This paper presents a FAIR-compliant etymological dataset of noun and verb lemmas from the Gallo-Italic variety spoken in the Sicilian towns of Nicosia and Sperlinga. Based on Trovato and Menza (2020) and various lexicographic materials, the dataset captures borrowing relations from Sicilian, modeled using the OntoLex-lemon framework and its etymological extension. Linguistic phenomena involved in these borrowings-so-called GalloSicilian features-are formalized in the Linguistic Phenomena Ontology (LiPh) as regular relations, enabling the automated generation of candidate derivations. Verified derivations, reviewed by lexicographers, are encoded using LiPh to reflect how individual features contribute to lexical transformations. The dataset is enriched with detailed metadata, geographic and linguistic contextualization, and is aligned with current standards in lexical resource publication, supporting future reuse and integration in the Linguistic Linked Open Data cloud.
In general, a lexical expression in any language may be either inherited from its parent language or
borrowed from a foreign one. Following [
Language features are essential tools for investigating contact-induced changes (see [
As reported in [
Here we document an efort to create a dataset of nouns and verbs from the Gallo-Sicilian variety spoken in Nicosia and Sperlinga, based on the lexical material in [7]. The dataset includes etymological information, expressed in terms of language features, for lexical expressions borrowed from Sicilian, thereby enabling automated analysis of borrowing processes.
From this perspective, we established the following representational goals for our dataset: [G1] To maximize findability , accessibility, interoperability and reusability, the dataset must comply with the FAIR principles, as outlined in [8]. [G2] To ensure findability, the entire lexicon must be accompanied by detailed and comprehensive metadata, facilitating indexing in authoritative linguistic resource catalogs such as the Linguistic Linked Open Data Cloud (LLOD) [9]. [G3] The dataset must include all lemmas for nouns and verbs in the Gallo-Sicilian variety spoken in Nicosia and Sperlinga, as these constitute the primary objects of analysis. [G4] Since our language-feature detection strategy is grounded in Sicilian, the dataset must include etymologies from Sicilian. [G5] Each etymology must be described through a derivation path listing the language features involved in transforming the Sicilian etymon into the Gallo-Sicilian lemma.
The knowledge domain items referenced in the goals above are summarized in the UML diagram (see [10]) in Figure 1. In the rest of the paper, all the examples are provided using the UML formalism.
The remainder of the paper describes the structure of the dataset and discusses the design choices made. Section 2 outlines the technologies and the knowledge representation formalisms adopted to ensure compliance with the FAIR principles. Section 3 details the metadata used to enhance the ifndability of the lexicon. Section 4 explains how the dictionary entries from [ 7] were incorporated into our dataset. Sections 5 and 6 describe how Sicilian etymons and the corresponding language features involved in borrowings are reported in the dataset, respectively. Finally, Section 7 presents our conclusions and outlines directions for future research.
Linked Data technologies (see [11, 12]) provide all the essential features for constructing and publishing a FAIR dataset. Entities in the domain of discourse are uniquely identified by resolvable IRIs (see [13]), which allow one to retrieve relevant information using the standard HTTP protocol (see [14]). Data are represented using the formalisms in Semantic Web stack such as RDF [15], RDFS [16], and OWL [17]. These technologies support the association of rich metadata with resources and the interlinking of datasets.
Although Linked Data technologies facilitate the creation of FAIR datasets, several design aspects require careful planning (see best practices reported in [18]). In particular, one must determine how the dataset will be published—specifying its IRI and defining a consistent IRI structure for internal resources.
The dataset is provided in the Turtle format [19], though the choice of serialization is not critical due to the availability of standard tools for converting between formats. Throughout this paper, Turtle syntax is used for namespaces and prefixes.
As part of the Gallo-Sicilian project, our dataset is provided under the project namespace.1 In addition, to accommodate future datasets for other Gallo-Sicilian languages, we organized our dataset IRI under the lexica sub-namespace, allowing additional vocabularies to be hosted within a coherent and extensible namespace structure.
We adopted hashed IRIs, meaning that each internal resource is identified by a fragment IRI within the following namespace:
@prefix: <https://gallosiciliani.unict.it/ns/lexica/nicosiaesperlinga#> In the remainder of this paper, this namespace will be considered the default, and omitted for brevity when referring to dataset entities.
A core principle of both FAIR and Linked Data approaches is that data about an item should be retrievable by dereferencing its IRI. In our case, since hashed IRIs are used, a client accessing the IRI of an item will retrieve the entire dataset, and then locate the desired resource via its fragment identifier (i.e., the IRI part after the hash).
Although Linked Data and Semantic Web technologies are general-purposes, they support the specification of domain-specific vocabularies and thesauri.
FAIR guidelines also emphasize the importance of domain-relevant community standards. In the domain of computational linguistics, [20] provides a comprehensive overview of best practices for representing lexical resources as Linked Data, along with a review of key initiatives and projects in the ifeld.
A widely adopted standard for lexical resource modeling is OntoLex-lemon, an OWL vocabulary
introduced in [21]. It has become the de facto standard for lexical information encoding (see, e.g.,
[22, 23, 24, 25, 26]). OntoLex-lemon is modular, with each module devoted to a specific aspect of lexical
information representation and defined in its own namespace. Several extensions have been proposed
to address specialized representational needs. In our case, we use:
• the Etymological Extension, introduced in [27] and further discussed in Section 5, which we use
to represent etymological relations between Sicilian etymons and Gallo-Sicilian borrowings; and
• the Liguistic Phenomena Ontology (LiPh) [
Our dataset is built upon OntoLex-lemon and the aforementioned extensions. The following sections illustrate how these vocabularies are used to fulfill the representational goals mentioned in Section 1.
While our resource leverages OWL and includes formal vocabularies (such as OntoLex-lemon and LiPh), it is more appropriately described as a dataset rather than using the general term ontology. This is because it primarily consists of structured lexical and etymological data—specific entries, forms, derivations, and annotations—rather than a general, reusable conceptual schema. Ontologies are indeed used within the dataset to ensure semantic precision and interoperability, but the core content is a collection of domain-specific instances derived from empirical linguistic sources.
As stated in Goal [G2], the first component to be described is the lexicon, along with its associated metadata. The OntoLex-lemon module responsible for representing lexica and their metadata is called lime. Conceptually, a lexicon is a collection of lexical entries for a particular language. Lexical entries will be discussed in detail in Section 4. For the purposes of the present section, it sufices to note that lexical entries correspond to the individual entries of a vocabulary.
The namespace of the lime module is as follows:
Lexica are represented in OntoLex-lemon as instances of the lime:Lexicon class. Each lime:Lexicon instance is linked to its constituent lexical entries via the lime:entry property. Accordingly, the individual lexicon in our dataset is declared as an instance of lime:Lexicon.
As reported in [21], various Semantic Web vocabularies can be used to provide basic metadata for instances of lime:Lexicon, such as title, author, publication date, and more (see also [28, 29]). In fact, lime:Lexicon extends the Dataset class from VoID, a vocabulary for providing metadata about RDF datasets, as described in [30, 31]. VoID, in turn, recommends the use of Dublin Core Metadata Terms (see [29]). Accordingly, metadata for our dataset—such as title, description, authorship, license, versioning, and creation date—is provided as properties of the lexicon individual. Furthermore, we cite [7] as source of our ontology, since the lexical entries in our lexicon were extracted from it. Lastly, information about project funding is expressed using the EUropean Research Information Ontology (EURIO), available at [32].
However, OWL itself provides a native mechanism for embedding metadata: every OWL ontology or dataset must include an individual whose IRI matches that of the ontology itself, and metadata must be attached directly to this individual. Since this is an intrinsic feature of OWL, metadata specified in this way is recognized by all Linked Data agents and is commonly used, for example, by platforms such as the Linked Open Data Cloud2 to index ontologies and datasets. For these reasons, metadata for our dataset is provided also as properties of the ontology individual, following the guidelines in [18].
The lime module provides the lime:language property to indicate the language of all entries in a lexicon. As is standard in OWL, languages are specified using language tags, as defined in [33]. These tags are built from subtags in the IANA Language Subtag Registry.3 Unfortunately, this registry does not include a specific tag for Gallo-Italic as a whole—only for certain varieties such as Piemontese. Therefore, we use the tag mis, which denotes an unencoded language.
Alongside the lime:language datatype property, the OntoLex-lemon specification recommends the use of the corresponding object property from the Dublin Core vocabulary (see [29]) dcterms:language, where the dcterms stands for the Dublin Core terms namespace. In our case, this property is used to link the lexicon to the Glottolog entry for Gallo-Italic languages. Glottolog, described in [34], is a repository of language identifiers and metadata, with a particular focus on lesser-resourced and under-documented languages.
The language of lexical expressions in our lexicon is further specified through geographic characterization. To this end, our lexicon is linked, via the Dublin Core dcterms:coverage property, to entities representing the towns of Nicosia and Sperlinga in the Geonames ontology (see [35]).
Finally, the lime:linguisticCatalog property is used to indicate the catalog of linguistic categories adopted in the lexicon to describe the grammatical and lexical properties of its entries. Since OntoLexlemon is agnostic with respect to category systems, it encourages the reuse of established external vocabularies. In our dataset, we adopt the category system provided by LexInfo, as described in [36] and reiterated in [21]. The following prefix declaration is used for referencing LexInfo elements:
Dictionary entries are represented in OntoLex-lemon as instances of ontolex:LexicalEntry, where ontolex is the namespace for elements of the OntoLex-lemon core module
For brevity, we will refer to instances of this class simply as “entries”. All entries in our dataset were retrieved from [7] by parsing the corresponding PDF file, excluding those that could not be classified as either nouns or verbs. An example of an entry from [7] is shown in Figure 2.
For our purposes, each entry must be associated with both a lemma and a part of speech.
A lemma is a specific grammatical form of a lexical expression, typically serving as the canonical or citation form in dictionaries. For instance, the lemma reported in the dictionary entry shown in Figure 2 is andè. The criteria for identifying lemmas vary by language. For example, in English, the singular form of a noun is conventionally used as the lemma.
Grammatical forms are represented in OntoLex-lemon as instances of the ontolex:Form class. Among the forms associated with a lexical entry, the one that corresponds to the lemma is specified using the ontolex:canonicalForm property. Each form must include at least one written representation, expressed via the datatype property ontolex:writtenRep. The range of this property is the datatype rdf:langString, as defined in [ 16].4 As a consequence, every written representation must be provided as a string with an accompanying language tag consistent with the language declared for the lexicon (in our case, mis).
As noted in Section 3, our lexicon adopts the LexInfo category system. Accordingly, each entry is assigned a part of speech by connecting the ontolex:LexicalEntry individual to the appropriate LexInfo one—lexinfo:noun for nouns and lexinfo:verb for verbs—using the lexinfo:partOfSpeech property.
Parts of speech are retrieved from [7] by examining the word class field ( classe di parola in Figure 2), which is visually distinguished in magenta. These word classes convey information beyond just part of speech—for example, they may also indicate mood, gender, number, and so on. The source dictionary contains approximately 400 distinct values for this field. Each of these has been manually recognized as indicating a noun, a verb, or another type of entry that falls outside the scope of this work (such as adjectives). For instance, the word class “verbo” in the entry shown in Figure 2 corresponds to the lexinfo:verb individual.
As a result of the parsing and conversion of the entries in [7], our dataset contains 10193 instances of ontolex:LexicalEntry. Figure 3 illustrates how the entry of Figure 2 has been encoded in our dataset. 4The prefix rdf refers to the canonical RDF namespace: http://www.w3.org/1999/02/22-rdf-syntax-ns#
As anticipated in Section 2, etymologies for Gallo-Sicilian lemmas borrowed from Sicilian are represented using the OntoLex-lemon Etymological Extension, lemonEty, introduced in [27]. Elements of the LemonEty vocabulary are defined within the following namespace:
@prefix lemonEty: <http://lari-datasets.ilc.cnr.it/lemonEty#> In this section, we briefly summarize the lemonEty representational features employed in our dataset. Etymologies are modeled as instances of the lemonEty:Etymology class. Lexical entries are linked to their corresponding etymologies via the lemonEty:etymology property. Note that multiple lemonEty:Etymology instances may be associated with the same entry, as they represent alternative “hypothesis about the history of a given lexical element,” in the sense of [27].
In lemonEty, etymologies can be represented as ordered lists, enabling the reconstruction of the full genealogy of a lexical expression through a sequence of borrowing and inheritance steps. These lists are modeled using the class lemonEty:EtyLink. Each lemonEty:EtyLink instance describes a single etimological step and is characterized by a source (the etymon) and a target (the derived expression in the recipient language), both of which must be instances of ontolex:LexicalEntry. These are linked to the lemonEty:EtyLink individual via the properties lemonEty:etySource and lemonEty:etyTarget, respectively. An etymology, modeled as an instance of lemonEty:Etymology, is then connected to the first step of the sequence through the property lemonEty:startingLink.
In general, it is also appropriate to relate also the forms involved in the inheritance or borrowing process represented by a lemonEty:EtyLink. For example, the Italian lemma “padre” originates from the borrowing of the Latin accusative form “patrem,” derived from “pater.” To express this, lemonEty provides two additional properties for lemonEty:EtyLink, which can be used to indicate the specific forms involved: lemonEty:etySubSource and lemonEty:etySubTarget.
Following Goal [G4], each etymology in our dataset corresponds to a single borrowing from a Sicilian expression. Thus, each entry representing a Gallo-Sicilian lemma identified as a borrowing from Sicilian is associated with a corresponding lemonEty:Etymology individual, which in turn is linked to a lemonEty:EtyLink individual representing the borrowing process. Each such lemonEty:EtyLink individual is structured as follows: • the dataset entry under consideration is specified as lemonEty:etyTarget; • lemonEty:etySubTarget is assigned the canonical form of the entry; • an ontolex:Form individual representing the Sicilian etymon is used as lemonEty:etySubSource. Notice that the Sicilian etymon is not necessarily a Sicilian lemma, as borrowings may involve any grammatical form. For this reason, the lexical entry corresponding to the etymon is left unspecified, and only the relevant form is included to serve our purposes.
Figure 4 illustrates how the borrowing of the Gallo-Sicilian lemma “còpela” from the Sicilian etymon “còppula” is represented in our dataset using the lemonEty vocabulary.
As discussed in Section 5, lexical expressions may undergo specific linguistic phenomena during
borrowing processes (in the sense of [
@prefix gs: <https://gallosiciliani.unict.it/ns/gs-features#>
These features are expressed using the Linguistic Phenomena Ontology (LiPh), introduced in [
For example, the feature gs:degem.5 replaces every occurrence of “pp” with a single “p”. The regular expression for gs:degem.5 is “pp” and the replacement “p”. This representation in LiPh is shown in Figure 5.
6Note that the Gallo-Sicilian Features Ontology is based on the latest version of LiPh, which slightly difers from the one
described in [
This modeling approach enables automation tasks such as the semi-automatic identification of candidate etymologies and derivations—that is, sequences of linguistic phenomena applied to transform a source expression into its borrowed form. Consider, for example, the borrowing of the Sicilian “còppula”, which becomes “còpela” in Gallo-Sicilian. This transformation involves the application of two Gallo-Sicilian features: • gs:degem.5, which degeminated “pp” into a single “p”, and • gs:vocal.8.a, which modifies the vowel “u” into “e”.
One of the (two) possible derivation paths of from “còppula” to “còpela” proceeds as follows: gs:degem.5 ifrst transforms “còppula” into the intermediate form “còpula,” and then gs:vocal.8.a changes “còpula” into the Gallo-Sicilian lemma “còpela.” This derivation is illustrated in Figure 6.
We leveraged the encoding of Gallo-Sicilian features as regular relations to generate a set of candidate derivations capable of transforming Sicilian expressions into Gallo-Italic lemmas included in our dataset. Specifically, we adopted a brute-force approach: all possible combinations of Gallo-Sicilian features were applied to the Sicilian lexical expressions extracted from [38, 39, 40, 41, 42].7 If a derivation resulted in a lemma present in our dataset, it was recorded as a candidate derivation for that lemma, and the originating Sicilian expression was identified as a candidate etymon. To reduce redundancy, only the shortest derivations—i.e., those with the minimal number of transformation steps—were retained for each lemma. All automatically generated derivations were subsequently reviewed by lexicographers, who evaluated their historical and linguistic plausibility, discarding those deemed implausible.
The set of verified derivations was incorporated into our dataset using LiPh. In addition to representing linguistic phenomena, LiPh provides classes and properties for describing, in fine detail, how these phenomena operate during borrowing and inheritance processes.
Each derivation step is modeled as an instance of the class liph:LinguisticPhenomenaOccurrence, which is connected to:
• the corresponding linguistic phenomena via the property liph:isOccurrenceOf , and
• the source and resulting lexical expressions—whether actual or intermediate forms—through the properties liph:source and liph:target, respectively.
Consider, for example, the first step in the derivation illustrated in Figure 6. The corresponding liph:LinguisticPhenomenaOccurrence instance is linked to the gs:degem.5 phenomenon. The liph:source property is set to the actual form representing the Sicilian etymon “còppula” (as shown in Figure 4), while the liph:target points to a new individual representing the intermediate form “còpula”, modeled as an instance of the class liph:LexicalObject , which is used for representing such intermediate forms. Notably, the language tag associated with “còpula” is set to und (undetermined), following the IANA language subtag registry, as it is not possible to ascertain whether this intermediate form was ever used by actual speakers during a specific historical period.
Figure 7 illustrates how the full derivation from “còppula” to “còpula”, as previously shown in Figure 6, is encoded in our dataset using LiPh.
We have introduced a structured, machine-readable dataset of noun and verb lemmas from the GalloItalic variety spoken in Nicosia and Sperlinga, focusing on borrowings from Sicilian and the linguistic phenomena involved. The dataset leverages Semantic Web technologies to ensure FAIR compliance and adopts community-driven standards such as OntoLex-lemon, lemonEty, and LiPh to support rich, interpretable representations of lexical structure and etymological derivations.
By encoding relevant features as regular relations, we enable semi-automated derivation generation, which ofers both practical utility and theoretical insight into contact-induced language changes. All candidate derivations were validated by expert lexicographers, ensuring their historical and linguistic plausibility.
This resource fills a critical gap in the digital documentation of under-resourced Gallo-Italic varieties and lays the groundwork for further computational and linguistic research on contact phenomena, diachronic processes, and dialectal variation within these speech communities in Sicily.
In addition, we argue that the methodology adopted to identify derivations, as well as the design choices made, can be efectively applied to the study of contact phenomena involving any pair of source and recipient languages.
Future work includes (i) extending the dataset to additional Gallo-Sicilian varieties and other grammatical categories, (ii) refining the brute-force derivation engine through heuristic pruning and statistical weighting, and (iii) integrating usage frequency and temporal data to distinguish attested forms from hypothetical intermediate.
We also plan to publish SPARQL endpoints, potentially equipped with a visual exploration tool such as that described in [43], to support comparative research in contact linguistics.
Finally, our dataset will be included in the Gallo-Sicilian project’s data catalog, which is provided using the DCAT vocabulary as described in [28].
This work was carried out within the PRIN 2022 PNRR project “Contact-induced change and sociolinguistics: an experimental study on the Gallo-Italic dialects of Sicily", funded by the European Union – Next Generation EU, Mission 4, Component 1 (CUP J53D23017360001 - ID P2022YWS8T) under the responsibility of the Research Unit of the University of Catania.
Domenico Cantone, Marianna Nicolosi Asmundo, and Daniele Francesco Santamaria acknowledges the Research Program PIAno di inCEntivi per la Ricerca di Ateneo 2024/2026 – Linea di Intervento I “Progetti di ricerca collaborativa" - Università di Catania - Progetto “Semantic Web of EveryThing through Ontological Protocols" (SWETOP).
During the preparation of this work, the authors used ChatGPT in order to grammar and spelling check, paraphrase and reword. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the publication’s content. [6] A. De Angelis, The strange case of the gallo-italic dialects of sicily: Preservation and innovation in contact-induced change, Languages 8 (2023). URL: https://www.mdpi.com/2226-471X/8/3/163. doi:10.3390/languages8030163. [7] S. C. Trovato, S. Menza, Vocabolario del dialetto galloitalico di Nicosia e Sperlinga, number 39 in Materiali e ricerche dell’Atlante Linguistico della Sicilia, Centro di studi filologici e linguistici siciliani, 2020. [8] M. D. Wilkinson, M. Dumontier, I. J. Aalbersberg, G. Appleton, M. Axton, A. Baak, N. Blomberg, J.-W. Boiten, L. B. da Silva Santos, P. E. Bourne, J. Bouwman, A. J. Brookes, T. Clark, M. Crosas, I. Dillo, O. Dumon, S. Edmunds, C. T. Evelo, R. Finkers, A. Gonzalez-Beltran, A. J. Gray, P. Groth, C. Goble, J. S. Grethe, J. Heringa, P. A. ’t Hoen, R. Hooft, T. Kuhn, R. Kok, J. Kok, S. J. Lusher, M. E. Martone, A. Mons, A. L. Packer, B. Persson, P. Rocca-Serra, M. Roos, R. van Schaik, S.-A. Sansone, E. Schultes, T. Sengstag, T. Slater, G. Strawn, M. A. Swertz, M. Thompson, J. van der Lei, E. van Mulligen, J. Velterop, A. Waagmeester, P. Wittenburg, K. Wolstencroft, J. Zhao, B. Mons, The FAIR guiding principles for scientific data management and stewardship, Scientific Data 3 (2016) 160018–. URL: https://doi.org/10.1038/sdata.2016.18. [9] C. Chiarcos, B. Klimek, C. Fäth, T. Declerck, J. P. McCrae, On the linguistic linked open data infrastructure, in: G. Rehm, K. Bontcheva, K. Choukri, J. Hajic, S. Piperidis, A. Vasiljevs (Eds.), IWLTP@LREC, European Language Resources Association, 2020, pp. 8–15. URL: http://dblp. uni-trier.de/db/conf/lrec/iwltp2020.html#ChiarcosKFDM20. [10] Object Management Group, OMG Unified Modeling Language Specification Version 2.5.1, https: //www.omg.org/spec/UML/2.5.1/, 2017. [11] World Wide Web Consortium, Linked data - design issues, 2006. URL: https://www.w3.org/
DesignIssues/LinkedData.html. [12] C. Bizer, T. Heath, T. Berners-Lee, Linked data - the story so far, International Journal on Semantic Web and Information Systems 5 (2009) 1–22. URL: http://www.igi-global.com/article/ linked-data-story-far/37496. doi:10.4018/jswis.2009081901. [13] M. Duerst, M. Suignard, Internationalized Resource Identifiers (IRIs), RFC 3987, 2005. URL: https: //rfc-editor.org/rfc/rfc3987.txt. doi:10.17487/RFC3987. [14] R. Fielding, J. Reschke, Hypertext Transfer Protocol (HTTP/1.1): Message Syntax and Routing,
RFC 7230, 2014. URL: https://rfc-editor.org/rfc/rfc7230.txt. doi:10.17487/RFC7230. [15] F. Manola, E. Miller (Eds.), RDF Primer, W3C Recommendation, World Wide Web Consortium, 2004. URL: http://www.w3.org/TR/rdf-primer/. [16] D. Brickley, R. Guha, RDF Schema 1.1 - W3C Recommendation 25 February 2014, Technical Report,
World Wide Web Consortium (W3C), 2014. URL: http://www.w3.org/TR/rdf-schema/. [17] P. Hitzler, M. Krötzsch, B. Parsia, P. Patel-Schneider, S. Rudolph, OWL 2 Web Ontology Language Primer, W3C Recommendation, World Wide Web Consortium, 2009. URL: http://www.w3.org/TR/ owl2-primer/. [18] D. Garijo, M. Poveda-Villalón, Best practices for implementing FAIR vocabularies and ontologies on the Web, CoRR abs/2003.13084 (2020). URL: http://dblp.uni-trier.de/db/journals/corr/corr2003. html#abs-2003-13084. [19] D. Beckett, T. Berners-Lee, G. Carothers, E. Prud’hommeaux, RDF 1.1 Turtle, W3C Recommendation, W3C, 2014. URL: http://www.w3.org/TR/2014/REC-turtle-20140225/. [20] A. F. Khan, C. Chiarcos, T. Declerck, D. Gifu, E. González-Blanco García, J. Gracia, M. Ionov, P. Labropoulou, F. Mambrini, J. P. McCrae, É. Pagé-Perron, M. Passarotti, S. Ros Muñoz, C.-O. Truica, When linguistics meets web technologies. Recent advances in modelling linguistic linked data, Semantic Web 13 (2022). URL: https://doi.org/10.5281/zenodo.7129494. doi:10.5281/zenodo. 7129494. [21] P. Cimiano, J. McCrae, P. Buitelaar, Lexicon Model for Ontologies: Community Report, Technical
Report, W3C, 2016. URL: https://www.w3.org/2016/05/ontolex/. [22] Y. M. Abgaz, Using OntoLex-lemon for representing and interlinking lexicographic collections of Bavarian dialects, in: M. Ionov, J. P. McCrae, C. Chiarcos, T. Declerck, J. Bosque-Gil, J. Gracia (Eds.), LDL@LREC, European Language Resources Association, 2020, pp. 61–69. URL: http://dblp.