From my PhD coursework, 2018
During my first year of PhD studies, I explored a fascinating question: What if English stress patterns aren't governed by abstract phonological rules, but instead emerge from how speakers actually use and process language?
This question led me to examine English stress patterns through the lens of Emergent Grammar, a usage-based framework that proposes grammatical structures aren't predetermined rules but rather patterns that emerge from language use. The paper I wrote then represents an early exploration of these ideas—rough around the edges, but capturing an approach to phonology that I still find compelling.
Traditional generative phonology explains English stress using abstract rules and underlying representations. For example, derivational suffixes like -ion, -ity, and -ic are described with rules that manipulate stress placement based on syllable weight, foot structure, and other abstract phonological units.
The Emergent Grammar perspective offers a radically different view: what if these patterns aren't rules at all, but rather emergent regularities that arise from:
In this term paper, I analyzed three common derivational suffix patterns in English:
Instead of positing abstract stress rules, I argued that speakers learn the stress pattern of -ion words through repeated exposure to high-frequency examples like education, information, and creation. The consistent antepenultimate stress pattern (stress on the third syllable from the end) emerges from analogical extension of these frequent forms.
Similarly, the preantepenultimate stress pattern in -ity words (stress four syllables from the end, as in ability, electricity) can be explained by analogy to stored exemplars rather than abstract foot-building rules.
The penultimate stress pattern in -ic words (democratic, economic) likewise emerges from frequency-based learning and analogical processes.
The paper drew heavily on exemplar theory, which proposes that speakers store detailed memories of linguistic experiences rather than abstract rules. When producing a new word, speakers access stored exemplars and make analogical extensions based on similarity.
For stress patterns, this means:
Looking at this work now, several years later, I can see both its ambitions and its limitations. The paper was exploratory—more of a thought experiment than a rigorous empirical study. I didn't provide corpus evidence, production experiments, or computational modeling to support the emergentist account.
However, the core intuition remains interesting: that phonological patterns, even seemingly systematic ones like English stress, might not require abstract generative machinery. Instead, they could emerge from domain-general cognitive processes like categorization, analogy, and frequency-sensitive learning.
This perspective has continued to influence my thinking about language—whether in my later work on Korean phonetics using big data approaches, or in considering how neural language models might be learning linguistic patterns through distributional statistics rather than explicit rules.
This term paper represents a snapshot of my thinking during the early years of my PhD. It's not polished research, but it captures an intellectual journey—grappling with fundamental questions about the nature of linguistic knowledge and whether the rules we linguists write down actually exist in speakers' minds.
The Emergent Grammar framework offers a compelling alternative to rule-based phonology, though much empirical work remains to fully substantiate its claims. Still, asking these questions—challenging the assumptions we inherit from established frameworks—is what research is all about.
Note: This blog post is based on a term paper I wrote during my first year of PhD studies in Linguistics at the University of Arizona. It represents my thinking at that time and is shared here as part of documenting my research journey.