This paper investigates the ability of Large Language Models (LLMs) to differentiate between canonical and non-canonical sentences in Italian, employing advanced neural architectures like LLaMA and its adaptations. Canonical sentences adhere to the standard Subject-Verb-Object (SVO) structure. We hypothesize that recent generative LLMs are influenced heavily by the English language, where non-canonical structures are very rare. Using the in-context learning technique, we probe these models and further fine-tune them for this specific task. Initial results indicate that these models continue to struggle with this task even after fine-tuning. Additionally, we introduce a new dataset comprising several hundred sentences from the poetry domain, which presents significant challenges for the canonical structure task.

In the use and creation of current Deep Learning Models the only number that is used for the overall computation is the frequency value associated with the current word form in the corpus, which is used to substitute it. Frequency values come in two forms: absolute and relative. Absolute frequency is used indirectly when selecting the vocabulary against which the word embeddings are created: the cutoff threshold is usually fixed at 30/50K entries of the most frequent words. Relative frequency comes in directly when computing word embeddings based on co-occurrence values of the tokens included in a window size 2/5 adjacent tokens. The latter values are then used to compute similarity, mostly based on cosine distance. In this paper we will evaluate the impact of these two frequency parameters on a small corpus of Italian sentences whose main features are two: presence of very rare words and of non-canonical structures. Rather than basing our evaluation on cosine measure alone, we propose a graded scale of scores which are linguistically motivated. The results computed on the basis of a perusal of BERT’s raw embeddings shows that the two parameters conspire to decide the level of predictability.

This paper presents work carried out to transform glosses of a fable in Italian Sign Language (LIS) into a text which is then read by a TTS synthesizer from an SSML modified version of the same text. Whereas many systems exist that generate sign language from a text, we decided to do the reverse operation and generate text from LIS. For that purpose we used a version of the fable The Tortoise and the Hare, signed and made available on Youtube by ALBA cooperativa sociale, which was annotated manually by second author for her master’s thesis. In order to achieve our goal, we converted the multilayer glosses into linear Prolog terms to be fed to the generator. In the paper we focus on the main problems encountered in the transformation of the glosses into a semantically and pragmatically consistent representation. The main problems have been caused by the complexities of a text like a fable which requires coreference mechanisms and speech acts to be implemented in the representation which are often unexpressed and constitute implicit information.

This paper introduces work carried out for the automatic generation of a written text in Italian starting from glosses of a fable in Italian Sign Language (LIS). The paper gives a brief overview of sign languages (SLs) and some peculiarities of SL fables such as the use of space, the strategy of Role Shift and classifiers. It also presents the annotation of the fable “The Tortoise and the Hare” - signed in LIS and made available by Alba Cooperativa Sociale -, which was annotated manually by first author for her master’s thesis. The annotation was the starting point of a generation process that allowed us to automatically generate a text in Italian starting from LIS glosses. LIS sentences have been transcribed with Italian words into tables on simultaneous layers, each of which contains specific linguistic or non-linguistic pieces of information. In addition, the present work discusses problems encountered in the annotation and generation process.

In this paper we will be dealing with different levels of complexity in the processing of Italian, a Romance language inheriting many properties from Latin which make it an almost free word order language . The paper is concerned with syntactic complexity as measurable on the basis of the cognitive parser that incrementally builds up a syntactic representation to be used by the semantic component. The theory behind will be LFG and parsing preferences will be used to justify one choice both from a principled and a processing point of view. LFG is a transformationless theory in which there is no deep structure separate from surface syntactic structure. This is partially in accordance with constructional theories in which noncanonical structures containing non-argument functions FOCUS/TOPIC are treated as multifunctional constituents. Complexity is computed on a processing basis following suggestions made by Blache and demonstrated by Kluender and Chesi

In this paper we will present work carried out to scale up the system for text understanding called GETARUNS, and port it to be used in dialogue understanding. The current goal is that of extracting automatically argumentative information in order to build argumentative structure. The long term goal is using argumentative structure to produce automatic summarization of spoken dialogues. Very much like other deep linguistic processing systems, our system is a generic text/dialogue understanding system that can be used in connection with an ontology ― WordNet - and other similar repositories of commonsense knowledge. We will present the adjustments we made in order to cope with transcribed spoken dialogues like those produced in the ICSI Berkeley project. In a final section we present preliminary evaluation of the system on two tasks: the task of automatic argumentative labeling and another frequently addressed task: referential vs. non-referential pronominal detection. Results obtained fair much higher than those reported in similar experiments with machine learning approaches.

This document reports the process of extending MorphoPro for Venetan, a lesser-used language spoken in the Nort-Eastern part of Italy. MorphoPro is the morphological component of TextPro, a suite of tools oriented towards a number of NLP tasks. In order to extend this component to Venetan, we developed a declarative representation of the morphological knowledge necessary to analyze and synthesize Venetan words. This task was challenging for several reasons, which are common to a number of lesser-used languages: although Venetan is widely used as an oral language in everyday life, its written usage is very limited; efforts for defining a standard orthography and grammar are very recent and not well established; despite recent attempts to propose a unified orthography, no Venetan standard is widely used. Besides, there are different geographical varieties and it is strongly influenced by Italian.

We present an experiment evaluating the contribution of a system called GReG for reranking the snippets returned by Google’s search engine in the 10 best links presented to the user, captured by the use of Google’s API. The evaluation aims at establishing whether or not the introduction of deep linguistic information may improve the accuracy of Google or rather it is the opposite case as maintained by the majority of people working in Information Retrieval, using a Bag Of Words approach. We used 900 questions, answers taken from TREC 8, 9 competitions, execute three different types of evaluation: one without any linguistic aid; a second one with tagging, syntactic constituency contribution; another run with what we call Partial Logical Form. Even though GReG is still work in progress, it is possible to draw clearcut conclusions: adding linguistic information to the evaluation process of the best snippet that can answer a question improves enormously the performance. In another experiment we used the actual associated to the Q/A pairs distributed by one of TREC’s participant, got even higher accuracy.

In this paper we propose a rule-based approach to extract dependency and grammatical functions from the Venice Italian Treebank, a Treebank of written text with PoS and constituent labels consisting of 10,200 utterances and about 274,000 tokens. As manual corpus annotation is expensive and time-consuming, we decided to exploit this existing constituency-based Treebank to derive dependency structures with lower effort. After describing the procedure to extract heads and dependents, based on a head percolation table for Italian, we introduce the rules adopted to add grammatical relation labels. To this purpose, we manually relabeled all non-canonical arguments, which are very frequent in Italian, then we automatically labeled the remaining complements or arguments following some syntactic restrictions based on the position of the constituents w.r.t to parent and sibling nodes. The final section of the paper describes evaluation results. Evaluation was carried out in two steps, one for dependency relations and one for grammatical roles. Results are in line with similar conversion algorithms carried out for other languages, with 0.97 precision on dependency arcs and F-measure for the main grammatical functions scoring 0.96 or above, except for obliques with 0.75.