A probabilistic assessment of the Indo-Aryan Inner–Outer Hypothesis

This paper uses a novel data-driven probabilistic approach to address the century-old Inner-Outer hypothesis of Indo-Aryan. I develop a Bayesian hierarchical mixed-membership model to assess the validity of this hypothesis using a large data set of automatically extracted sound changes operating between Old Indo-Aryan and Modern Indo-Aryan speech varieties. I employ different prior distributions in order to model sound change, one of which, the Logistic Normal distribution, has not received much attention in linguistics outside of Natural Language Processing, despite its many attractive features. I find evidence for cohesive dialect groups that have made their imprint on contemporary Indo-Aryan languages, and find that when a Logistic Normal prior is used, the distribution of dialect components across languages is largely compatible with a core-periphery pattern similar to that proposed under the Inner-Outer hypothesis.

A Correlated Topic Model Using Word Embeddings

Conventional correlated topic models are able to capture correlation structure among latent topics by replacing the Dirichlet prior with the logistic normal distribution. Word embeddings have been proven to be able to capture semantic regularities in language. Therefore, the semantic relatedness and correlations between words can be directly calculated in the word embedding space, for example, via cosine values. In this paper, we propose a novel correlated topic model using word embeddings. The proposed model enables us to exploit the additional word-level correlation information in word embeddings and directly model topic correlation in the continuous word embedding space. In the model, words in documents are replaced with meaningful word embeddings, topics are modeled as multivariate Gaussian distributions over the word embeddings and topic correlations are learned among the continuous Gaussian topics. A Gibbs sampling solution with data augmentation is given to perform inference. We evaluate our model on the 20 Newsgroups dataset and the Reuters-21578 dataset qualitatively and quantitatively. The experimental results show the effectiveness of our proposed model.

A Logistic Normal Mixture Model for Compositional Data Allowing Essential Zeros

The usual candidate distributions for modeling compositions, the Dirichlet and the logistic normal distribution, do not include zero components in their support. Methods have been developed and refined for dealing with zeros that are rounded, or due to a value being below a detection level. Methods have also been developed for zeros in compositions arising from count data. However, essential zeros, cases where a component is truly absent, in continuous compositions are still a problem.The most promising approach is based on extending the logistic normal distribution to model essential zeros using a mixture of additive logistic normal distributions of different dimension, related by common parameters. We continue this approach, and by imposing an additional constraint, develop a likelihood, and show ways of estimating parameters for location and dispersion. The proposed likelihood, conditional on parameters for the probability of zeros, is a mixture of additive logistic normal distributions of different dimensions whose location and dispersion parameters are projections of a common location or dispersion parameter. For some simple special cases, we contrast the relative efficiency of different location estimators.