10
Dan Wright Developing Algorithms for Computational Comparative Diachronic Historical Linguistics

Dan Wright Developing Algorithms for Computational Comparative Diachronic Historical Linguistics

Embed Size (px)

Citation preview

Page 1: Dan Wright Developing Algorithms for Computational Comparative Diachronic Historical Linguistics

Dan Wright

Developing Algorithms for Computational Comparative Diachronic Historical Linguistics

Page 2: Dan Wright Developing Algorithms for Computational Comparative Diachronic Historical Linguistics

Historical Linguistics

● Historical linguistics is the study of how language changes over time.

● Languages split into groups, forming a hierarchy or web of languages, each related to its ancestors

● All changes in language are completely regular, so they can be analyzed and to a degree discovered from the current state of the descendant languages.

Page 3: Dan Wright Developing Algorithms for Computational Comparative Diachronic Historical Linguistics

Phonetics

● The fundamental unit of language is the phoneme.

● In order to analyze language, one must first devise a method to deal with phonemes.

● Phonemes can be classified on five axes, using the separations of the International Phonetic Alphabet.

Page 4: Dan Wright Developing Algorithms for Computational Comparative Diachronic Historical Linguistics

Phoneme Categorization

Page 5: Dan Wright Developing Algorithms for Computational Comparative Diachronic Historical Linguistics

Phoneme Storage

Vowels

RoundednessOpennessFrontness

Offset

Consonants

VoicednessPlace of Articulation

Method of ArticulationNot used

Vowel or Consonant

Page 6: Dan Wright Developing Algorithms for Computational Comparative Diachronic Historical Linguistics

Correspondence

● My first attempts to analyze the web-structures of languages was by measuring correspondence between languages.

● I ran lists of words through algorithms which measured how much certain phonemes and axial structures matched up.

● I attempted to build a web of languages from the bottom up, connecting languages through correspondence.

Page 7: Dan Wright Developing Algorithms for Computational Comparative Diachronic Historical Linguistics

● But there is a better way!

Page 8: Dan Wright Developing Algorithms for Computational Comparative Diachronic Historical Linguistics

From the Top Down!

● My second approach to web formation was to start with all of the languages in one organization.

● I then separated them into languages which are more related to each other than a regressed hypothetical ancestor language.

● This was recursively applied to the new families.

Page 9: Dan Wright Developing Algorithms for Computational Comparative Diachronic Historical Linguistics

Conclusion

● My top-down approach was able to somewhat reliably separate languages into their actual categories based on phonetics alone.

Page 10: Dan Wright Developing Algorithms for Computational Comparative Diachronic Historical Linguistics