The development of language technology (LT) for an endangered language is often identified as a goal in language revitalization efforts, but developing such technologies is typically subject to additional methodological challenges as well as social and ethical concerns. In particular, LT development has too often taken on colonialist qualities, extracting language data, relying on outside experts, and denying the speakers of a language sovereignty over the technologies produced.We seek to avoid such an approach through the development of the Building Endangered Language Technology (BELT) website, an educational resource designed for speakers and community members with limited technological experience to develop LTs for their own language. Specifically, BELT provides interactive lessons on basic Python programming, coupled with projects to develop specific language technologies, such as spellcheckers or word games. In this paper, we describe BELT’s design, the motivation underlying many key decisions, and preliminary responses from learners.

Distributional approaches have proven effective in modeling semantics and phonology through vector embeddings. We explore whether distributional representations can also effectively model morphological information. We train static vector embeddings over morphological sequences. Then, we explore morpheme categories for fusional morphemes, which encode multiple linguistic dimensions, and often have close relationships to other morphemes. We study whether the learned vector embeddings align with these linguistic dimensions, finding strong evidence that this is the case. Our work uses two low-resource languages, Uspanteko and Tsez, demonstrating that distributional morphological representations are effective even with limited data.

Uniform Meaning Representation (UMR) is a semantic labeling system in the AMR family designed to be uniformly applicable to typologically diverse languages. The UMR labeling system is quite thorough and can be time-consuming to execute, especially if annotators are starting from scratch. In this paper, we focus on methods for bootstrapping UMR annotations for a given language from existing resources, and specifically from typical products of language documentation work, such as lexical databases and interlinear glossed text (IGT). Using Arapaho as our test case, we present and evaluate a bootstrapping process that automatically generates UMR subgraphs from IGT. Additionally, we describe and evaluate a method for bootstrapping valency lexicon entries from lexical databases for both the target language and English. We are able to generate enough basic structure in UMR graphs from the existing Arapaho interlinearized texts to automate UMR labeling to a significant extent. Our method thus has the potential to streamline the process of building meaning representations for new languages without existing large-scale computational resources.

This paper reports the first release of the UMR (Uniform Meaning Representation) data set. UMR is a graph-based meaning representation formalism consisting of a sentence-level graph and a document-level graph. The sentence-level graph represents predicate-argument structures, named entities, word senses, aspectuality of events, as well as person and number information for entities. The document-level graph represents coreferential, temporal, and modal relations that go beyond sentence boundaries. UMR is designed to capture the commonalities and variations across languages and this is done through the use of a common set of abstract concepts, relations, and attributes as well as concrete concepts derived from words from invidual languages. This UMR release includes annotations for six languages (Arapaho, Chinese, English, Kukama, Navajo, Sanapana) that vary greatly in terms of their linguistic properties and resource availability. We also describe on-going efforts to enlarge this data set and extend it to other genres and modalities. We also briefly describe the available infrastructure (UMR annotation guidelines and tools) that others can use to create similar data sets.

Despite Uniform Meaning Representation’s (UMR) potential for cross-lingual semantics, limited annotated data has hindered its adoption. There are large datasets of English AMRs (Abstract Meaning Representations), but the process of converting AMR graphs to UMR graphs is non-trivial. In this paper we address a complex piece of that conversion process, namely cases where one AMR role can be mapped to multiple UMR roles through a non-deterministic process. We propose a neuro-symbolic method for role conversion, integrating animacy parsing and logic rules to guide a neural network, and minimizing human intervention. On test data, the model achieves promising accuracy, highlighting its potential to accelerate AMR-to-UMR conversion. Future work includes expanding animacy parsing, incorporating human feedback, and applying the method to broader aspects of conversion. This research demonstrates the benefits of combining symbolic and neural approaches for complex semantic tasks.

Canonical morphological segmentation is the process of analyzing words into the standard (aka underlying) forms of their constituent morphemes.This is a core task in endangered language documentation, and NLP systems have the potential to dramatically speed up this process. In typical language documentation settings, training data for canonical morpheme segmentation is scarce, making it difficult to train high quality models. However, translation data is often much more abundant, and, in this work, we present a method that attempts to leverage translation data in the canonical segmentation task. We propose a character-level sequence-to-sequence model that incorporates representations of translations obtained from pretrained high-resource monolingual language models as an additional signal. Our model outperforms the baseline in a super-low resource setting but yields mixed results on training splits with more data. Additionally, we find that we can achieve strong performance even without needing difficult-to-obtain word level alignments. While further work is needed to make translations useful in higher-resource settings, our model shows promise in severely resource-constrained settings.

This paper describes the LECS Lab submission to the AmericasNLP 2024 Shared Task on the Creation of Educational Materials for Indigenous Languages. The task requires transforming a base sentence with regards to one or more linguistic properties (such as negation or tense). We observe that this task shares many similarities with the well-studied task of word-level morphological inflection, and we explore whether the findings from inflection research are applicable to this task. In particular, we experiment with a number of augmentation strategies, finding that they can significantly benefit performance, but that not all augmented data is necessarily beneficial. Furthermore, we find that our character-level neural models show high variability with regards to performance on unseen data, and may not be the best choice when training data is limited.

In this work, we present the results of the AmericasNLP 2023 Shared Task on Machine Translation into Indigenous Languages of the Americas. This edition of the shared task featured eleven language pairs, one of which – Chatino-Spanish – uses a newly collected evaluation dataset, consisting of professionally translated text from the legal domain. Seven teams participated in the shared task, with a total of 181 submissions. Additionally, we conduct a human evaluation of the best system outputs, and compare them to the best submissions from the prior shared task. We find that this analysis agrees with the quantitative measures used to rank submissions, which shows further improvements of 9.64 ChrF on average across all languages, when compared to the prior winning system.

Generalization is of particular importance in resource-constrained settings, where the available training data may represent only a small fraction of the distribution of possible texts. We investigate the ability of morpheme labeling models to generalize by evaluating their performance on unseen genres of text, and we experiment with strategies for closing the gap between performance on in-distribution and out-of-distribution data. Specifically, we use weight decay optimization, output denoising, and iterative pseudo-labeling, and achieve a 2% improvement on a test set containing texts from unseen genres. All experiments are performed using texts written in the Mayan language Uspanteko.

Aspectual meaning refers to how the internal temporal structure of situations is presented. This includes whether a situation is described as a state or as an event, whether the situation is finished or ongoing, and whether it is viewed as a whole or with a focus on a particular phase. This survey gives an overview of computational approaches to modeling lexical and grammatical aspect along with intuitive explanations of the necessary linguistic concepts and terminology. In particular, we describe the concepts of stativity, telicity, habituality, perfective and imperfective, as well as influential inventories of eventuality and situation types. Aspect is a crucial component of semantics, especially for precise reporting of the temporal structure of situations, and future NLP approaches need to be able to handle and evaluate it systematically.

State-of-the-art models for identifying offensive language often fail to generalize over more nuanced or implicit cases of offensive and hateful language. Understanding model performance on complex cases is key for building robust models that are effective in real-world settings. To help researchers efficiently evaluate their models, we introduce OLEA, a diagnostic, open-source, extensible Python library that provides easy-to-use tools for error analysis in the context of detecting offensive language in English. OLEA packages analyses and datasets proposed by prior scholarship, empowering researchers to build effective, explainable and generalizable offensive language classifiers.

This paper presents detailed mappings between the structures used in Abstract Meaning Representation (AMR) and those used in Uniform Meaning Representation (UMR). These structures include general semantic roles, rolesets, and concepts that are largely shared between AMR and UMR, but with crucial differences. While UMR annotation of new low-resource languages is ongoing, AMR-annotated corpora already exist for many languages, and these AMR corpora are ripe for conversion to UMR format. Rather than focusing on semantic coverage that is new to UMR (which will likely need to be dealt with manually), this paper serves as a resource (with illustrated mappings) for users looking to understand the fine-grained adjustments that have been made to the representation techniques for semantic categoriespresent in both AMR and UMR.

Linguistic analysis is a core task in the process of documenting, analyzing, and describing endangered and less-studied languages. In addition to providing insight into the properties of the language being studied, having tools to automatically label words in a language for grammatical category and morphological features can support a range of applications useful for language pedagogy and revitalization. At the same time, most modern NLP methods for these tasks require both large amounts of data in the language and compute costs well beyond the capacity of most research groups and language communities. In this paper, we present a gloss-to-gloss (g2g) model for linguistic analysis (specifically, morphological analysis and part-of-speech tagging) that is lightweight in terms of both data requirements and computational expense. The model is designed for the interlinear glossed text (IGT) format, in which we expect the source text of a sentence in a low-resource language, a translation of that sentence into a language of wider communication, and a detailed glossing of the morphological properties of each word in the sentence. We first produce silver standard parallel glossed data by automatically labeling the high-resource translation. The model then learns to transform source language morphological labels into output labels for the target language, mediated by a structured linguistic representation layer. We test the model on both low-resource and high-resource languages, and find that our simple CNN-based model achieves comparable performance to a state-of-the-art transformer-based model, at a fraction of the computational cost.

This paper presents the findings of the SIGMORPHON 2023 Shared Task on Interlinear Glossing. This first iteration of the shared task explores glossing of a set of six typologically diverse languages: Arapaho, Gitksan, Lezgi, Natügu, Tsez and Uspanteko. The shared task encompasses two tracks: a resource-scarce closed track and an open track, where participants are allowed to utilize external data resources. Five teams participated in the shared task. The winning team Tü-CL achieved a 23.99%-point improvement over a baseline RoBERTa system in the closed track and a 17.42%-point improvement in the open track.

Rooted in AMR, Uniform Meaning Representation (UMR) is a graph-based formalism with nodes as concepts and edges as relations between them. When used to represent natural language semantics, UMR maps words in a sentence to concepts in the UMR graph. Multiword expressions (MWEs) pose a particular challenge to UMR annotation because they deviate from the default one-to-one mapping between words and concepts. There are different types of MWEs which require different kinds of annotation that must be specified in guidelines. This paper discusses the specific treatment for each type of MWE in UMR.

Pretrained multilingual models are able to perform cross-lingual transfer in a zero-shot setting, even for languages unseen during pretraining. However, prior work evaluating performance on unseen languages has largely been limited to low-level, syntactic tasks, and it remains unclear if zero-shot learning of high-level, semantic tasks is possible for unseen languages. To explore this question, we present AmericasNLI, an extension of XNLI (Conneau et al., 2018) to 10 Indigenous languages of the Americas. We conduct experiments with XLM-R, testing multiple zero-shot and translation-based approaches. Additionally, we explore model adaptation via continued pretraining and provide an analysis of the dataset by considering hypothesis-only models. We find that XLM-R’s zero-shot performance is poor for all 10 languages, with an average performance of 38.48%. Continued pretraining offers improvements, with an average accuracy of 43.85%. Surprisingly, training on poorly translated data by far outperforms all other methods with an accuracy of 49.12%.

For the task of classifying verbs in context as dynamic or stative, current models approach human performance, but only for particular data sets. To better understand the performance of such models, and how well they are able to generalize beyond particular test sets, we apply the contrast set (Gardner et al., 2020) methodology to stativity classification. We create nearly 300 contrastive pairs by perturbing test set instances just enough to change their labels from one class to the other, while preserving coherence, meaning, and well-formedness. Contrastive evaluation shows that a model with near-human performance on an in-distribution test set degrades substantially when applied to transformed examples, showing that the stative vs. dynamic classification task is more complex than the model performance might otherwise suggest. Code and data are freely available.

Language documentation encompasses translation, typically into the dominant high-resource language in the region where the target language is spoken. To make data accessible to a broader audience, additional translation into other high-resource languages might be needed. Working within a project documenting Kotiria, we explore the extent to which state-of-the-art machine translation (MT) systems can support this second translation – in our case from Portuguese to English. This translation task is challenging for multiple reasons: (1) the data is out-of-domain with respect to the MT system’s training data, (2) much of the data is conversational, (3) existing translations include non-standard and uncommon expressions, often reflecting properties of the documented language, and (4) the data includes borrowings from other regional languages. Despite these challenges, existing MT systems perform at a usable level, though there is still room for improvement. We then conduct a qualitative analysis and suggest ways to improve MT between high-resource languages in a language documentation setting.

This paper presents the results of the 2021 Shared Task on Open Machine Translation for Indigenous Languages of the Americas. The shared task featured two independent tracks, and participants submitted machine translation systems for up to 10 indigenous languages. Overall, 8 teams participated with a total of 214 submissions. We provided training sets consisting of data collected from various sources, as well as manually translated sentences for the development and test sets. An official baseline trained on this data was also provided. Team submissions featured a variety of architectures, including both statistical and neural models, and for the majority of languages, many teams were able to considerably improve over the baseline. The best performing systems achieved 12.97 ChrF higher than baseline, when averaged across languages.

This paper presents two different systems for unsupervised clustering of morphological paradigms, in the context of the SIGMORPHON 2021 Shared Task 2. The goal of this task is to correctly cluster words in a given language by their inflectional paradigm, without any previous knowledge of the language and without supervision from labeled data of any sort. The words in a single morphological paradigm are different inflectional variants of an underlying lemma, meaning that the words share a common core meaning. They also - usually - show a high degree of orthographical similarity. Following these intuitions, we investigate KMeans clustering using two different types of word representations: one focusing on orthographical similarity and the other focusing on semantic similarity. Additionally, we discuss the merits of randomly initialized centroids versus pre-defined centroids for clustering. Pre-defined centroids are identified based on either a standard longest common substring algorithm or a connected graph method built off of longest common substring. For all development languages, the character-based embeddings perform similarly to the baseline, and the semantic embeddings perform well below the baseline. Analysis of the systems’ errors suggests that clustering based on orthographic representations is suitable for a wide range of morphological mechanisms, particularly as part of a larger system.

As machine translation (MT) systems progress at a rapid pace, questions of their adequacy linger. In this study we focus on negation, a universal, core property of human language that significantly affects the semantics of an utterance. We investigate whether translating negation is an issue for modern MT systems using 17 translation directions as test bed. Through thorough analysis, we find that indeed the presence of negation can significantly impact downstream quality, in some cases resulting in quality reductions of more than 60%. We also provide a linguistically motivated analysis that directly explains the majority of our findings. We release our annotations and code to replicate our analysis here: https://github.com/mosharafhossain/negation-mt.

Despite recent advances in natural language processing and other language technology, the application of such technology to language documentation and conservation has been limited. In August 2019, a workshop was held at Carnegie Mellon University in Pittsburgh, PA, USA to attempt to bring together language community members, documentary linguists, and technologists to discuss how to bridge this gap and create prototypes of novel and practical language revitalization technologies. The workshop focused on developing technologies to aid language documentation and revitalization in four areas: 1) spoken language (speech transcription, phone to orthography decoding, text-to-speech and text-speech forced alignment), 2) dictionary extraction and management, 3) search tools for corpora, and 4) social media (language learning bots and social media analysis). This paper reports the results of this workshop, including issues discussed, and various conceived and implemented technologies for nine languages: Arapaho, Cayuga, Inuktitut, Irish Gaelic, Kidaw’ida, Kwak’wala, Ojibwe, San Juan Quiahije Chatino, and Seneca.

This paper presents WikiPossessions, a new benchmark corpus for the task of temporally-oriented possession (TOP), or tracking objects as they change hands over time. We annotate Wikipedia articles for 90 different well-known artifacts paintings, diamonds, and archaeological artifacts), producing 799 artifact-possessor relations with associated attributes. For each article, we also produce a full possession timeline. The full version of the task combines straightforward entity-relation extraction with complex temporal reasoning, as well as verification of textual support for the relevant types of knowledge. Specifically, to complete the full TOP task for a given article, a system must do the following: a) identify possessors; b) anchor possessors to times/events; c) identify temporal relations between each temporal anchor and the possession relation it corresponds to; d) assign certainty scores to each possessor and each temporal relation; and e) assemble individual possession events into a global possession timeline. In addition to the corpus, we release evaluation scripts and a baseline model for the task.

The focus of a negation is the set of tokens intended to be negated, and a key component for revealing affirmative alternatives to negated utterances. In this paper, we experiment with neural networks to predict the focus of negation. Our main novelty is leveraging a scope detector to introduce the scope of negation as an additional input to the network. Experimental results show that doing so obtains the best results to date. Additionally, we perform a detailed error analysis providing insights into the main error categories, and analyze errors depending on whether the model takes into account scope and context information.

Our system for the PropEval task explores the ability of semantic features to detect and label propagandistic rhetorical techniques in English news articles. For Subtask 2, labeling identified propagandistic fragments with one of fourteen technique labels, our system attains a micro-averaged F1 of 0.40; in this paper, we take a detailed look at the fourteen labels and how well our semantically-focused model detects each of them. We also propose strategies to fill the gaps.

This paper outlines our approach to Tasks A & B for the English Language track of SemEval-2020 Task 12: OffensEval 2: Multilingual Offensive Language Identification in Social Media. We use a Linear SVM with document vectors computed from pre-trained word embeddings, and we explore the effectiveness of lexical, part of speech, dependency, and named entity (NE) features. We manually annotate a subset of the training data, which we use for error analysis and to tune a threshold for mapping training confidence values to labels. While document vectors are consistently the most informative features for both tasks, testing on the development set suggests that dependency features are an effective addition for Task A, and NE features for Task B.

Negation often conveys implicit positive meaning. In this paper, we present a corpus of negations and their underlying positive interpretations. We work with negations from Simple Wikipedia, automatically generate potential positive interpretations, and then collect manual annotations that effectively rewrite the negation in positive terms. This procedure yields positive interpretations for approximately 77% of negations, and the final corpus includes over 5,700 negations and over 5,900 positive interpretations. We also present baseline results using seq2seq neural models.

This paper presents the UNT HiLT+Ling system for the Sigmorphon 2019 shared Task 2: Morphological Analysis and Lemmatization in Context. Our core approach focuses on the morphological tagging task; part-of-speech tagging and lemmatization are treated as secondary tasks. Given the highly multilingual nature of the task, we propose an approach which makes minimal use of the supplied training data, in order to be extensible to languages without labeled training data for the morphological inflection task. Specifically, we use a parallel Bible corpus to align contextual embeddings at the verse level. The aligned verses are used to build cross-language translation matrices, which in turn are used to map between embedding spaces for the various languages. Finally, we use sets of inflected forms, primarily from a high-resource language, to induce vector representations for individual UniMorph tags. Morphological analysis is performed by matching vector representations to embeddings for individual tokens. While our system results are dramatically below the average system submitted for the shared task evaluation campaign, our method is (we suspect) unique in its minimal reliance on labeled training data.

This paper presents models to predict event durations. We introduce aspectual features that capture deeper linguistic information than previous work, and experiment with neural networks. Our analysis shows that tense, aspect and temporal structure of the clause provide useful clues, and that an LSTM ensemble captures relevant context around the event.

This paper focuses on a particular type of abusive language, targeting expressions in which typically neutral adjectives take on pejorative meaning when used as nouns - compare ‘gay people’ to ‘the gays’. We first collect and analyze a corpus of hand-curated, expert-annotated pejorative nominalizations for four target adjectives: female, gay, illegal, and poor. We then collect a second corpus of automatically-extracted and POS-tagged, crowd-annotated tweets. For both corpora, we find support for the hypothesis that some adjectives, when nominalized, take on negative meaning. The targeted constructions are non-standard yet widely-used, and part-of-speech taggers mistag some nominal forms as adjectives. We implement a tool called NomCatcher to correct these mistaggings, and find that the same tool is effective for identifying new adjectives subject to transformation via nominalization into abusive language.

While there is wide acknowledgement in NLP of the utility of document characterization by genre, it is quite difficult to determine a definitive set of features or even a comprehensive list of genres. This paper addresses both issues. First, with prototype semantics, we develop a hierarchical taxonomy of discourse functions. We implement the taxonomy by developing a new text genre corpus of contemporary German to perform a text based comparative register analysis. Second, we extract a host of style features, both deep and shallow, aiming beyond linguistically motivated features at situational correlates in texts. The feature sets are used for supervised text genre classification, on which our models achieve high accuracy. The combination of the corpus typology and feature sets allows us to characterize types of communicative purpose in a comparative setup, by qualitative interpretation of style feature loadings of a regularized discriminant analysis. Finally, to determine the dependence of genre on topics (which are arguably the distinguishing factor of sub-genre), we compare and combine our style models with Latent Dirichlet Allocation features across different corpus settings with unstable topics.

Detecting aspectual properties of clauses in the form of situation entity types has been shown to depend on a combination of syntactic-semantic and contextual features. We explore this task in a deep-learning framework, where tuned word representations capture lexical, syntactic and semantic features. We introduce an attention mechanism that pinpoints relevant context not only for the current instance, but also for the larger context. Apart from implicitly capturing task relevant features, the advantage of our neural model is that it avoids the need to reproduce linguistic features for other languages and is thus more easily transferable. We present experiments for English and German that achieve competitive performance. We present a novel take on modeling and exploiting genre information and showcase the adaptation of our system from one language to another.

Modal verbs have different interpretations depending on their context. Their sense categories – epistemic, deontic and dynamic – provide important dimensions of meaning for the interpretation of discourse. Previous work on modal sense classification achieved relatively high performance using shallow lexical and syntactic features drawn from small-size annotated corpora. Due to the restricted empirical basis, it is difficult to assess the particular difficulties of modal sense classification and the generalization capacity of the proposed models. In this work we create large-scale, high-quality annotated corpora for modal sense classification using an automatic paraphrase-driven projection approach. Using the acquired corpora, we investigate the modal sense classification task from different perspectives.

We present LQVSumm, a corpus of about 2000 automatically created extractive multi-document summaries from the TAC 2011 shared task on Guided Summarization, which we annotated with several types of linguistic quality violations. Examples for such violations include pronouns that lack antecedents or ungrammatical clauses. We give details on the annotation scheme and show that inter-annotator agreement is good given the open-ended nature of the task. The annotated summaries have previously been scored for Readability on a numeric scale by human annotators in the context of the TAC challenge; we show that the number of instances of violations of linguistic quality of a summary correlates with these intuitively assigned numeric scores. On a system-level, the average number of violations marked in a system’s summaries achieves higher correlation with the Readability scores than current supervised state-of-the-art methods for assigning a single readability score to a summary. It is our hope that our corpus facilitates the development of methods that not only judge the linguistic quality of automatically generated summaries as a whole, but which also allow for detecting, labeling, and fixing particular violations in a text.

n this paper we investigate the potential of answer clustering for semi-automatic scoring of short answer questions for German as a foreign language. We use surface features like word and character n-grams to cluster answers to listening comprehension exercises per question and simulate having human graders only label one answer per cluster and then propagating this label to all other members of the cluster. We investigate various ways to select this single item to be labeled and find that choosing the item closest to the centroid of a cluster leads to improved (simulated) grading accuracy over random item selection. Averaged over all questions, we can reduce a teacher’s workload to labeling only 40% of all different answers for a question, while still maintaining a grading accuracy of more than 85%.

This paper reports on an ongoing project that uses varied language resources and advanced NLP tools for a linguistic classification task in discourse semantics. The system we present is designed to assign a "situation entity" class label to each predicator in English text. The project goal is to achieve the best-possible identification of situation entities in naturally-occurring written texts by implementing a robust system that will deal with real corpus material, rather than just with constructed textbook examples of discourse. In this paper we focus on the combination of multiple information sources, which we see as being vital for a robust classification system. We use a deep syntactic grammar of English to identify morphological, syntactic, and discourse clues, and we use various lexical databases for fine-grained semantic properties of the predicators. Experiments performed to date show that enhancing the output of the grammar with information from lexical resources improves recall but lowers precision in the situation entity classification task.