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.

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

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.

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.

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.

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.

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.

“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.

Abstract Most compositional distributional semantic models represent sentence meaning with a single vector. In this paper, we propose a structured distributional model (SDM) that combines word embeddings with formal semantics and is based on the assumption that sentences represent events and situations. The semantic representation of a sentence is a formal structure derived from discourse representation theory and containing distributional vectors. This structure is dynamically and incrementally built by integrating knowledge about events and their typical participants, as they are activated by lexical items. Event knowledge is modelled as a graph extracted from parsed corpora and encoding roles and relationships between participants that are represented as distributional vectors. SDM is grounded on extensive psycholinguistic research showing that generalized knowledge about events stored in semantic memory plays a key role in sentence comprehension.We evaluate SDMon two recently introduced compositionality data sets, and our results show that combining a simple compositionalmodel with event knowledge constantly improves performances, even with dif ferent types of word embeddings.

In this paper, we introduce a new distributional method for modeling predicateargument 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-theart 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 verbspecific semantic roles.

In this paper, we claim that vector cosine – which is generally considered among the most efficient unsupervised measures for identifying word similarity in Vector Space Models – can be outperformed by an unsupervised measure that calculates the extent of the intersection among the most mutually dependent contexts of the target words. To prove it, we describe and evaluate APSyn, a variant of the Average Precision that, without any optimization, outperforms the vector cosine and the co-occurrence on the standard ESL test set, with an improvement ranging between +9.00% and +17.98%, depending on the number of chosen top contexts.

In this paper, we describe ROOT13, a supervised system for the classification of hypernyms, co-hyponyms and random words. The system relies on a Random Forest algorithm and 13 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, ROOT13 achieves an F1 score of 88.3%, against a baseline of 57.6% (vector cosine). When the classification is binary, ROOT13 achieves the following results: hypernyms-co-hyponyms (93.4% vs. 60.2%), hypernyms-random (92.3% vs. 65.5%) and co-hyponyms-random (97.3% vs. 81.5%). Our results are competitive with state-of-the-art models.

Several studies on sentence processing suggest that the mental lexicon keeps track of the mutual expectations between words.Current DSMs, however, represent context words as separate features, thereby loosing important information for word expectations, such as word interrelations.In this paper, we present a DSM that addresses this issue by defining verb contexts as joint syntactic dependencies.We test our representation in a verb similarity task on two datasets, showing that joint contexts achieve performances comparable to single dependencies or even better.Moreover, they are able to overcome the data sparsity problem of joint feature spaces, in spite of the limited size of our training corpus.

This paper investigates the effects of data size and frequency range on distributional semantic models. We compare the performance of a number of representative models for several test settings over data of varying sizes, and over test items of various frequency. Our results show that neural network-based models underperform when the data is small, and that the most reliable model over data of varying sizes and frequency ranges is the inverted factorized model.

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.

Research into corpus-based semantics has focused on the development of ad hoc models that treat single tasks, or sets of closely related tasks, as unrelated challenges to be tackled by extracting different kinds of distributional information from the corpus. As an alternative to this “one task, one model” approach, the Distributional Memory framework extracts distributional information once and for all from the corpus, in the form of a set of weighted word-link-word tuples arranged into a third-order tensor. Different matrices are then generated from the tensor, and their rows and columns constitute natural spaces to deal with different semantic problems. In this way, the same distributional information can be shared across tasks such as modeling word similarity judgments, discovering synonyms, concept categorization, predicting selectional preferences of verbs, solving analogy problems, classifying relations between word pairs, harvesting qualia structures with patterns or example pairs, predicting the typical properties of concepts, and classifying verbs into alternation classes. Extensive empirical testing in all these domains shows that a Distributional Memory implementation performs competitively against task-specific algorithms recently reported in the literature for the same tasks, and against our implementations of several state-of-the-art methods. The Distributional Memory approach is thus shown to be tenable despite the constraints imposed by its multi-purpose nature.

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