This work explores the potential of Transformer models focusing on the translation of ancient Egyptian hieroglyphs. We present a novel Hieroglyphic Transformer model, built upon the powerful M2M-100 multilingual translation framework and trained on a dataset we customised from the Thesaurus Linguae Aegyptiae database. Our experiments demonstrate promising results, with the model achieving significant accuracy in translating hieroglyphics into both German and English. This work holds significant implications for Egyptology, potentially accelerating the translation process and unlocking new research approaches.

Multimodal metaphorical interpretation of abstract concepts has always been a debated problem in many research fields, including cognitive linguistics and NLP. With the dramatic improvements of Large Language Models (LLMs) and the increasing attention toward multimodal Vision-Language Models (VLMs), there has been pronounced attention on the conceptualization of abstracts. Nevertheless, a systematic scientific investigation is still lacking. This work introduces a framework designed to shed light on the indirect grounding mechanisms that anchor the meaning of abstract concepts to concrete situations (e.g. ability - a person skating), following the idea that abstracts acquire meaning from embodied and situated simulation. We assessed human and LLMs performances by a situation generation task. Moreover, we assess the figurative richness of images depicting concrete scenarios, via a text-to-image retrieval task performed on LAION-400M.

Semantic plausibility (e.g. knowing that “the actor won the award” is more likely than “the actor won the battle”) serves as an effective proxy for general world knowledge. Language models (LMs) capture vast amounts of world knowledge by learning distributional patterns in text, accessible via log probabilities (LogProbs) they assign to plausible vs. implausible outputs. The new generation of instruction-tuned LMs can now also provide explicit estimates of plausibility via prompting. Here, we evaluate the effectiveness of LogProbs and basic prompting to measure semantic plausibility, both in single-sentence minimal pairs (Experiment 1) and short context-dependent scenarios (Experiment 2). We find that (i) in both base and instruction-tuned LMs, LogProbs offers a more reliable measure of semantic plausibility than direct zero-shot prompting, which yields inconsistent and often poor results; (ii) instruction-tuning generally does not alter the sensitivity of LogProbs to semantic plausibility (although sometimes decreases it); (iii) across models, context mostly modulates LogProbs in expected ways, as measured by three novel metrics of context-sensitive plausibility and their match to explicit human plausibility judgments. We conclude that, even in the era of prompt-based evaluations, LogProbs constitute a useful metric of semantic plausibility, both in base and instruction-tuned LMs.

The field of Distributional Semantics has recently undergone important changes, with the contextual representations produced by Transformers taking the place of static word embeddings models. Noticeably, previous studies comparing the two types of vectors have only focused on the English language and a limited number of models. In our study, we present a comparative evaluation of static and contextualized distributional models for Mandarin Chinese, focusing on a range of intrinsic tasks. Our results reveal that static models remain stronger for some of the classical tasks that consider word meaning independent of context, while contextualized models excel in identifying semantic relations between word pairs and in the categorization of words into abstract semantic classes.

An open question in language comprehension studies is whether non-compositional multiword expressions like idioms and compositional-but-frequent word sequences are processed differently. Are the latter constructed online, or are instead directly retrieved from the lexicon, with a degree of entrenchment depending on their frequency? In this paper, we address this question with two different methodologies. First, we set up a self-paced reading experiment comparing human reading times for idioms and both highfrequency and low-frequency compositional word sequences. Then, we ran the same experiment using the Surprisal metrics computed with Neural Language Models (NLMs). Our results provide evidence that idiomatic and high-frequency compositional expressions are processed similarly by both humans and NLMs. Additional experiments were run to test the possible factors that could affect the NLMs’ performance.

Ellipsis is a linguistic phenomenon characterized by the omission of one or more sentence elements. Solving such a linguistic construction is not a trivial issue in natural language processing since it involves the retrieval of non-overtly expressed verbal material, which might in turn require the model to integrate human-like syntactic and semantic knowledge. In this paper, we explored the issue of how the prototypicality of event participants affects the ability of Language Models (LMs) to handle elliptical sentences and to identify the omitted arguments at different degrees of thematic fit, ranging from highly typical participants to semantically anomalous ones. With this purpose in mind, we built ELLie, the first dataset composed entirely of utterances containing different types of elliptical constructions, and structurally suited for evaluating the effect of argument thematic fit in solving ellipsis and reconstructing the missing element. Our tests demonstrated that the probability scores assigned by the models are higher for typical events than for atypical and impossible ones in different elliptical contexts, confirming the influence of prototypicality of the event participants in interpreting such linguistic structures. Finally, we conducted a retrieval task of the elided verb in the sentence in which the low performance of LMs highlighted a considerable difficulty in reconstructing the correct event.

In psycholinguistics, semantic attraction is a sentence processing phenomenon in which a given argument violates the selectional requirements of a verb, but this violation is not perceived by comprehenders due to its attraction to another noun in the same sentence, which is syntactically unrelated but semantically sound. In our study, we use autoregressive language models to compute the sentence-level and the target phrase-level Surprisal scores of a psycholinguistic dataset on semantic attraction. Our results show that the models are sensitive to semantic attraction, leading to reduced Surprisal scores, although none of them perfectly matches the human behavioral pattern.

A central quest of probing is to uncover how pre-trained models encode a linguistic property within their representations. An encoding, however, might be spurious—i.e., the model might not rely on it when making predictions. In this paper, we try to find an encoding that the model actually uses, introducing a usage-based probing setup. We first choose a behavioral task which cannot be solved without using the linguistic property. Then, we attempt to remove the property by intervening on the model’s representations. We contend that, if an encoding is used by the model, its removal should harm the performance on the chosen behavioral task. As a case study, we focus on how BERT encodes grammatical number, and on how it uses this encoding to solve the number agreement task. Experimentally, we find that BERT relies on a linear encoding of grammatical number to produce the correct behavioral output. We also find that BERT uses a separate encoding of grammatical number for nouns and verbs. Finally, we identify in which layers information about grammatical number is transferred from a noun to its head verb.

Word order, an essential property of natural languages, is injected in Transformer-based neural language models using position encoding. However, recent experiments have shown that explicit position encoding is not always useful, since some models without such feature managed to achieve state-of-the art performance on some tasks. To understand better this phenomenon, we examine the effect of removing position encodings on the pre-training objective itself (i.e., masked language modelling), to test whether models can reconstruct position information from co-occurrences alone. We do so by controlling the amount of masked tokens in the input sentence, as a proxy to affect the importance of position information for the task. We find that the necessity of position information increases with the amount of masking, and that masked language models without position encodings are not able to reconstruct this information on the task. These findings point towards a direct relationship between the amount of masking and the ability of Transformers to capture order-sensitive aspects of language using position encoding.

Although transformer-based Neural Language Models demonstrate impressive performance on a variety of tasks, their generalization abilities are not well understood. They have been shown to perform strongly on subject-verb number agreement in a wide array of settings, suggesting that they learned to track syntactic dependencies during their training even without explicit supervision. In this paper, we examine the extent to which BERT is able to perform lexically-independent subject-verb number agreement (NA) on targeted syntactic templates. To do so, we disrupt the lexical patterns found in naturally occurring stimuli for each targeted structure in a novel fine-grained analysis of BERT’s behavior. Our results on nonce sentences suggest that the model generalizes well for simple templates, but fails to perform lexically-independent syntactic generalization when as little as one attractor is present.

Abstract concepts, notwithstanding their lack of physical referents in real world, are grounded in sensorimotor experience. In fact, images depicting concrete entities may be associated to abstract concepts, both via direct and indirect grounding processes. However, what are the links connecting the concrete concepts represented by images and abstract ones is still unclear. To investigate these links, we conducted a preliminary study collecting word association data and image-abstract word pair ratings, to identify whether the associations between visual and verbal systems rely on the same conceptual mappings. The goal of this research is to understand to what extent linguistic associations could be confirmed with visual stimuli, in order to have a starting point for multimodal analysis of abstract and concrete concepts.

Usage-based constructionist approaches consider language a structured inventory of constructions, form-meaning pairings of different schematicity and complexity, and claim that the more a linguistic pattern is encountered, the more it becomes accessible to speakers. However, when an expression is unavailable, what processes underlie the interpretation? While traditional answers rely on the principle of compositionality, for which the meaning is built word-by-word and incrementally, usage-based theories argue that novel utterances are created based on previously experienced ones through analogy, mapping an existing structural pattern onto a novel instance. Starting from this theoretical perspective, we propose here a computational implementation of these assumptions. As the principle of compositionality has been used to generate distributional representations of phrases, we propose a neural network simulating the construction of phrasal embedding as an analogical process. Our framework, inspired by word2vec and computer vision techniques, was evaluated on tasks of generalization from existing vectors.

Automatic Readability Assessment aims at assigning a complexity level to a given text, which could help improve the accessibility to information in specific domains, such as the administrative one. In this paper, we investigate the behavior of a Neural Pairwise Ranking Model (NPRM) for sentence-level readability assessment of Italian administrative texts. To deal with data scarcity, we experiment with cross-lingual, cross- and in-domain approaches, and test our models on Admin-It, a new parallel corpus in the Italian administrative language, containing sentences simplified using three different rewriting strategies. We show that NPRMs are effective in zero-shot scenarios (~0.78 ranking accuracy), especially with ranking pairs containing simplifications produced by overall rewriting at the sentence-level, and that the best results are obtained by adding in-domain data (achieving perfect performance for such sentence pairs). Finally, we investigate where NPRMs failed, showing that the characteristics of the training data, rather than its size, have a bigger effect on a model’s performance.

Both humans and neural language models are able to perform subject verb number agreement (SVA). In principle, semantics shouldn’t interfere with this task, which only requires syntactic knowledge. In this work we test whether meaning interferes with this type of agreement in English in syntactic structures of various complexities. To do so, we generate both semantically well-formed and nonsensical items. We compare the performance of BERT-base to that of humans, obtained with a psycholinguistic online crowdsourcing experiment. We find that BERT and humans are both sensitive to our semantic manipulation: They fail more often when presented with nonsensical items, especially when their syntactic structure features an attractor (a noun phrase between the subject and the verb that has not the same number as the subject). We also find that the effect of meaningfulness on SVA errors is stronger for BERT than for humans, showing higher lexical sensitivity of the former on this task.

Contextual embeddings build multidimensional representations of word tokens based on their context of occurrence. Such models have been shown to achieve a state-of-the-art performance on a wide variety of tasks. Yet, the community struggles in understanding what kind of semantic knowledge these representations encode. We report a series of experiments aimed at investigating to what extent one of such models, BERT, is able to infer the semantic relations that, according to Dowty’s Proto-Roles theory, a verbal argument receives by virtue of its role in the event described by the verb. This hypothesis were put to test by learning a linear mapping from the BERT’s verb embeddings to an interpretable space of semantic properties built from the linguistic dataset by White et al. (2016). In a first experiment we tested whether the semantic properties inferred from a typed version of the BERT embeddings would be more linguistically plausible than those produced by relying on static embeddings. We then move to evaluate the semantic properties inferred from the contextual embeddings both against those available in the original dataset, as well as by assessing their ability to model the semantic properties possessed by the agent of the verbs participating in the so-called causative alternation.

An intelligent system is expected to perform reasonable inferences, accounting for both the literal meaning of a word and the meanings a word can acquire in different contexts. A specific kind of inference concerns the connective and, which in some cases gives rise to a temporal succession or causal interpretation in contrast with the logic, commutative one (Levinson, 2000). In this work, we investigate the phenomenon by creating a new dataset for evaluating the interpretation of and by NLI systems, which we use to test three Transformer-based models. Our results show that all systems generalize patterns that are consistent with both the logical and the pragmatic interpretation, perform inferences that are inconsistent with each other, and show clear divergences with both theoretical accounts and humans’ behavior.

Prior research has explored the ability of computational models to predict a word semantic fit with a given predicate. While much work has been devoted to modeling the typicality relation between verbs and arguments in isolation, in this paper we take a broader perspective by assessing whether and to what extent computational approaches have access to the information about the typicality of entire events and situations described in language (Generalized Event Knowledge). Given the recent success of Transformers Language Models (TLMs), we decided to test them on a benchmark for the dynamic estimation of thematic fit. The evaluation of these models was performed in comparison with SDM, a framework specifically designed to integrate events in sentence meaning representations, and we conducted a detailed error analysis to investigate which factors affect their behavior. Our results show that TLMs can reach performances that are comparable to those achieved by SDM. However, additional analysis consistently suggests that TLMs do not capture important aspects of event knowledge, and their predictions often depend on surface linguistic features, such as frequent words, collocations and syntactic patterns, thereby showing sub-optimal generalization abilities.

Eye-tracking psycholinguistic studies have revealed that context-word semantic coherence and predictability influence language processing. In this paper we show our approach to predict eye-tracking features from the ZuCo dataset for the shared task of the Cognitive Modeling and Computational Linguistics (CMCL2021) workshop. Using both cosine similarity and surprisal within a regression model, we significantly improved the baseline Mean Absolute Error computed among five eye-tracking features.

Word embeddings are vectorial semantic representations built with either counting or predicting techniques aimed at capturing shades of meaning from word co-occurrences. Since their introduction, these representations have been criticized for lacking interpretable dimensions. This property of word embeddings limits our understanding of the semantic features they actually encode. Moreover, it contributes to the “black box” nature of the tasks in which they are used, since the reasons for word embedding performance often remain opaque to humans. In this contribution, we explore the semantic properties encoded in word embeddings by mapping them onto interpretable vectors, consisting of explicit and neurobiologically motivated semantic features (Binder et al. 2016). Our exploration takes into account different types of embeddings, including factorized count vectors and predict models (Skip-Gram, GloVe, etc.), as well as the most recent contextualized representations (i.e., ELMo and BERT). In our analysis, we first evaluate the quality of the mapping in a retrieval task, then we shed light on the semantic features that are better encoded in each embedding type. A large number of probing tasks is finally set to assess how the original and the mapped embeddings perform in discriminating semantic categories. For each probing task, we identify the most relevant semantic features and we show that there is a correlation between the embedding performance and how they encode those features. This study sets itself as a step forward in understanding which aspects of meaning are captured by vector spaces, by proposing a new and simple method to carve human-interpretable semantic representations from distributional vectors.

Metaphor is a widespread linguistic and cognitive phenomenon that is ruled by mechanisms which have received attention in the literature. Transformer Language Models such as BERT have brought improvements in metaphor-related tasks. However, they have been used only in application contexts, while their knowledge of the phenomenon has not been analyzed. To test what BERT knows about metaphors, we challenge it on a new dataset that we designed to test various aspects of this phenomenon such as variations in linguistic structure, variations in conventionality, the boundaries of the plausibility of a metaphor and the interpretations that we attribute to metaphoric expressions. Results bring out some tendencies that suggest that the model can reproduce some human intuitions about metaphors.

The domain-specialised application of Named Entity Recognition (NER) is known as Biomedical NER (BioNER), which aims to identify and classify biomedical concepts that are of interest to researchers, such as genes, proteins, chemical compounds, drugs, mutations, diseases, and so on. The BioNER task is very similar to general NER but recognising Biomedical Named Entities (BNEs) is more challenging than recognising proper names from newspapers due to the characteristics of biomedical nomenclature. In order to address the challenges posed by BioNER, seven machine learning models were implemented comparing a transfer learning approach based on fine-tuned BERT with Bi-LSTM based neural models and a CRF model used as baseline. Precision, Recall and F1-score were used as performance scores evaluating the models on two well-known biomedical corpora: JNLPBA and BIOCREATIVE IV (BC-IV). Strict and partial matching were considered as evaluation criteria. The reported results show that a transfer learning approach based on fine-tuned BERT outperforms all others methods achieving the highest scores for all metrics on both corpora.

Eye-tracking psycholinguistic studies have suggested that context-word semantic coherence and predictability influence language processing during the reading activity. In this study, we investigate the correlation between the cosine similarities computed with word embedding models (both static and contextualized) and eye-tracking data from two naturalistic reading corpora. We also studied the correlations of surprisal scores computed with three state-of-the-art language models. Our results show strong correlation for the scores computed with BERT and GloVe, suggesting that similarity can play an important role in modeling reading times.

Our paper offers a computational model of the semantic recoverability of verb arguments, tested in particular on direct objects and Instruments. Our fully distributional model is intended to improve on older taxonomy-based models, which require a lexicon in addition to the training corpus. We computed the selectional preferences of 99 transitive verbs and 173 Instrument verbs as the mean value of the pairwise cosines between their arguments (a weighted mean between all the arguments, or an unweighted mean with the topmost k arguments). Results show that our model can predict the recoverability of objects and Instruments, providing a similar result to that of taxonomy-based models but at a much cheaper computational cost.

In this paper, we propose FRAQUE, a question answering system for factoid questions in the Public administration domain. The system is based on semantic frames, here intended as collections of slots typed with their possible values. FRAQUE queries unstructured textual data and exploits the potential of different approaches: it extracts pattern elements from texts which are linguistically analyzed through statistical methods.FRAQUE allows Italian users to query vast document repositories related to the domain of Public Administration. Given the statistical nature of most of its components such as word embeddings, the system allows for a flexible domain and language adaptation process. FRAQUE’s goal is to associate questions with frames stored into a Knowledge Graph along with relevant document passages, which are returned as the answer.

In this work, we carry out two experiments in order to assess the ability of BERT to capture the meaning shift associated with metonymic expressions. We test the model on a new dataset that is representative of the most common types of metonymy. We compare BERT with the Structured Distributional Model (SDM), a model for the representation of words in context which is based on the notion of Generalized Event Knowledge. The results reveal that, while BERT ability to deal with metonymy is quite limited, SDM is good at predicting the meaning of metonymic expressions, providing support for an account of metonymy based on event knowledge.

In linguistics and cognitive science, Logical metonymies are defined as type clashes between an event-selecting verb and an entity-denoting noun (e.g. The editor finished the article), which are typically interpreted by inferring a hidden event (e.g. reading) on the basis of contextual cues. This paper tackles the problem of logical metonymy interpretation, that is, the retrieval of the covert event via computational methods. We compare different types of models, including the probabilistic and the distributional ones previously introduced in the literature on the topic. For the first time, we also tested on this task some of the recent Transformer-based models, such as BERT, RoBERTa, XLNet, and GPT-2. Our results show a complex scenario, in which the best Transformer-based models and some traditional distributional models perform very similarly. However, the low performance on some of the testing datasets suggests that logical metonymy is still a challenging phenomenon for computational modeling.

“Voices of the Great War” is the first large corpus of Italian historical texts dating back to the period of First World War. This corpus differs from other existing resources in several respects. First, from the linguistic point of view it gives account of the wide range of varieties in which Italian was articulated in that period, namely from a diastratic (educated vs. uneducated writers), diaphasic (low/informal vs. high/formal registers) and diatopic (regional varieties, dialects) points of view. From the historical perspective, through a collection of texts belonging to different genres it represents different views on the war and the various styles of narrating war events and experiences. The final corpus is balanced along various dimensions, corresponding to the textual genre, the language variety used, the author type and the typology of conveyed contents. The corpus is fully annotated with lemmas, part-of-speech, terminology, and named entities. Significant corpus samples representative of the different “voices” have also been enriched with meta-linguistic and syntactic information. The layer of syntactic annotation forms the first nucleus of an Italian historical treebank complying with the Universal Dependencies standard. The paper illustrates the final resource, the methodology and tools used to build it, and the Web Interface for navigating it.

While neural embeddings represent a popular choice for word representation in a wide variety of NLP tasks, their usage for thematic fit modeling has been limited, as they have been reported to lag behind syntax-based count models. In this paper, we propose a complete evaluation of count models and word embeddings on thematic fit estimation, by taking into account a larger number of parameters and verb roles and introducing also dependency-based embeddings in the comparison. Our results show a complex scenario, where a determinant factor for the performance seems to be the availability to the model of reliable syntactic information for building the distributional representations of the roles.

Is it possible to use images to model verb semantic similarities? Starting from this core question, we developed two textual distributional semantic models and a visual one. We found particularly interesting and challenging to investigate this Part of Speech since verbs are not often analysed in researches focused on multimodal distributional semantics. After the creation of the visual and textual distributional space, the three models were evaluated in relation to SimLex-999, a gold standard resource. Through this evaluation, we demonstrate that, using visual distributional models, it is possible to extract meaningful information and to effectively capture the semantic similarity between verbs.

In this paper, we propose a new type of semantic representation of Construction Grammar that combines constructions with the vector representations used in Distributional Semantics. We introduce a new framework, Distributional Construction Grammar, where grammar and meaning are systematically modeled from language use, and finally, we discuss the kind of contributions that distributional models can provide to CxG representation from a linguistic and cognitive perspective.

Distributional Semantic Models have been successfully used for modeling selectional preferences in a variety of scenarios, since distributional similarity naturally provides an estimate of the degree to which an argument satisfies the requirement of a given predicate. However, we argue that the performance of such models on rare verb-argument combinations has received relatively little attention: it is not clear whether they are able to distinguish the combinations that are simply atypical, or implausible, from the semantically anomalous ones, and in particular, they have never been tested on the task of modeling their differences in processing complexity. In this paper, we compare two different models of thematic fit by testing their ability of identifying violations of selectional restrictions in two datasets from the experimental studies.

This paper describes the SemEval 2018 Task 10 on Capturing Discriminative Attributes. Participants were asked to identify whether an attribute could help discriminate between two concepts. For example, a successful system should determine that ‘urine’ is a discriminating feature in the word pair ‘kidney’, ‘bone’. The aim of the task is to better evaluate the capabilities of state of the art semantic models, beyond pure semantic similarity. The task attracted submissions from 21 teams, and the best system achieved a 0.75 F1 score.

In theoretical linguistics, logical metonymy is defined as the combination of an event-subcategorizing verb with an entity-denoting direct object (e.g., The author began the book), so that the interpretation of the VP requires the retrieval of a covert event (e.g., writing). Psycholinguistic studies have revealed extra processing costs for logical metonymy, a phenomenon generally explained with the introduction of new semantic structure. In this paper, we present a general distributional model for sentence comprehension inspired by the Memory, Unification and Control model by Hagoort (2013,2016). We show that our distributional framework can account for the extra processing costs of logical metonymy and can identify the covert event in a classification task.

In this paper, we introduce a new distributional method for modeling predicate-argument thematic fit judgments. We use a syntax-based DSM to build a prototypical representation of verb-specific roles: for every verb, we extract the most salient second order contexts for each of its roles (i.e. the most salient dimensions of typical role fillers), and then we compute thematic fit as a weighted overlap between the top features of candidate fillers and role prototypes. Our experiments show that our method consistently outperforms a baseline re-implementing a state-of-the-art system, and achieves better or comparable results to those reported in the literature for the other unsupervised systems. Moreover, it provides an explicit representation of the features characterizing verb-specific semantic roles.

In this paper, we introduce for the first time a Distributional Model for computing semantic complexity, inspired by the general principles of the Memory, Unification and Control framework(Hagoort, 2013; Hagoort, 2016). We argue that sentence comprehension is an incremental process driven by the goal of constructing a coherent representation of the event represented by the sentence. The composition cost of a sentence depends on the semantic coherence of the event being constructed and on the activation degree of the linguistic constructions. We also report the results of a first evaluation of the model on the Bicknell dataset (Bicknell et al., 2010).

Notwithstanding the success of the notion of construction, the computational tradition still lacks a way to represent the semantic content of these linguistic entities. Here we present a simple corpus-based model implementing the idea that the meaning of a syntactic construction is intimately related to the semantics of its typical verbs. It is a two-step process, that starts by identifying the typical verbs occurring with a given syntactic construction and building their distributional vectors. We then calculated the weighted centroid of these vectors in order to derive the distributional signature of a construction. In order to assess the goodness of our approach, we replicated the priming effect described by Johnson and Golberg (2013) as a function of the semantic distance between a construction and its prototypical verbs. Additional support for our view comes from a regression analysis showing that our distributional information can be used to model behavioral data collected with a crowdsourced elicitation experiment.

The shared task of the 5th Workshop on Cognitive Aspects of the Lexicon (CogALex-V) aims at providing a common benchmark for testing current corpus-based methods for the identification of lexical semantic relations (synonymy, antonymy, hypernymy, part-whole meronymy) and at gaining a better understanding of their respective strengths and weaknesses. The shared task uses a challenging dataset extracted from EVALution 1.0, which contains word pairs holding the above-mentioned relations as well as semantically unrelated control items (random). The task is split into two subtasks: (i) identification of related word pairs vs. unrelated ones; (ii) classification of the word pairs according to their semantic relation. This paper describes the subtasks, the dataset, the evaluation metrics, the seven participating systems and their results. The best performing system in subtask 1 is GHHH (F1 = 0.790), while the best system in subtask 2 is LexNet (F1 = 0.445). The dataset and the task description are available at https://sites.google.com/site/cogalex2016/home/shared-task.

The present paper investigates the phenomenon of antonym canonicity by providing new behavioural and distributional evidence on Italian adjectives. Previous studies have showed that some pairs of antonyms are perceived to be better examples of opposition than others, and are so considered representative of the whole category (e.g., Deese, 1964; Murphy, 2003; Paradis et al., 2009). Our goal is to further investigate why such canonical pairs (Murphy, 2003) exist and how they come to be associated. In the literature, two different approaches have dealt with this issue. The lexical-categorical approach (Charles and Miller, 1989; Justeson and Katz, 1991) finds the cause of canonicity in the high co-occurrence frequency of the two adjectives. The cognitive-prototype approach (Paradis et al., 2009; Jones et al., 2012) instead claims that two adjectives form a canonical pair because they are aligned along a simple and salient dimension. Our empirical evidence, while supporting the latter view, shows that the paradigmatic distributional properties of adjectives can also contribute to explain the phenomenon of canonicity, providing a corpus-based correlate of the cognitive notion of salience.

This paper introduces LexFr, a corpus-based French lexical resource built by adapting the framework LexIt, originally developed to describe the combinatorial potential of Italian predicates. As in the original framework, the behavior of a group of target predicates is characterized by a series of syntactic (i.e., subcategorization frames) and semantic (i.e., selectional preferences) statistical information (a.k.a. distributional profiles) whose extraction process is mostly unsupervised. The first release of LexFr includes information for 2,493 verbs, 7,939 nouns and 2,628 adjectives. In these pages we describe the adaptation process and evaluated the final resource by comparing the information collected for 20 test verbs against the information available in a gold standard dictionary. In the best performing setting, we obtained 0.74 precision, 0.66 recall and 0.70 F-measure.

In this paper we compare different context selection approaches to improve the creation of Emotive Vector Space Models (VSMs). The system is based on the results of an existing approach that showed the possibility to create and update VSMs by exploiting crowdsourcing and human annotation. Here, we introduce a method to manipulate the contexts of the VSMs under the assumption that the emotive connotation of a target word is a function of both its syntagmatic and paradigmatic association with the various emotions. To study the differences among the proposed spaces and to confirm the reliability of the system, we report on two experiments: in the first one we validated the best candidates extracted from each model, and in the second one we compared the models’ performance on a random sample of target words. Both experiments have been implemented as crowdsourcing tasks.

This paper proposes a new method for Italian verb classification -and a preliminary example of resulting classes- inspired by Levin (1993) and VerbNet (Kipper-Schuler, 2005), yet partially independent from these resources; we achieved such a result by integrating Levin and VerbNet’s models of classification with other theoretic frameworks and resources. The classification is rooted in the constructionist framework (Goldberg, 1995; 2006) and is distribution-based. It is also semantically characterized by a link to FrameNet’ssemanticframesto represent the event expressed by a class. However, the new Italian classes maintain the hierarchic “tree” structure and monotonic nature of VerbNet’s classes, and, where possible, the original names (e.g.: Verbs of Killing, Verbs of Putting, etc.). We therefore propose here a taxonomy compatible with VerbNet but at the same time adapted to Italian syntax and semantics. It also addresses a number of problems intrinsic to the original classifications, such as the role of argument alternations, here regarded simply as epiphenomena, consistently with the constructionist approach.

ROOT9 is a supervised system for the classification of hypernyms, co-hyponyms and random words that is derived from the already introduced ROOT13 (Santus et al., 2016). It relies on a Random Forest algorithm and nine unsupervised corpus-based features. We evaluate it with a 10-fold cross validation on 9,600 pairs, equally distributed among the three classes and involving several Parts-Of-Speech (i.e. adjectives, nouns and verbs). When all the classes are present, ROOT9 achieves an F1 score of 90.7%, against a baseline of 57.2% (vector cosine). When the classification is binary, ROOT9 achieves the following results against the baseline. hypernyms-co-hyponyms 95.7% vs. 69.8%, hypernyms-random 91.8% vs. 64.1% and co-hyponyms-random 97.8% vs. 79.4%. In order to compare the performance with the state-of-the-art, we have also evaluated ROOT9 in subsets of the Weeds et al. (2014) datasets, proving that it is in fact competitive. Finally, we investigated whether the system learns the semantic relation or it simply learns the prototypical hypernyms, as claimed by Levy et al. (2015). The second possibility seems to be the most likely, even though ROOT9 can be trained on negative examples (i.e., switched hypernyms) to drastically reduce this bias.

In this paper, we claim that Vector Cosine ― which is generally considered one of the most efficient unsupervised measures for identifying word similarity in Vector Space Models ― can be outperformed by a completely unsupervised measure that evaluates the extent of the intersection among the most associated contexts of two target words, weighting such intersection according to the rank of the shared contexts in the dependency ranked lists. This claim comes from the hypothesis that similar words do not simply occur in similar contexts, but they share a larger portion of their most relevant contexts compared to other related words. To prove it, we describe and evaluate APSyn, a variant of Average Precision that ― independently of the adopted parameters ― outperforms the Vector Cosine and the co-occurrence on the ESL and TOEFL test sets. In the best setting, APSyn reaches 0.73 accuracy on the ESL dataset and 0.70 accuracy in the TOEFL dataset, beating therefore the non-English US college applicants (whose average, as reported in the literature, is 64.50%) and several state-of-the-art approaches.

This paper presents the design and results of a crowdsourcing experiment on the recognition of Italian event nominals. The aim of the experiment was to assess the feasibility of crowdsourcing methods for a complex semantic task such as distinguishing the eventive interpretation of polysemous nominals taking into consideration various types of syntagmatic cues. Details on the theoretical background and on the experiment set up are provided together with the final results in terms of accuracy and inter-annotator agreement. These results are compared with the ones obtained by expert annotators on the same task. The low values in accuracy and Fleiss’ kappa of the crowdsourcing experiment demonstrate that crowdsourcing is not always optimal for complex linguistic tasks. On the other hand, the use of non-expert contributors allows to understand what are the most ambiguous patterns of polysemy and the most useful syntagmatic cues to be used to identify the eventive reading of nominals.

The goal of this paper is to propose a classification of the syntactic alternations admitted by the most frequent Italian verbs. The data-driven two-steps procedure exploited and the structure of the identified classes of alternations are presented in depth and discussed. Even if this classification has been developed with a practical application in mind, namely the semi-automatic building of a VerbNet-like lexicon for Italian verbs, partly following the methodology proposed in the context of the VerbNet project, its availability may have a positive impact on several related research topics and Natural Language Processing tasks

This paper empirically evaluates the performances of different state-of-the-art distributional models in a nominal lexical semantic classification task. We consider models that exploit various types of distributional features, which thereby provide different representations of nominal behavior in context. The experiments presented in this work demonstrate the advantages and disadvantages of each model considered. This analysis also considers a combined strategy that we found to be capable of leveraging the bottlenecks of each model, especially when large robust data is not available.

The aim of this paper is to introduce LexIt, a computational framework for the automatic acquisition and exploration of distributional information about Italian verbs, nouns and adjectives, freely available through a web interface at the address http://sesia.humnet.unipi.it/lexit. LexIt is the first large-scale resource for Italian in which subcategorization and semantic selection properties are characterized fully on distributional ground: in the paper we describe both the process of data extraction and the evaluation of the subcategorization frames extracted with LexIt.

The paper describes the design and the results of a manual annotation methodology devoted to enrich the ISST--TANL Corpus, derived from the Italian Syntactic--Semantic Treebank (ISST), with Semantic Frames information. The main issues encountered in applying the English FrameNet annotation criteria to a corpus of Italian language are discussed together with the choice of anchoring the semantic annotation layer to the underlying dependency syntactic structure. The results of a case study aimed at extending and specialising this methodology for the annotation of a corpus of legislative texts are also discussed.

As the interest of the NLP community grows to develop several treebanks also for languages other than English, we observe efforts towards evaluating the impact of different annotation strategies used to represent particular languages or with reference to particular tasks. This paper contributes to the debate on the influence of resources used for the training and development on the performance of parsing systems. It presents a comparative analysis of the results achieved by three different dependency parsers developed and tested with respect to two treebanks for the Italian language, namely TUT and ISST--TANL, which differ significantly at the level of both corpus composition and adopted dependency representations.

A SuperSense Tagger is a tool for the automatic analysis of texts that associates to each noun, verb, adjective and adverb a semantic category within a general taxonomy. The developed tagger, based on a statistical model (Maximum Entropy), required the creation of an Italian annotated corpus, to be used as a training set, and the improvement of various existing tools. The obtained results significantly improved the current state-of-the art for this particular task.

n this paper, we outline the methodology we adopted to develop a FrameNet for Italian. The main element of novelty with respect to the original FrameNet is represented by the fact that the creation and annotation of Lexical Units is strictly grounded in distributional information (statistical distribution of verbal subcategorization frames, lexical and semantic preferences of each frame) automatically acquired from a large, dependency-parsed corpus. We claim that this approach allows us to overcome some of the shortcomings of the classical lexicographic method used to create FrameNet, by complementing the accuracy of manual annotation with the robustness of data on the global distributional patterns of a verb. In the paper, we describe our method for extracting distributional data from the corpus and the way we used it for the encoding and annotation of LUs. The long-term goal of our project is to create an electronic lexicon for Italian similar to the original English FrameNet. For the moment, we have developed a database of syntactic valences that will be made freely accessible via a web interface. This represents an autonomous resource besides the FrameNet lexicon, of which we have a beginning nucleus consisting of 791 annotated sentences.

There is by now widespread agreement that the most realistic way to construct the large-scale commonsense knowledge repositories required by natural language and artificial intelligence applications is by letting machines learn such knowledge from large quantities of data, like humans do. A lot of attention has consequently been paid to the development of increasingly sophisticated machine learning algorithms for knowledge extraction. However, the nature of the input that humans are exposed to while learning commonsense knowledge has received much less attention. The BabyExp project is collecting very dense audio and video recordings of the first 3 years of life of a baby. The corpus constructed in this way will be transcribed with automated techniques and made available to the research community. Moreover, techniques to extract commonsense conceptual knowledge incrementally from these multimodal data are also being explored within the project. The current paper describes BabyExp in general, and presents pilot studies on the feasibility of the automated audio and video transcriptions.

Verb lexical semantic properties are only one of the factors that contribute to the determination of the event type expressed by a sentence, which is instead the result of a complex interplay between the verb meaning and its linguistic context. We report on two computational models for the automatic identification of event type in Italian. Both models use linguistically-motivated features extracted from Italian corpora. The main goal of our experiments is to evaluate the contribution of different types of linguistic indicators to identify the event type of a sentence, as well as to model various cases of context-driven event type shift. In the first model, event type identification has been modelled as a supervised classification task, performed with Maximum Entropy classifiers. In the second model, Self-Organizing Maps have been used to define and identify event types in an unsupervised way. The interaction of various contextual factors in determining the event type expressed by a sentence makes event type identification a highly challenging task. Computational models can help us to shed new light on the real structure of event type classes as well as to gain a better understanding of context-driven semantic shifts.

In this paper, we reported experiments of unsupervised automatic acquisition of Italian and English verb subcategorization frames (SCFs) from general and domain corpora. The proposed technique operates on syntactically shallow-parsed corpora on the basis of a limited number of search heuristics not relying on any previous lexico-syntactic knowledge about SCFs. Although preliminary, reported results are in line with state-of-the-art lexical acquisition systems. The issue of whether verbs sharing similar SCFs distributions happen to share similar semantic properties as well was also explored by clustering verbs that share frames with the same distribution using the Minimum Description Length Principle (MDL). First experiments in this direction were carried out on Italian verbs with encouraging results.

The paper reports on a detailed quantitative analysis of distributional language data of both Italian and Czech, highlighting the relative contribution of a number of distributed grammatical factors to sentence-based identification of subjects and direct objects. The work is based on a Maximum Entropy model of stochastic resolution of grammatical conflicting constraints, and is demonstrably capable of putting explanatory theoretical accounts to the challenging test of an extensive, usage-based empirical verification.

In this paper we present an original approach to natural language query interpretation which has been implemented withinthe FuLL (Fuzzy Logic and Language) Italian project of BC S.r.l. In particular, we discuss here the creation of linguisticand ontological resources, together with the exploitation of existing ones, for natural language-driven database access andretrieval. Both the database and the queries we experiment with are Italian, but the methodology we broach naturally extends to other languages.

The ISLE project is a continuation of the long standing EAGLES initiative, carried out under the Human Language Technology (HLT) programme in collaboration between American and European groups in the framework of the EU-US International Research Co-operation, supported by NSF and EC. In this paper we concentrate on the current position of the ISLE Computational Lexicon Working Group (CLWG), whose activities aim at defining a general schema for a multilingual lexical entry (MILE), as the basis for a standard framework for multilingual computational lexicons. The needs and features of existing Machine Translation systems provide the main reference points for the process of consensual definition of the MILE. The overall structure of the MILE will be illustrated with particular attention to some of the issues raised for multilingual lexicons by the need of expressing complex transfer conditions among translation equivalents