What do you work on?

I work on a mixture of soci­olin­guis­tics and seman­tics, through the lens of lan­guage per­cep­tion. More specif­i­cal­ly, I look at how inter­nal lin­guis­tic vari­a­tion in swear­ing influ­ences person perception.

What did you do using Gorilla?

I used Gorilla to run a variant cat­e­go­riza­tion task using a mixture of regular words, swear­words and legal non-words all ending with the mor­pheme ‘-ing’. Tokens of these words were arti­fi­cial­ly manip­u­lat­ed in a MATLAB based program called TANDEM-STRAIGHT to create pho­net­ic con­tin­ua between the two common pro­nun­ci­a­tions of ‘-ing’: the stan­dard [ɪŋ] variant (e.g. in ‘playing’) and the more non-stan­dard [ɪn] variant (e.g. in ‘playin’). The 7‑step con­tin­ua created some poten­tial­ly ambigu­ous tokens of each item.

In a forced-choice variant cat­e­go­riza­tion task, par­tic­i­pants heard 30 test stimuli (plus 20 dis­trac­tors). On each trial, par­tic­i­pants had to use their key­boards to select whether they heard an ‘-ing’ word or an ‘-in’ word. I hypoth­e­sised that, for swear­words, lis­ten­ers would be biased to select the more infor­mal [ɪn] variant when the token was acousti­cal­ly ambigu­ous – the same bias shouldn’t occur for pho­net­i­cal­ly match­ing regular words and legal non-words. This pre­dic­tion was moti­vat­ed by the fact that, like the [ɪn] variant, swear­words are also more fre­quent in infor­mal contexts.

What did you find?

Con­trary to this hypoth­e­sis, swear­words actu­al­ly biased lis­ten­ers towards select­ing the [ɪŋ] variant on ambigu­ous tokens, com­pared with regular words and non-words. The results suggest that, rather than swear­words cueing the alve­o­lar form, as pre­dict­ed, swear­words inhib­it­ed per­for­mance. Swear­words are shown to take up more atten­tion­al resources than neutral words, for example in Stroop tasks. With par­tic­i­pants’ atten­tion on the ‘-ing’ token, the swear­words dis­tract­ed them from this task. As a result, they default­ed to the under­ly­ing form (the [ɪŋ] variant) as the most likely option. The results have con­se­quences for how soci­olin­guists inter­est­ed in cog­ni­tion think about atten­tion­al load in perception.

Are your mate­ri­als avail­able in Gorilla Open Materials?

Yes, you can find them here!

For you, what is the stand-out feature in Gorilla?

The best aspect of Gorilla is the func­tion­al­i­ty of the experiment builder for people with little to no coding expe­ri­ence. The process is very simple and allows for a large amount of customization.

How did Gorilla make your life or research better, easier or faster?

Gorilla allowed me to auto-reject people that failed catch trials during my experiment. It was impor­tant that par­tic­i­pants were able to dif­fer­en­ti­ate between unam­bigu­ous tokens of ‘-ing’ and ‘-in’ – to test this I used a series of prac­tice trials. Gorilla then reject­ed anyone who failed to reach this thresh­old by includ­ing their per­for­mance as embed­ded data.

How do you think online research is going to change your field?

Pre­vi­ous­ly, most lan­guage per­cep­tion studies required people to come into the lab in person, which was both costly and time-con­sum­ing – the same is true of gram­mat­i­cal­i­ty judg­ment tasks.

“Online research means this work can be done much faster and at lower cost.”

When you’re not working what do you enjoy doing?

In my spare time I work as a special con­sta­ble in the met­ro­pol­i­tan police in London – I spend time polic­ing the South East bor­oughs of London on a vol­un­tary basis. Apart from that, I am a keen rugby player and I have recent­ly started playing a lot of chess.

Who or what orig­i­nal­ly inspired you to work in your field of research?

I have oddly been inter­est­ed in swear­ing for a long time, back to my under­grad­u­ate studies. During that time I worked part-time as a night­club doorman in Southamp­ton, which exposed me to some less-than-savoury lan­guage on a fairly regular basis.

“I have a par­tic­u­lar anec­dote about a time a cus­tomer referred to me, in on sen­tence, using two seem­ing­ly con­tra­dic­to­ry swear­words for the male and female gen­i­talia respec­tive­ly. This amused me so much that I then decided to start research­ing swear­ing in my studies and, well, here we are.”

What are the main ways people mis­un­der­stand your thesis?

Some of my work focuses on the gram­mat­i­cal­i­ty of swear­ing. For example, I ran an experiment on Gorilla looking at infix­ing e.g. fan‑f*cking-tastic. Accord­ing to phono­log­i­cal theory, this is con­strained by prosody, meaning that it is ill-formed to say fanta‑f*cking-stic.

I think some people take this as me saying that you should always swear accord­ing to the grammar. But frankly, I don’t care how you swear – the beauty of swear­ing, par­tic­u­lar­ly in English, is that is so per­mis­sive and so end­less­ly cre­ative. I often encounter non-lin­guists more gen­er­al­ly that think I am going to be a grammar-nazi, but again, no self-respect­ing lin­guist would actu­al­ly spend their time cor­rect­ing people – it goes direct­ly against what we stand for.

If you could inter­view any researcher (alive or dead), who would it be and why?

I’d liked to have inter­viewed Simone de Beau­voir, for the simple fact that she was a bril­liant writer and thinker. I have another project about phrases like ‘the wife’, which is so much more common than ‘the husband’, and I’d be keen to get her take on it.

Why did you choose Gorilla?

A pre­vi­ous attempt at running an online experiment had used free soft­ware – while cheaper, their server couldn’t sustain the level of activ­i­ty, to the experiment would reg­u­lar­ly crash. I chose Gorilla on advice from my PhD super­vi­sor, another regular user.

Did you include any special fea­tures in your study to ensure good quality data? If so, what did you do?

I used a head­phone check task to ensure that par­tic­i­pants were indeed using head­phones and not just the speak­ers on their com­put­ers. This task was created by another research group. It is a 3‑AFC “Which tone is qui­etest?” task with 200Hz pure tones. Unbe­knownst to the par­tic­i­pant, a random one of the tones is in antiphase across the stereo chan­nels, result­ing in heavy atten­u­a­tion only when heard over loud­speak­ers (but not over head­phones). This results in very poor per­for­mance if the task is attempt­ed without headphones.

What is the biggest advan­tage of online research methods?

Ensur­ing quality results that would match those that you could collect in the lab.

What advice would you give to someone start­ing out in behav­iour­al science research?

The only way to learn is trial and error. You can read all the open mate­ri­als and advice columns, but ulti­mate­ly you won’t learn what works and what doesn’t until you have tried to create an experiment from start to finish. Also, pilot pilot pilot – I can’t stress that enough.

What’s your favourite science inter­net meme?

Any meme that exem­pli­fies the strug­gles of sta­tis­tics and R in particular.

Are there any online courses, pod­casts, dis­cus­sion groups or resources that you’d rec­om­mend to others?

For anyone inter­est­ed in seman­tics (and who can speak German), there is a great series on YouTube by Daniel Gutz­mann.

What science book have your read recent­ly that you’d rec­om­mend to others.

A great new book about lin­guis­tics for the general reader is Lan­guage Unlim­it­ed: The Science Behind Our Most Cre­ative Power by David Adger.