Presentation is loading. Please wait.

Presentation is loading. Please wait.

SE367 Course Project Shourya Sonkar Roy Burman (Y8487) Learning Grammatical Gender in an Artificial Language Based on Hindi.

Similar presentations


Presentation on theme: "SE367 Course Project Shourya Sonkar Roy Burman (Y8487) Learning Grammatical Gender in an Artificial Language Based on Hindi."— Presentation transcript:

1 SE367 Course Project Shourya Sonkar Roy Burman (Y8487) Learning Grammatical Gender in an Artificial Language Based on Hindi

2 Introduction Is grammatical gender an arbitrarily defined categorisation? Position: No, grammatical gender is acquired by an individual based on two cues: Distributional Cues: Co-occurrence with gender marked articles, verbs, adjective etc. in a certain manner Phonological Cues: Similar sounding words are acquired as same gender Objective of the study is to support this position using an artificial language constructed using Hindi syllables and Hindi sentence structures

3 Experiment Design

4 The Artificial Language Design similar to that of Mirkovic J., Forrest S. & Gaskell M. G. (2011) Based on Hindi pronounceable syllables Masculine verbs: पातु and जीमु Feminine Verbs: वज and फोस Masculine Nouns: Words with 2-3 syllables, which have no meaning in Hindi, as checked in Google Translate. Feminine Nouns: Words with 2-3 syllables ending with ू or ो, which have no meaning in Hindi, as checked in Google Translate. Mirkovic J., Forrest S. & Gaskell M. G. (2011). Semantic Regularities in Grammatical Categories: Learning Grammatical Gender in an Artificial Language. Proceedings of the 33rd Annual Conference of the Cognitive Science Society

5 The Artificial Language Sentence constructions are [Noun] [verb] है ! कबू वज है ! The sentences were then recorded in the voice of a UP resident (Bimodal learning) Modifications to the experiment suggested by Prof. Achla Raina, Department of Humanities and Social Sciences

6 Schedule of Tasks DayTask(s) 1Word-picture Matching 2 Verb selection Word-picture Matching 3 Word-picture Matching (on Generalisation set) Verb selection Participants: Five employees in the Hall 1 Mess

7 Examples of Tasks Size of word-picture matching dataset: 44 Size of the generalisation set: 8

8

9

10 पातु / जीमु वज / फोस

11 Results

12 Word-Picture Matching (Day 1)

13 Discussion The participants are learning the artificial nouns. The learning rates differed between individuals.

14 Thank you!


Download ppt "SE367 Course Project Shourya Sonkar Roy Burman (Y8487) Learning Grammatical Gender in an Artificial Language Based on Hindi."

Similar presentations


Ads by Google