I See Dead Code

I was a student of computational linguistics in Tübingen and finished my BA in 2006.

BA Thesis

Title: Analysis of German Compounds Using Weighted Finite State Transducers

As the title aptly says, my BA thesis was about implementing a finite state compound splitter. For the thesis, I half-automatically split ~120,000 German compounds into their parts [download]. The code I wrote for creating the splitter automaton is also available [download]. Usage instructions and format descriptions are given in the thesis paper, but I’d be glad to answer any questions.

Term Paper

Title: Predicting Linking Elements in German Compounds: A Stochastic Approach

German compounds usually take a semantically empty linking element (or morpheme) between head and complement (Fugenelement, like the -s- in Mittag-s-zeit) (lunchtime). Some analysises about the choice of different linking elements (like -en-, -s-, -e-) in compounds are based on semantic features or inflexion classes of head and complement. Such knowledge is generally not available when the prediction is to be made by a computer program, an easier theory is desirable.

Baayen 2003 [1] shows a probabilistic approach on predicting linking elements relying only on phon ological features of head and complements for Dutch, which produces quite useful results. This paper tries to apply Baayen’s theory on German compounding. The new formalism’s performance will be evalu ated to find out if it is similar to the performance observed for Dutch.

[1] R. Harald Baayen. Probabilistic approaches to morphology. In Probabilistic linguistics, MIT Press, 2003
Parsing

Phonology

Together with Armin Schmidt.

In the course “Introduction to Phonetics and Phonology (SS 2004)”, we had to write three papers on different aspects of the phonology of a freely chosen language. We chose Polish.

The sound inventory of Polish

The first part [download] deals with the sound inventory of the language and outlines the differences to the German sound inventory

Acoustic analysis of Polish fricatives

The second part [download] contains an analysis of two minimally different Polish fricatives (and there are many fricatives in Polish)

Two theories on Polish syllable structure

The third part [download] is a comparison of two approaches for syllable structure assignment to Polish words, where one is taken from the literature and the other one is made up by us, the authors.

Languages of the world

In the course “Languages of the world (WS 03/04)” taught by Dr. Monika Rathert, a part of the grade was to write an essay about the structure of a non-Indoeuropean language. My homework was about Korean [download].