Seeing both the forest and the trees

Humanity’s greatest achievements are contingent on our capacity for language: it allows us to communicate an unbounded number of ideas to each other by changing the order and form of words we know. This is only possible because of structure in the lexical system, which we effortlessly exploit to create and understand words we’ve never heard before: upon hearing “I went zicking yesterday” one will intuitively and effortlessly parse the unknown word and reply “have you zicked before?”. Morphologists characterise systematicity in word systems by abstracting away from idiosyncratic properties of individual words in favour of finding generalisations. We employ features of various kinds to group together words based on their lexical or grammatical meaning, resulting in lexemes and paradigm cells respectively.

The lexeme ZICK, organised by features into paradigm cells:⠀⠀⠀⠀⠀

⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀.         -PAST⠀⠀ +PAST

-3.SINGULAR (I/you/we/they)⠀ zick ⠀⠀⠀ zicked

+3.SINGULAR (he/she/it)⠀⠀⠀⠀ zicks ⠀⠀ zicked

The abstractions of traditional theory fueled progress by pushing idiosyncrasies into the background, thereby making the intense complexity of language tractable. However, significant advances in computational methods now allow this complexity to be interrogated while also keeping theidiosyncrasies in view. This project will use computational tools to build a representation of the lexicon that retains information about the idiosyncratic properties of words, and look for morphological structure in this space. This will enable us to find additional structure that will shed light on morphological phenomena, in addition to providing a more nuanced characterisation of lexemes, cells and features. Experiments will then be performed to test how speakers are using this complex space in language comprehension and production. Specifically, the project will test the hypothesis that morphological category membership is not inherent to words, but rather that speakers "shine a spotlight" on the portion and dimensions of lexical space that are relevant for the task they are performing.

The project will provide us with a new metalanguage to discuss morphological phenomena, and an implementation that does not abstract away from the idiosyncratic behaviour of individual words. It will contribute to a better understanding of lexical organisation, a core question in morphology and crucial for disciplines that rely on morphologists’ understanding of lexical structure to answer their own questions.