Downloads & Free Reading Options - Results

We are developing a scalable and cost-efficient approach for generating synthetic datasets using Large Language Models (LLMs). Initially, we are focusing on creating a dataset that systematically evaluates the social reasoning capabilities of LLMs. To assess quality, we have crowdworkers (recruited via Prolific Academic) review a subset of the model-generated data (100 random items) and compare ratings with 25 random items from a crowdsourced data set and 25 random items from datasets developed by researchers (experts).

A considerable amount of research in moral psychology has examined how people resolve moral dilemmas pitting adherence to moral norms against overall consequences for the greater good (e.g., Greene, Sommerville, Nystrom, Darley, & Cohen, 2001; Greene, Nystrom, Engell, Darley, & Cohen, 2004). Within this paradigm, several accounts have been proposed that aim to capture the mental process(es) that (1) cause judgments that adhere to relevant moral norms (i.e., characteristically deontological judgments) and (2) cause judgments that maximize overall consequences in terms of welfare (i.e., characteristically utilitarian judgments). One straightforward possibility is that characteristically deontological judgments are produced by reference to salient moral norms, while characteristically utilitarian judgments are produced by cost-benefit analyses of aggregate well-being. Alternatively, dual-process theory (Greene, 2007) proposes that, while characteristically utilitarian judgments are indeed produced by cost-benefit analyses of aggregate well-being, characteristically deontological judgments are caused by the automatic emotional response to the idea of causing harm. In contrast to these two possibilities, some researchers suggest that both characteristically utilitarian and characteristically deontological judgments may be caused by processes of cost-benefit analyses of aggregate well-being (e.g., Baron & Cohen, 2020; Hennig & Hütter, 2020). Yet, other researchers suggest that both characteristically utilitarian and characteristically deontological judgments may be caused by coherence-based reasoning processes about norms (Holyoak & Powell, 2016). The purpose of the current research is to distinguish between these possibilities by investigating how moral dilemma judgments are associated with conscious reasoning processes. To this end, we will examine associations between factors influencing moral dilemma judgments and the content of conscious reasoning in language. Using the CNI model (Gawronski, Armstrong, Conway, Friesdorf, & Hütter, 2017), we will quantify sensitivity to consequences (in a characteristically utilitarian sense), sensitivity to moral norms (in a characteristically deontological sense), and general preference for inaction over action in responses to moral dilemmas. Using Linguistic Inquiry and Word Count 2015 (LIWC2015; Pennebaker, Boyd, Jordan, & Blackburn, 2015), we will quantify frequency of language related to the cost-benefit analyses of aggregate well-being (consequentialism), adherence to moral norms and rules (deontology), and expressions of morally relevant emotions. If the first possibility is correct, there should be (1) a positive association between sensitivity to consequences in responses to moral dilemmas and reasoning in terms of consequentialist language and (2) a positive association between sensitivity to moral norms in responses to moral dilemmas and reasoning in terms of deontological language. If the second possibility is correct, there should be (1) a positive association between sensitivity to consequences in responses to moral dilemmas and reasoning in terms of consequentialist language and (2) a positive association between sensitivity to moral norms in responses to moral dilemmas and reasoning in terms of emotional language. If the third possibility is correct, there should be (1) a positive association between sensitivity to consequences in responses to moral dilemmas and reasoning in terms of consequentialist language and (2) a positive association between sensitivity to moral norms in responses to moral dilemmas and reasoning in terms of consequentialist language. Finally, if the fourth possibility is correct, there should be (1) a positive association between sensitivity to consequences in responses to moral dilemmas and reasoning in terms of deontological language and (2) a positive association between sensitivity to moral norms in responses to moral dilemmas and reasoning in terms of deontological language.

A considerable amount of research in moral psychology has examined how people resolve moral dilemmas pitting adherence to moral norms against overall consequences for the greater good (e.g., Greene, Sommerville, Nystrom, Darley, & Cohen, 2001; Greene, Nystrom, Engell, Darley, & Cohen, 2004). Within this paradigm, several accounts have been proposed that aim to capture the mental process(es) that (1) cause judgments that adhere to relevant moral norms (i.e., characteristically deontological judgments) and (2) cause judgments that maximize overall consequences in terms of welfare (i.e., characteristically utilitarian judgments). One straightforward possibility is that characteristically deontological judgments are produced by reference to salient moral norms, while characteristically utilitarian judgments are produced by cost-benefit analyses of aggregate well-being. Alternatively, dual-process theory (Greene, 2007) proposes that, while characteristically utilitarian judgments are indeed produced by cost-benefit analyses of aggregate well-being, characteristically deontological judgments are caused by the automatic emotional response to the idea of causing harm. In contrast to these two possibilities, some researchers suggest that both characteristically utilitarian and characteristically deontological judgments may be caused by processes of cost-benefit analyses of aggregate well-being (e.g., Baron & Cohen, 2020; Hennig & Hütter, 2020). Yet, other researchers suggest that both characteristically utilitarian and characteristically deontological judgments may be caused by coherence-based reasoning processes about norms (Holyoak & Powell, 2016). The purpose of the current research is to distinguish between these possibilities by investigating how moral dilemma judgments are associated with conscious reasoning processes. To this end, we will examine associations between factors influencing moral dilemma judgments and the content of conscious reasoning in language. Using the CNI model (Gawronski, Armstrong, Conway, Friesdorf, & Hütter, 2017), we will quantify sensitivity to consequences (in a characteristically utilitarian sense), sensitivity to moral norms (in a characteristically deontological sense), and general preference for inaction over action in responses to moral dilemmas. Using Linguistic Inquiry and Word Count 2015 (LIWC2015; Pennebaker, Boyd, Jordan, & Blackburn, 2015), we will quantify frequency of language related to the cost-benefit analyses of aggregate well-being (consequentialism), adherence to moral norms and rules (deontology), and expressions of morally relevant emotions. If the first possibility is correct, there should be (1) a positive association between sensitivity to consequences in responses to moral dilemmas and reasoning in terms of consequentialist language and (2) a positive association between sensitivity to moral norms in responses to moral dilemmas and reasoning in terms of deontological language. If the second possibility is correct, there should be (1) a positive association between sensitivity to consequences in responses to moral dilemmas and reasoning in terms of consequentialist language and (2) a positive association between sensitivity to moral norms in responses to moral dilemmas and reasoning in terms of emotional language. If the third possibility is correct, there should be (1) a positive association between sensitivity to consequences in responses to moral dilemmas and reasoning in terms of consequentialist language and (2) a positive association between sensitivity to moral norms in responses to moral dilemmas and reasoning in terms of consequentialist language. Finally, if the fourth possibility is correct, there should be (1) a positive association between sensitivity to consequences in responses to moral dilemmas and reasoning in terms of deontological language and (2) a positive association between sensitivity to moral norms in responses to moral dilemmas and reasoning in terms of deontological language.

The comprehensive report of the activities of the 1953 interdisciplinary seminar in language and culture held at Ann Arbor, Michigan appears here with the working bibliography compiled at the time of the seminar. The report discusses existing text materials, general perspectives, popular conceptions of nationality, objective criteria, implementation, and evaluation, and concludes with recommendations and a summary report. The accompanying bibliography, not originally published with the report, covers general and specific treatments of culture (French, German, Hispanic, and American), national stereotypes, patterning in culture, and pedagogical materials and methods. (AF)

The objective of this systematic review is to examine the literature that has investigated the impact of code-switching on language processing, and to identify whether its influence differs across methodologies, and how participant- and item-level factors may affect this. Early research focused on code-switching production and the associated reaction-time costs in cued language-switching tasks. Although some work on code-switched reading and spoken language processing has identified similar costs, recent work has suggested that such costs are absent for common code-switches and among frequent code-switchers. This review aims to synthesise these findings to understand whether code-switching hinders or benefits reading and spoken language processing. It also investigates whether these costs or benefits may depend on the study design, the switches participants are exposed to, as well as the language or code-switching background of the participants. This review will be focused on non-clinical bilingual populations, with no age range of participants or date range of studies imposed. Three databases will be searched: Web of Science, SCOPUS, and PsychInfo. Studies included must be experimental where participants are either exposed to unilingual and code-switching stimuli, or to different types of code-switching stimuli, and must include a numeric outcome measure that assesses the impact of code-switching on participants’ language processing. Studies only investigating code-switching production or only investigating code-switching’s impact on language control will be excluded. This protocol is based on the JBI methodology for systematic reviews and the PRISMA-P checklist.

Whole Language has become a visible and strong movement in language instruction. Yet there is still considerable disagreement as to what Whole Language entails and there are different degrees of knowledge held by various self-professed Whole Language teachers. Some are at a beginning stage of knowledge regarding Whole Language as philosophy and methodology. This has serious implications for the children entrusted to them. Teachers of whole language should have attained an advanced level of knowledge. Seven biases that may interfere with the attainment of knowledge at this level are: (1) oversimplification and overregularization; (2) overreliance on a single basis for mental representation; (3) overreliance on top-down processing; (4) context-independent conceptual representation; (5) overreliance on precompiled knowledge structures; (6) rigid compartmentalization of knowledge; and (7) passive transmission of knowledge. (Eighteen references are attached.) (RS)

This is a guide for teachers and administrators to familiarize them with the Laotian people, language and culture. The first section contains a brief geography and history of Laos, a discussion of the ethnic and lingustic groups of Laos, and information on the economic and religious life of these groups. Section two describes the Laotian refugee experience and considers some of the adjustments Laotians must make for their new life in the United States. This section also explains elements of the international, national and local support systems which assist Indochinese refugees. Section three gives a brief history of the educational system in Laos, and the implications for educational needs of Laotians now residing in the United States. Suggestions for working with Laotians in the schools and some potential problem areas are also covered. The last section presents an analysis of the Laotian language. Emphasis is placed on the problems Laotians have with English, and suggestions are made for overcoming these difficulties. A brief bibliography is included. (Author/APM)

During this contract period, we introduced new representation schemes called event shape diagrams for events and actions; developed mechanisms for explanatory schema acquisition, a process which allows a program to learn new plans and scripts (i.e., action sequences) by reading articles or stories based on the plans or scripts; completed work on computer programs which can understand complex noun phrases (such as water pump pulley adjustment screw thread damage report summary); made substantial progress toward identifying and cataloging cognitive universals.

In current mathematics reform, assessment is integral to effective instruction. However, accurate assessment of over three million limited English proficient (LEP) students in today's schools is of great concern. How does students' language proficiency affect the accuracy of teachers' assessment of students' mathematical understanding? This research quantitatively and qualitatively examined teachers' assessment of students' mathematical understanding through analysis of teachers' prediction of students' performance on a fraction test, videotaped tutoring sessions with one LEP student and one non-LEP student, and subsequent individual interviews with teachers and students. This research found that teachers' different instructional scripts created differential opportunities for teachers to assess students' mathematical understanding. Results indicate that teachers know less about LEP students' mathematical understanding than non-LEP students' understanding. Teachers need assistance in closing the gaps between how they believe they teach and assess LEP students' understanding and how they actually do. Contains 16 references. (Author/MKR)

Book digitized by Google from the library of Harvard University and uploaded to the Internet Archive by user tpb.

This overview is organized within a historical framework, although time limitations have forced me to invent a version of history that is necessarily incomplete. The title of the talk was given to me by the AAAI program Committee, who wisely restricted the scope of my task by including the descriptor knowledge-based. This mercifully allowed me to ignore a large body of work that focuses exclusively on the syntactic structures of natural language. Even so, the body of work that can accurately be described as knowledge-based natural language understanding is large, and difficult to cover in the space of one hour. To maintain continuity, I utilized the recurring theme of weak methods vs. strong methods. This foundational theme helped me pare down my view of history and serves as my only defense against otherwise unforgivable omissions in the overview. Even so, it was difficult to pick and choose from the corpus of potentially relevant research, and the usual disclaimers about intelligible brevity at the cost of comprehensive coverage must be piously invoked to ward off inevitable accusations of ignorance, prejudice, and other sins associated with warped thinking.

The present experiment asks whether language may support toddlers’ understanding of others’ perspectives and experiences. For full details, please see "PreregDUCK-220221.docx".

Body language, also called kinesics, is the discipline concerned with the study of all bodily motions that are communicative. An understanding of kinesics across cultures necessitates a close look at posture, movement, facial expression, eye management, gestures, and proxemics (distancing). The popularity of one posture over another and the emotion conveyed by a given posture seems to be largely determined by culture. Recent studies of rhythm and dance as they relate to body movement have revealed astonishing new insights into human interaction. Facial expressions are very revealing and would be an important aspect of language learners attempting to master the nonverbal system of another culture. The role of eye contact in a conversational exchange between two Americans is well defined and is intensely important in interpersonal communication. An understanding of the role gestures play (autistic, technical and folk) within a culture is critical to sensitive communication. In each culture, the use of space (proxemics) depends upon the nature of the social interaction, but all cultures distinguish the four basic categories of intimate, personal, social, and public distance. While it is clear that all cultures make use of kinesic behaviors in communication, scholars do not agree on the precise nature of the role they play and are searching for some kinesic universals. An appendix is included with suggested activities for sensitizing students to aspects of nonverbal communication. A bibliography is also appended. (NCR)

During this contract period the authors have: (a) continued investigation of events and actions by means of representation schemes called 'event shape diagrams'; (b) written a parsing program which selects appropriate word and sentence meanings by a parallel process known as activation and inhibition; (c) begun investigation of the point of a story or event by modeling the motivations and emotional behaviors of story characters; (d) started work on combining and translating two machine-readable dictionaries into a lexicon and knowledge base which will form an integral part of our natural language understanding programs; (e) made substantial progress toward a general model for the representation of cognitive relations by comparing English scene and event descriptions with similar descriptions in other languages; (f) constructed a general model for the representation of tense and aspect of verbs; (g) made progress toward the design of an integrated robotics system which accepts English requests, and uses visual and tactile inputs in making decisions and learning new tasks. (Author)

During this contract period the authors have: (a) continued investigation of events and actions by means of representation schemes called 'event shape diagrams'; (b) written a parsing program which selects appropriate word and sentence meanings by a parallel process known as activation and inhibition; (c) begun investigation of the point of a story or event by modeling the motivations and emotional behaviors of story characters; (d) started work on combining and translating two machine-readable dictionaries into a lexicon and knowledge base which will form an integral part of our natural language understanding programs; (e) made substantial progress toward a general model for the representation of cognitive relations by comparing English scene and event descriptions with similar descriptions in other languages; (f) constructed a general model for the representation of tense and aspect of verbs; (g) made progress toward the design of an integrated robotics system which accepts English requests, and uses visual and tactile inputs in making decisions and learning new tasks. (Author)

During this contract period the authors have: (a) continued investigation of events and actions by means of representation schemes called 'event shape diagrams'; (b) written a parsing program which selects appropriate word and sentence meanings by a parallel process known as activation and inhibition; (c) begun investigation of the point of a story or event by modeling the motivations and emotional behaviors of story characters; (d) started work on combining and translating two machine-readable dictionaries into a lexicon and knowledge base which will form an integral part of our natural language understanding programs; (e) made substantial progress toward a general model for the representation of cognitive relations by comparing English scene and event descriptions with similar descriptions in other languages; (f) constructed a general model for the representation of tense and aspect of verbs; (g) made progress toward the design of an integrated robotics system which accepts English requests, and uses visual and tactile inputs in making decisions and learning new tasks. (Author)

This paper presents an exploration of low resource languages and the specific challenges that arise in natural language understanding of these by a voice assistant. While voice assistants have made significant strides when it comes to their understanding of mainstream languages, this paper focuses on extending this understanding to low resource languages in order to maintain diversity of linguistics and also delight the customer. In this paper, the specific nuances of natural language understanding when it comes to these low resource languages has been discussed. The paper also proposes techniques to overcome some of the challenges in voice assistants understanding low resource language models. The proposed methods and future direction presented in this doc are poised to drive advancements in voice technology and promote inclusivity by ensuring that voice assistants are accessible to speakers of underrepresented languages.

Children’s language ability is important for multiple outcomes across the lifespan. Children living in lower socioeconomic circumstances often have lower language development. Several mechanisms have been proposed to explain this, including parent linguistic responsiveness. However, this research often fails to recognise the constraints parents living in lower socioeconomic circumstances are under. The present study will aim to investigate whether parents’ psychological distress, both currently and in the postpartum period, mediates the relationship between socioeconomic circumstances and language development using a longitudinal design, measuring language at three points between 30 and 48 months. We will also investigate whether parents’ linguistic responsiveness moderates the relationship between psychological distress and language development, examining whether responsiveness is protective to children’s language when parents are experiencing distress.

In this study, researchers will leverage a nationally representative sample of children attending Head Start, the Family and Child Experiences Survey, 2014, to provide a more nuanced understanding of dual language learners' (DLLs) bilingual vocabulary strengths. We will implement latent profile analyses, a person-centered, data-based approach to empirically derive bilingual vocabulary groups (i.e., latent profiles) in the fall and spring of Head Start, among Spanish-speaking DLLs. Profiles will reveal DLLs’ relative strengths in each language. We will then examine how family characteristics, such as household income and parents’ educational attainment relate to fall bilingual vocabulary profiles. Next, we will conduct latent transition analysis to assess how children’s membership in bilingual vocabulary groups changes between the fall and spring. Finally, we will examine how measures of classroom quality predict changes among profiles between the fall and spring, to examine the role of high-quality classrooms in supporting bilingual vocabulary growth.

Research in computer understanding of natural language has led to the construction of programs which can handle a number of different types of language, including questions about the contents of data bases, stories and news articles, dialogues, and scene descriptions. This research draws on and has in turn had an effect on many other research areas, including software engineering, linguistics, psychology, philosophy, and knowledge representation. This paper provides a brief history and overview of the field, along with examples and explanations of the operation of several natural language understanding programs. The limitations of our current technology are discussed, and assessments are given of the most promising current research directions. (Author)

We study understanding of sentences in children within the framework of the good-enough theory of language comprehension developed for adults (Ferreira, Bailey and Ferraro, 2002). Our main hypothesis is that while children are able to understand language in a very good way, they may only understand it in a good enough way like adults do. This type of comprehension would allow children to communicate while keeping the cognitive cost at a decent level. We investigated such comprehension by exposing children to questions with or without a mistake. For example, in the question “In the Lion King, what colors are the tigers ?”, we exchanged “lions” with “tigers”. Children who understand sentences in a good enough way could, under certain circumstances, not consider the mistakes and answer the question as if we stated the right word. We choose to present our questions under two conditions: the Answer condition where children had to answer the questions with time pressure and the Check condition where children have to find errors in the questions without any time pressure.

We study understanding of sentences in children within the framework of the good-enough theory of language comprehension developed for adults (Ferreira, Bailey and Ferraro, 2002). Our main hypothesis is that while children are able to understand language in a very good way, they may only understand it in a good enough way like adults do. This type of comprehension would allow children to communicate while keeping the cognitive cost at a decent level. We investigated such comprehension by exposing children to questions with or without a mistake. For example, in the question “In the Lion King, what colors are the tigers ?”, we exchanged “lions” with “tigers”. Children who understand sentences in a good enough way could, under certain circumstances, not consider the mistakes and answer the question as if we stated the right word. We choose to present our questions under two conditions: the Answer condition where children had to answer the questions with time pressure and the Check condition where children have to find errors in the questions without any time pressure.

Ellipsis in English can occur in many different forms. A large class of ellipses is isolated, encompassing such traditionally distinct linguistic rules as Right-node Raising, Gapping, VP-Deletion, Coordinate Reduction, and other phenomena which crosses sentence boundaries. The members of this class of ellipses involve implicit or explicit coordination. A uniform algorithm implemented in the UC (Unix Consultant) natural language system resolves the ellipses in this class, and constraints are given which govern the production of those ellipses. Justification is provided for the approach taken, and its limitations and applications are explored.

This is volume one of a three volume final report. This volume, Volume 1, provides a technical overview of the effort. Chapter 1 is the Executive Summary. Chapter 2 overviews the effort as a whole reporting on (1) Our Integration of software and results from other Strategic Computing contractors, (2) Our participation in Fleet Command Center Battle Management Program, (3) Distribution of the resulting software to other sites, and efforts, and (4) Contributions of the work to the state of the art. The problem of acquiring linguistic knowledge is the chief obstacle to widespread use of natural language technology. Chapters 3 and 6 report results of five to tenfold increase in our productivity in moving the natural language shells to new application domains. The ability of natural language systems to cooperatively handle novel, errorful, or incomplete forms is also critical; chapters 5 and 7 report new techniques to intelligently and gracefully respond to such forms. Chapter 4 reports on an implementation of a discourse module for understanding definite reference. (Author) (kr)

Intelligent Information Access and Extraction suggest significant challenges for Natural Language analysis. Tasks of interest include semantic role labeling (determining who did what to whom, when and where), information extraction (identifying entities, relations and events), following natural language instructions, and textual entailment (determining whether one utterance is a likely consequence of another). A computational approach to these challenges often involves assigning values to sets of interdependent variables and thus frequently necessitates performing global inference that accounts for these interdependencies. This talk presents research on Constrained Conditional Models (CCMs), a framework that augments probabilistic models with declarative constraints as a way to support such decisions. We will present a framework we introduced a few years ago, formulating decision problems in NLP as Integer Linear Programming problems, but focus on new algorithms for training these global models using indirect supervision signals. Learning models for structured tasks is difficult partly since generating supervision signals is costly. We show that it is often easy to obtain a related indirect supervision signal, and discuss several options for deriving this supervision signal, including inducing it from the world's response to the model's actions. Our learning framework is “Constraints Driven” in the sense that it allows and even gains from global inference that guided by expressive declarative knowledge (encoded as constraints). Experimental results show the significant contribution of easy-to-get indirect supervision on NLP tasks such as information extraction, Transliteration and Textual Entailment. ©2011 Microsoft Corporation. All rights reserved.

We introduce a novel framework for evaluating multimodal deep learning models with respect to their language understanding and generalization abilities. In this approach, artificial data is automatically generated according to the experimenter's specifications. The content of the data, both during training and evaluation, can be controlled in detail, which enables tasks to be created that require true generalization abilities, in particular the combination of previously introduced concepts in novel ways. We demonstrate the potential of our methodology by evaluating various visual question answering models on four different tasks, and show how our framework gives us detailed insights into their capabilities and limitations. By open-sourcing our framework, we hope to stimulate progress in the field of multimodal language understanding.

This report presents an evaluation of the Excellence and Success through Career Understanding and Enriched Language Activities (Project ESCUELA), an Elementary and Secondary Education Act Title VII-funded project in its second year of operation at five schools in Brooklyn, New York. The project served 415 Spanish-speaking and 178 Yiddish-speaking students of limited English proficiency. Participating students received instruction in English as a Second Language (ESL), native language arts (NLA) and content subjects, as well as computer training and career education. Staff development programs, parent participation workshops and ESL instruction for parents were integral to the project. An evaluation of Project ESCUELA found that it met its objectives for Spanish NLA, Yiddish NLA, staff development, and parent involvement. It failed to meet its objectives for English development, the content area subjects of science and social studies, or mainstreaming. Recommendations on program improvement are included. Two appendixes include a list of instructional materials used and class schedules. (MDM)

This report includes a collection of reprints reporting natural language research. The problem of acquiring linguistic knowledge is the chief obstacle to widespread use of natural language technology. Chapter 1 and 4 report results of five to tenfold increase in our productivity in moving the Janus and Parlance natural language shells to new application domains. The ability of natural language systems to cooperatively handle novel, errorful, or incomplete forms is also critical; chapter 3 and 5 report new techniques for intelligently and graceful respond to such forms. Chapter 2 reports on an implementation of a discourse module for understanding definite reference.

This report includes a collection of reprints reporting natural language research. The problem of acquiring linguistic knowledge is the chief obstacle to widespread use of natural language technology. Chapter 1 and 4 report results of five to tenfold increase in our productivity in moving the Janus and Parlance natural language shells to new application domains. The ability of natural language systems to cooperatively handle novel, errorful, or incomplete forms is also critical; chapter 3 and 5 report new techniques for intelligently and graceful respond to such forms. Chapter 2 reports on an implementation of a discourse module for understanding definite reference.

The purpose of this paper is to present a framework and definitions that may be useful in considering the role of advocacy in counseling. Using this framework as a foundation, it reviews advocacy in counseling and discusses examples in a range of advocacy actions. In addition, it proposes some of the possible reasons that advocacy has historically experienced marginalization in counseling. This paper intends to facilitate counselors in identifying the range of appropriate ways they may advocate for clients and client groups. This paper also discusses supportive literature in advocacy, empowerment, and social action as well as issues influencing the implementation of advocacy roles in counseling. (Contains 18 references.) (MKA)

Language understanding work at Paramax focuses on applying general-purpose language understanding technology to spoken language understanding, text understanding, and document processing, integrating language understanding with speech recognition, knowledge-based information retrieval and image understanding.

A discussion of the application of artificial intelligence to natural language processing looks at several problems in language comprehension, involving semantic ambiguity, anaphoric reference, and metonymy. Examples of these problems are cited, and the importance of the computational approach in analyzing them is explained. The approach applies specific mechanisms or models to natural language to make explicit the complex system of assumptions, knowledge, and processes used in understanding language. It is proposed that by examining language at the micro level it is possible to see how much a learner--whether a computer, a child, or a second language learner--must master. (MSE)

A discussion of the application of artificial intelligence to natural language processing looks at several problems in language comprehension, involving semantic ambiguity, anaphoric reference, and metonymy. Examples of these problems are cited, and the importance of the computational approach in analyzing them is explained. The approach applies specific mechanisms or models to natural language to make explicit the complex system of assumptions, knowledge, and processes used in understanding language. It is proposed that by examining language at the micro level it is possible to see how much a learner--whether a computer, a child, or a second language learner--must master. (MSE)

This report presents findings of the implementation of Communities in Schools (CIS) Success Coaches within five Wake County Public School System (WCPSS) middle schools during the 2019-20 and 2020-21 school years. End-of-year outcomes, including attendance rates and English language arts and mathematics grades, are examined and compared between the students on the Success Coaches' caseloads and similar students. Trends in outcomes across the years are largely inconclusive due to the onset of the COVID-19 pandemic during the 2019-20 school year, which significantly impacted the way students received services and connected with the school environment. Data collected from Success Coaches, students on their caseloads, and school principals suggest variability in implementation and services that could be due to a coach's skill set and the school's needs. Individual student reports indicate ways Success Coaches have made positive impacts, though the quantitative analyses show no discernable differences in outcomes between students on the coaches' caseloads and their counterparts. Recommendations are presented to continue progress monitoring and create a Memorandum of Understanding between CIS and WCPSS. [Cover title varies: "CIS Success Coaches in WCPSS, Implementing School Years 2019-20 and 2020-21."]

This report contains an annual summary of BBN's ARPA-sponsored natural language understanding project and two Technical Notes: 'KLONE's Progress,' by R.J. Brachman, and 'Generalizations of ATN Grammars,' by W.A. Woods. The Annual summary includes a list of publications of the project and an overview of the first year's work. The first Technical Note focuses on the progress in the development of the KLONE knowledge representation system, an implementation of Brachman's Structured Inheritance Networks. The second describes some generalizations of the notion of ATN grammar, including Cascaded ATN's(a sequence of nondeterministic ATN transducers with each successive ATN taking its input from the output of the previous), Generalized Transition Networks (a generalization of the notion of ATN to handle arbitrary structured input rather than only linear strings), and an interpretation of KLONE networks as a kind of ATN with inheritance.

This manual provides teachers and administrators with information regarding Korean people, language and culture. Discussion of Korean culture includes these topics: (1) social and cultural changes; (2) family structure; (3) holidays and festivals; (4) discipline for children; (5) home customs; (6) educational system; and (7) learning and teaching styles. Characteristics of the Korean language are examined and English and Korean phonology are compared. The problems that Korean children might have with English articles, tenses, singular and plural nouns, auxiliaries, and comparative adjectives and adverbs are reviewed. Strategies for English as a Second Language instruction are offered along with a list of resource materials on ESL and Korean culture. (APM)

This user's manual for the Irus-II natural language system describes the capabilities of the different modes of the end user interface for Irus-II, including entering and editing queries the history list, control of processing, viewing of the intermediate results, window management and the use of files of queries. Irus-II, is part of a larger system called Janus, which is composed of a number of seamless interface utilities, including understanding (provided by Irus-II), generation (provided by Spokesman), maps, and other graphics tools. This is the second volume of a three volume final report. The different applications of IRUS require that the interface provide configurations for three different classes of users; (1) an end-user configuration that supports easy and flexible input, editing, and processing of typed or mouse-selected queries and commands; (2) developer configurations for natural language experts who are developing, testing, debugging, or updating IRUS; (3) pop up windows for query-response interactions that may be used in conjunction with other interfaces. The available IRUS interface configurations use the Symbolics Lisp Machine window system to provide multiple pane displays and mouse and keyboard input. The interface can be developed from the generic capabilities and illustrated below and tailored to the needs of a particular application. (Author) (kr)

Improving Language Understanding by Generative Pre-Training

A series of experiments explored the use of television in Israel to teach English to Hebrew-Speaking students. The emphasis of the experiments was on assessing the ways in which television can be used to fulfill specific tasks in language instruction that are not easily fulfilled by the classroom teachers. Ninth graders were divided into two groups: One group received the standard course of English-language instruction; a second group had English-language instruction supplemented by a television program which presented the spoken language in natural context and in every day situations. The ninth graders who had the television experience showed a better understanding of English-language films and audio tapes, as well as better achievement in active speech. The contribution of television to understanding from audio tapes proved to be particularly great in pupils whose IQ was less than 110. The greater amount of exposure to oral presentations did not have a negative effect on reading comprehension. (JY)

A series of experiments explored the use of television in Israel to teach English to Hebrew-Speaking students. The emphasis of the experiments was on assessing the ways in which television can be used to fulfill specific tasks in language instruction that are not easily fulfilled by the classroom teachers. Ninth graders were divided into two groups: One group received the standard course of English-language instruction; a second group had English-language instruction supplemented by a television program which presented the spoken language in natural context and in every day situations. The ninth graders who had the television experience showed a better understanding of English-language films and audio tapes, as well as better achievement in active speech. The contribution of television to understanding from audio tapes proved to be particularly great in pupils whose IQ was less than 110. The greater amount of exposure to oral presentations did not have a negative effect on reading comprehension. (JY)

This report discussed an abstract parallel machine organization and a class of 'marker passing' algorithms for a family of important operations required for intelligent manipulation and use of knowledge. These operations have the characteristic that they involve a considerable amount of 'non- deterministic' programming and/or search. The major such problem is that of 'high level perception' or 'situation recognition' - the problem of finding which of a large collection of situation-action rules are satisfied at any given moment. The combinatoric costs of various algorithms for large collections of rules are discussed, and a proposed parallel architecture for performing such operations in real time is presented. An example marker passing algorithm for the problem of finding the most specific concepts that subsume a given input description is presented, and issues relative to the implementation of such algorithms on physical machines are discussed.

This monograph points out the difficulties in language development of which teachers should be aware, suggests some methods through which children may be helped to overcome comprehension problems, and presents bibliographies helpful for persons who wish to undertake further reading. Chapters include "Some Possible Origins of the Prevalence of Verbalism," which defines verbalism and discusses the symbolic nature of language; "Reading and Understanding," which presents basic facts relating to understanding in reading, examines sources of difficulty in acquiring meaning, and suggests several necessary conditions for efficient teaching of reading; "How the Curriculum May Contribute to Understanding," which stresses the importance of first-hand experience in learning, presents studies on the significance of experience, and suggests approaches toward an experience-centered curriculum; and "The Development of Thinking and of Concepts," which stresses developing sensitivity toward word meanings and presents research acquainting teachers with the difficulties children encounter while acquiring language. Each chapter contains a bibliography. (RB)

While highly constrained language can be used for robot control, robots that can operate as fully autonomous subordinate agents communicating via rich language remain an open challenge. Toward this end, the central goal of the SUBTLE MURI project was to develop an autonomous system that supports natural, continuous interaction with the operator through language before, during, and after mission execution. The operator communicates instructions to the system through natural language and is given feedback on how each instruction was understood as the system constructs a logical representation of its orders using linear temporal logic. While the plan is executed, the operator is updated on relevant progress via language and images and can change the robots orders. Unlike many other integrated systems of this type, the language interface is built using robust, general purpose parsing and semantics systems that do not rely on domain-specific grammars. The natural language system uses domain-general components that can easily be adapted to cover the vocabulary of new applications. We demonstrate the robustness of the natural language understanding system through a user study where participants interacted with a simulated robot in a search and rescue scenario.

Under this effort, SRI has developed spoken-language technology for interactive problem solving, featuring real-time performance for up to several thousand word vocabularies, high semantic accuracy, habitability within the domain, and robustness to many sources of variability. Although the technology is suitable for many applications, efforts to date have focussed on developing an Air Travel Information System (ATIS) prototype application. SRI's ATIS system has been evaluated in four ARPA benchmark evaluations, and has consistently been at or near the top in performance. These achievements are the result of SRI's technical progress in speech recognition, natural-language processing, and speech and natural-language integration.

This paper describes several experiments combining natural language and acoustic constraints to improve overall performance of the MIT VOYAGER spoken language system. This system couples the SUMMIT speech recognition system with the TINA language understanding system to answer spoken queries about navigational assistance in the Cambridge, MA, area. The overall goal of our research is to combine acoustic, syntactic and semantic knowledge sources. Our first experiment showed improvement by combining acoustic score and parse probability normalized for number of terminals. Results were further improved by the use of an explicit rejection criterion based on normalized parse probabilities. The use of the combined parse/acoustic score, together with the rejection criterion, gave an improvement in overall score of more than 33% on both training and test data, where score is defined as percent correct minus percent incorrect. Experiments on a fully integrated system which uses the parser to predict possible next words to the recognizer are now underway.

In this, our second Dr. J Show with Dr. Oas, she discusses the manipulation of words by organizations intent on increasing abortion, artificial contraception, and other forms of so-called "Family Planning." It is vitally important that we know what is meant by such misnomers as "reproductive health" and a myriad of other misused phrases. Dr. Oas also explains how scientific research can be manipulated for desired outcomes by organizations seeking to show an "unmet need" for abortion. She notes that the United Nations’ "non-binding" resolutions against marriage, families, and children, and in favor of promiscuous sexual "freedom," are in fact binding on the U.N. many international organizations, such as UNICEF and the U.N. Population Fund, and thus have a major impact worldwide, particularly in the Third World. This is an audio podcast of The Dr J Show (http://www.ruthinstitute.org/dr-j-show) . Full video episode is available here (http://www.ruthinstitute.org/dr-j-show/dont-use-their-language-understanding-how-words-are-redefined) , with readings & resources.

Bibliography: p. 182-191

VICE article. Researchers have identified new elements of whale vocalizations that they propose are analogous to human speech, including vowels and pitch.

VICE article. Researchers have identified new elements of whale vocalizations that they propose are analogous to human speech, including vowels and pitch.