The purpose of this paper is to provide guidelines for building a word alignment evaluation scheme. The notion of word alignment quality depends on the application: here we review standard scoring metrics for full text alignment and give explanations on how to use them better. We discuss strategies to build a reference corpus, and show that the ratio between ambiguous and unambiguous links in the reference has a great impact on scores measured with these metrics. In particular, automatically computed alignments with higher precision or higher recall can be favoured depending on the value of this ratio. Finally, we suggest a strategy to build a reference corpus particularly adapted to applications where recall plays a significant role, like in machine translation. The manually aligned corpus we built for the Spanish-English European Parliament corpus is also described. This corpus is freely available.

In this paper, we propose a segment-based annotation tool providing appropriate interactivity between a human annotator and an automatic parser. The proposed annotation tool provides the preview of a complete sentence structure suggested by the parser, and updates the preview whenever the annotator cancels or selects each segmentation point. Thus, the annotator can select the proper sentence segments maximizing parsing accuracy and minimizing human intervention. Experimental results show that the proposed tool allows the annotator to be able to reduce human intervention by approximately 39% compared with manual annotation. Sejong Korean treebank, one of the large scale treebanks, was constructed with the proposed annotation tool.

This paper describes a multi-modal corpus of hand-annotated meeting dialogues that was designed for studying addressing behaviour in face-to-face conversations. The corpus contains annotated dialogue acts, addressees, adjacency pairs and gaze direction. First, we describe the corpus design where we present the meetings collection, annotation scheme and annotation tools. Then, we present the analysis of the reproducibility and stability of the annotation scheme.

We describe a Chinese lexical semantic resource that consists of 11,765 predicates (mostly verbs and their nominalizations) analyzed with coarse-grained senses and semantic roles. We show that distinguishing senses at a coarse-grained level is a necessary part of specifying the semantic roles and describe our strategies for sense determination for purposes of predicate-argument structure specification. The semantic roles are postulated to account for syntactic variations, the different ways in which the semantic roles of a predicate are realized. The immediate purpose for this lexical semantic resource is to support the annotation of the Chinese PropBank, but we believe it can also serve as stepping stone for higher-level semantic generalizations.

Only very recently have Vietnamese researchers begun to be involved in the domain of Natural Language Processing (NLP). As there does not exist any published work in formal linguistics nor any recognizable standard for Vietnamese word definition and word categories, the fundamental tasks for automatic Vietnamese language processing, such as part-of-speech tagging, parsing, etc., are very difficult tasks for computer scientists. The fact that all necessary linguistic resources have to be built from scratch by each research team is a real obstacle to the development of Vietnamese language processing. The aim of our projects is thus to build a common linguistic database that is freely and easily exploitable for the automatic processing of Vietnamese. In this paper, we present our work on creating a Vietnamese lexicon for NLP applications. We emphasize the standardization aspect of the lexicon representation. We especially propose an extensible set of Vietnamese syntactic descriptions that can be used for tagset definition and morphosyntactic analysis. These descriptors are established in such a way as to be a reference set proposal for Vietnamese in the context of ISO subcommittee TC 37/SC 4 (Language Resource Management).

In a 12-month project we have developed a new, register-diverse, 55-million-word bilingual corpus—the New Corpus for Ireland (NCI)—to support the creation of a new English-to-Irish dictionary. The paper describes the strategies we employed, and the solutions to problems encountered. We believe we have a good model for corpus creation for lexicography, and others may find it useful as a blueprint. The corpus has two parts, one Irish, the other Hiberno-English (English as spoken in Ireland). We describe its design, collection and encoding.

We briefly discuss the origin and development of WordNet, a large lexical database for English. We outline its design and contents as well as its usefulness for Natural Language Processing. Finally, we discuss crosslinguistic WordNets and complementary lexical resources. [ABSTRACT FROM AUTHOR], Copyright of Language Resources & Evaluation is the property of Springer Nature and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Much electronic text in the languages of South Asia has been published on the Internet. However, while Unicode has emerged as the favoured encoding system of corpus and computational linguists, most South Asian language data on the web uses one of a wide range of non-standard legacy encodings. This paper describes the difficulties inherent in converting text in these encodings to Unicode. Among the various legacy encodings for South Asian scripts, the most problematic are 8-bit fonts based on graphical principles (as opposed to the logical principles of Unicode). Graphical fonts typically encode several features in ways highly incompatible with Unicode. For instance, half-form glyphs used to construct conjunct consonants are typically separate code points in 8-bit fonts; in Unicode they are represented by the full consonant followed by virama. There are many more such cases. The solution described here is an approach to text conversion based on mapping rules. A small number of generalised rules (plus the capacity for more specialised rules) captures the behaviour of each character in a font, building up a conversion algorithm for that encoding. This system is embedded in a font-mapping program, outputting CES-compliant SGML Unicode. This program, a generalised text-conversion tool, has been employed extensively in corpus-building for South Asian languages.

TimeBank is the only reference corpus for TimeML, an expressive language for annotating complex temporal information. It is a rich resource for a broad range of research into various aspects of the expression of time and temporally related events. This paper traces the development of TimeBank from its initial—and somewhat noisy—version (1.1) to a substantially revised release (1.2), now available via the Linguistic Data Consortium. The development path is motivated by the encouraging empirical results of TimeML-compliant annotators developed on the basis of TimeBank 1.1, and is informed by a detailed study of the characteristics of that initial release, which guides a clean-up process turning TimeBank 1.2 into a consistent and robust community resource.

The analysis of lectures and meetings inside smart rooms has recently attracted much interest in the literature, being the focus of international projects and technology evaluations. A key enabler for progress in this area is the availability of appropriate multimodal and multi-sensory corpora, annotated with rich human activity information during lectures and meetings. This paper is devoted to exactly such a corpus, developed in the framework of the European project CHIL, “Computers in the Human Interaction Loop”. The resulting data set has the potential to drastically advance the state-of-the-art, by providing numerous synchronized audio and video streams of real lectures and meetings, captured in multiple recording sites over the past 4 years. It particularly overcomes typical shortcomings of other existing databases that may contain limited sensory or monomodal data, exhibit constrained human behavior and interaction patterns, or lack data variability. The CHIL corpus is accompanied by rich manual annotations of both its audio and visual modalities. These provide a detailed multi-channel verbatim orthographic transcription that includes speaker turns and identities, acoustic condition information, and named entities, as well as video labels in multiple camera views that provide multi-person 3D head and 2D facial feature location information. Over the past 3 years, the corpus has been crucial to the evaluation of a multitude of audiovisual perception technologies for human activity analysis in lecture and meeting scenarios, demonstrating its utility during internal evaluations of the CHIL consortium, as well as at the recent international CLEAR and Rich Transcription evaluations. The CHIL corpus is publicly available to the research community.

In this paper we present an annotated audio–video corpus of multi-party meetings. The multimodal corpus provides for each subject involved in the experimental sessions six annotation dimensions referring to group dynamics; speech activity and body activity. The corpus is based on 11 audio and video recorded sessions which took place in a lab setting appropriately equipped with cameras and microphones. Our main concern in collecting this multimodal corpus was to explore the possibility of providing feedback services to facilitate group processes and to enhance self awareness among small groups engaged in meetings. We therefore introduce a coding scheme for annotating relevant functional roles that appear in a small group interaction. We also discuss the reliability of the coding scheme and we present the first results for automatic classification.

The AMI Meeting Corpus contains 100 h of meetings captured using many synchronized recording devices, and is designed to support work in speech and video processing, language engineering, corpus linguistics, and organizational psychology. It has been transcribed orthographically, with annotated subsets for everything from named entities, dialogue acts, and summaries to simple gaze and head movement. In this written version of an LREC conference keynote address, I describe the data and how it was created. If this is “killer” data, that presupposes a platform that it will “sell”; in this case, that is the NITE XML Toolkit, which allows a distributed set of users to create, store, browse, and search annotations for the same base data that are both time-aligned against signal and related to each other structurally.

As part of a project to construct an interactive program which would encourage children to play with language by building jokes, we developed a lexical database, starting from WordNet. To the existing information about part of speech, synonymy, hyponymy, etc., we have added phonetic representations and phonetic similarity ratings for pairs of words/phrases. [ABSTRACT FROM AUTHOR], Copyright of Language Resources & Evaluation is the property of Springer Nature and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Lexical classifications have proved useful in supporting various natural language processing (NLP) tasks. The largest verb classification for English is Levin’s (1993) work which defines groupings of verbs based on syntactic and semantic properties. VerbNet (VN) (Kipper et al. 2000; Kipper-Schuler 2005)—an extensive computational verb lexicon for English—provides detailed syntactic-semantic descriptions of Levin classes. While the classes included are extensive enough for some NLP use, they are not comprehensive. Korhonen and Briscoe (2004) have proposed a significant extension of Levin’s classification which incorporates 57 novel classes for verbs not covered (comprehensively) by Levin. Korhonen and Ryant (unpublished) have recently proposed another extension including 53 additional classes. This article describes the integration of these two extensions into VN. The result is a comprehensive Levin-style classification for English verbs providing over 90% token coverage of the Proposition Bank data (Palmer et al. 2005) and thus can be highly useful for practical applications.

In this paper we describe the current state of a new Japanese lexical resource: the Hinoki treebank. The treebank is built from dictionary definitions, examples and news text, and uses an HPSG based Japanese grammar to encode both syntactic and semantic information. It is combined with an ontology based on the definition sentences to give a detailed sense level description of the most familiar 28,000 words of Japanese.

In this paper, we present the building of various language resources for a multi-engine bi-directional English-Filipino Machine Translation (MT) system. Since linguistics information on Philippine languages are available, but as of yet, the focus has been on theoretical linguistics and little is done on the computational aspects of these languages, attempts are reported here on the manual construction of these language resources such as the grammar, lexicon, morphological information, and the corpora which were literally built from almost non-existent digital forms. Due to the inherent difficulties of manual construction, we also discuss our experiments on various technologies for automatic extraction of these resources to handle the intricacies of the Filipino language, designed with the intention of using them for the MT system. To implement the different MT engines and to ensure the improvement of translation quality, other language tools (such as the morphological analyzer and generator, and the part of speech tagger) were developed.

Since emotions are expressed through a combination of verbal and non-verbal channels, a joint analysis of speech and gestures is required to understand expressive human communication. To facilitate such investigations, this paper describes a new corpus named the “interactive emotional dyadic motion capture database” (IEMOCAP), collected by the Speech Analysis and Interpretation Laboratory (SAIL) at the University of Southern California (USC). This database was recorded from ten actors in dyadic sessions with markers on the face, head, and hands, which provide detailed information about their facial expressions and hand movements during scripted and spontaneous spoken communication scenarios. The actors performed selected emotional scripts and also improvised hypothetical scenarios designed to elicit specific types of emotions (happiness, anger, sadness, frustration and neutral state). The corpus contains approximately 12 h of data. The detailed motion capture information, the interactive setting to elicit authentic emotions, and the size of the database make this corpus a valuable addition to the existing databases in the community for the study and modeling of multimodal and expressive human communication.

We describe a suite of standards, resources and tools for computational encoding and processing of Modern Hebrew texts. These include an array of XML schemas for representing linguistic resources; a variety of text corpora, raw, automatically processed and manually annotated; lexical databases, including a broad-coverage monolingual lexicon, a bilingual dictionary and a WordNet; and morphological processors which can analyze, generate and disambiguate Hebrew word forms. The resources are developed under centralized supervision, so that they are compatible with each other. They are freely available and many of them have already been used for several applications, both academic and industrial.

The rapid development of language resources and tools using machine learning techniques for less computerized languages requires appropriately tagged corpus. A tagged Bengali news corpus has been developed from the web archive of a widely read Bengali newspaper. A web crawler retrieves the web pages in Hyper Text Markup Language (HTML) format from the news archive. At present, the corpus contains approximately 34 million wordforms. Named Entity Recognition (NER) systems based on pattern based shallow parsing with or without using linguistic knowledge have been developed using a part of this corpus. The NER system that uses linguistic knowledge has performed better yielding highest F-Score values of 75.40%, 72.30%, 71.37%, and 70.13% for person, location, organization, and miscellaneous names, respectively.

This paper concerns metaphor resource creation. It provides an account of methods used, problems discovered, and insights gained at the Hamburg Metaphor Database project, intended to inform similar resource creation initiatives, as well as future metaphor processing algorithms. After introducing the project, the theoretical underpinnings that motivate the subdivision of represented information into a conceptual and a lexical level are laid out. The acquisition of metaphor attestations from electronic corpora is explained, and annotation practices as well as database contents are evaluated. The paper concludes with an overview of related projects and an outline of possible future work.

This paper is a contribution to the discussion on compiling computational lexical resources from conventional dictionaries. It describes the theoretical as well as practical problems that are encountered when reusing a conventional dictionary for compiling a lexical-semantic resource in terms of a wordnet. More specifically, it describes the methodological issues of compiling a wordnet for Danish, DanNet, from a monolingual basis, and not—as is often seen—by applying the translational expansion method with Princeton WordNet as the English source. Thus, we apply as our basis a large, corpus-based printed dictionary of modern Danish. Using this approach, we discuss the issues of readjusting inconsistent and/or underspecified hyponymy hierarchies taken from the conventional dictionary, sense distinctions as opposed to the synonym sets of wordnets, generating semantic wordnet relations on the basis of sense definitions, and finally, supplementing missing or implicit information.

Some phrases can be interpreted in their context either idiomatically (figuratively) or literally. The precise identification of idioms is essential in order to achieve full-fledged natural language processing. Because of this, the authors of this paper have created an idiom corpus for Japanese. This paper reports on the corpus itself and the results of an idiom identification experiment conducted using the corpus. The corpus targeted 146 ambiguous idioms, and consists of 102,856 examples, each of which is annotated with a literal/idiomatic label. All sentences were collected from the World Wide Web. For idiom identification, 90 out of the 146 idioms were targeted and a word sense disambiguation (WSD) method was adopted using both common WSD features and idiom-specific features. The corpus and the experiment are both, as far as can be determined, the largest of their kinds. It was discovered that a standard supervised WSD method works well for idiom identification and it achieved accuracy levels of 89.25 and 88.86%, with and without idiom-specific features, respectively. It was also found that the most effective idiom-specific feature is the one that involves the adjacency of idiom constituents.

The aim of the French Media project was to define a protocol for the evaluation of speech understanding modules for dialog systems. Accordingly, a corpus of 1,257 real spoken dialogs related to hotel reservation and tourist information was recorded, transcribed and semantically annotated, and a semantic attribute-value representation was defined in which each conceptual relationship was represented by the names of the attributes. Two semantic annotation levels are distinguished in this approach. At the first level, each utterance is considered separately and the annotation represents the meaning of the statement without taking into account the dialog context. The second level of annotation then corresponds to the interpretation of the meaning of the statement by taking into account the dialog context; in this way a semantic representation of the dialog context is defined. This paper discusses the data collection, the detailed definition of both annotation levels, and the annotation scheme. Then the paper comments on both evaluation campaigns which were carried out during the project and discusses some results.

Recent work in computational linguistics points out the need for systems to be sensitive to the veracity or factuality of events as mentioned in text; that is, to recognize whether events are presented as corresponding to actual situations in the world, situations that have not happened, or situations of uncertain interpretation. Event factuality is an important aspect of the representation of events in discourse, but the annotation of such information poses a representational challenge, largely because factuality is expressed through the interaction of numerous linguistic markers and constructions. Many of these markers are already encoded in existing corpora, albeit in a somewhat fragmented way. In this article, we present FactBank, a corpus annotated with information concerning the factuality of events. Its annotation has been carried out from a descriptive framework of factuality grounded on both theoretical findings and data analysis. FactBank is built on top of TimeBank, adding to it an additional level of semantic information.

This article introduces ukWaC, deWaC and itWaC, three very large corpora of English, German, and Italian built by web crawling, and describes the methodology and tools used in their construction. The corpora contain more than a billion words each, and are thus among the largest resources for the respective languages. The paper also provides an evaluation of their suitability for linguistic research, focusing on ukWaC and itWaC. A comparison in terms of lexical coverage with existing resources for the languages of interest produces encouraging results. Qualitative evaluation of ukWaC versus the British National Corpus was also conducted, so as to highlight differences in corpus composition (text types and subject matters). The article concludes with practical information about format and availability of corpora and tools.

This paper describes a multichannel acoustic data collection recorded under the European DICIT project, during Wizard of Oz (WOZ) experiments carried out at FAU and FBK-irst laboratories. The application of interest in DICIT is a distant-talking interface for control of interactive TV working in a typical living room, with many interfering devices. The objective of the experiments was to collect a database supporting efficient development and tuning of acoustic processing algorithms for signal enhancement. In DICIT, techniques for sound source localization, multichannel acoustic echo cancellation, blind source separation, speech activity detection, speaker identification and verification as well as beamforming are combined to achieve a maximum possible reduction of the user speech impairments typical of distant-talking interfaces. The collected database permitted to simulate at preliminary stage a realistic scenario and to tailor the involved algorithms to the observed user behaviors. In order to match the project requirements, the WOZ experiments were recorded in three languages: English, German and Italian. Besides the user inputs, the database also contains non-speech related acoustic events, room impulse response measurements and video data, the latter used to compute three-dimensional positions of each subject. Sessions were manually transcribed and segmented at word level, introducing also specific labels for acoustic events.

We describe annotation of multiword expressions (MWEs) in the Prague dependency treebank, using several automatic pre-annotation steps. We use subtrees of the tectogrammatical tree structures of the Prague dependency treebank to store representations of the MWEs in the dictionary and pre-annotate following occurrences automatically. We also show a way to measure reliability of this type of annotation.

This article describes the design and implementation of a Dutch Electronic Lexicon of Multiword Expressions (DuELME). DuELME describes the core properties of over 5,000 Dutch multiword expressions. This article gives an overview of the decisions made in order to come to a standard lexical representation and discusses the description fields this representation comprises. We discuss the approach taken, which is innovative since it is based on the Equivalence Class Method (ECM). It is shown that introducing parameters to the ECM optimizes the method. The selection of the lexical entries and their properties is corpus-based. We describe the extraction of candidate expressions from corpora and discuss the selection criteria of the lexical entries. Moreover, we present the results of an evaluation of the standard representation in Alpino, a Dutch dependency parser.

In this paper the transcription and evaluation of the corpus DIMEx100 for Mexican Spanish is presented. First we describe the corpus and explain the linguistic and computational motivation for its design and collection process; then, the phonetic antecedents and the alphabet adopted for the transcription task are presented; the corpus has been transcribed at three different granularity levels, which are also specified in detail. The corpus statistics for each transcription level are also presented. A set of phonetic rules describing phonetic context observed empirically in spontaneous conversation is also validated with the transcription. The corpus has been used for the construction of acoustic models and a phonetic dictionary for the construction of a speech recognition system. Initial performance results suggest that the data can be used to train good quality acoustic models.

This article describes the enrichment of the AnCora corpora of Spanish and Catalan (400 k each) with coreference links between pronouns (including elliptical subjects and clitics), full noun phrases (including proper nouns), and discourse segments. The coding scheme distinguishes between identity links, predicative relations, and discourse deixis. Inter-annotator agreement on the link types is 85–89% above chance, and we provide an analysis of the sources of disagreement. The resulting corpora make it possible to train and test learning-based algorithms for automatic coreference resolution, as well as to carry out bottom-up linguistic descriptions of coreference relations as they occur in real data.

This paper describes a recently completed common resource for the study of spoken discourse, the NXT-format Switchboard Corpus. Switchboard is a long-standing corpus of telephone conversations (Godfrey et al. in SWITCHBOARD: Telephone speech corpus for research and development. In Proceedings of ICASSP-92, pp. 517–520, 1992). We have brought together transcriptions with existing annotations for syntax, disfluency, speech acts, animacy, information status, coreference, and prosody; along with substantial new annotations of focus/contrast, more prosody, syllables and phones. The combined corpus uses the format of the NITE XML Toolkit, which allows these annotations to be browsed and searched as a coherent set (Carletta et al. in Lang Resour Eval J 39(4):313–334, 2005). The resulting corpus is a rich resource for the investigation of the linguistic features of dialogue and how they interact. As well as describing the corpus itself, we discuss our approach to overcoming issues involved in such a data integration project, relevant to both users of the corpus and others in the language resource community undertaking similar projects.

We present the MATCH corpus, a unique data set of 447 dialogues in which 26 older and 24 younger adults interact with nine different spoken dialogue systems. The systems varied in the number of options presented and the confirmation strategy used. The corpus also contains information about the users’ cognitive abilities and detailed usability assessments of each dialogue system. The corpus, which was collected using a Wizard-of-Oz methodology, has been fully transcribed and annotated with dialogue acts and “Information State Update” (ISU) representations of dialogue context. Dialogue act and ISU annotations were performed semi-automatically. In addition to describing the corpus collection and annotation, we present a quantitative analysis of the interaction behaviour of older and younger users and discuss further applications of the corpus. We expect that the corpus will provide a key resource for modelling older people’s interaction with spoken dialogue systems.

The Europe Media Monitor (EMM) family of applications is a set of multilingual tools that gather, cluster and classify news in currently fifty languages and that extract named entities and quotations (reported speech) from twenty languages. In this paper, we describe the recent effort of adding the African Bantu language Swahili to EMM. EMM is designed in an entirely modular way, allowing plugging in a new language by providing the language-specific resources for that language. We thus describe the type of language-specific resources needed, the effort involved, and ways of boot-strapping the generation of these resources in order to keep the effort of adding a new language to a minimum. The text analysis applications pursued in our efforts include clustering, classification, recognition and disambiguation of named entities (persons, organisations and locations), recognition and normalisation of date expressions, as well as the identification of reported speech quotations by and about people.

Semantic interpretation of language requires extensive and rich lexical knowledge bases (LKB). The Basque WordNet is a LKB based on WordNet and its multilingual counterparts EuroWordNet and the Multilingual Central Repository. This paper reviews the theoretical and practical aspects of the Basque WordNet lexical knowledge base, as well as the steps and methodology followed in its construction. Our methodology is based on the joint development of wordnets and annotated corpora. The Basque WordNet contains 32,456 synsets and 26,565 lemmas, and is complemented by a hand-tagged corpus comprising 59,968 annotations.

We describe the Lwazi corpus for automatic speech recognition (ASR), a new telephone speech corpus which contains data from the eleven official languages of South Africa. Because of practical constraints, the amount of speech per language is relatively small compared to major corpora in world languages, and we report on our investigation of the stability of the ASR models derived from the corpus. We also report on phoneme distance measures across languages, and describe initial phone recognisers that were developed using this data. We find that a surprisingly small number of speakers (fewer than 50) and around 10 to 20 h of speech per language are sufficient for the purposes of acceptable phone-based recognition.

We present a set of language resources and tools—a morphological parser, a morphological disambiguator, and a text corpus—for exploiting Turkish morphology in natural language processing applications. The morphological parser is a state-of-the-art finite-state transducer-based implementation of Turkish morphology. The disambiguator is based on the averaged perceptron algorithm and has the best accuracy reported for Turkish in the literature. The text corpus has been compiled from the web and contains about 500 million tokens. This is the largest Turkish web corpus published.

The role of the Web for text corpus construction is becoming increasingly significant. However, the contribution of the Web is largely confined to building a general virtual corpus or low quality specialised corpora. In this paper, we introduce a new technique called SPARTAN for constructing specialised corpora from the Web by systematically analysing website contents. Our evaluations show that the corpora constructed using our technique are independent of the search engines employed. In particular, SPARTAN-derived corpora outperform all corpora based on existing techniques for the task of term recognition.

Structural metadata extraction (MDE) research aims to develop techniques for automatic conversion of raw speech recognition output to forms that are more useful to humans and downstream automatic processes. The MDE annotation includes inserting boundaries of sentence-like units to the flow of speech, labeling non-content words like filled pauses and discourse markers for optional removal, and identifying sections of disfluent speech. This paper describes design, creation, and analysis of data resources for structural MDE from spoken Czech. The annotation is based on the LDC’s MDE annotation standard for English, with changes applied to accommodate specific phenomena of Czech. In addition to the necessary language-dependent modifications, we further proposed and applied several language-independent modifications slightly refining the original annotation scheme. We created two Czech MDE speech corpora—one in the domain of broadcast news and the other in the domain of broadcast conversations. Both corpora have already been published at LDC. The analysis section of this paper presents a variety of statistics about fillers, edit disfluencies, and sentence-like units. The two Czech corpora are not only compared with each other, but also with statistics relating to the available English MDE corpora. We also report the statistics indicating that edit disfluencies have a different part of speech (POS) distribution in comparison with the overall POS distribution. The findings from the corpus analysis should help guide strategies for developing automatic MDE systems.

The blog phenomenon is universal. Blogs are characterized by their evaluative use, in that they enable Internet users to express their opinion on a given subject. From this point of view, they are an ideal resource for the constitution of an annotated sentiment analysis corpus, crossing the subject and the opinion expressed on this subject. This paper presents the Blogoscopy corpus for the French language which was built up with personal thematic blogs. The annotation was governed by three principles: theoretical, as opinion is grounded in a linguistic theory of evaluation, practical, as every opinion is linked to an object, and methodological as annotation rules and successive phases are defined to ensure quality and thoroughness.

Verb Phrase Ellipsis (VPE) has been studied in great depth in theoretical linguistics, but empirical studies of VPE are rare. We extend the few previous corpus studies with an annotated corpus of VPE in all 25 sections of the Wall Street Journal corpus (WSJ) distributed with the Penn Treebank. We annotated the raw files using a stand-off annotation scheme that codes the auxiliary verb triggering the elided verb phrase, the start and end of the antecedent, the syntactic type of antecedent (VP, TV, NP, PP or AP), and the type of syntactic pattern between the source and target clauses of the VPE and its antecedent. We found 487 instances of VPE (including predicative ellipsis, antecedent-contained deletion, comparative constructions, and pseudo-gapping) plus 67 cases of related phenomena such as do so anaphora. Inter-annotator agreement was high, with a 0.97 average F-score for three annotators for one section of the WSJ. Our annotation is theory neutral, and has better coverage than earlier efforts that relied on automatic methods, e.g. simply searching the parsed version of the Penn Treebank for empty VP’s achieves a high precision (0.95) but low recall (0.58) when compared with our manual annotation. The distribution of VPE source–target patterns deviates highly from the standard examples found in the theoretical linguistics literature on VPE, once more underlining the value of corpus studies. The resulting corpus will be useful for studying VPE phenomena as well as for evaluating natural language processing systems equipped with ellipsis resolution algorithms, and we propose evaluation measures for VPE detection and VPE antecedent selection. The stand-off annotation is freely available for research purposes.

Research in machine translation and corpus annotation has greatly benefited from the increasing availability of word-aligned parallel corpora. This paper presents ongoing research on the development and application of the sawa corpus, a two-million-word parallel corpus English—Swahili. We describe the data collection phase and zero in on the difficulties of finding appropriate and easily accessible data for this language pair. In the data annotation phase, the corpus was semi-automatically sentence and word-aligned and morphosyntactic information was added to both the English and Swahili portion of the corpus. The annotated parallel corpus allows us to investigate two possible uses. We describe experiments with the projection of part-of-speech tagging annotation from English onto Swahili, as well as the development of a basic statistical machine translation system for this language pair, using the parallel corpus and a consolidated database of existing English—Swahili translation dictionaries. We particularly focus on the difficulties of translating English into the morphologically more complex Bantu language of Swahili.

Plagiarism is widely acknowledged to be a significant and increasing problem for higher education institutions (McCabe 2005; Judge 2008). A wide range of solutions, including several commercial systems, have been proposed to assist the educator in the task of identifying plagiarised work, or even to detect them automatically. Direct comparison of these systems is made difficult by the problems in obtaining genuine examples of plagiarised student work. We describe our initial experiences with constructing a corpus consisting of answers to short questions in which plagiarism has been simulated. This corpus is designed to represent types of plagiarism that are not included in existing corpora and will be a useful addition to the set of resources available for the evaluation of plagiarism detection systems.

In this paper, we describe tools and resources for the study of African languages developed at the Collaborative Research Centre 632 “Information Structure”. These include deeply annotated data collections of 25 sub-Saharan languages that are described together with their annotation scheme, as well as the corpus tool ANNIS, which provides unified access to a broad variety of annotations created with a range of different tools. With the application of ANNIS to several African data collections, we illustrate its suitability for the purpose of language documentation, distributed access, and the creation of data archives.

This paper addresses some of the issues learned during the course of building a written language resource, called ‘Peykare’, for the contemporary Persian. After defining five linguistic varieties and 24 different registers based on these linguistic varieties, we collected texts for Peykare to do a linguistic analysis, including cross-register differences. For tokenization of Persian, we propose a descriptive generalization to normalize orthographic variations existing in texts. To annotate Peykare, we use EAGLES guidelines which result to have a hierarchy in the part-of-speech tags. To this aim, we apply a semi-automatic approach for the annotation methodology. In the paper, we also give a special attention to the Ezafe construction and homographs which are important in Persian text analyses.

The focus of this article is on the creation of a collection of sentences manually annotated with respect to their sentence structure. We show that the concept of linear segments—linguistically motivated units, which may be easily detected automatically—serves as a good basis for the identification of clauses in Czech. The segment annotation captures such relationships as subordination, coordination, apposition and parenthesis; based on segmentation charts, individual clauses forming a complex sentence are identified. The annotation of a sentence structure enriches a dependency-based framework with explicit syntactic information on relations among complex units like clauses. We have gathered a collection of 3,444 sentences from the Prague Dependency Treebank, which were annotated with respect to their sentence structure (these sentences comprise 10,746 segments forming 6,341 clauses). The main purpose of the project is to gain a development data—promising results for Czech NLP tools (as a dependency parser or a machine translation system for related languages) that adopt an idea of clause segmentation have been already reported. The collection of sentences with annotated sentence structure provides the possibility of further improvement of such tools.

Arabic is the most widely spoken language in the Arab World. Most people of the Islamic World understand the Classic Arabic language because it is the language of the Qur’an. Despite the fact that in the last decade the number of Arabic Internet users (Middle East and North and East of Africa) has increased considerably, systems to analyze Arabic digital resources automatically are not as easily available as they are for English. Therefore, in this work, an attempt is made to build a real time Named Entity Recognition system that can be used in web applications to detect the appearance of specific named entities and events in news written in Arabic. Arabic is a highly inflectional language, thus we will try to minimize the impact of Arabic affixes on the quality of the pattern recognition model applied to identify named entities. These patterns are built up by processing and integrating different gazetteers, from DBPedia (http://dbpedia.org/About, 2009) to GATE (A general architecture for text engineering, 2009) and ANERGazet (http://users.dsic.upv.es/grupos/nle/?file=kop4.php).

In this paper, we present first results from annotating abstract (discourse-deictic) anaphora in German. Our annotation guidelines provide linguistic tests for identifying the antecedent, and for determining the semantic types of both the antecedent and the anaphor. The corpus consists of selected speaker turns from the Europarl corpus. To date, 100 texts have been annotated according to these guidelines. The annotations show that anaphoric personal and demonstrative pronouns differ with respect to the distance to their antecedents. A semantic analysis reveals that, contrary to suggestions put forward in the literature, referents of anaphors do not tend to be more abstract than the referents of their antecedents.

Although superlatives are commonly used in natural language, so far there has been no large-scale computational investigation of the types of comparisons they express. This article describes a comprehensive annotation scheme for superlatives, which classifies superlatives according to their surface forms and motivates an initial focus on so-called “ISA superlatives”. This type of superlative comparison is especially suitable for a computational approach because both their targets and comparison sets are explicitly realised in the text, and the proposed annotation scheme offers guidelines for annotating the spans of such comparative elements. The annotations are tested and evaluated on 500 tokens of superlatives with good inter-annotator agreement. In addition to providing a platform for investigating superlatives on a larger scale, this research also introduces a new text-based Wikipedia corpus in which all superlative instances have been annotated according to the proposed annotation scheme, and which has been used to develop a tool that can reliably distinguish between different superlative types, and identify the comparative components of ISA superlatives.