥ |
Padamādyaṃ pṛthaksarvaṃ padaṃ sākāṅkṣamityapi
Vākyaṃ prati matirbhinnā bahudhā nyāyavādinam || (VP.II.1-2)
The meaning of these definitions is as follows:
Ākhyātaśabdaḥ - The verb
Saṅghātaḥ - A combination of words
Jātiḥ saṅghātavartinī - The universal in the combination of words
Eko’navayavaḥ śabdaḥ - An utterance which is one and devoid of parts
Kramaḥ - A sequence of words
Buddhyanusaṃhṛtiḥ - The single whole meaning principle in the mind
Padamādyam - The first word
Pṛthak sarvam padam sākāṇkṣam – Each word having expectancy for one another.
These eight views on the sentence are held by earlier grammarians and Mīmāṃsakas. They look at the sentence from different angles depending upon the mental dispositions formed due to their discipline in different Śāstras.
More on Speech and consciousness view below video
Bhartrihari's conception of utterance and understanding can be grasped with the following schema under the rubric of:
Considering these studies as the motivation, we test the definition of the verb by using an experimental method i.e. by using readers’ Eye Movement Behavior
Studying Eye Movements
The eyes are often quoted as being ‘windows to our soul’. This sentiment reveals the deep connection between the eyes and being human. They provide us with vision, the most dominant of our five senses that uses over a third of our brain , and they are also an essential part of social interaction . The eyes also provide a link to the cognitive and perceptual processes that are taking place ‘under the bonnet’, within our brains. What we look at, when we look, how we look and how long we look all have implications for how we process, interpret and interact with the environment around us. For example, from earliest infancy we look at things that grab our attention and interest us, and this can result in ‘sticky fixations’ where an infant under the age of 3 months finds it difficult to look away from a central stimulus. It is only from 3 months onwards that we find it easier to disengage from something that initially holds our attention . This very early developmental behaviour highlights the ingrained connection between our eyes and the outside world. There is also a strong connection between where we look, the outside world and our actions in that world. When performing a routine task, such as making a cup of tea or preparing a sandwich, we tend to look at objects involved in that task just before they are needed . These ‘just in time’ fixations mean we might only look at the knife we will use to butter our bread half-a-second before actually picking it up. The link between attention and where we look is highlighted by how difficult it is to pay attention to one location whilst moving our eyes to a different location . Try it yourself.
Gaze-Tracking: Applicability
Till date, there has been lots of research which have been carried out using eye movement data on various tasks such as reading (texts, poetry, musical notes, numerals), typing, scene perception, face perception, mathematics, physics, analogies, arithmetic problem-solving and various other dynamic situations (driving, basketball foul shooting, golf putting, table tennis, baseball, gymnastics, walking on an uneven terrain, mental rotation, interacting with the computer screens, video game playing etc.) and media communication etc.
Reading researchers have applied eye-tracking for behavioral studies as surveyed by Rayner. Recently, some researchers have even used this technique to explore learning processes in complex learning contexts such as emergent literacy, multimedia learning, and science problem-solving strategies.
Dataset prepartion
Dataset is prepared out of twenty documents consisting of either a prose (Total 13) or a poetry (a subhāṣita) (Total 7) in the Sanskrit language. Prose documents mainly contain the stories taken from the texts such as Pañcatantra, Vaṃśavṛkṣaḥ and Bālanītikathāmālā while as Subhāṣitas are taken from the text Subhāṣitamañjūṣā. The stories are comprised of 10-15 lines each, and
each subhāṣita is 2 - 4 verse long
Method
Feature Description
The eye-tracking device records the activity of the participant’s eye on the screen and records various features through gaze data. We do not use all the feature values provided by the device for our analysis, but only the ones which can provide us with the prominence of a word (interestarea) and in turn, show us the importance of words which belong to the same category. These are features which are calculated based on the gaze behavior of the participant, and are used for analysis:
1. Fixation-based features -
Studies have shown that attentional movements and fixations are obligatorily coupled. More fixations on a word are because of incomplete lexical processes. More cognitive load will lead to more time spent on the respective word. There are some variables that affect the time spent on the word such as word frequency, word predictability, number of meanings of a word or word familiarity etc. (Rayner, 1998). We consider Fixation duration, Total fixation, Fixation Count for the analysis. These are motivated by Mishra et al. (2016a)
(a) Fixation Duration (or First Fixation Duration)-
First fixations are fixations occurring during the first pass reading. Intuitively, an increased first fixation duration is associated with more time spent on the words, which accounts for lexical complexity.
(b) Total Fixation Duration (or Gaze Duration)-
This is a sum of all fixation durations on the interest areas. Sometimes, when there is syntactic ambiguity, a reader re-reads the already read part of the text in order to disambiguate the text. Total fixations duration accounts for sum of all such fixation duration's occurring during the overall reading span.
(c) Fixation Count
This is the number of fixations on the interest area. If the reader reads fast, the first fixation duration may not be high even if the lexical complexity is more. But the number of fixations may increase on the text. So, fixation count may help capture lexical complexity in such cases.
2. Regression-based feature -
Regressions are very common in complicated sentences and many regressions are due to comprehension failures. Short saccade to the left is done to read efficiently. Short within word saccades show that a reader is processing the currently fixated word. Longer regression occur because the reader did not understand the text. Syntactic ambiguity (such as Garden Path sentences etc.), syntactic violation (missing words, replaced words) and syntactic unpredictability leads to shorter saccades and longer regressions. considered the feature Regression Count i.e. a total number of gaze regressions around the area of interest.
3. Skip Count -
Our brain doesn’t read every letter by itself. While reading people keep on jumping to next word. Predictable target word is more likely to be skipped than an unpredictable one. Skip count was taken as a feature to calculate the results. Skip count means whether an interest area was skipped or not fixated on while reading. This is calculated as number of words skipped divided by total word count. Intuitively, higher skip count should correspond to lesser semantic processing requirement
4. Run Count -
Run count is the number of times an interest-area was read.
5. Dwell Time-based feature -
Dwell time and Dwell Time percentage i.e. the amount of time spent on an interest-area, and the percentage of time spent on it given the total number of words.
Results
Accuracy Scores
Conclusion
Paper presents a fresh view to study Bhartṛhari’s ‘Vākyapadīya’, especially the definitions given by him on the syntactic and the semantic level. They picked sentence definition one viz. Ākhyātaśabdaḥ, that the “verb” can also be considered as a sentence. We discuss his work in brief and perform an experiment to study this definition in cognitive point of view. We employ
eye-tracking technique and follow the methodology of silent-reading of Sanskrit paragraphs to perform the above-mentioned experiment in order to have the better understanding of the definition. Authors aim to extend our work under the purview of Cognitive NLP and use it to resolve computational problems. With this work, we open a new vista for studying sentence definitions in the cognitive point of view by following an investigational technique.
Comments