This article comprises a brief overview of my PhD research proposal investigating the ontological approach of Modern Greek (MG) morphology. Its main objective is to study contemporary onto-linguistic models in order to form an onto-morphological tool for MG morphological analysis. The research was motivated by the lack of an ontologically holistic approach based on the Semantic Web (SW) paradigm to represent MG morphology. After a brief review on the current ontological setting within the Semantic Web, the respective morphological framework is determined and placed into the Strong Lexicalist theory justified by MG morpheme-based nature. Following this, main research questions are defined and the methodology of the research is presented as an itinerary process between ontological development, theory and lexical data testing. Finally, the article concludes with some preliminary research results based on a morpheme-based analysis of indicative MG lexical data in the MMoOn ontological model.
Morphology focuses on the least meaningful entities within words, called morphemes2, as well as
how words are composed (word formation) or inflected. Except for the extensive linguistic framework
that also includes empirical analysis, the study of language has been triggered by informational models,
among which is the ontological paradigm [
For MG morphology, in particular, the major attempts of representation to date have been mostly
machine readable dictionaries (MRDs) [
On the other hand, more purposeful steps towards language ontological analysis within the SW have been made in the last decade by the LLOD community, resulting in the creation of representational models such as the Ontolex-lemon [14], [15], OLiA [16], GOLD [17] and LexInfo [18], [19]. However, even though all of these models deal with morphological information, they are far from granular nor do they focus on sub-lexical analysis or on derivational morphology. To fill this gap, the MMoOn3 model has been formed lately, focusing exclusively on language morphology and taking a morpheme-based approach that puts the morpheme concept at the center of analysis [20].
In terms of theory, the research is situated within the Lexicon and is sufficiently interpreted by Strong Lexicalism [21]–[27], that regards morphology as a separate and autonomous field in relation to syntax. Additionally, it adopts a binary relational pattern [28] of combinatorial morphology [24] as well as the Lexical Morphology theory, which explains word formation as a layered hierarchical process [29], [30]. The adoption of Lexicalism has been due to the morpheme prioritization as a steady meaningful entity within words and in the Lexicon, a fact that also aligns with the complexity MG creates lexical structures and its rich morphemic typology. The emphasis on the morpheme concept lies, additionally, in the need to justify non-or-hard-transparent words due to their origin from Ancient Greek (AG) (e.g. rίγnimi ‘to break’ > rίγ-ma4 ‘breach’). This intrinsic relation reveals the MG language allomorphic nature, which also requires the management of words as strings of distinct morphemes.
From the previous discussion the following research questions are due to be explored: ● the coverage of the available models and especially of the selected base-model on indicative areas such as: morphemic typology and limits, diachronic analysis, interoperability between morphological levels (inflection, derivation, composition), allomorphy, semantic and grammatical meaning, morphemic relations, morpho-phonological processes, models of word formation, theory representation etc. ● the consistency of the base-model as well as of the MG ontological instance to the former and how this can be expressed in an ontology language e.g. OWL ● Usability issues via realistic use cases. How for example SPARQL can be used to form simple or complex queries for postulating morphological axioms (e.g. the extent of productivity of specific morphemes, which the most frequent derivational pattern is etc.) ● How would the ontology be populated so that it is overall tested and evaluated towards MG lexical data? Can an automated or semi-automated way be leveraged according to related implementations? 3 https://github.com/MMoOn-Project/MMoOn.
4 The phonological transcriptions are based https://www.internationalphoneticassociation.org/content/ipa-chart). on the
International
Phonetic
Alphabet
The methodology of this research moves iteratively between morphological theory, primary lexical data and the MG ontological instance and will prompt continuous tests for the extension, reforming and functionality of the latter. The research will go through the following stages: ● Review of current ontological models for language morphology ● Review of MG contemporary morphological theory with arguable viewpoints in the field. It is possible that this check-up will redefine morphological areas in view of the ontological representation ● Development of the ontological model through: a) schematic or tabular depiction of MG morphological peculiarities b) extension and development of the model according to coverage, consistency and usability criteria with the assistance of an ontology editor (e.g. Protégé) c) assessment of the ontological model with sufficient lexical data via query expansion (QE) or inferencing
In the research so far, the morpheme-based approach has been explored for MG morphology and tested on the MMoOn model, which was selected as most appropriate to host the MG ontological instance [31]. In Figure 1, we ontologically analyze the structural units participating in a MG common concatenative pattern, i.e. -τη-ς (-ti-s) > -τ-ικ-ος (t-ik-os), applied to two different lexical bases: καλλιεργη- (kallierji-) 〜 καλλιεργ- (kallierγ-)5 and χορευ- (xoref-) 〜 χορευ- (xorev-) of the derived words καλλιεργητής ‘cultivator’ (kallierjitίs) > καλλιεργητικός ‘cultivating’ (kallierjitikόs) and χορευτής ‘dancer’ (xoreftίs) > χορευτικός ‘dancing’ (xoreftikόs). We do this by just using the classes mmoon:Word and mmoon:Morph (and their subclasses) in binary formation structures and leveraging the inverse consistsOf ↔ belongsTo object properties (OP) [31]. 5 The 〜 symbol denotes the allomorphic relation (given by the isAllomorphTo object property) between two morphemes.
As to the following steps in the research, other approaches to word formation are to be explored (e.g. word-based) and contrasted to the morpheme-based.
This research was supported by the project: “Activities of the Laboratory on Digital Libraries and Electronic Publishing of the Department of Archives, Library Science and Museology”. [14] B. Klimek, J. P. McCrae, J. Bosque-Gil, M. Ionov, J. K. Tauber, and C. Chiarcos, “Challenges for the Representation of Morphology in Ontology Lexicons,” in Proceedings of the eLex 2019 conference. 1-3 October 2019, Sintra, Portugal. Brno: Lexical Computing CZ, s.r.o., 2019, pp. 570–591. [Online]. Available: https://elex.link/elex2019/wpcontent/uploads/2019/09/eLex_2019_33.pdf [15] J. P. McCrae, J. Bosque-Gil, J. Gracia, and P. Buitelaar, “The OntoLex-Lemon Model: Development and Applications,” in Electronic lexicography in the 21st century: Proceedings of eLex 2017 conference : Lexicography from Scratch, 2017, pp. 587–597. [16] C. Chiarcos and M. Sukhareva, “OLiA – Ontologies of Linguistic Annotation,” Semantic Web, vol. 6, no. 4, pp. 379–386, Aug. 2015, doi: 10.3233/SW-140167. [17] S. Farrar and D. T. Langendoen, “An OWL-DL Implementation of Gold,” in Linguistic Modeling of Information and Markup Languages: Contributions to Language Technology, A. Witt and D. Metzing, Eds. Dordrecht: Springer Netherlands, 2010, pp. 45–66. doi: 10.1007/978-90-481-3331-4_3. [18] P. Buitelaar, P. Cimiano, P. Haase, and M. Sintek, “Towards Linguistically Grounded Ontologies,” in The Semantic Web: Research and Applications, Berlin, Heidelberg, 2009, pp. 111–125. doi: 10.1007/978-3-642-02121-3_12. [19] P. Cimiano, P. Buitelaar, J. McCrae, and M. Stintek, “LexInfo: A declarative model for the lexicon-ontology interface | Elsevier Enhanced Reader,” J. Web Semant., vol. 9, no. 1, pp. 29– 51, 2011, doi: 10.1016/j.websem.2010.11.001. [20] B. Klimek, M. Ackermann, M. Brümmer, and S. Hellmann, “MMoOn Core - The Multilingual
Morpheme Ontology,” Semantic Web, vol. 4, pp. 1–30, 2020. [21] A. M. Di Sciullo and E. Williams, On the Definition of the Word. Cambridge, Mass: MIT Press, 1987. [22] S. Lapointe, “A Theory of Grammatical Agreement,” Ph.D. Diss., University of Massachusetts,
Amherst, 1980. [23] R. Lieber, “On the Organization of the Lexicon,” Ph.D. Diss., MIT, 1980. [24] A. Ralli, Morfologia (in Greek). Athens: Patakis, 2005. [25] A. Ralli, “Morphology in Greek Linguistics: A State-of-the Art,” J. Greek Linguist., vol. 4, pp.
77–130, 2003. [26] E. Selkirk, The syntax of Words. Cambridge, Mass: MIT Press, 1982. [27] Α. Ράλλη, “Η μορφολογία ως αυτόνομο τμήμα της γραμματικής,” in Γλώσσης Χάριν. Τόμος αφιερωμένος από τον Τομέα Γλωσσολογίας στον καθηγητή Γεώργιο Μπαμπινιώτη, Α. Μόζερ, Ed. Ελληνικά Γράμματα, 2008, pp. 141–156. Accessed: Jun. 09, 2020. [Online]. Available: https://www.angelaralli.gr/sites/default/files/Morphology%20as%20an%20%20Independent% 20%20Grammatical%20%20Module%20%28in%20Greek%29.%20%20.pdf [28] R. Jackendoff, “Morphological and Semantic Regularities in the Lexicon,” Linguist. Inq., vol.
7, pp. 89–150, 1975. [29] P. Kiparsky, Lexical Morphology and Phonology. 1982. Accessed: Oct. 25, 2020. [Online].
Available: https://web.stanford.edu/~kiparsky/Papers/Lexical%20Morphology%20and%20Phonology.pd f [30] K. P. Mohanan, The Theory of Lexical Phonology. Studies in Natural Language and Linguistic
Theory. Dordrecht: D. Reidel, 1986. [31] N. Vasilogamvrakis and M. Sfakakis, “A Morpheme-Based Paradigm for the Ontological Analysis of Modern Greek Derivational Morphology,” in Metadata and Semantic Research, Cham, 2022, pp. 389–400. doi: 10.1007/978-3-030-98876-0_34.