Starting my PhD
Why I chose to do a PhD in linguistics over an MSc in Speech Technology
This fall, I am beginning a PhD in linguistics at Université Laval.
That was not the plan a few months ago.
I had been accepted into the MSc in Speech Technology at the University of Groningen. I had even found an apartment in the Netherlands. The programme brought together language, programming, and data, and it seemed like a natural continuation of the computer-science studies I had recently undertaken.
It was also a more clearly professional path. Speech technology is a growing field, and the programme would have prepared me to work directly on systems involving speech recognition and language processing.
Yet, as the move approached, I became increasingly unsure that it was the life I wanted.
The problem was not that I had lost interest in computation or speech technology. It was that I could not imagine giving up language science as the central part of my work. I did not want only to become technically proficient at building or improving linguistic technologies. I wanted to continue asking questions about language itself.
Ultimately, I chose the PhD because I could not imagine spending the rest of my working life as a technician rather than as a researcher.
Choosing research
This is not my first experience beginning a doctorate. I previously started a PhD in neuroscience, but eventually realized that it was not the right fit for me.
Leaving was difficult. It also made me understandably cautious about beginning another doctoral programme. A PhD involves years of uncertain work, and an academic position at the end is far from guaranteed.
The MSc in Groningen therefore had a great deal in its favour. It offered specialized technical training, strong employment prospects, and a relatively direct path into an established industry. The PhD offered considerably less certainty.
But the choice was not simply between two degrees. It was between two different relationships with language.
I enjoy programming, data analysis, and computational methods. What motivates me most, however, is using those tools to investigate how language works. I want to formulate questions, construct studies, interpret evidence, and contribute—even modestly—to what is known about language.
I do not know whether completing a PhD will eventually lead to a tenure-track position. It may not. But a permanent academic appointment is not the only condition under which research can be done. Whatever position I ultimately occupy, doctoral training will allow me to continue developing research projects, collaborating with other scholars, publishing, and participating in language science.
That possibility mattered more to me than the security of a more narrowly technical path.
Returning to Quebec French
Once I had decided that research needed to remain central to my life, the next question was what I actually wanted to study.
I kept returning to Quebec French.
More specifically, I wanted to investigate Quebec French using the computational skills I had been developing.
My first idea was to build a large corpus of Quebec French. There was a clear motivation for doing so. Researchers interested in Quebec French often work with corpora created for particular purposes, limited to particular genres, or covering only certain periods. A large and reusable corpus could potentially support many different kinds of linguistic research.
When I discussed the idea with my supervisor, Bruno Courbon, he cautioned that it was probably too ambitious for a doctoral project. Building a corpus of that scale would involve much more than collecting texts. It would also raise questions of representativeness, copyright, metadata, annotation, storage, access, and long-term maintenance.
The corpus risked becoming the thesis rather than a tool for answering a linguistic question.
I therefore needed to reverse the order of my thinking. Instead of beginning with the dataset I wanted to construct, I needed to begin with the phenomenon I wanted to understand.
Bruno suggested that I turn toward neology: the study of new words and new meanings.
The suggestion brought several of my interests together. Neology could be studied using large collections of textual data, but it also raised questions about linguistic structure, social behaviour, institutional intervention, and language change. It would allow me to work on Quebec French while drawing on corpus linguistics, lexicology, terminology, sociolinguistics, statistics, and computational methods.
The timing was also favourable. Bruno knew that ENEOLI, a European network devoted to lexical innovation, was preparing an application to become a COST Action. A project on neology could connect research grounded in Quebec French with a broader international community.
The more I read, the more the topic began to make sense.
Why do some new words spread?
The question at the centre of the project is relatively simple to formulate:
Why do some new words spread while others do not?
Like all living languages, Quebec French is continually enriched by new words and new meanings. Some emerge spontaneously among speakers. Others are proposed deliberately, particularly by the Office québécois de la langue française, which regularly recommends French alternatives to English terms.
Some of these innovations become widely used. Others remain marginal, coexist with competing forms, or disappear.
Their trajectories may depend partly on their linguistic characteristics. A new word may be relatively short or long, transparent or opaque, simple or structurally complex. It may resemble familiar words or fit existing patterns in the language. Certain forms may simply be easier to understand, remember, and reuse.
Diffusion may also depend on extralinguistic factors, including speakers’ attitudes, social networks, age, geographic mobility, the prestige of early users, institutional support, or the types of texts in which a word appears.
Many such explanations have been proposed. They are dispersed, however, across fields including lexicology, terminology, sociolinguistics, psychology, and computer science. They have also rarely been evaluated together using very large collections of naturally occurring language data.
My doctoral project will examine the linguistic and extralinguistic factors associated with the diffusion—or non-diffusion—of institutional and non-institutional neologisms in Quebec French.
This is the kind of question that made the PhD feel necessary to me. It is fundamentally a question about language, but answering it will require the computational methods that first drew me toward speech technology.
Three complementary studies
The thesis is currently organized around three studies.
The first will be a scoping review of the factors that researchers have associated with neological diffusion. Its purpose will be to gather ideas and evidence distributed across several disciplines and produce a structured inventory of the variables that may influence whether a new word succeeds.
This review will provide the conceptual foundation for the two empirical studies that follow.
The second study will examine terms proposed by the OQLF as alternatives to English forms. I plan to compare the use of recommended French terms with that of their English competitors in Quebec newspaper and social-media data.
Why does one institutional recommendation become common while another is rarely encountered outside official contexts? Do recommendations circulate differently in journalism and on Twitter? Which linguistic and social characteristics appear to favour their adoption?
The third study will examine non-institutional neologisms: words that emerge without being introduced through an official language-planning process. Using large corpora of Quebec newspapers and Twitter posts, I hope to identify new forms, trace their circulation, and investigate the factors associated with their spread.
Together, the studies should make it possible to compare institutional and spontaneous lexical innovation across two very different communicative settings.
What I was really choosing
Choosing linguistics did not mean rejecting computation.
The project will require the collection and management of large quantities of textual data, text mining, statistical modelling, and possibly machine-learning methods. My computer-science studies have given me tools that I can now apply to linguistic problems.
The difference is that I want the research question to come first.
An MSc in Speech Technology would have prepared me to solve technical problems involving spoken language. That remains an interesting and valuable form of work. But I realized that I did not want my relationship with language to be limited to the development of systems and applications.
I wanted to study how language changes. I wanted to understand how words enter a language, how they circulate among speakers, why some innovations become established, and how deliberate language planning interacts with actual usage.
I wanted to remain involved in language science for the rest of my life.
A PhD does not guarantee that I will become a professor. It does not guarantee a tenure-track position, or even a conventional academic career. But it gives me the training, the time, and the intellectual foundation to become an independent researcher. That is something I can carry into different kinds of employment and continue practising whether or not academia ultimately offers me a permanent place.
The current plan is ambitious. It involves three studies, several large datasets, and methods that I will continue learning throughout the doctorate. The proposal I submitted for admission includes a detailed schedule leading to a thesis submission in 2025 and a defence in 2026.
I am aware that doctoral projects evolve. Some questions will become more precise, some methods will change, and some parts of the plan may prove more difficult than they currently appear.
But the underlying decision now feels clear.
I gave up an apartment in the Netherlands and a place in a promising technical programme because I did not merely want to work with language. I wanted to continue studying it.
Whatever career follows, that is the work I want to be able to do.