Search

We make one of the first attempts to build working models for intra-sentential code-switching based on the Equivalence-Constraint (Poplack 1980) and Matrix-Language (Myers-Scotton 1993) theories. We conduct a detailed theoretical analysis, and a small-scale empirical study of the two models for Hindi-English CS. Our analyses show that the models are neither sound nor complete. Taking insights from the errors made by the models, we propose a new model that combines features of both the theories., Comment: 13 pages

This paper presents the first evaluation framework for Web search query segmentation based directly on IR performance. In the past, segmentation strategies were mainly validated against manual annotations. Our work shows that the goodness of a segmentation algorithm as judged through evaluation against a handful of human annotated segmentations hardly reflects its effectiveness in an IR-based setup. In fact, state-of the-art algorithms are shown to perform as good as, and sometimes even better than human annotations -- a fact masked by previous validations. The proposed framework also provides us an objective understanding of the gap between the present best and the best possible segmentation algorithm. We draw these conclusions based on an extensive evaluation of six segmentation strategies, including three most recent algorithms, vis-a-vis segmentations from three human annotators. The evaluation framework also gives insights about which segments should be necessarily detected by an algorithm for achieving the best retrieval results. The meticulously constructed dataset used in our experiments has been made public for use by the research community.

It is a well-known fact that the degree distribution (DD) of the nodes in a partition of a bipartite network influences the DD of its one-mode projection on that partition. However, there are no studies exploring the effect of the DD of the other partition on the one-mode projection. In this article, we show that the DD of the other partition, in fact, has a very strong influence on the DD of the one-mode projection. We establish this fact by deriving the exact or approximate closed-forms of the DD of the one-mode projection through the application of generating function formalism followed by the method of iterative convolution. The results are cross-validated through appropriate simulations., Comment: 9 pages, 1 figure, Accepted for publication in Physica A: Statistical Mechanics and its Applications

We study the global topology of the syntactic and semantic distributional similarity networks for English through the technique of spectral analysis. We observe that while the syntactic network has a hierarchical structure with strong communities and their mixtures, the semantic network has several tightly knit communities along with a large core without any such well-defined community structure., Comment: In the proceedings of ACL 2009 (short paper)

n this paper, we attempt to explain the emergence of the linguistic diversity that exists across the consonant inventories of some of the major language families of the world through a complex network based growth model. There is only a single parameter for this model that is meant to introduce a small amount of randomness in the otherwise preferential attachment based growth process. The experiments with this model parameter indicates that the choice of consonants among the languages within a family are far more preferential than it is across the families. The implications of this result are twofold -- (a) there is an innate preference of the speakers towards acquiring certain linguistic structures over others and (b) shared ancestry propels the stronger preferential connection between the languages within a family than across them. Furthermore, our observations indicate that this parameter might bear a correlation with the period of existence of the language families under investigation., Comment: In EACL 2009 Workshop on Cognitive Aspects of Computational Language Acquisition

The paper is being withdrawn since the authors felt that the submission is a little premature after a careful reading by some of the experts in this field., Comment: This paper has been withdrawn by the authors. The submission is a little premature. The new submission shall be made to arXiv soon after some important revisions that were suggested by some of the experts in this field

Recent research has shown that language and the socio-cognitive phenomena associated with it can be aptly modeled and visualized through networks of linguistic entities. However, most of the existing works on linguistic networks focus only on the local properties of the networks. This study is an attempt to analyze the structure of languages via a purely structural technique, namely spectral analysis, which is ideally suited for discovering the global correlations in a network. Application of this technique to PhoNet, the co-occurrence network of consonants, not only reveals several natural linguistic principles governing the structure of the consonant inventories, but is also able to quantify their relative importance. We believe that this powerful technique can be successfully applied, in general, to study the structure of natural languages., Comment: In the proceedings of EACL 2009

Life and language are discrete combinatorial systems (DCSs) in which the basic building blocks are finite sets of elementary units: nucleotides or codons in a DNA sequence and letters or words in a language. Different combinations of these finite units give rise to potentially infinite numbers of genes or sentences. This type of DCS can be represented as an Alphabetic Bipartite Network ($\alpha$-BiN) where there are two kinds of nodes, one type represents the elementary units while the other type represents their combinations. There is an edge between a node corresponding to an elementary unit $u$ and a node corresponding to a particular combination $v$ if $u$ is present in $v$. Naturally, the partition consisting of the nodes representing elementary units is fixed, while the other partition is allowed to grow unboundedly. Here, we extend recently analytical findings for $\alpha$-BiNs derived in [Peruani et al., Europhys. Lett. 79, 28001 (2007)] and empirically investigate two real world systems: the codon-gene network and the phoneme-language network. The evolution equations for $\alpha$-BiNs under different growth rules are derived, and the corresponding degree distributions computed. It is shown that asymptotically the degree distribution of $\alpha$-BiNs can be described as a family of beta distributions. The one-mode projections of the theoretical as well as the real world $\alpha$-BiNs are also studied. We propose a comparison of the real world degree distributions and our theoretical predictions as a means for inferring the mechanisms underlying the growth of real world systems.

We study the self-organization of the consonant inventories through a complex network approach. We observe that the distribution of occurrence as well as cooccurrence of the consonants across languages follow a power-law behavior. The co-occurrence network of consonants exhibits a high clustering coefficient. We propose four novel synthesis models for these networks (each of which is a refinement of the earlier) so as to successively match with higher accuracy (a) the above mentioned topological properties as well as (b) the linguistic property of feature economy exhibited by the consonant inventories. We conclude by arguing that a possible interpretation of this mechanism of network growth is the process of child language acquisition. Such models essentially increase our understanding of the structure of languages that is influenced by their evolutionary dynamics and this, in turn, can be extremely useful for building future NLP applications., Comment: 8pages, 3 figures, In the proceedings of COLING 2008

We study the growth of bipartite networks in which the number of nodes in one of the partitions is kept fixed while the other partition is allowed to grow. We study random and preferential attachment as well as combination of both. We derive the exact analytical expression for the degree-distribution of all these different types of attachments while assuming that edges are incorporated sequentially, i.e., a single edge is added to the growing network in a time step. We also provide an approximate expression for the case when more than one edge are added in a time step. We show that depending on the relative weight between random and preferential attachment, the degree-distribution of this type of network falls into one of four possible regimes which range from a binomial distribution for pure random attachment to an u-shaped distribution for dominant preferential attachment.

The difficulties involved in spelling error detection and correction in a language have been investigated in this work through the conceptualization of SpellNet - the weighted network of words, where edges indicate orthographic proximity between two words. We construct SpellNets for three languages - Bengali, English and Hindi. Through appropriate mathematical analysis and/or intuitive justification, we interpret the different topological metrics of SpellNet from the perspective of the issues related to spell-checking. We make many interesting observations, the most significant among them being that the probability of making a real word error in a language is propotionate to the average weighted degree of SpellNet, which is found to be highest for Hindi, followed by Bengali and English., Comment: 8 pages, 3 figures, to be published in the proceedings of HLT-NAACL Workshop - TextGraphs 2

In this work, we attempt to capture patterns of co-occurrence across vowel systems and at the same time figure out the nature of the force leading to the emergence of such patterns. For this purpose we define a weighted network where the vowels are the nodes and an edge between two nodes (read vowels) signify their co-occurrence likelihood over the vowel inventories. Through this network we identify communities of vowels, which essentially reflect their patterns of co-occurrence across languages. We observe that in the assortative vowel communities the constituent nodes (read vowels) are largely uncorrelated in terms of their features and show that they are formed based on the principle of maximal perceptual contrast. However, in the rest of the communities, strong correlations are reflected among the constituent vowels with respect to their features indicating that it is the principle of feature economy that binds them together. We validate the above observations by proposing a quantitative measure of perceptual contrast as well as feature economy and subsequently comparing the results obtained due to these quantifications with those where we assume that the vowel inventories had evolved just by chance., Comment: Submitted to Advances in Complex Systems

The sound inventories of the world's languages self-organize themselves giving rise to similar cross-linguistic patterns. In this work we attempt to capture this phenomenon of self-organization, which shapes the structure of the consonant inventories, through a complex network approach. For this purpose we define the occurrence and co-occurrence networks of consonants and systematically study some of their important topological properties. A crucial observation is that the occurrence as well as the co-occurrence of consonants across languages follow a power law distribution. This property is arguably a consequence of the principle of preferential attachment. In order to support this argument we propose a synthesis model which reproduces the degree distribution for the networks to a close approximation. We further observe that the co-occurrence network of consonants show a high degree of clustering and subsequently refine our synthesis model in order to incorporate this property. Finally, we discuss how preferential attachment manifests itself through the evolutionary nature of language., Comment: 18 pages, 15 figures

Cross-linguistic similarities are reflected by the speech sound systems of languages all over the world. In this work we try to model such similarities observed in the consonant inventories, through a complex bipartite network. We present a systematic study of some of the appealing features of these inventories with the help of the bipartite network. An important observation is that the occurrence of consonants follows a two regime power law distribution. We find that the consonant inventory size distribution together with the principle of preferential attachment are the main reasons behind the emergence of such a two regime behavior. In order to further support our explanation we present a synthesis model for this network based on the general theory of preferential attachment., Comment: 8 pages, 8 figures, To be published in COLING-ACL(2006)

Speech sounds of the languages all over the world show remarkable patterns of cooccurrence. In this work, we attempt to automatically capture the patterns of cooccurrence of the consonants across languages and at the same time figure out the nature of the force leading to the emergence of such patterns. For this purpose we define a weighted network where the consonants are the nodes and an edge between two nodes (read consonants) signify their co-occurrence likelihood over the consonant inventories. Through this network we identify communities of consonants that essentially reflect their patterns of co-occurrence across languages. We test the goodness of the communities and observe that the constituent consonants frequently occur in such groups in real languages also. Interestingly, the consonants forming these communities reflect strong correlations in terms of their features, which indicate that the principle of feature economy acts as a driving force towards community formation. In order to measure the strength of this force we propose an information theoretic definition of feature economy and show that indeed the feature economy exhibited by the consonant communities are substantially better than those if the consonant inventories had evolved just by chance., Comment: 15 pages, 9 figures, To be published in Int. Jour. of Modern Phy. C