Research Apprenticeships

The goal of a research apprenticeship is to convey practical experience in the focal research topics of our institute and to lead students through individual guidance to independent research. Students are expected to work independently and to be self-motivated. Mentors will offer guidance, feedback and advice in regular meetings or other appropriate form.

Below is a list of possible topics offered by various members of our institute. Interested students should contact potential mentors for further information.

 

title

Information and language

areas

language variation and change, language structure

contact

Christian Bentz

Human language is an information encoding device. This apprenticeship aims to familiarize students with the theory of information and how it relates to research in linguistics. The RA offers several topics: 1) language variation: information-theoretic accounts of linguistic diversity, 2) language change: the evolution of information-encoding in human languages, 3) language structure: investigating the difference between language and other communication systems. Depending on preferences and skills, these topics can entail detailed literature reviews, (small scale) corpus building, basic statistical analyses and visualizations with R.

 

title

Automatic cognate identification

areas

language variation & change, language structure

contact

Gerhard Jäger

One of the big challenges in computational historical linguistics is to infer language trees from raw word lists in phonetic transcription, without the aid of additional expert knowledge such as manual cognate classification or identification of loan words. The dominant approach so far is to start by assessing the phonetic similarity of words from different languages. Kirby & Ellison (2006) make an interesting alternative proposal. Starting from language-internal comparison of word forms, they construct a lexical network for each language under consideration. The degree of relatedness between two languages is then assessed by comparing the structures of their lexical networks. In the apprenticeship project, this method should be applied to the various cross-linguistic resources that have become available since 2006. In a second step, the method should be modified in such a way that it can be used as a heuristics to identify cognate words and loan words across related languages. Programming skills are essential. Some background/interest in historical linguistics and/or phonology and phonetics will be a plus.

 

title

Studying dialect data

areas

language variation & change, language structure

contact

Igor Yanovich

Spoken languages vary from town to town and village to village. That variation has been recorded by dialectologists in impressive surveys published as dialectological atlases. Such atlases represent a snapshop of the living language in its diversity, and can even allow researchers a good view into the past. In this project, you will learn more about how dialects can differ from each other and how they can be structured linguistically and geographically. The practical part of the internship will involve working with a dialectological atlas, both by hand and electronically, in order to better understand the information hidden in the atlas’s maps.

 

title

Computational phylogenetics

areas

language processing, language structure

contact

Taraka Rama

Students can take up the following projects. 1) Identify a language group and then collect word lists for the language group and document the decisions made through data collection. The end result would be to make the student acquainted with a language family and then apply phylogenetic methods and publish a paper. 2) The project is to implement a paper about sequence alignment or a machine learning paper and apply it for phylogenetics. 3) Phylogenetic from web. There are many resources on the web that have to be harvested for phylogenetic purposes. The student will write scripts to process the data and convert the data into machine readable formats. Then, the student can apply ML techniques for analyzing the data.

 

title

Bayesian cognitive modeling of experimental data

areas

language processing, language structure

contact

Michael Franke

We will study probabilistic cognitive modeling from a Bayesian point of view, in the style of Lee & Wagenmakers (2015). The apprenticeship will familiarize students with the basics of Bayesian data analysis and its main tools. We will then go through some of the key examples in Lee & Wagenmakers’ textbook and round the project off with a novel and individual case study based on linguistic data.

 

title

Online replication of a psycholinguistic experiment

areas

language structure, language change, language processing

contact

Michael Franke

In this project, students will attempt to program and run a browser-based replication of an existing psycholinguistic experiment. Most simple experiments, including reaction time studies or self-paced reading, can be implemented for and issued via the internet. Students will familiarize themselves with the possibilities of recruiting experimental subjects online via crowd-sourcing platforms, such as Amazon’s Mechanical Turk or CrowdFlower. Implementing the experiment will require acquisition of basic familiarity with HTML, CSS and Javascript. Depending on the success of the implementation/replication, the project ideally entails at least cursory data visualization and analysis.

 

title

Language evolution in the lab

areas

language structure, language processing

contact

Michael Franke

[NB: This is a more advanced version of the previous project. Sufficient skills in Python and Javascript are required; or otherwise a keen sense of adventure.]

Experimental approaches to language evolution are increasingly popular. Such experiments usually use dynamic and interactive designs. For example, in iterated learning experiments, the responses from one participant may be the input stimuli for another. Alternatively, experiments on the evolution of conventions may feature a small group of participants communicating with each other, possibly in a tightly regulated network structure. Wallace is an open programming platform for programming such dynamic and interactive experiments to run online, allowing the recruitment of experimental subjects via Amazon’s Mechanical Turk. This project consists in implementing a basic experiment (ideally a replication of an existing study) in Wallace.

 

For further information, please send an email to michael.franke@remove-this.uni-tuebingen.de.