Modelling Maltese noun plural classes without morphemes

Abstract

Word-based models of morphology propose that complex words are stored without reference to morphemes. One of the questions that arises is whether information about word forms alone is enough to determine a noun's number from its form. We take up this question by modelling the classification and production of the Maltese noun plural system, using models that do not assume morphemic representations. We use the Tilburg Memory-Based Learner, a computational implementation of exemplar theory and the Naive Discriminative Learner, an implementation of Word and Paradigm, for classification. Both models classify Maltese nouns well. In their current implementations, TiMBL and NDL cannot concatenate sequences of phones that result in word forms. We used two neural networks architectures (LSTM and GRU) to model the production of plurals. We conclude that the Maltese noun plural system can be modelled on the basis of whole words without morphemes, supporting word-based models of morphology.

A Discriminative Lexicon Approach to Word Comprehension, Production and Processing: Maltese Plurals

“…For the held-out data, the model understands and produces unseen form with satisfactory accuracy, around 70%, an accuracy that actually is surprisingly high for a noun system that is irregular in many ways and only semi-productive. Compared to previous modeling results obtained within the framework of Word and Paradigm morphology (Nieder, Tomaschek, et al, 2021), accuracy is much higher than that of an Encoder-Decoder deep learning model, but lower than the exemplar-based model implemented with TiMBL. This model, however, was given a much simpler task, namely, to predict classes of form changes, including classes bringing together many low-frequency patterns of change.…”

“…It remains unclear, however, how the model would have performed if it had been trained on both broken plurals and sound plurals jointly. Nieder, Tomaschek, et al (2021) compared three different computational models to investigate whether it is in principle possible to account for the form-based relations in Maltese nominal paradigms without making recourse to the construct of the morpheme: the Tilburg Memory-Based Learner (TiMBL) (Daelemans et al, 2004), the Naive Discriminative Learner (NDL) (Baayen, 2011), and an Encoder-Decoder network. TiMBL and NDL are classifiers, the Encoder-Decoder network is a model generating actual plural forms.…”

“…For Semitic languages such as Arabic and Maltese, predicting the plural of a noun is set up as a classification problem by Dawdy-Hesterberg and Pierrehumbert (2014), focusing on Arabic, and by Nieder, Tomaschek, et al (2021), focusing on Maltese. The former study used the Generalized Context Model (Nosofsky, 1986), the latter study applied Memory-Based learning (Daelemans et al, 2001), Naive Discriminative Learning (Baayen, 2011), as well as an Encoder-Decoder deep learning architecture (McCoy et al, 2020) to generate plurals from singulars.…”

“…For second language learners, in pedagogical contexts, instructions for how to create the forms of a paradigm from its principal parts can be quite helpful. But whether native speakers derive forms via other forms is still an open question (Blevins, 2016;Nieder, Tomaschek, et al, 2021).…”

"All mimsy were the borogoves” – A discriminative learning model of morphological knowledge in pseudo-word inflection

“…The model proposed by Skousen (1989) works in a similar way, but has a different computational implementation than TiMBL. Despite TiMBL's success in classifying Maltese nouns, as reported in Nieder, Tomaschek, et al (2021), there are two issues with a memory-based analogical model to capture the knowledge of the Maltese noun plural system. The first issue is that the similarity of words is assessed by comparing words in an edge-aligned fashion.…”