Artificial Texts Research Articles

On Artificial and Post-artificial Texts: Machine Learning and the Reader's Expectations of Literary and Non-literary Writing

Abstract With the advent of ChatGPT and other large language models, the number of artificial texts we encounter on a daily basis is about to increase substantially. This essay asks how this new textual situation may influence what one can call the “standard expectation of unknown texts,” which has always included the assumption that any text is the work of a human being. As more and more artificial writing begins to circulate, the essay argues, this standard expectation will shift—first, from the immediate assumption of human authorship to, second, a creeping doubt: did a machine write this? In the wake of what Matthew Kirschenbaum has called the “textpocalypse,” however, this state cannot be permanent. The author suggests that after this second transitional period, one may suspend the question of origins and, third, take on a post-artificial stance. One would then focus only on what a text says, not on who wrote it; post-artificial writing would be read with an agnostic attitude about its origins. This essay explores the implications of such post-artificiality by looking back to the early days of text synthesis, considering the limitations of aesthetic Turing tests, and indulging in reasoned speculation about the future of literary and nonliterary text generation.

Artificial and Post-Artificial Texts

In contrast to the times of Höllerer and Zemanek, we are now truly on the threshold of being surrounded by texts that are entirely artificial—while at the same time we continue to merge with our language technologies in our own writing, so that our text production is increasingly supported, extended, and partially taken over by assistance systems. Therefore, I want to discuss—quite speculatively, but always with an eye on the state of the art—two questions: first, what happens when we are confronted with artificial texts in addition to natural ones? How do we read a text that we can no longer be sure was not written by an AI? And second, what direction might this development take if, at some point, the distinction between natural and artificial itself becomes obsolete, so that we no longer even ask about it and instead read post-artificial texts?

Шляхи вдосконалення комунікативно-орієнтованого усного мовлення студентів англійською мовою (за автентичними матеріалами)

The presented article briefly examines the problems that first and second course students confront while speaking. The article emphasises that English has become a fashionable language to learn in Azerbaijan in the past decades. First course students know grammar rules well enough and they possess rich vocabulary also. Yet most students (not all of them, of course) are unable to take part in simple conversations either with a teacher or with each other. The lack of self-confidence in students lead them to express their view points in Azerbaijani. The presented paper emphasizes the importance of the use of authentic materials to improve first and second course students’ communication-oriented skills in English. It is indicated in the paper that according to the requirements of the communicative approach, the usage of authentic texts and language materials in the English language lesson are better than the presentation of artificial texts. Speaking is the productive use of language and improving students’ speaking skills which is considered to be the most difficult part of language teaching and is directly connected with motivation, communicative exercises, listening activities. In the article, it is stated that habit forming provides an excellent incentive so speak English fluently. With regard to the selection of authentic materials, the interests of the students should be taken into consideration. Those authentic materials should be presented to the students that meet their needs and requirements. In addition, the article outlines the types of authentic materials. Teachers are encouraged to use all the types authentic materials in class. Authentic reading and listening materials used in English classes and the activities to be carried out on them are also grouped. The paper points out that authentic materials have become an important element of the English language lessons noting that students should use them in English more often in the contexts in which real oral speech occurs naturally than in the context in which artificial speech occurs. Therefore, it is important to use authentic materials in the process of teaching English

Natural Language Statistical Features of LSTM-Generated Texts.

Long short-term memory (LSTM) networks have recently shown remarkable performance in several tasks that are dealing with natural language generation, such as image captioning or poetry composition. Yet, only few works have analyzed text generated by LSTMs in order to quantitatively evaluate to which extent such artificial texts resemble those generated by humans. We compared the statistical structure of LSTM-generated language to that of written natural language, and to those produced by Markov models of various orders. In particular, we characterized the statistical structure of language by assessing word-frequency statistics, long-range correlations, and entropy measures. Our main finding is that while both LSTM- and Markov-generated texts can exhibit features similar to real ones in their word-frequency statistics and entropy measures, LSTM-texts are shown to reproduce long-range correlations at scales comparable to those found in natural language. Moreover, for LSTM networks, a temperature-like parameter controlling the generation process shows an optimal value-for which the produced texts are closest to real language-consistent across different statistical features investigated.

Keepin' it real: Linguistic models of authenticity judgments for artificially generated rap lyrics.

Through advances in neural language modeling, it has become possible to generate artificial texts in a variety of genres and styles. While the semantic coherence of such texts should not be over-estimated, the grammatical correctness and stylistic qualities of these artificial texts are at times remarkably convincing. In this paper, we report a study into crowd-sourced authenticity judgments for such artificially generated texts. As a case study, we have turned to rap lyrics, an established sub-genre of present-day popular music, known for its explicit content and unique rhythmical delivery of lyrics. The empirical basis of our study is an experiment carried out in the context of a large, mainstream contemporary music festival in the Netherlands. Apart from more generic factors, we model a diverse set of linguistic characteristics of the input that might have functioned as authenticity cues. It is shown that participants are only marginally capable of distinguishing between authentic and generated materials. By scrutinizing the linguistic features that influence the participants’ authenticity judgments, it is shown that linguistic properties such as ‘syntactic complexity’, ‘lexical diversity’ and ‘rhyme density’ add to the user’s perception of texts being authentic. This research contributes to the improvement of the quality and credibility of generated text. Additionally, it enhances our understanding of the perception of authentic and artificial art.

THE DICTIONARY STRUCTURE FOR EFFECTIVE WORD SEARCH

In the paper some issues connected with indexing documents in the Polish language are discussed. Algorithms for stemming and part of speech tagging, important in the process of text analysis and indexing are shortly described. Next their suitability to the Polish language, which has a very extensive inflection, is discussed. The usefulness for stemming and part of speech tagging of large dictionaries with inflected forms, like WordNet and open-source dictionary of Polish language is also described. Two dictionary structures enabling effective word searching are presented. In the final part, some tests of implemented two dictionary structures are described. Tests were made on the six actual and three crafted artificial texts. At the end conclusions of performed tests are formulated.

Method of the Artificial Texts Identification based on the Calculation of the Belonging Measure to the Invariants

The work is devoted to the identification of texts generated automatically (artificially) with the use of software algorithms. This is an important and topical issue, because such texts are being widely spread on the Internet. Created «copies» of the web pages are used to attract readers to online resources as well as to disseminate a large number of unique copies of pages with content specific orientation. This article describes the features of determining the origin of the text by the example of working on texts generated by synonymization as the most common method of generating artificial web content. The author provides an invariant of artificial texts as a set of the values of text characteristics, which allows classification of texts according to the process of their creation. The article proposes a method of the artificial texts identification based on the calculation of the belonging measure to the invariants, which allows making a decision about the origin of the text. The article also presents values obtained from the experiments on identifying artificial texts.

Finding and Characterizing Repeats in Plant Genomes.

Plant genomes contain a particularly high proportion of repeated structures of various types. This chapter proposes a guided tour of available software that can help biologists to look for these repeats and check some hypothetical models intended to characterize their structures. Since transposable elements are a major source of repeats in plants, many methods have been used or developed for this large class of sequences. They are representative of the range of tools available for other classes of repeats and we have provided a whole section on this topic as well as a selection of the main existing software. In order to better understand how they work and how repeats may be efficiently found in genomes, it is necessary to look at the technical issues involved in the large-scale search of these structures. Indeed, it may be hard to keep up with the profusion of proposals in this dynamic field and the rest of the chapter is devoted to the foundations of the search for repeats and more complex patterns. The second section introduces the key concepts that are useful for understanding the current state of the art in playing with words, applied to genomic sequences. This can be seen as the first stage of a very general approach called linguistic analysis that is interested in the analysis of natural or artificial texts. Words, the lexical level, correspond to simple repeated entities in texts or strings. In fact, biologists need to represent more complex entities where a repeat family is built on more abstract structures, including direct or inverted small repeats, motifs, composition constraints as well as ordering and distance constraints between these elementary blocks. In terms of linguistics, this corresponds to the syntactic level of a language. The last section introduces concepts and practical tools that can be used to reach this syntactic level in biological sequence analysis.

Filtering artificial texts with statistical machine learning techniques

Fake content is flourishing on the Internet, ranging from basic random word salads to web scraping. Most of this fake content is generated for the purpose of nourishing fake web sites aimed at biasing search engine indexes: at the scale of a search engine, using automatically generated texts render such sites harder to detect than using copies of existing pages. In this paper, we present three methods aimed at distinguishing natural texts from artificially generated ones: the first method uses basic lexicometric features, the second one uses standard language models and the third one is based on a relative entropy measure which captures short range dependencies between words. Our experiments show that lexicometric features and language models are efficient to detect most generated texts, but fail to detect texts that are generated with high order Markov models. By comparison our relative entropy scoring algorithm, especially when trained on a large corpus, allows us to detect these "hard" text generators with a high degree of accuracy.

A Critical Stance in Language Education: A Reply to Alan Waters

In his recent Forum article on ideology in applied linguistics, Alan Waters (2009) takes up arms against what he perceives as a damaging critical tendency. Ideas about language teaching, he claims, are promoted (e.g. learner centredness) or proscribed (e.g. artificial texts) ‘on the basis of ideological belief rather than pedagogical value’. By making this distinction, Waters fails to recognize that the relationship between ideology and pedagogy is inextricable: ideologies are constructed, reproduced, and made manifest in social practices, such as language teaching. Furthermore, in certain language learning and teaching situations, an uncritical stance—one which views language teaching as a neutral and value-free activity—is incompatible with students’ language learning and broader life concerns. In this response article, I maintain that in such contexts, the field of applied linguistics has an obligation to mediate in a way that is both critical and pedagogically relevant.

Dictionary-based methods for information extraction

In this paper, we present a general method for information extraction that exploits the features of data compression techniques. We first define and focus our attention on the so-called dictionary of a sequence. Dictionaries are intrinsically interesting and a study of their features can be of great usefulness to investigate the properties of the sequences they have been extracted from e.g. DNA strings. We then describe a procedure of string comparison between dictionary-created sequences (or artificial texts) that gives very good results in several contexts. We finally present some results on self-consistent classification problems.

Computing the Type Token Relation From the A Priori Distribution of Types

For homogeneous texts, the dependence of the vocabulary size y(x) on text length x is completely determined by the distribution D of the type probabilities. The function y(x) is derived from a simple difference equation. This solution is checked with artificial texts as well as with several German, English and Italian texts. For the natural texts, the distributions D are approximated by an interpolation equation: wj =const/jp. This expression is adjusted by a weighted least square fit for each text separately. The values obtained are in the range 0.36≤p≤0.81.

Numerical Analysis of Word Frequencies in Artificial and Natural Language Texts

We perform a numerical study of the statistical properties of natural texts written in English and of two types of artificial texts. As statistical tools we use the conventional Zipf analysis of the distribution of words and the inverse Zipf analysis of the distribution of frequencies of words, the analysis of vocabulary growth, the Shannon entropy and a quantity which is a nonlinear function of frequencies of words, the frequency "entropy". Our numerical results, obtained by investigation of eight complete books and sixteen related artificial texts, suggest that, among these analyses, the analysis of vocabulary growth shows the most striking difference between natural and artificial texts. Our results also suggest that, among these analyses, those who give a greater weight to low frequency words succeed better in distinguishing between natural and artificial texts. The inverse Zipf analysis seems to succeed better than the conventional Zipf analysis and the frequency "entropy" better than the usual word entropy. By studying the scaling behavior of both entropies as a function of the total number of words T of the investigated text, we find that the word relative entropy scales with the same functional form for both natural and artificial texts but with a different parameter, while the frequency relative "entropy" decreases monotonically with T for the artificial texts while having a minimum at T≈104 for the natural texts.

"We Shall Not Speak of It": Nazism and the Holocaust in the Elementary College German Course

Adorno's dictum that it is impossible adequately to the is certainly as compelling as Elie Wiesel's admonition that we must never forget.1 Yet too often German teachers-and authors of German language textbooks-respond to the difficulties of historical representation rather than to the duty of remembrance. There is a genuine dilemma in attempting both tasks simultaneously; but playing it safe by failing to mention (or treating only superficially) Nazism, the Second World War, and the Holocaust in particular is a far greater failing than the unavoidable pitfall of oversimplification. It is my contention that authentic materials can and should be introduced in the elementary curriculum in order to redress the lacunae and distortions evident in even the best introductory textbooks. Since there is an abundance of WWII/ Holocaust material available to intermediate and advanced readers,2 I will restrict myself to the elementary level. Drawing on sample texts for illustration, as well as invaluable input from colleagues and students, I hope to reverse the widespread assumption that language skills at the beginning level are too limited to approach a topic as complex as the Holocaust. Since curricular reform of any type depends upon the cooperation and commitment of instructors, I focus in Part 1 on teacher attitudes toward the inclusion of materials on Nazism and the Holocaust in the elementary course. Those already convinced of the merits of such a project may wish to proceed directly to Part 2, which considers the major introductory textbooks from the perspective of their treatment of the Third Reich. Part 3 (along with the appendices) is intended to compensate for the dearth of documentary materials available in these first year books: Here I include an original interview with the Jewish German emigres Ruth and Sam Dreifus (1993), Ernst Toller's open letter to Joseph Goebbels (1933), and a short piece of testimony given in 1946 by the commandant of Auschwitz, Rudolph H68. These texts open up perspectives and raise issues that are often passed over in silence in the introductory German textbook. Additionally, they represent documents which are authen tic both linguistically and in terms of historical accuracy-a welcome improvement over those overburdened artificial texts intended both to demonstrate some grammatical lesson as well as treat a major historical era.

The empirical study of figurative language in literature

This article addresses the study of figurative language in literary texts. The psychological research on eight forms of figurative language is reviewed. The majority of this research has utilized short artificial texts which may not reflect the psychological processes involved in comprehending figurative language in longer, more literary texts. Other limitations of figurative language research include: (1) a bias towards metaphor research at the expense of other common forms of figurative language; (2) insufficient examination of how authorial intent, genre, and reader knowledge can affect figurative language comprehension; (3) a bias towards research on comprehension at the expense of production; and (4) insufficient understanding of how discourse goals can be fulfilled by figurative language. It is suggested that attention to these four issues could greatly improve our understanding of figurative language in ecologically valid contexts.