@article {bib_87, title = {Man or Machine: Evaluations of Human and Machine-Generated Movie Reviews}, year = {2023}, address = {San Francisco, CA}, abstract = {Recent advances in generative language models, such as ChatGPT have demonstrated an uncanny ability to produce texts that appear to be comparable to those produced by humans. Nevertheless, machine generated texts differ from those produced by humans in important aspects, such as routinely including references to nonexistent sources. In this paper, we use both psycholinguistic measurements and participant responses to compare texts generated by machine with equivalent texts generated by humans. Our analysis demonstrates some of the ways in which machine-generated texts differ from human-generated ones in both style (e.g., increased use of positive connectives) and content (e.g., increased confidence). We also note multiple ways in which texts generated by these models are similar to those generated by humans (e.g., their use of emotion words). We believe this research provides insights that can be useful to understanding how language is generated by both humans and machines.}, author = {Eyal Sagi and Hadar Y Jabotinsky} } @article {bib_86, title = {Supreme Court Oral Arguments by Amici Curiae: A comparative analysis}, year = {2023}, abstract = {The legal system is concerned with interpreting and applying existing laws to real world circumstances. In the U.S., the final arbiters of these laws are the nine justices of the U.S. Supreme Court. The cases presented before the court generally involve disputes between two parties, the petitioner and the respondent. However, in some cases third parties provide additional information to the court as amici curiae ({\textquoteleft}friends of the court{\textquoteright}). The present study uses corpus statistics and moral foundation theory to explore and compare the types of arguments brought up by the different parties during oral arguments over the past 6 decades, with a focus on amicus curiae arguments and how they affect the results of the case. Overall, the results show that both amici and respondents rely on moral arguments more frequently than petitioners do. }, author = {Eyal Sagi} } @article {bib_81, title = {Taming big data: Applying the experimental method to naturalistic data sets}, journal = {Behavior Research Methods}, volume = {51}, year = {2019}, pages = {1619-1635}, abstract = {Psychological researchers have traditionally focused on lab-based experiments to test their theories and hypotheses. Although the lab provides excellent facilities for controlled testing, some questions are best explored by collecting information that is difficult to obtain in the lab. The vast amounts of data now available to researchers can be a valuable resource in this respect. By incorporating this new realm of data and translating it into traditional laboratory methods, we can expand the reach of the lab into the wilderness of human society. This study demonstrates how the troves of linguistic data generated by humans can be used to test theories about cognition and representation. It also suggests how similar interpretations can be made of other research in cognition. The first case tests a long-standing prediction of Gentner{\textquoteright}s natural partition hypothesis: that verb meaning is more subject to change due to the textual context in which it appears than is the meaning of nouns. Within a diachronic corpus, verbs and other relational words indeed showed more evidence of semantic change than did concrete nouns. In the second case, corpus statistics were employed to empirically support the existence of phonesthemes{\textemdash}nonmorphemic units of sound that are associated with aspects of meaning. A third study also supported this measure, by demonstrating that it corresponds with performance in a lab experiment. Neither of these questions can be adequately explored without the use of big data in the form of linguistic corpora.}, keywords = {Big data, Corpus statistics, Phonesthemes, Representation, Semantic change}, doi = {10.3758/s13428-018-1185-6}, url = {http://link.springer.com/10.3758/s13428-018-1185-6}, author = {Sagi, Eyal} } @article {bib_84, title = {Cognition and Emotion in Narratives of Redemption: An Automated Analysis}, year = {2018}, address = {New Orleans, LA}, abstract = {Redemptive narratives are stories of challenge, failure, or adversity that in some way acknowledge the goodness or personal growth that came of the recounted difficult event. In this paper we use a corpus-statistic based approach to explore the role of cognition and emotion in these narrative arcs. In particular, we trace the shift from negative to positive sentiment (a change in the emotional valence) and vice to virtue (evidence of cognitive, moral processing) within the narrative and compare these with similar narratives that do not present a redemptive arc. Our results suggest that the shift to goodness and growth that is at the core of redemptive narratives is driven by prior cognitive processes more so than emotional ones. We believe this type of analysis can also be used to trace and classify similar narrative arcs and assist with the coding of autobiographical narratives in general.}, author = {Sagi, Eyal and Jones, Brady K.} } @conference {bib_82, title = {Cognition and Emotion in Narratives of Redemption: An Automated Analysis}, booktitle = {Proceedings of the 40th Annual Conference of the Cognitive Science Society}, year = {2018}, pages = {2382-2387}, publisher = {Cognitive Science Society}, organization = {Cognitive Science Society}, address = {Austin, TX}, abstract = {Redemptive narratives are stories of challenge, failure, or adversity that in some way acknowledge the goodness or personal growth that came of the recounted difficult event. In this paper we use a corpus-statistic based approach to explore the role of cognition and emotion in these narrative arcs. In particular, we trace the shift from negative to positive sentiment (a change in the emotional valence) and vice to virtue (evidence of cognitive, moral processing) within the narrative. Our results suggest that cognitive processes, more than emotion, drive the shift to goodness and growth that is at the core of redemptive narratives. We discuss the implications of these results to both narrative psychology and cognitive psychology.}, author = {Sagi, Eyal and Jones, Brady K.}, editor = {Rogers, T. T. and Rau, M. and Zhu, X. and Kalish, C. W.} } @article {bib_72, title = {Developing a new method for psychological investigation using text as data}, journal = {SAGE Research Methods Cases}, volume = {2}, year = {2018}, abstract = {Over the past few decades, the study of psychology has been undergoing a methodological transformation. The increasing availability and quantity of real-world big data has prompted researchers to look for new ways to use this data to gain psychological insights. In this case study, I trace the process of developing a new method for testing psychological theories using corpora. Instead of bringing participants to the lab, I analyze statistical patterns of co-occurrence in naturally-occurring texts obtained from a variety of sources, including political speeches, literature, and the internet. The basic assumption I make is that these patterns reflect the representations and cognitive processes of their author. I present 3 different applications of the method and use them to describe how such data can be analyzed and used to answer a range of questions in psychology and social science. The first application examines the linguistic question of the relationship between word form and meaning. The second application identifies cognitive frames as they are found in text. The final application uses texts to measure the style of moral reasoning individuals apply in particular contexts. This case study provides insight into the process of developing new methodologies for hypothesis testing, as well as demonstrating how to formulate hypotheses that can be tested using corpora. In addition, several key pitfalls in the process of adapting statistical methods to new uses are identified and discussed.}, isbn = {9781526442604}, doi = {10.4135/9781526442604}, author = {Eyal Sagi} } @article {bib_80, title = {Embodied concept mapping: Blending structure-mapping and embodiment theories}, journal = {Pragmatics \& Cognition}, volume = {24}, year = {2018}, pages = {164-185}, abstract = {Metaphors are cognitive and linguistic tools that allow reasoning. They enable the understanding of abstract domains via elements borrowed from concrete ones. The underlying mechanism in metaphorical mapping is the manipulation of concepts. This article proposes another view on what concepts are and their role in metaphor and reasoning. That is, based on current neuroscientific and behavioural evidence, it is argued that concepts are grounded in perceptual and motor experience with physical and social environments. This definition of concepts is then embedded in the Structure-Mapping Theory (SMT), a model for metaphorical processing and reasoning. The blended view of structure-mapping and embodied cognition offers an insight into the processes through which the target domain of a metaphor is embodied or realised in terms of its base domain. The implications of the proposed embodied SMT model are then discussed and future topics of investigation are outlined.}, issn = {0929-0907}, doi = {10.1075/pc.17013.mar}, url = {http://www.jbe-platform.com/content/journals/10.1075/pc.17013.mar}, author = {Marmolejo-Ramos, Fernando and Khatin-Zadeh, Omid and Yazdani-Fazlabadi, Babak and Tirado, Carlos and Eyal Sagi} } @article {bib_85, title = {Embodied concept mapping: redefining concepts}, year = {2018}, address = {Sydney, Australia}, author = {Marmolejo-Ramos, Fernando and Khatin-Zadeh, Omid and Yazdani-Fazlabadi, Babak and Tirado, Carlos and Sagi, Eyal} } @article {bib_83, title = {Language Dynamics in Supreme Court Oral Arguments}, year = {2018}, address = {Madison, WI}, abstract = {During conversations, it is not uncommon to notice that interlocutors start using similar words and grammatical structures. This alignment of language use is thought to help comprehension, as well as lead to an alignment in underlying representations. In the context of negotiations, the degree to which parties exhibit such an alignment can indicate the likelihood of reaching an agreement. The present study expands this notion to the courts and uses corpus statistics to examine the relationship between the alignment of semantic content during oral arguments and the decision reached by the justices. The analysis demonstrates that lawyers that align their language with that of the justices are more likely to have a decision in their favor. Additionally, as befits the power dynamic between justices and lawyers, lawyers are more likely to align their language with the justices than the justices are to align their language to that of the lawyers.}, author = {Sagi, Eyal} } @conference {bib_73, title = {Is it fair? Textual effects on the salience of moral foundations}, booktitle = {Proceedings of the 39th Annual Conference of the Cognitive Science Society}, year = {2017}, pages = {3028-3032}, abstract = {Many of the important decisions we make have moral implications. Moral Foundations Theory (Haidt \& Joseph, 2004) identifies 5 distinct styles of moral reasoning that may be applied to such decisions. This paper explores how reading text that emphasizes one of these styles might affect our reasoning. After participants read a series of tweets that emphasized the Fairness/Cheating foundation they exhibited an increased reliance on this style compared to when they read tweets emphasizing the Care/Harm foundation. This affected participants{\textquoteright} answers to a questionnaire designed to measure the perceived importance of the different foundations, as well as in their rating of the foundations evident in other tweets. Interestingly, this effect was short lived and was not observed for the Care/Harm foundation. These results suggest that exposure to the moral reasoning of others might temporarily influence what moral arguments we are likely to accept and employ.}, author = {Eyal Sagi}, editor = {Gunzelmann, G. and Howes, A. and Tenbrink, T. and Davelaar, E. J.} } @article {bib_74, title = {Language Dynamics in Supreme Court Oral Arguments}, year = {2017}, type = {Poster}, address = {Vancouver, BC}, abstract = {During conversations, it is not uncommon to notice that interlocutors start using similar words and grammatical structures. This alignment of language use is thought to help comprehension, as well as indicate an alignment in underlying representations. It has also been shown that, in the context of negotiations, the degree to which parties exhibit such an alignment is an indication of the likelihood of reaching an agreement. The present study expands this notion to the courts and uses corpus statistics to examine the relationship between the alignment of semantic content during oral arguments and the decision reached by the justices. The analysis demonstrates that lawyers that align their language with that of the justices are more likely to have a decision in their favor. Additionally, as befits the power dynamic between justices and lawyers, lawyers are more likely to align their language with the justices than the justices are to align their language to that of the lawyers.}, author = {Eyal Sagi} } @article {bib_70, title = {Language Use and Coalition Formation in Multiparty Negotiations}, journal = {Cognitive Science}, volume = {41}, year = {2017}, pages = {259-271}, abstract = {The alignment of bargaining positions is crucial to a successful negotiation. Prior research has shown that similarity in language use is indicative of the conceptual alignment of interlocutors. We use latent semantic analysis to explore how the similarity of language use between negotiating parties develops over the course of a three-party negotiation. Results show that parties that reach an agreement show a gradual increase in language similarity over the course of the negotiation. Furthermore, reaching the most financially efficient outcome is dependent on similarity in language use between the parties that have the most to gain from such an outcome.}, keywords = {Coalition formation, Latent semantic analysis, Linguistic entrainment, Negotiation, Psycholinguistics}, issn = {0364-0213}, doi = {10.1111/cogs.12325}, author = {Eyal Sagi and Daniel Diermeier} } @article {bib_77, title = {Moral Rhetoric in the Basel Accords: A Quantitative Analysis}, year = {2016}, type = {Poster}, address = {Philadelphia, PA}, author = {Eyal Sagi and Hadar Y Jabotinsky} } @article {bib_75, title = {Moral Rhetoric in the Basel Accords: A Quantitative Analysis}, year = {2016}, type = {Oral presentation}, address = {Durham, NC}, author = {Hadar Y Jabotinsky and Eyal Sagi} } @article {bib_78, title = {Moral Rhetoric in the Basel Accords: A Quantitative Analysis}, year = {2016}, address = {Amsterdam, Netherlands}, author = {Hadar Y Jabotinsky and Eyal Sagi} } @article {bib_79, title = {Moral Rhetoric in the Basel Accords: A Quantitative Analysis}, year = {2016}, type = {Oral presentation}, address = {Jerusalem, Israel}, author = {Hadar Y Jabotinsky and Eyal Sagi} } @article {bib_76, title = {Priming Moral Reasoning Using Text}, year = {2016}, type = {Poster}, address = {Boston, MA}, abstract = {Many of the important decisions we make have moral implications. Moral Foundations Theory (Haidt \& Joseph, 2004) identifies 5 distinct styles of moral reasoning that may be applied to such decisions. A study explores how reading text that emphasizes one of these styles might affect our reasoning. In particular, after participants read a series of tweets that emphasized the Fairness/Cheating foundation they exhibited an increase reliance on this style compared to when they read tweets emphasizing the Care/Harm foundation. This affected participants{\textquoteright} answers to a questionnaire designed to measure the perceived importance of the different foundations, as well as in their rating of the foundations evident in other tweets. Interestingly, this effect was short lived and was not observed for the Care/Harm foundation. These results suggest that exposure to the moral reasoning of others might temporarily influence what moral arguments we are likely to employ.}, author = {Eyal Sagi} } @article {bib_71, title = {Purity Homophily in Social Networks.}, journal = {Journal of Experimental Psychology: General}, volume = {145}, year = {2016}, pages = {366-375}, abstract = {Does sharing moral values encourage people to connect and form communities? The importance of moral homophily (love of same) has been recognized by social scientists, but the types of moral similarities that drive this phenomenon are still unknown. Using both large-scale, observational social-media analyses and behavioral lab experiments, the authors investigated which types of moral similarities influence tie formations. Analysis of a corpus of over 700,000 tweets revealed that the distance between 2 people in a social-network can be predicted based on differences in the moral purity content{\textemdash}but not other moral content{\textemdash}of their messages. The authors replicated this finding by experimentally manipulating perceived moral difference (Study 2) and similarity (Study 3) in the lab and demonstrating that purity differences play a significant role in social distancing. These results indicate that social network processes reflect moral selection, and both online and offline differences in moral purity concerns are particularly predictive of social distance. This research is an attempt to study morality indirectly using an observational big-data study complemented with 2 confirmatory behavioral experiments carried out using traditional social-psychology methodology.}, doi = {10.1037/xge0000139}, url = {http://dx.doi.org/10.1037/xge0000139}, author = {Morteza Dehghani and Kate Johnson and Joe Hoover and Eyal Sagi and Justin Garten and Niki Jitendra Parmar and Stephen Vaisey and Rumen Iliev and Jesse Graham} } @article {56, title = {Automated Text Analysis in Psychology: Methods, Applications and Future Developments}, journal = {Language and Cognition}, volume = {7}, year = {2015}, pages = {265-290}, abstract = {Recent years have seen rapid developments in automated text analysis methods focused on measuring psychological and demographic properties. While this development has mainly been driven by computer scientists and computational linguists, such methods can be of great value for social scientists in general, and for psychologists in particular. In this paper, we review some of the most popular approaches to automated text analysis from the perspective of social scientists, and give examples of their applications in different theoretical domains. After describing some of the pros and cons of these methods, we speculate about future methodological developments, and how they might change social sciences. We conclude that despite the fact that current methods have many disadvantages and pitfalls compared to more traditional methods of data collection, the constant increase of computational power and the wide availability of textual data will inevitably make automated text analysis a common tool for psychologists.}, doi = {10.1017/langcog.2014.30}, author = {Rumen Iliev and Morteza Dehghani and Eyal Sagi} } @article {bib_68, title = {The Influence of Causal Information on Pronoun Disambiguation}, year = {2015}, type = {Poster}, address = {Chicago, IL}, abstract = {The disambiguation of pronouns is a complicated process that has been shown to be influenced by many syntactic and grammatical factors. Here I present evidence that non-linguistic knowledge, specifically causal information, informs these processes. For example, in the sentence pair {\textquoteleft}John accused Mark of stealing a car. He called the police{\textquoteright}, the antecedent of {\textquoteleft}he{\textquoteright} is more likely to be John than Mark because of the perceived causal link between the accusation and calling the police. A series of experiments explored the implication of this link between causality and pronoun disambiguation. The results demonstrated that the underlying process is similar to that used to track the identities of individuals and objects over time. This suggests that the process of pronoun disambiguation makes use general cognitive processes in addition to psycholinguistic ones.}, author = {Eyal Sagi} } @article {bib_66, title = {Moral Rhetoric in the Basel Accords: A Quantitative Analysis}, year = {2015}, type = {Talk}, address = {Vienna, Austria}, author = {Hadar Y Jabotinsky and Eyal Sagi} } @article {bib_65, title = {Moral Rhetoric in the Basel Accords: A Quantitative Analysis}, year = {2015}, type = {Talk}, address = {Santo Domingo, Dominican Republic}, author = {Hadar Y Jabotinsky and Eyal Sagi} } @conference {bib_67, title = {The Moral Rhetoric of Climate Change}, booktitle = {Proceedings of the 37th Annual Conference of the Cognitive Science Society}, year = {2015}, pages = {2063-2068}, abstract = {Communication in the media about climate change in the United States is complicated by the intensely ideologically polarized state of the debate surrounding the issue; moral rhetoric is an important dimension of how ideology is communicated. In this study we examined how moral rhetoric regarding this issue differs on the basis of a publication{\textquoteright}s perceived ideological lean. To address the question, we built a corpus from a diverse group of online news media that were rated for their perceived ideological lean. Using Latent Semantic Analysis we calculated the average loading for the five moral domains identified in Haidt{\textquoteright}s Moral Foundations Theory (Haidt \& Joseph, 2004) on the terms "climate change" and "global warming." We found that there were higher moral loadings overall for "climate change" with a greater difference seen among the more progressive media.}, author = {Eyal Sagi and Timothy M. Gann and Teenie Matlock}, editor = {D. C. Noelle and Rick Dale and A. S. Warlaumont and J. Yoshimi and Teenie Matlock and C. D. Jennings and P. P. Maglio} } @article {55, title = {Identity, Causality, and Pronoun Ambiguity}, journal = {Topics in Cognitive Science}, volume = {6}, year = {2014}, pages = {663-680}, abstract = {This article looks at the way people determine the antecedent of a pronoun in sentence pairs, such as: Albert invited Ron to dinner. He spent hours cleaning the house. The experiment reported here is motivated by the idea that such judgments depend on reasoning about identity (e.g., the identity of the he who cleaned the house). Because the identity of an individual over time depends on the causal-historical path connecting the stages of the individual, the correct antecedent will also depend on causal connections. The experiment varied how likely it is that the event of the first sentence (e.g., the invitation) would cause the event of the second (the house cleaning) for each of the two individuals (the likelihood that if Albert invited Ron to dinner, this would cause Albert to clean the house, vs. cause Ron to clean the house). Decisions about the antecedent followed causal likelihood. A mathematical model of causal identity accounted for most of the key aspects of the data from the individual sentence pairs.}, doi = {10.1111/tops.12105}, author = {Eyal Sagi and Lance J Rips} } @article {bib_1, title = {Measuring Moral Rhetoric in Text}, journal = {Social Science Computer Review}, volume = {32}, year = {2014}, pages = {132-144}, abstract = {In this paper we present a computational text analysis technique for measuring the moral loading of concepts as they are used in a corpus. This method is especially useful for the study of online corpora as it allows for the rapid analysis of moral rhetoric in texts such as blogs and tweets as events unfold. We use Latent Semantic Analysis to compute the semantic similarity between concepts and moral keywords taken from the Moral Foundation Dictionary. This measure of semantic similarity represents the loading of these concepts on the five moral dimensions identified by Moral Foundation Theory. We demonstrate the efficacy of this method using three different concepts and corpora.}, issn = {0894-4393}, doi = {10.1177/0894439313506837}, url = {http://ssc.sagepub.com/cgi/doi/10.1177/0894439313506837}, author = {Eyal Sagi and Morteza Dehghani} } @article {bib_64, title = {Moral Rhetoric in the Basel Accords: A Quantitative Analysis}, year = {2014}, type = {Talk}, address = {Rome, Italy}, author = {Hadar Y Jabotinsky and Eyal Sagi} } @conference {bib_53, title = {Moral Rhetoric in Twitter: A Case Study of the U.S. Federal Shutdown of 2013}, booktitle = {Proceedings of the 36th Annual Conference of the Cognitive Science Society}, year = {2014}, pages = {1347-1352}, abstract = {In this paper we apply a computational text analysis technique used for measuring moral rhetoric in text to analyze the moral loadings of tweets. We focus our analysis on tweets regarding the 2013 federal government shutdown; a topic that was at the forefront of U.S. politics in late 2013. Our results demonstrate that the positions of the members of the two major political parties are mirrored by the positions taken by the Twitter communities that are aligned with them. We also analyze retweeting behavior by examining the differences in the moral loadings of intra-community and inter-community retweets. We find that retweets in our corpus favor rhetoric that enhances the cohesion of the community, and emphasize content over moral rhetoric. We argue that the method proposed in this paper contributes to the general study of moral cognition and social behavior.}, author = {Eyal Sagi and Morteza Dehghani}, editor = {Paul Bello and Marcello Guarini and Marjorie McShane and Brian Scassellati} } @article {bib_69, title = {Semantic Vector Spaces as a Tool for Psychological Investigation}, year = {2014}, address = {Long Beach, CA}, abstract = {The dramatic increase in the availability of language data through the internet over the past two decades, and the corresponding increase in computational power, provide a rich source of information for psychological research. However, analyzing this information and using it to test psychological theories is not always easy or straightforward. In this presentation I describe a method that uses semantic vector spaces (such as those generated by Latent Semantic Analysis and Topic Models) to quantify patterns of word co-occurrence and test hypotheses. Using this method, I explore how differences in the representation of nouns and verbs affect the stability of their meaning, compare the levels and types of moral rhetoric associated with various concepts and debates, and track the convergence of language use as a measure of reaching agreement in a negotiation.}, author = {Eyal Sagi} } @article {bib_2, title = {Identifying Issue Frames in Text}, journal = {PLoS ONE}, volume = {8}, year = {2013}, pages = {e69185}, abstract = {Framing, the effect of context on cognitive processes, is a prominent topic of research in psychology and public opinion research. Research on framing has traditionally relied on controlled experiments and manually annotated document collections. In this paper we present a method that allows for quantifying the relative strengths of competing linguistic frames based on corpus analysis. This method requires little human intervention and can therefore be efficiently applied to large bodies of text. We demonstrate its effectiveness by tracking changes in the framing of terror over time and comparing the framing of abortion by Democrats and Republicans in the U.S.}, doi = {10.1371/journal.pone.0069185}, url = {http://dx.plos.org/10.1371/journal.pone.0069185}, author = {Eyal Sagi and Daniel Diermeier and Stefan Kaufmann} } @conference {jamrozik2013relational, title = {Relational words have high metaphoric potential}, booktitle = {Proceedings of the First Workshop on Metaphor in NLP}, year = {2013}, pages = {21-26}, abstract = {What influences the likelihood that a word will be used metaphorically? We tested whether the likelihood of metaphorical use is related to the relationality of a word{\textquoteright}s meaning. Relational words name relations between entities. We predicted that relational words, such as verbs (e.g., speak) and relational nouns (e.g., marriage) would be more likely to be used metaphorically than words that name entities (e.g., item). In two experiments, we collected expert ratings of metaphoricity for uses of verbs, relational nouns, and entity nouns collected from a corpus search. As predicted, uses of relational words were rated as more metaphorical than uses of entity words. We discuss how these findings could inform NLP models of metaphor.}, url = {http://www.aclweb.org/anthology/W/W13/W13-09.pdf$\#$page=31}, author = {Jamrozik, Anja and Eyal Sagi and Goldwater, Micah and Dedre Gentner}, editor = {E. Shutova and Beata Beigman-Klebanov and J. Testreault and Z. Kozareva} } @conference {53, title = {The Time Course of Language Use in Multiparty Negotiations}, booktitle = {Proceedings of the 35th Annual Conference of the Cognitive Science Society}, year = {2013}, pages = {3343-3347}, author = {Eyal Sagi and Daniel Diermeier}, editor = {M. Knauff and M. Pauen and N. Sebanz and I. Wachsmuth} } @article {49, title = {A quantitative approach to framing in political speech}, year = {2012}, type = {Talk}, address = {Oxford, UK}, url = {http://www.ncrm.ac.uk/TandE/video/RMF2012/filmed.php?id=24bc879}, author = {Eyal Sagi and Daniel Diermeier and Stefan Kaufmann} } @article {48, title = {Representational Form and Metaphorical Word Use}, year = {2012}, type = {Poster}, address = {Sapporo, Japan}, abstract = {How does semantic representation influence the likelihood that a word will be used metaphorically? We explore whether words whose meanings are defined by relations among entities (e.g., marriage, forget), are more likely to be used metaphorically than words whose meanings are defined by features of entities (e.g., bird). Verbs are generally more relational than nouns (Gentner, 1981). Relationality can also distinguish different kinds of nouns: specifically, relational nouns (e.g., marriage) vs. entity nouns (e.g., bird) (Gentner \& Kurtz, 2005; Goldwater, Markman, \& Stilwell, 2011; Markman \& Stilwell, 2001). Prior studies have shown that the meanings of relational words are more mutable across contexts than those of entity words (Gentner \& France, 1988; Asmuth \& Gentner, under review). Extending this work, we find that uses of relational words (both verbs and relational nouns) tend to be more metaphorical than uses of entity nouns in natural language corpora.}, author = {Jamrozik, Anja and Goldwater, Micah and Eyal Sagi and Dedre Gentner} } @article {50, title = {Using large corpora to explore the framing of concepts}, year = {2012}, type = {Talk}, address = {Helsinki, Finland}, abstract = {Psychologists and Social Scientists have long observed that the way in which a question or problem is presented to people can impact their attitudes and decisions. Framing is a widely discussed instance of this phenomenon: The choice of words and metaphors in talking about a given issue can affect hearers{\textquoteright} interpretations and biases, making some actions or strategies appear more plausible than others (1). The framing of issues is thus of great practical importance to those with an interest in steering the course of decision processes or justifying actions to a wider audience. For instance, different ways of dealing with drug abuse can become salient depending on whether the problem is presented as one of social policy or of law enforcement (2). Consequently, the framing of issues is of interest to decision makers and is a prominent topic of research in Political Science, Sociology, Economics, Psychology and related fields. Researchers interested in these questions tend to rely on controlled experiments and manually annotated document collections. The recent explosion in the amount of textual data available electronically provides an opportunity for investigating framing on a large scale and tracking how it influences, and is influenced by, decisions made by governments and businesses. However, analyzing this massive amount of data requires tools to facilitate a fast and efficient analysis of the available data sets. To meet this demand, researchers have turned to machine-learning methods from computational linguistics for applications in topic analysis and opinion classification. Most of these methods rely in some way on word co-occurrence patterns. A prominent example is Latent Semantic Analysis (LSA) (3), which has been applied to a wide range of tasks, including word sense discrimination, text summarization, and identifying semantic change (4-6). Here we present an LSA-based approach designed to observe and quantify variation in the framing of concepts across time or speaker/author populations. We illustrate this methodology using two examples of framing in political debates in the US senate: the rise and time-course of the framing of terror as a military struggle following the events of September 11th, 2001, and the different framings of abortion by democrats and republicans. References 1. Tversky, A. \& Kahneman, D. (1981) The Framing of decisions and the psychology of choice. Science 211 (4481): 453{\textendash}458. 2. Mark, E. (2003) War on Drugs: Legislation in the 108th Congress and Related Development, Congressional Research Service Report IB10113 ( http://opencrs.com/document/IB10113/ ). 3. Landauer, T. K., \& Dumais, S. T. (1997). A solution to Plato{\textquoteright}s problem: The Latent Semantic Analysis theory of the acquisition, induction, and representation of knowledge. Psychological Review, 104, 211-240. 4. Sch{\"u}tze, H. (1998) Automatic word sense discrimination. Computational Linguistics 24(1):97-124. 5. Marcu, D (2003) Automatic Abstracting, In Darke, M. A. (ed.) Encyclopedia of Library and Information, 245-256. 6. Sagi, E., Kaufmann, S., \& Clark, B. (2009). Semantic Density Analysis: Comparing Word Meaning across Time and Phonetic Space. In Basili R., and Pennacchiotti M. (Eds.), Proceedings of the EACL 2009 Workshop on GEMS: Geometrical Models of Natural Language Semantics. Athens, Greece.}, author = {Eyal Sagi and Daniel Diermeier and Stefan Kaufmann} } @article {bib_3, title = {What Difference Reveals About Similarity}, journal = {Cognitive Science}, volume = {36}, year = {2012}, pages = {1019 - 1050}, abstract = {Detecting that two images are different is faster for highly dissimilar images than for highly similar images. Paradoxically, we showed that the reverse occurs when people are asked to describe how two images differ {\textemdash} that is, to state a difference between two images. Following structure-mapping theory, we propose that this disassociation arises from the multistage nature of the comparison process. Detecting that two images are different can be done in the initial (local-matching) stage, but only for pairs with low overlap; thus, {\textquoteleft}{\textquoteleft}different{\textquoteright}{\textquoteright} responses are faster for low-similarity than for high-similarity pairs. In contrast, identifying a specific difference generally requires a full structural alignment of the two images, and this alignment process is faster for high-similarity pairs. We described four experiments that demonstrate this dissociation and show that the results can be simulated using the Structure-Mapping Engine. These results pose a significant challenge for nonstructural accounts of similarity comparison and suggest that structural alignment processes play a significant role in visual comparison.}, doi = {10.1111/j.1551-6709.2012.01250.x}, url = {http://doi.wiley.com/10.1111/j.1551-6709.2012.01250.x}, author = {Eyal Sagi and Dedre Gentner and Andrew Lovett} } @article {51, title = {Quantitative and Qualitative approaches for tracing changes to the framing of issues in political speeches}, year = {2011}, address = {University of Chicago, Chicago, IL}, author = {Eyal Sagi and Stefan Kaufmann and Daniel Diermeier} } @inbook {german2011role, title = {The role of speaker beliefs in determining accent placement}, booktitle = {Language, games, and evolution}, year = {2011}, pages = {92{\textendash}116}, publisher = {Springer}, organization = {Springer}, address = {New York, NY}, author = {James German and Eyal Sagi and Stefan Kaufmann and Brady Clark}, editor = {A. Bentz and C. Ebert and G. Jager and R. van Rooij} } @inbook {42, title = {Tracing semantic change with Latent Semantic Analysis}, booktitle = {Current Methods in Historical Semantics}, year = {2011}, pages = {161-183}, publisher = {Mouton de Gruyter}, organization = {Mouton de Gruyter}, address = {Berlin, Germany}, author = {Eyal Sagi and Stefan Kaufmann and Brady Clark}, editor = {Kathryn Allan and Justyna A Robinson} } @mastersthesis {52, title = {The Use of Causal Coherence Information in Disambiguating Pronouns}, volume = {PhD}, year = {2011}, month = {12/2011}, pages = {142}, school = {Northwestern University}, type = {Doctoral}, address = {Evanston, IL}, abstract = {This dissertation looks at the people use causal coherence information is used to determine the referent of a pronoun in sentence pairs, such as John accused Mark of stealing a car. He called the police. The experiments reported here test the idea that the process of pronoun disambiguation is informed by the causal connection between the person denoted by the pronoun (e.g., the one who called the police) and each of its potential referents (John and Mark). Experiment 1 varied how likely it is that the event of the first sentence (e.g., the accusation) would cause the event of the second (the notification) for each of the two individuals, and required participants to choose the appropriate referent. Participants selected the antecedent that had high causal strength or split their votes if neither did. Experiment 2 used explicit connectives that were either causal (and so, because) or temporal (before, after) and demonstrated that participants were more confident in their use of causal information when the connective was causal than when it was temporal. Experiment 3 explored the influence of negation on participants{\textquoteright} choice of referent. Participants{\textquoteright} ratings were affected by negation, but their choices of referent and their confidence in their choices were not. This indicates that participants use causal information when choosing the appropriate referent rather than relying on likelihood of occurrence. Finally, Experiment 4 used a self-paced reading paradigm to explore the effect pronoun ambiguity and reference has on participants reading time of two types of causal coherence relations {\textendash} result and explanation. Participants read sentence pairs connected by result relations fastest when an unambiguous pronoun referred to the object. They read sentence pairs connected by explanation fastest when an unambiguous pronoun referred to the subject. The results of these four experiments suggest that the process of causal coherence information and the process of pronoun disambiguation interact. These findings are in line with models of intersentential coherence in which discourse coherence is achieved by relating adjacent discourse segments by means of coherence relations (e.g., Hobbs, 1985; Kehler, 2002).}, isbn = {9781249907060}, author = {Eyal Sagi} } @conference {38, title = {Discourse Structure Effects on the Global Coherence of Texts}, booktitle = {Computational Models of Narrative, Papers from the 2010 AAAI Fall Symposium}, volume = {AAAI Technical report FS-10-04}, year = {2010}, abstract = {Many theories of discourse structure rely on the idea that the segments comprising the discourse are linked through inferred relations such as causality and temporal contiguity. These theories suggest that the resulting discourse is represented hierarchically. Two experiments examine some of the implications of these hierarchical structures on the perceived coherence of texts. Experiment 1 shows that texts with more levels to their hierarchical structure are judged to be more coherent. Experiment 2 demonstrates that these effects are sensitive to the genre of the text. Specifically, narratives seem to be more affected by manipulation of the discourse structure than procedural texts.}, url = {https://aaai.org/papers/02275-2275-discourse-structure-effects-on-the-global-coherence-of-texts/}, author = {Eyal Sagi}, editor = {Mark Finlayson} } @article {36, title = {Nouns are more stable than Verbs: Patterns of semantic change in 19th century English}, year = {2010}, type = {Poster}, address = {Portland, OR}, abstract = {It has been hypothesized in the literature that nouns are acquired earlier than verbs because they are more concrete and involve fewer relations. This hypothesis also predicts that the meaning of nouns should be more stable over time and across speakers. In this paper I use Latent Semantic Analysis of a 19th century literary corpus containing works from British and American authors to test this prediction. I examined the variability in the vector representations of frequently used nouns and verbs based on the culture of the author and the time period. The results show the nouns vary less than verbs between the two cultures and across time. Moreover, these differences still exist when the concreteness of the words is taken into account. These results are consistent with the hypothesis that the relational nature of verbs contributes to their difficulty and variability beyond its effect on the verb{\textquoteright}s concreteness.}, author = {Eyal Sagi} } @article {37, title = {Tracking Changes in Word Meaning and Use}, year = {2010}, type = {Talk}, address = {Stuttgart, Germany}, author = {Eyal Sagi and Stefan Kaufmann and Brady Clark and Daniel Diermeier} } @article {43, title = {Word Vector Spaces as a Tool for Investigating Semantic Change}, year = {2010}, address = {L2C2, Universit{\'e} Lyon, Lyon, France}, author = {Eyal Sagi} } @conference {21, title = {Culture in the Mirror of Language: A Latent Semantic Analysis Approach to Culture}, booktitle = {Proceedings of the 31st Annual Conference of the Cognitive Science Society}, year = {2009}, pages = {637-642}, address = {Amsterdam, The Netherlands}, abstract = {In the social sciences, culture is often explored via the use of knowledgeable informants and direct observation. In this paper we present a novel approach for cultural investigation that focuses on the statistical analysis of texts. We use Latent Semantic Analysis to generate a semantic space representing one or more cultures based on a corpus of texts. By comparing the vector representation of texts within this corpus it is possible to gain insights into cultural change. We demonstrate this method by exploring the divergence of British and American societies following the Revolutionary War. Possible uses of this method for exploratory and experimental research are discussed.}, author = {Eyal Sagi and Stefan Kaufmann and Brady Clark}, editor = {Niels Taatgen and Hedderik van Rijn and Lambert Schomaker and John Nerbonne} } @conference {22, title = {Modeling Perceptual Similarity as Analogy Resolves the Paradox of Difference Detection}, booktitle = {Proceedings of the 2nd International Analogy Conference}, year = {2009}, pages = {320-329}, address = {Sofia, Bulgaria}, abstract = {There is a paradoxical dissociation between recognizing that two stimuli are different and recognizing how they are different. We show that this dissociation can be captured by modeling perceptual similarity as a species of analogical processes. Using SME to model comparison, we show that the dissociation arises naturally from different stages in the analogical mapping process. Rather than relying on hand-coded input representations, our model uses an automatic, incremental encoding process to generate representations from the same stimuli as given to human participants.}, author = {Andrew Lovett and Dedre Gentner and Eyal Sagi and Kenneth Forbus}, editor = {Boicho Kokinov and Keith Holyoak and Dedre Gentner} } @article {23, title = {The perceived importance of explanations in narrative and procedural texts}, year = {2009}, type = {Poster}, address = {Rotterdam, Netherlands}, author = {Eyal Sagi} } @conference {35, title = {Phonaesthemic and Etymological effects on the Distribution of Senses in Statistical Models of Semantics}, booktitle = {Proceedings of the Cognitive Science Conference 2009 Workshop on Distributional Semantics beyond Concrete Concepts}, year = {2009}, pages = {35-40}, address = {Amsterdam, Netherlands}, abstract = {This paper uses methods based on corpus statistics and synonymy to explore the role language history and sound/form relationships play in conceptual organization through a case study relating the phonaestheme gl- to its prevalent Proto-Indo European root, *ghel. The results of both methods point to a strong link between the phonaestheme and the historical root, suggesting that the lineage of a language plays an important role in the distribution of linguistic meaning. The implications of these findings are discussed.}, author = {Armelle Boussidan and Eyal Sagi and Sabine Ploux}, editor = {Yves Peirsman and Yannick Versley and Tim Van de Cruys} } @conference {20, title = {Semantic Density Analysis: Comparing Word Meaning across Time and Phonetic Space}, booktitle = {Proceedings of the EACL 2009 Workshop on GEMS: Geometrical Models of Natural Language Semantics}, year = {2009}, pages = {104-111}, address = {Athens, Greece}, abstract = {This paper presents a new statistical method for detecting and tracking changes in word meaning, based on Latent Semantic Analysis. By comparing the density of semantic vector clusters this method allows researchers to make statistical inferences on questions such as whether the meaning of a word changed across time or if a phonetic cluster is associated with a specific meaning. Possible applications of this method are then illustrated in tracing the semantic change of "dog","do", and "deer" in early English and examining and comparing phonaesthemes.}, author = {Eyal Sagi and Stefan Kaufmann and Brady Clark}, editor = {Roberto Basili and Marco Pennacchiotti} } @article {41, title = {Tracing semantic change with Latent Semantic Analysis}, year = {2009}, address = {Northwestern Institute on Complex Systems (NICO) weekly seminar series, Evanston, IL}, author = {Eyal Sagi and Stefan Kaufmann and Brady Clark} } @article {24, title = {Tracing semantic change with Latent Semantic Analysis}, year = {2009}, type = {Talk}, address = {San Francisco, CA}, author = {Eyal Sagi and Stefan Kaufmann and Brady Clark} } @article {19, title = {Using Analogical Mapping to Simulate Time-Course Phenomena in Perceptual Similarity}, journal = {Cognitive Systems Research}, volume = {10}, year = {2009}, pages = {216-228}, abstract = {We present a computational model of visual similarity. The model is based upon the idea that perceptual comparisons may utilize the same mapping processes as are used in analogy. We use the Structure Mapping Engine (SME), a model of Gentner{\textquoteright}s structure-mapping theory of analogy, to perform comparison on representations that are automatically generated from visual input. By encoding visual scenes incrementally and sampling the output of SME at multiple stages in its processing, we are able to model not only the output of similarity judgments, but the time course of the comparison process. We demonstrate the model{\textquoteright}s effectiveness by replicating the results from three psychological studies that bear on the time course of comparison.}, author = {Andrew Lovett and Kenneth Forbus and Dedre Gentner and Eyal Sagi} } @article {28, title = {Discourse Relations in Context: Structural effects in the comprehension of texts}, year = {2008}, type = {Poster}, address = {Washington, DC}, author = {Eyal Sagi} } @article {29, title = {Effects of Hierarchical Discourse Structures on the Processing and Comprehension of Texts}, year = {2008}, type = {Poster}, address = {Chicago, IL}, author = {Eyal Sagi} } @conference {18, title = {Phonaesthemes: A Corpus-Based Analysis}, booktitle = {Proceedings of the 30th Annual Meeting of the Cognitive Science Society}, year = {2008}, pages = {65-70}, author = {Katya Otis and Eyal Sagi}, editor = {Bradley C Love and Ken McRae and Vladimir M Sloutsky} } @article {25, title = {Semantic Glimmers: Phonaesthemes facilitate access to sentence meaning}, year = {2008}, type = {Talk}, address = {Cleveland, OH}, author = {Eyal Sagi and Katya Otis} } @article {26, title = {Tracing semantic change with Latent Semantic Analysis}, year = {2008}, type = {Talk}, address = {Munich, Germany}, author = {Eyal Sagi and Stefan Kaufmann and Brady Clark} } @article {27, title = {Using Latent Semantic Analysis to Examine the Relationship between Sound and Meaning}, year = {2008}, type = {Talk}, address = {Washington, DC}, author = {Eyal Sagi and Katya Otis} } @conference {31, title = {Analogy as a mechanism for comparison}, booktitle = {Proceedings of Analogies: Integrating Multiple Cognitive Abilities}, year = {2007}, pages = {27-30}, author = {Andrew Lovett and Eyal Sagi and Dedre Gentner}, editor = {Angela Schwering and Ulf Krumnack and Kai-Uwe K{\"u}hnberger and Helmar Gust} } @article {32, title = {Does {\textquotedblleft}Different{\textquotedblright} Imply a Difference? A Comparison of Two Tasks}, year = {2007}, type = {Talk}, address = {Evanston, IL}, author = {Eyal Sagi and Dedre Gentner} } @article {30, title = {The Effect of the Speaker{\textquoteright}s Motivation on the Interpretation of Logical Connectives}, year = {2007}, type = {Poster}, address = {Nashville, TN}, author = {James German and Eyal Sagi and Stefan Kaufmann and Brady Clark and Min-Joo Kim} } @article {17, title = {Are Coherence relations computed during reading? Evidence from a priming experiment}, year = {2006}, type = {Poster}, address = {Minneapolis, MN}, author = {Eyal Sagi} } @conference {15, title = {Context and the Processing of Discourse: Priming and Genre Effects on Discourse Comprehension}, booktitle = {Proceedings of the 28th Annual Conference of the Cognitive Science Society}, year = {2006}, pages = {2071-2076}, author = {Eyal Sagi}, editor = {Ron Sun and Naomi Miyake} } @conference {14, title = {Does "Different" Imply a Difference? A Comparison of Two Tasks}, booktitle = {Proceedings of the 28th Annual Conference of the Cognitive Science Society}, year = {2006}, pages = {261-266}, address = {Vancouver, BC}, author = {Dedre Gentner and Eyal Sagi}, editor = {Ron Sun and Naomi Miyake} } @article {16, title = {The role of hearer{\textquoteright}s beliefs in the interpretation of logical connectives}, year = {2006}, type = {Talk}, address = {Berlin, Germany}, author = {James German and Eyal Sagi and Brady Clark and Min-Joo Kim and Stefan Kaufmann} } @article {33, title = {Coherence: A psychological perspective}, year = {2005}, type = {Talk}, address = {Evanston, IL}, author = {Eyal Sagi} }