Literary and Linguistic Computing - current issue

A chronometric approach to Indian alchemical literature

Hellwig, O. — Mon, 26 Oct 2009 21:31:55 PDT

Indian alchemy, a branch of traditional Indian medicine (Ayurveda), has produced a corpus of texts that are difficult to date using regular philological techniques. This article describes a contents-based computational method that is capable of calculating the relative chronology of these texts. Central parts of alchemical literature are encoded in a language model that can be understood by a computer and then compared with an alignment algorithm. Phylogenetic trees derived from these alignments show regularities in the ordering of alchemical texts, and these may be interpreted as temporal patterns. Processing these patterns with a minimization algorithm, we are able to compute a relative chronology of the corpus, which is largely consistent with results obtained using traditional philological techniques.

Assessing frequency changes in multistage diachronic corpora: Applications for historical corpus linguistics and the study of language acquisition

Hilpert, M., Gries, S. Th. — Mon, 26 Oct 2009 21:31:55 PDT

The use of corpora that are divided into temporally ordered stages is becoming increasingly wide-spread in historical corpus linguistics. This development is partly due to the fact that more and more resources of this kind are being developed. Since the assessment of frequency changes over multiple periods of time is a relatively recent practice, there are few agreed-upon standards of how such trends should be statistically interpreted. This article addresses the need for a basic analytical toolbox that is specifically tailored to the interpretation of frequency changes in multistage diachronic corpora. We present a number of suggestions for the analysis of data that analysts commonly face in historical studies, but also in the study of language acquisition.

Untangling the derivatives: points for clarification in the findings of the Shakespeare Clinic

Merriam, T. — Mon, 26 Oct 2009 21:31:55 PDT

The work of the Shakespeare Clinic of Claremont McKenna College, led by Ward E.Y. Elliott and Robert J. Valenza, is recognized for its pioneering computer analysis of many early modern texts to determine whether William Shakespeare (1564–1616) wrote the works traditionally ascribed to him. The Clinic achieved its primary objective of eliminating all other known candidates and thus confirming that Shakespeare wrote them. Two general methods of analysis were applied to whole plays and variable-sized large texts: Discrete Composite Analysis and Continuous Composite Analysis.. The first uses univariate analysis to determine acceptance or rejection of forty-eight stylometric tests for each text. The second uses a multi-dimensional composite mean for Shakespeare derived from all forty-eight in order to determine acceptance or rejection for each text. This article notes the omission of Discrete Analysis to take into consideration statistical dependencies between the forty-eight tests, the partly arbitrary ‘handfitting’ of acceptance–rejection boundaries for each of the forty-eight tests, the failure to take into full account the factor of chronology, and the absence of discussion of the part played by prior probabilities as to existing beliefs concerning attribution. By this last point, I mean the role played by the existing traditional consensus as to Shakespeare attribution, prior to linguistic analysis. For Continuous Analysis, it is noted that the stated probabilities are not true probabilities as acknowledged, and that the resulting acceptance–rejection levels for them are calibrated in line with prior beliefs. Principal component analysis is shown to give improved results in dealing with co-authored Shakespeare plays, Henry VIII, Timon of Athens, and Pericles. This does not invalidate the overall aim of the Shakespeare Clinic.

Evaluating methods for computer-assisted stemmatology using artificial benchmark data sets

Roos, T., Heikkila, T. — Mon, 26 Oct 2009 21:31:55 PDT

Given a collection of imperfect copies of a textual document, the aim of stemmatology is to reconstruct the history of the text, indicating for each variant the source text from it was copied. We describe an experiment involving three artificial benchmark data sets to which a number of computer-assisted stemmatology methods were applied. Contrary to earlier similar experiments, we propose and use a numerical criterion to evaluate all the solutions. Moreover, our primary data set is significantly larger than used before. The results suggest the superiority of two computer-assisted methods amongst those tested: the maximum parsimony method implemented in the PAUP* software package and a related compression-based method we have proposed in earlier work.

Lexical Diversity in a Literary Genre: A Corpus Study of the Rgveda

Sotov, A. — Mon, 26 Oct 2009 21:31:55 PDT

This research¹ evaluates the extent to which lexical diversity, measured by frequent content words, hapax legomena, and type-token ratios (TTRs), is dependent on three features of the genre of the oral Indo-Aryan cultic poetry represented by the literary corpus of the Rgveda (ca. 165,000 tokens): characteristic choice of subject matter, usage of refrains, and the attribution of hymns to distinct poetic collectives. Analysis of 255 texts of 200 tokens showed that hymns on popular topics and where refrains were attested have a significantly higher rate of high-frequency content words and a lower ratio of once-occurring types. A higher TTR is observed in the hymns of specific family origin. Complexity of genre can be interpreted as a result of different discourse strategies of the poets. Overall, conservative mythological texts are characterized by regularity in word usage. Occurrence of content words, in the entire corpus, with lexemes denoting ‘deities’ on the one side and ‘nature’ on the other is accounted for by the factor of semantics, which deals with the structure of narrative.

Dictionary generation for less-frequent language pairs using WordNet

Varga, I., Yokoyama, S., Hashimoto, C. — Mon, 26 Oct 2009 21:31:55 PDT

Bilingual dictionaries are vital resources in many areas of natural language processing. Numerous methods of machine translation require bilingual dictionaries of large coverage, but less-frequent language pairs rarely have any digitalized resources of such kind. Since the need for these resources is increasing, but the human resources are scarce for less represented languages, efficient automatized methods are imperative. This article presents a fully automated, robust intermediate language-based bilingual dictionary generation method that uses the WordNet of the intermediate language to build a new bilingual dictionary. We propose the usage of WordNet in order to increase accuracy; we also introduce a bidirectional selection method with a flexible threshold to maximize recall. The evaluations showed 79% accuracy and 51% weighted recall, outperforming representative pivot language-based methods. A dictionary generated with this method will still need manual post-editing, but the improved recall and precision decrease the work of human correctors.

An exercise in non-ideal authorship attribution: the mysterious Maria Ward

Hoover, D. L., Hess, S. — Mon, 26 Oct 2009 21:31:55 PDT

The dangers of computational approaches to authorship attribution in the absence of an adequate set of training texts for the claimant authors are well known. This study aims to show, however, that significant progress can be made even where conditions are quite problematic. We investigate a difficult authorship question involving three texts, ostensibly by three authors, each of whom wrote nothing else. Only one of the texts can be unquestionably ascribed to a known author, and this author has been suggested as the true author of one of the two remaining texts. We investigate these three texts, along with similar texts by other authors, using cluster analysis, Delta analysis, t-testing, and PCA. We also create simulations of our authorship problem using sets of three texts of known authorship by one, two, and three authors. We test these sets using correct and incorrect assumptions of authorial difference, and then compare the results with analyses of our three texts based on the same range of assumptions. By combining information from all of these tests, we achieve what we believe is a persuasive, if not conclusive, solution to a significant and long-standing question concerning the authorship of Maria Warda's violently anti-Mormon Female Life Among the Mormons. At the same time, we demonstrate methods for making progress in cases where conditions are less than ideal.

Digital Images for the Information Professional.: Melissa M. Terras.

van Horik, R. — Mon, 26 Oct 2009 21:31:55 PDT

Cyberculture, Cyborgs and Science Fiction: Consciousness and the Posthuman. * William S. Haney.

Tripp, M. — Mon, 26 Oct 2009 21:31:55 PDT