Abstract While existing concreteness norms primarily target words in isolation, little attention has been paid to concreteness in context. To address this, we systematically collect multilingual concreteness ratings using Best-Worst Scaling (BWS) for 5,814 verb-direct object noun expressions in three languages with different degrees of resource availability: English, German, and Slovene. We identify consistent patterns where the concreteness of verb-noun combinations is more strongly influenced by the nominal object than the verb. Through comparative analyses on an English subset, we demonstrate that BWS guarantees more reliable concreteness judgments than traditional rating scales. Expanding beyond our human-generated data, we use traditional and LLM-based automatic extrapolation methods to generate a large-scale multilingual resource of over 430,000 expressions. Additionally, we conduct a study examining the interaction between concreteness and literal vs. figurative judgments for a subset of 1,800 expressions in all three languages, along with example usage sentences. Our findings show that lower concreteness ratings correlate with figurative language, thus reinforcing the link between abstractness and figurativeness. All resources are available from https://github.com/urbikn/multilingual-concreteness-vo.

We analyze the evolution of English noun compounds, which we represent as vectors of timespecific values.We implement a wide array of methods to create a rich set of features, using them to classify compounds for present-day compositionality and to assess the informativeness of the corresponding linguistic patterns.Our best results use BERT -reflecting the similarity of compounds and sentence contextsand we further capture relevant and complementary information across approaches.Leveraging these feature differences, we find that the development of low-compositional meanings is reflected by a parallel drop in compositionality and sustained semantic change.The same distinction is echoed in transformer processing: compositionality estimates require far less contextualization than semantic change estimates.

Abstract concepts like mercy and peace often lack clear visual grounding, and thus challenge humans and models to provide suitable image representations. To address this challenge, we introduce AbsVis – a dataset of 675 images annotated with 14,175 concept–explanation attributions from humans and two Vision-Language Models (VLMs: Qwen and LLaVA), where each concept is accompanied by a textual explanation. We compare human and VLM attributions in terms of diversity, abstractness, and alignment, and find that humans attribute more varied concepts. AbsVis also includes 2,680 human preference judgments evaluating the quality of a subset of these annotations, showing that overlapping concepts (attributed by both humans and VLMs) are most preferred. Explanations clarify and strengthen the perceived attributions, both from humans and VLMs. Explanations clarify and strengthen the perceived attributions, both from human and VLMs. Finally, we show that VLMs can approximate human preferences and use them to fine-tune VLMs via Direct Preference Optimization (DPO), yielding improved alignments with preferred concept–explanation pairs.

We present a comprehensive computational study of the under-investigated phenomenon of personal name compounds (PNCs) in German such as Willkommens-Merkel (‘Welcome-Merkel’). Prevalent in news, social media, and political discourse, PNCs are hypothesized to exhibit an evaluative function that is reflected in a more positive or negative perception as compared to the respective personal full name (such as Angela Merkel). We model 321 PNCs and their corresponding full names at discourse level, and show that PNCs bear an evaluative nature that can be captured through a variety of computational methods. Specifically, we assess through valence information whether a PNC is more positively or negatively evaluative than the person’s name, by applying and comparing two approaches using (i) valence norms and (ii) pre-trained language models (PLMs). We further enrich our data with personal, domain-specific, and extra-linguistic information and perform a range of regression analyses revealing that factors including compound and modifier valence, domain, and political party membership influence how a PNC is evaluated.

Predicting the compositionality of noun compounds such as climate change and tennis elbow is a vital component in natural language understanding. While most previous computational methods that automatically determine the semantic relatedness between compounds and their constituents have applied a synchronic perspective, the current study investigates what diachronic changes in contexts and semantic topics of compounds and constituents reveal about the compounds’ present-day degrees of compositionality. We define a binary classification task that utilizes two diachronic vector spaces based on contextual co-occurrences and semantic topics, and demonstrate that diachronic changes in cosine similarities – measured over context or topic distributions – uncover patterns that distinguish between compounds with low and high present-day compositionality. Despite fewer dimensions in the topic models, the topic space performs on par with the co-occurrence space and captures rather similar information. Temporal similarities between compounds and modifiers as well as between compounds and their prepositional paraphrases predict the compounds’ present-day compositionality with accuracy >0.7.

Dominik Schlechtweg, Pierluigi Cassotti, Bill Noble, David Alfter, Sabine Schulte Im Walde, Nina Tahmasebi. Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. 2024.

The visual representation of a concept varies significantly depending on its meaning and the context where it occurs; this poses multiple challenges both for vision and multimodal models. Our study focuses on concreteness, a well-researched lexical-semantic variable, using it as a case study to examine the variability in visual representations. We rely on images associated with approximately 1,000 abstract and concrete concepts extracted from two different datasets: Bing and YFCC. Our goals are: (i) evaluate whether visual diversity in the depiction of concepts can reliably distinguish between concrete and abstract concepts; (ii) analyze the variability of visual features across multiple images of the same concept through a nearest neighbor analysis; and (iii) identify challenging factors contributing to this variability by categorizing and annotating images. Our findings indicate that for classifying images of abstract versus concrete concepts, a combination of basic visual features such as color and texture is more effective than features extracted by more complex models like Vision Transformer (ViT). However, ViTs show better performances in the nearest neighbor analysis, emphasizing the need for a careful selection of visual features when analyzing conceptual variables through modalities other than text.

Abstract Multiword expressions (MWEs) are composed of multiple words and exhibit variable degrees of compositionality. As such, their meanings are notoriously difficult to model, and it is unclear to what extent this issue affects transformer architectures. Addressing this gap, we provide the first in-depth survey of MWE processing with transformer models. We overall find that they capture MWE semantics inconsistently, as shown by reliance on surface patterns and memorized information. MWE meaning is also strongly localized, predominantly in early layers of the architecture. Representations benefit from specific linguistic properties, such as lower semantic idiosyncrasy and ambiguity of target expressions. Our findings overall question the ability of transformer models to robustly capture fine-grained semantics. Furthermore, we highlight the need for more directly comparable evaluation setups.

Humans tend to strongly agree on ratings on a scale for extreme cases (e.g., a CAT is judged as very concrete), but judgements on mid-scale words exhibit more disagreement.Yet, collected rating norms are heavily exploited across disciplines.Our study focuses on concreteness ratings and (i) implements correlations and supervised classification to identify salient multimodal characteristics of mid-scale words, and (ii) applies a hard clustering to identify patterns of systematic disagreement across raters.Our results suggest to either fine-tune or filter midscale target words before utilising them.

We propose a novel approach to learn domainspecific plausible materials for components in the vehicle repair domain by probing Pretrained Language Models (PLMs) in a cloze task style setting to overcome the lack of annotated datasets.We devise a new method to aggregate salient predictions from a set of cloze query templates and show that domainadaptation using either a small, high-quality or a customized Wikipedia corpus boosts performance.When exploring resource-lean alternatives, we find a distilled PLM clearly outperforming a classic pattern-based algorithm.Further, given that 98% of our domain-specific components are multiword expressions, we successfully exploit the compositionality assumption as a way to address data sparsity.

To date, transformer-based models such as BERT have been less successful in predicting compositionality of noun compounds than static word embeddings.This is likely related to a suboptimal use of the encoded information, reflecting an incomplete grasp of how the models represent the meanings of complex linguistic structures.This paper investigates variants of semantic knowledge derived from pretrained BERT when predicting the degrees of compositionality for 280 English noun compounds associated with human compositionality ratings.Our performance strongly improves on earlier unsupervised implementations of pretrained BERT and highlights beneficial decisions in data preprocessing, embedding computation, and compositionality estimation.The distinct linguistic roles of heads and modifiers are reflected by differences in BERT-derived representations, with empirical properties such as frequency, productivity, and ambiguity affecting model performance.The most relevant representational information is concentrated in the initial layers of the model architecture.

We investigate the effect of sub-word tokenization on representations of German noun compounds: single orthographic words which are composed of two or more constituents but often tokenized into units that are not morphologically motivated or meaningful.Using variants of BERT models and tokenization strategies on domain-specific restricted diachronic data, we introduce a suite of evaluations relying on the masked language modelling task and compositionality prediction.We obtain the most consistent improvements by pre-splitting compounds into constituents.

We provide a novel dataset – DiaWUG – with judgements on diatopic lexical semantic variation for six Spanish variants in Europe and Latin America. In contrast to most previous meaning-based resources and studies on semantic diatopic variation, we collect annotations on semantic relatedness for Spanish target words in their contexts from both a semasiological perspective (i.e., exploring the meanings of a word given its form, thus including polysemy) and an onomasiological perspective (i.e., exploring identical meanings of words with different forms, thus including synonymy). In addition, our novel dataset exploits and extends the existing framework DURel for annotating word senses in context (Erk et al., 2013; Schlechtweg et al., 2018) and the framework-embedded Word Usage Graphs (WUGs) – which up to now have mainly be used for semasiological tasks and resources – in order to distinguish, visualize and interpret lexical semantic variation of contextualized words in Spanish from these two perspectives, i.e., semasiological and onomasiological language variation.

Agenda-setting is a widely explored phenomenon in political science: powerful stakeholders (governments or their financial supporters) have control over the media and set their agenda: political and economical powers determine which news should be salient. This is a clear case of targeted manipulation to divert the public attention from serious issues affecting internal politics (such as economic downturns and scandals) by flooding the media with potentially distracting information. We investigate agenda-setting in the Russian social media landscape, exploring the relation between economic indicators and mentions of foreign geopolitical entities, as well as of Russia itself. Our contributions are at three levels: at the level of the domain of the investigation, our study is the first to substructure the Russian media landscape in state-controlled vs. independent outlets in the context of strategic distraction from negative economic trends; at the level of the scope of the investigation, we involve a large set of geopolitical entities (while previous work has focused on the U.S.); at the qualitative level, our analysis of posts on Ukraine, whose relationship with Russia is of high geopolitical relevance, provides further insights into the contrast between state-controlled and independent outlets.

Research on metaphorical language has shown ties between abstractness and emotionality with regard to metaphoricity; prior work is however limited to the word and sentence levels, and up to date there is no empirical study establishing the extent to which this is also true on the discourse level. This paper explores which textual and perceptual features human annotators perceive as important for the metaphoricity of discourses and expressions, and addresses two research questions more specifically. First, is a metaphorically-perceived discourse more abstract and more emotional in comparison to a literally- perceived discourse? Second, is a metaphorical expression preceded by a more metaphorical/abstract/emotional context than a synonymous literal alternative? We used a dataset of 1,000 corpus-extracted discourses for which crowdsourced annotators (1) provided judgements on whether they perceived the discourses as more metaphorical or more literal, and (2) systematically listed lexical terms which triggered their decisions in (1). Our results indicate that metaphorical discourses are more emotional and to a certain extent more abstract than literal discourses. However, neither the metaphoricity nor the abstractness and emotionality of the preceding discourse seem to play a role in triggering the choice between synonymous metaphorical vs. literal expressions. Our dataset is available at https://www.ims.uni-stuttgart.de/data/discourse-met-lit.

Sinan Kurtyigit, Maike Park, Dominik Schlechtweg, Jonas Kuhn, Sabine Schulte im Walde. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

While automatic term extraction is a wellresearched area, computational approaches to distinguish between degrees of technicality are still understudied.We semi-automatically create a German gold standard of technicality across four domains, and illustrate the impact of a web-crawled general-language corpus on predicting technicality.When defining a classification approach that combines general-language and domain-specific word embeddings, we go beyond previous work and align vector spaces to gain comparative embeddings.We suggest two novel models to exploit general-vs.domain-specific comparisons: a simple neural network model with pre-computed comparative-embedding information as input, and a multi-channel model computing the comparison internally.Both models outperform previous approaches, with the multi-channel model performing best.

Modelling language change is an increasingly important area of interest within the fields of sociolinguistics and historical linguistics. In recent years, there has been a growing number of publications whose main concern is studying changes that have occurred within the past centuries. The Corpus of Historical American English (COHA) is one of the most commonly used large corpora in diachronic studies in English. This paper describes methods applied to the downloadable version of the COHA corpus in order to overcome its main limitations, such as inconsistent lemmas and malformed tokens, without compromising its qualitative and distributional properties. The resulting corpus CCOHA contains a larger number of cleaned word tokens which can offer better insights into language change and allow for a larger variety of tasks to be performed.

Predicting the degree of compositionality of noun compounds such as “snowball” and “butterfly” is a crucial ingredient for lexicography and Natural Language Processing applications, to know whether the compound should be treated as a whole, or through its constituents, and what it means. Computational approaches for an automatic prediction typically represent and compare compounds and their constituents within a vector space and use distributional similarity as a proxy to predict the semantic relatedness between the compounds and their constituents as the compound’s degree of compositionality. This paper provides a systematic evaluation of vector-space reduction variants across kinds, exploring reductions based on part-of-speech next to and also in combination with Principal Components Analysis using Singular Value and word2vec embeddings. We show that word2vec and nouns only dimensionality reductions are the most successful and stable vector space variants for our task.

We present a dataset with difficulty ratings for 1,030 German closed noun compounds extracted from domain-specific texts for do-it-ourself (DIY), cooking and automotive. The dataset includes two-part compounds for cooking and DIY, and two- to four-part compounds for automotive. The compounds were identified in text using the Simple Compound Splitter (Weller-Di Marco, 2017); a subset was filtered and balanced for frequency and productivity criteria as basis for manual annotation and fine-grained interpretation. This study presents the creation, the final dataset with ratings from 20 annotators and statistics over the dataset, to provide insight into the perception of domain-specific term difficulty. It is particularly striking that annotators agree on a coarse, binary distinction between easy vs. difficult domain-specific compounds but that a more fine grained distinction of difficulty is not meaningful. We finally discuss the challenges of an annotation for difficulty, which includes both the task description as well as the selection of the data basis.

We perform a comparative study for automatic term extraction from domain-specific language using a PageRank model with different edge-weighting methods. We vary vector space representations within the PageRank graph algorithm, and we go beyond standard co-occurrence and investigate the influence of measures of association strength and first- vs. second-order co-occurrence. In addition, we incorporate meaning shifts from general to domain-specific language as personalized vectors, in order to distinguish between termhood strengths of ambiguous words across word senses. Our study is performed for two domain-specific English corpora: ACL and do-it-yourself (DIY); and a domain-specific German corpus: cooking. The models are assessed by applying average precision and the roc score as evaluation metrices.

We perform an interdisciplinary large-scale evaluation for detecting lexical semantic divergences in a diachronic and in a synchronic task: semantic sense changes across time, and semantic sense changes across domains.Our work addresses the superficialness and lack of comparison in assessing models of diachronic lexical change, by bringing together and extending benchmark models on a common state-of-the-art evaluation task.In addition, we demonstrate that the same evaluation task and modelling approaches can successfully be utilised for the synchronic detection of domain-specific sense divergences in the field of term extraction.

Marco Del Tredici, Diego Marcheggiani, Sabine Schulte im Walde, Raquel Fernández. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

Sentiment analysis in low-resource languages suffers from a lack of annotated corpora to estimate high-performing models.Machine translation and bilingual word embeddings provide some relief through cross-lingual sentiment approaches.However, they either require large amounts of parallel data or do not sufficiently capture sentiment information.We introduce Bilingual Sentiment Embeddings (BLSE), which jointly represent sentiment information in a source and target language.This model only requires a small bilingual lexicon, a source-language corpus annotated for sentiment, and monolingual word embeddings for each language.We perform experiments on three language combinations (Spanish, Catalan, Basque) for sentencelevel cross-lingual sentiment classification and find that our model significantly outperforms state-of-the-art methods on four out of six experimental setups, as well as capturing complementary information to machine translation.Our analysis of the resulting embedding space provides evidence that it represents sentiment information in the resource-poor target language without any annotated data in that language.

Domain adaptation for sentiment analysis is challenging due to the fact that supervised classifiers are very sensitive to changes in domain. The two most prominent approaches to this problem are structural correspondence learning and autoencoders. However, they either require long training times or suffer greatly on highly divergent domains. Inspired by recent advances in cross-lingual sentiment analysis, we provide a novel perspective and cast the domain adaptation problem as an embedding projection task. Our model takes as input two mono-domain embedding spaces and learns to project them to a bi-domain space, which is jointly optimized to (1) project across domains and to (2) predict sentiment. We perform domain adaptation experiments on 20 source-target domain pairs for sentiment classification and report novel state-of-the-art results on 11 domain pairs, including the Amazon domain adaptation datasets and SemEval 2013 and 2016 datasets. Our analysis shows that our model performs comparably to state-of-the-art approaches on domains that are similar, while performing significantly better on highly divergent domains. Our code is available at https://github.com/jbarnesspain/domain_blse

This paper explores the informationtheoretic measure entropy to detect metaphoric change, transferring ideas from hypernym detection to research on language change.We also build the first diachronic test set for German as a standard for metaphoric change annotation.Our model shows high performance, is unsupervised, language-independent and generalizable to other processes of semantic change.

Distinguishing between antonyms and synonyms is a key task to achieve high performance in NLP systems.While they are notoriously difficult to distinguish by distributional co-occurrence models, pattern-based methods have proven effective to differentiate between the relations.In this paper, we present a novel neural network model AntSynNET that exploits lexico-syntactic patterns from syntactic parse trees.In addition to the lexical and syntactic information, we successfully integrate the distance between the related words along the syntactic path as a new pattern feature.The results from classification experiments show that AntSyn-NET improves the performance over prior pattern-based methods.

We present a novel neural model HyperVec to learn hierarchical embeddings for hypernymy detection and directionality.While previous embeddings have shown limitations on prototypical hypernyms, HyperVec represents an unsupervised measure where embeddings are learned in a specific order and capture the hypernym-hyponym distributional hierarchy.Moreover, our model is able to generalize over unseen hypernymy pairs, when using only small sets of training data, and by mapping to other languages.Results on benchmark datasets show that HyperVec outperforms both state-of-theart unsupervised measures and embedding models on hypernymy detection and directionality, and on predicting graded lexical entailment.

Word embeddings have been demonstrated to benefit NLP tasks impressively. Yet, there is room for improvements in the vector representations, because current word embeddings typically contain unnecessary information, i.e., noise. We propose two novel models to improve word embeddings by unsupervised learning, in order to yield word denoising embeddings. The word denoising embeddings are obtained by strengthening salient information and weakening noise in the original word embeddings, based on a deep feed-forward neural network filter. Results from benchmark tasks show that the filtered word denoising embeddings outperform the original word embeddings.

This paper provides a binary, token-based classification of German particle verbs (PVs) into literal vs. non-literal usage.A random forest improving standard features (e.g., bagof-words; affective ratings) with PV-specific information and abstraction over common nouns significantly outperforms the majority baseline.In addition, PV-specific classification experiments demonstrate the role of shared particle semantics and semantically related base verbs in PV meaning shifts.

Enrico Santus, Alessandro Lenci, Qin Lu, Sabine Schulte im Walde. Proceedings of the 14th Conference of the European Chapter of the Association for Computational Linguistics, volume 2: Short Papers. 2014.

Recent investigations into grounded models of language have shown that holistic views of language and perception can provide higher performance than independent views.In this work, we improve a two-dimensional multimodal version of Latent Dirichlet Allocation (Andrews et al., 2009) in various ways.(1) We outperform text-only models in two different evaluations, and demonstrate that low-level visual features are directly compatible with the existing model.(2) We present a novel way to integrate visual features into the LDA model using unsupervised clusters of images.The clusters are directly interpretable and improve on our evaluation tasks.(3) We provide two novel ways to extend the bimodal models to support three or more modalities.We find that the three-, four-, and five-dimensional models significantly outperform models using only one or two modalities, and that nontextual modalities each provide separate, disjoint knowledge that cannot be forced into a shared, latent structure.

We present a study on the automatic acquisition of semantic classes for Catalan adjectives from distributional and morphological information, with particular emphasis on polysemous adjectives. The aim is to distinguish and characterize broad classes, such as qualitative (gran ‘big’) and relational (pulmonar ‘pulmonary’) adjectives, as well as to identify polysemous adjectives such as econòmic (‘economic ∣ cheap’). We specifically aim at modeling regular polysemy, that is, types of sense alternations that are shared across lemmata. To date, both semantic classes for adjectives and regular polysemy have only been sparsely addressed in empirical computational linguistics. Two main specific questions are tackled in this article. First, what is an adequate broad semantic classification for adjectives? We provide empirical support for the qualitative and relational classes as defined in theoretical work, and uncover one type of adjective that has not received enough attention, namely, the event-related class. Second, how is regular polysemy best modeled in computational terms? We present two models, and argue that the second one, which models regular polysemy in terms of simultaneous membership to multiple basic classes, is both theoretically and empirically more adequate than the first one, which attempts to identify independent polysemous classes. Our best classifier achieves 69.1% accuracy, against a 51% baseline.

This article presents clustering experiments on German verbs: A statistical grammar model for German serves as the source for a distributional verb description at the lexical syntax-semantics interface, and the unsupervised clustering algorithm k-means uses the empirical verb properties to perform an automatic induction of verb classes. Various evaluation measures are applied to compare the clustering results to gold standard German semantic verb classes under different criteria. The primary goals of the experiments are (1) to empirically utilize and investigate the well-established relationship between verb meaning and verb behavior within a cluster analysis and (2) to investigate the required technical parameters of a cluster analysis with respect to this specific linguistic task. The clustering methodology is developed on a small-scale verb set and then applied to a larger-scale verb set including 883 German verbs.

The paper describes the application of k-Means, a standard clustering technique, to the task of inducing semantic classes for German verbs. Using probability distributions over verb subcategorisation frames, we obtained an intuitively plausible clustering of 57 verbs into 14 classes. The automatic clustering was evaluated against independently motivated, hand-constructed semantic verb classes. A series of post-hoc cluster analyses explored the influence of specific frames and frame groups on the coherence of the verb classes, and supported the tight connection between the syntactic behaviour of the verbs and their lexical meaning components.

We describe and evaluate the application of a spectral clustering technique (Ng et al., 2002) to the unsupervised clustering of German verbs. Our previous work has shown that standard clustering techniques succeed in inducing Levin-style semantic classes from verb subcategorisation information. But clustering in the very high dimensional spaces that we use is fraught with technical and conceptual difficulties. Spectral clustering performs a dimensionality reduction on the verb frame patterns, and provides a robustness and efficiency that standard clustering methods do not display in direct use. The clustering results are evaluated according to the alignment (Christianini et al., 2002) between the Gram matrix defined by the cluster output and the corresponding matrix defined by a gold standard.