Grammatical optionality, qu’est-ce que c’est ça?

To briefly illustrate the phenomenon we’re interested in, we can just use two short examples. Consider the following pairs of sentences.

1. The witch gave the poison to the assassin.
   
2. The witch gave the assassin the poison.

Or

3. The witch’s magic wand.
   
4. The magic wand of the witch.

These expressions are tantamount in English, and are respectively called the dative alternation and the genitive alternation. They have been a professional fascination of Benedikt’s for the last fifteen years or so of his career.

Such variation is in principle present in the grammar of English (in this case). In other words, it is available to any speaker. But the deal is, when speaking, you have to make a choice. Which of these two (or more) grammatical options makes its way out of your mouth?

Now the issue is that there is a foundational assumption in many linguistic frameworks and theories that having to deal with such grammatical options is messy, cumbersome, annoying. In short, bad ☠️❌. Benedikt often refers to the prescriptive writing guide The elements of style by Strunk and White (first published in 1918). Or there is the idea that, if at some point, a linguistic system develops multiple options to kinda express the same thing (insert obligatory reference to Bill Labov RIP), it’s probably a temporary thing…

The only problem with such thinking is that (a) e.g. the dative alternation has been around for a while, (b) some of these phenomena can be found in wildly different languages, (c) there is a wide array of such phenomena that has been discussed in an ever increasing pile of research literature.

Still, there are difficulties involved in such alternations. For instance, when speaking, people have to mentally scan ahead to kind of figure out how long their linguistic constituents (the little groups of words that go together) will be. We tend to place longer constituents at the end of a dative or genitive alternation, in different grammatical optionality contexts. Try using the other alternative for the following two sentences. Feels awkward, huh?!

5. The witch gave the poisson [to the assassin who just came back from holiday.]
   
6. The witch gave the assassin [the poisson that had been prepared for a year and a day.]

And this way, previous work has found many such “language-internal constraints”. But also so-called language-external constraints, like dialect variation, or gender, or age (although we can’t really find effects for those things in our data).

So what’s the big idea?

Given what I’ve tried to summarize before, you might think: gee whiz, scanning ahead, choosing an option (#keuzestress); that all sounds like a lot of work. Maybe having grammatical optionality in your linguistic system really is a drag. Wouldn’t it be better if one form corresponded to one meaning (disregarding all the evidence of the contrary)? And if it’s bad to have multiple ways to express kinda the same thing, wouldn’t people then go uh and um the whole time, or need this time to reflect before uttering the choice they made?

THAT’S WHAT THE RESEARCH IS ABOUT.

So, Benedikt Szmrecsanyi and a former postdoc here at KU Leuven, Matt Hunt Gardner made this really cool paper in 2021 about how in spoken data they do not find this purported / expected correlation between having optionality contexts in a spoken turn and so-called * speech dysfluencies *.

That was a great seminal paper, but there were still many questions left unanswered. And that’s where I enter the picture.

I was very happy to work in my second postdoc with Benedikt (and, remotely, Matt!) on outstanding issues that had to do with potential confounds to that absence of positive correlation. Of course, this is good scientific practice: maybe it was the kind of grammatical alternation that influenced the results? (No, see Van Hoey et al. 2025 in Linguistic Typology at the Crossroads). Maybe it was the number of options one could choose from? (No, see Ma et al. 2025 in English Language and Linguistics). Maybe the number of constraints that govern those choices? (No, see Ma et al. 2025 in English Language and Linguistics). Hmm. How about the cues present when making a choice? Like, surely choices that are more constrained are easier, because more signals point toward one option over the other?? No??? - See Van Hoey et al. 2026 in Language Variation and Change.

Anyway, that was a short summary of the collaborations over the past few years. Now it’s time for some short reflections.

✨ Reflections ✨

(I’m sorry, I just like the sparkles emoji.)

In my first postdoc in Hong Kong, I learned how to conduct lab-based experiments, and I guess I got more confidence in the idea that you can really have multiple projects going on at the same time. In this second postdoc, I really got to witness what it’s like to build a research program. Based on that 2021 paper, Benedikt acquired an internal KU Leuven grant (which I got hired on).

This led to a number of MA theses, and a PhD student, MA Ruiming, who I am co-supervising. She works on the replication of our findings in Mandarin spoken data, which involves identifying, describing, and annotating such grammatical optionality contexts in Mandarin, alongside the dysfluencies.

It also led to another PhD student (who I’m in the committee of), HUANG Fen, who takes a slightly more experimental approach. At the same time, we have (had) other excellent PhD students, like GUAN Sumin and LI Yi, but also TIAN Xiaoyu and LIU Meili, who have (or are) provided detailed studies of particular alternations in Mandarin, their constraints, and their so-called envelopes of variation, stamps included. And honestly, it has been good for my Chinese knowledge to keep involved with this too. I certainly learned a lot.

Then another thought I had related to our project is its dissemination strategy. I have often felt that we kind of deliver similar messages at different conferences. After having come across the term salami slicing, I have made the reflection whether our research is like that. In the end, I don’t think so. Because, as you see, I certainly have reviewed papers that were like “here’s a micro detail, see recent publications (Author 1, Author 2, Author 3, Author 4, all in 2026 – anonymized for review)”. And that’s decidedly not what has been going on with our project.

In fact, every “maybe” question in the previous section involved a lot of manual annotation labor, rethinking what the best analysis method is for the particular question, reframing our findings in light of what was already known etc. And that’s very different from, to take a random example from the paradigm of pokémonastics having a fixed dataset, throwing it in a n-to-n mapping statistical software, seeing what correlations have p < 0.05 and then “motivating” how that should relate to natural language (see an upcoming book chapter), and spreading those findings over multiple papers (the hate is strong).

But yeah, in doing so, I’ve witnessed that a multi-year spread makes sense. I guess it’s a form of slow science, although that is partly due to the slow movements of the peer review juggernaut. For example, the paper in Van Hoey et al. 2025 (Linguistic Typology at the Crossroads) was actually the first paper I wrote together with Benedikt and Matt, back in 2024. Now, we kinda submitted another one first (the Ma et al. one) but that was also because this one was submitted to a special issue. And it’s a great special issue, I recommend interested readers to have a look, also fully open access so NO EXCUSES. That said, confusingly, it was published in 2026 but belongs to the 2025 special issue.

The other paper, Van Hoey et al. 2026 (the real 2026, in Language Variation and Change) was a bit tougher. We had some reviewers that approached the topic from a position that at the time seemed diametrically opposed to us. In the meantime, we have found that our perspectives are more complementary and perhaps involve more overlap than initially thought. Perhaps the dance party at last year’s SLE conference in Bordeaux Montaigne helped to straighten out any kinks in the cable (somehow I feel like I’m mixing metaphors and languages here, hmm, let’s run with #wordsmith and #yolo), demonstrating again that science involves a research community of scholars, and that communication is key.

The future - a step and a leap forward

In fact, those reviewers organized a great workshop (Benoît and Cameron, I want to expressly state that again). The next step forward for our bigger grammatical optionality project is then to work on other types of dysfluencies, with the hope of featuring in the special issue edited by Benoît Leclerq and Cameron Morin.

However, there’s more exciting news. THE NEXT LEAP FORWARD is that Benedikt and I (in collaboration with Matt) have obtained funding from the Research Foundation Flanders (FWO) to build the next addition to our optionality house.

That means WE ARE HIRING. We are extending the research program by investigating phonological and lexical optionality as well. After all, if the dysfluencies in speech aren’t triggered by grammatical optionality, perhaps there are other types of optionality that make them come out.

So yeah, here’s the job ad. Feel free to spread among your network.

Finally, some pictures

Weird right, having a blog update without pics thus far. I guess I can share two key figures from the paper.

From Van Hoey et al. (2025, Linguistic Typology at the Crossroads) there is this diagram. Basically, it’s a way to quickly visualize how two basic and fundamental starting points can influence how one deals with the type of variation we have been investigating. Basically, both groups recognize that alternations exist. But the left group, adhering to these Principles (I won’t go super deep into it, that’s what the papers are for) focuses on what sets two or more options apart. The right group, variationists, instead wallow, but also revel, in the mess that is (near-)synonymy.

Two roads diverged in a yellow wood,

and the one you take here has significant downstream consequences for the kinds of research you do. Nothing more, nothing less.

The second figure is this one, from Van Hoey et al. (2026, Language Variation and Change). Here, we found a way to quantify the relative difficulty of different types of alternations. Future temporal reference (the choice between choosing will, shall, be going to, gonna etc.) is rather hard. Conversely, the dative alternation (give you the soup vs. give the soup to you) is rather “easy”. This probably has to do with the (type of) constraints that have been identified that govern these alternations, and also the possibility space, but it deserves follow-up work. Interestingly, I heard from a colleague in Gothenburg (Sweden) that she found something similar, but I need to know more before I say more.

And there you have it.

Two new papers, a small step forward, a big leap forward, and a #ScienceCommunication bulletin to with capital SC.

Please spread the PhD vacancy among interested parties.

IcoLL (Nagoya)

It’s been the third time that I participate in the Iconicity in Language and Literature conferences. Each time has provided great opportunities to meet with leading scholars in my field, and more importantly, to gauge what the state of the field is. What edgy and new research topics are being pursued, and what can I learn from them? I especially appreciate that this conference’s goal is to explicitly also include a literature dimension. Too often, people stay within the boundaries of their own discipline. And while I don’t think we are all going up the same mountain (in fact, some are going down into the valley if you’d ask me), the means of mountaineering and the travel reports give rise to opportunities for reassessing your own methods. This is a very convoluted metaphor to say that I went to some literature talks and enjoyed them. I even dared to ask some questions:

And I was happy to present a paper I’m doing together with Xiaoyu YU, Shuhao ZHANG, Youngah DO (all HKU) and Dan DEWEY (Brigham Young University) entitled “Behavior mirrored in the brain: An fNIRS study of Chinese ideophone modal exclusivity”. Here is me presenting that bit.

In the paper, currently under review so if any of the reviewers is reading this, hurry up, we basically compare new behaviorally constructed multisensory profiles for collocate-ideophone constructions of the ABB type in Chinese, with neurolinguistic methods. Drop me a message if you want more info or the preprint.

This conference also proved to be the ideal occasion to reunite with my PhD supervisor, Chiarung LU. We had a sort of yuánfèn 緣分 on the first day. As I descended into the depths of the metro system, I suddenly heard a voice behind me “Thomas, 阿智”. And lu and behold, there was Chiarung with her colleague. We made our way to the venue together and got ready for a nice conference.

Later I saw Chiarung present, which was very huíwèi 回味 for me (there’s probably a German word that captures that feeling).

The conference also provided ample opportunity to (re)connect with other iconicity scholars. Most of the good stuff happens during dinners and post-dinner activities. Since this is Japan, that meant karaoke. Unfortunately, I will not be posting videos or pictures of said karaoke. But feel free to take me singing if you want to experience the tragedy of Total eclipse of the heart sung live by me, given that “every now and then I fall apart” as well.

I was very grateful then to meet the rest of the Taiwanese delegation, pictured here at the soon-to-be-eaten conference dinner.

Turns out that Japanese conference dinners are often walking dinners. While that was, uhm, unfortunate last summer at SLE, this time it didn’t bother me at all. In fact, it provided opportunity to take an updated picture with Kimi Akita, one of the most productive ideophone researchers in Japan.

And I also am happy to have seen Kiyoko Toratani in action. Here she is walking us through the somewhat awkward FORCE component in Ibarretxe-Antuñano’s (2019) Motion Semantic Grid. I have also had to find ways to come to terms with that term, so I’m happy other people share the same intellectual struggles.

Of course, Japan being Japan, had dedicated a special day to the 22nd of February. That is called ni-ni-ni (2/22), which is reminiscent of the sound a cat makes in Japanese, nyaa. So of course, 2/22 was CAT DAY! Everybody strike a cat pose!!

In those cat-pictures we see Luis-Miguel Rojas-Borscia, who is just getting into iconicity; Hinano Iida, a promising PhD student of Kimi Akita’s working on iconicity in lexical work; le mao [moi]; and then an up-and-coming group of Germany-based PhD students (see this blog): Josiah Nii Ashie Neequaye, Vanessa Wing Yan Tsang, and Marta Herget. I’ve run into them a few times now and, let’s just say, I am hopeful for the near future!

To top the conference off, we went to the Ghibli park in Aichi, as a further post-conference bonding activity. Here are Kiyoko and me as Howl and girl in Howl Sophie.

This was preceded by a visit to the Valley of the Witches. For most things you needed to have an extra ticket or pay extra, so we didn’t really stay long, although it was nostalgic to see Howl’s (un)moving castle in real life.

You could also pose with other scenes from the Ghibli movies. Here I am helping out with the rescue of Laputa (Castle in the sky), touching an angel (On your mark), and taking a ride on the cat bus (My neighbor Totoro).

And here is the picture I queued more than 30 minutes for: a trainride with No Face (Kaonashi 顔なし, Spirited away). Revel in it.

All in all it was a great IcoLL, huge thanks to the organisers: Ian Joo, Kimi Akita, Hinano Iida, and Christina Ljungberg.

Now it’s time to post-process the talks I saw there. I think my favorite plenary that bears on what I’m interest in right now was Pamela Perniss’s overview of advances in sign language and gesture-based iconicity.

Breaktime

Let your eyes rest a sec. Reposition yourself on your (toilet) seat. Let’s take this moment to congratulate my colleague Chiara Paolini for successfully defending her PhD thesis on semantic vector spaces and grammatical optionality. Here she is depicted with sword and hat, just as the Finnish tradition prescribes. Unfortunately, Belgium is not Finland, so we just made do with toy variants. Hip hip hooray for Dr. Paolini.

Hokugen (Otaru)

After the wonderful conference in Nagoya, it was time to head north. Ian Joo and I flew northward and he was so great as to show me around Sapporo, home of beer (in Japan).

He also took me to his campus in Otaru, where we would be holding Hokugen 1 in the near, now past, future.

If, like me, you had never really heard of Otaru before, I can recommend watching an anime called Golden Kamuy. It’s about a search for a massive Ainu treasure. I guess it sits somewhere between Prison Break (lots of tattooed bodies that form a map), culture appreciation (not appropriation, I hasten to add, before my friend Cedric spews more of his j’accuses in my direction), adventure, and silliness. Highly recommended.

Also, it’s become my current micro-obsession.

That picture is from the National Aina Museum in Upopoy, which is a sort of light Disneyland interactive theme park that lets you experience some Ainu culture, and also educates you. Honestly, it was really fun to do this with the group that would make up the workshop a day later. And maybe the lesson here is that a social programme can also fit before a conference starts. Those bonds allow for casual work-talk while you kinda navigate who would be good conference allies.

Anyway, here we are as Ainu people, and below we are learning how to play an instrument called the mukkur(i). It will take a while to become as proficient as the video embedded below, but never say never.

And here I am with a children’s toy that is meant to scare kids by making sounds. Praise the kamuy.

Later that day, I bought an Ainu language course. And a few days later, I bought a book describing how Golden Kamuy can help us learn more about Ainu culture. It’s in Japanese, so it’ll take a while to read but I’m excited for it.

What’s that? Oh, yeah, I got to give an invited talk at the workshop. For a whole hour I got to talk about “Accommodating path and manner in the lexical aspect of iconic words”. You see, lexical aspect is something I have been trying to wrap my brain around since last summer and I finally feel like I got somewhere. It wouldn’t be a Thomas talk if there were no diagrams. The feedback I got was really good, thanks in particular to Ian, Luis-Miguel, Rodolfo, Jiyeon, and Shawn (Sean?).

Tokyo

The final days in Tōkyō I mostly spent with Luis-Miguel (but final evening with my friend Tomo who especially took his annual leave one day early for me, ありがとう). Anyway, since it was LM’s first time in Tokyo, I thought we should visit the Sensō-ji　浅草寺 in Asakusa　浅草, one of the most well-known temple complexes in Tokyo. I kinda had the cheeky idea to maybe get a rickshaw ride, as I’ve never been brave enough to do that by myself or on previous trips. Luckily, LM was in, and he was also early – as I would have been if I hadn’t jumped on the train in the wrong direction, yet somehow I still arrived exactly on the agreed upon hour, what is this magic public transport system Japan, please Belgium, take notes – and already fixed us the best rickshaw guy in the hood. You might not say it from the picture, but the guy (“Genie”) – who is pursuing a new career in the beer industry – was strong, funny, entertaining, and great at taking pictures.

Honestly, I wouldn’t mind taking a rickshaw in the future. My original opinion was one of compassion, because they have to do this kind of work. But actually, it’s just fit guys (and girls!! we have seen them) who act as guides and know the area really well.

Anyway, after that stint, we went to the capybara cafe, where they wouldn’t have us because of course you needed to reserve online and we didn’t do that, whomp whomp 😕. So instead, we went to Meiji jingū 明治神宮, a shrine dedicated to the Meiji emperor, known from the Meiji restoration. It’s a pretty big area in Tokyo, quiet, yet well-visited.

I really love some core idea of Shintō 神道. For example, a lot of shinto revolves around purification. So as you enter a shrine, there is typically a ritual wash basin, to clean yourself in this holy place. The holy place is also not dedicated to capital God, but rather to one (or multiple) of the kami spirits. Japanese people love throwing five yen (0.027 EUR) coins as an offering. I somewhat tacitly assumed the hole in the coins had to do with it but it’s just because 5 yen in Japanese is go en 五円 which sounds like, you guessed it, go en ご縁 ‘good luck’. You just commune with the kami, clap once for them, clap once more for the divine spark in yourself, bow once for the kami, bow again for yourself, and clap again. If this is done respectfully, this is a great ritual, and the Japanese welcome foreigners to participate.

Back in Nagoya, I also made a visit to Atsuta 熱田神社 shrine, a very old shrine that supposedly holds one of the three treasures of the imperial household (a magical sword). My friend Veerle just happened to ask me to make a short clip about “reading the air” (de lucht lezen) in East Asian societies. So I decided to film there (again, respectfully). You can watch it here, it’s in Dutch though. But you can kind of get the feeling of what it’s like to move through a shrine.

Outro

I think we’ve covered enough. This Japan Twenty Twenty Sixxx trip was great. Intellectually stimulating (during the conference and workshop, but also outside of them). Thought-provoking (the ways in which ‘iconicity’ is interpreted shows that it’s really in the eye of the beholder). Experiential (the Ainu museum opened up a new interest in me, as did the rickshaw ride, some bars we visited, the National Tokyo Museum (pics on request) and the Ghibli park). In sum, it was the whole package and I can’t wait for the next time I get to visit Japan.

Here is a picture of me and a beautiful Japanese taxus in the Ainu Museum. Apunno paye yan!

Invited talk in Taiwan (June 2025)

I was fortunate to return to National Taiwan University, my Alma Mater, to meet up with my former supervisor Chiarung Lu, some old friends and classmates, and the newer cohorts of graduate students. That always starts with good food (and very often also ends with good food).

Dinner with the NTU GIL professors.

Lunch with Iju Hsu.

Friends from the Summer+ programme I attended at NTU in 2014!

A drink with Powei Li.

Visiting a scenic village with A-Sheng, my cohort’s most dependable friend.

A goodbye karaoke with CJ Young.

Streetfood in Taipei.

The main reason I wanted to go back to Taiwan was to attend the 30th anniversary of the Graduate Institute of Linguistics at National Taiwan University, where I did my PhD. It was really great to see again all my fellow students, but also GIL’ers past and present. Can you spot me in the pictures below?

I was also able to finally make a bucket list item come true, namely visiting the Ju Ming 朱銘 (1938-2023) museum. Ju Ming was a Taiwanese artist who, to me at least, is mostly known for his Tai Chi series of stone sculptures. We had one in front the Lèxuéguǎn that used to house the GIL. Fun fact, I was attacked by a protected bird species, the Taiwan blue magpie, but don’t worry: I’m a survivor.

Finally, I also delivered a talk here, on the “Empirical accountability meets theorizing about language variation: The Principle of Optionality.” This is part of a research project that I was actively working on with Benedikt Szmrecsanyi at KU Leuven. Basically, we keep finding that grammatical optionality (of the type I gave the glass to you vs. I gave you the glass) do not lead to more filled pauses or unfilled pauses, that is, things like uh and um or speech planning time. This is somewhat problematic for stubborn interpretations in theoretical linguistics, that have an idea of isomorphism baked into their foundations as a some sort of holy principle. Conversely, in actual language production, you know the thing we can observe, such grammatical optionality is not a problem. So the conclusion this points to is that there is at least a counter principle, namely a Principle of Optionality, that provides a number of benefits for developing and keeping around more than one way to express the same meaning.

Invited talk in Hong Kong (June 2025)

Next stop, Hong Kong.

Here too, it was great to meet up with all the friends and ex-colleagues that made the two years I spent there worth the initial three-week covid _lock-up quarantine. I look back at my time at the University of Hong Kong with bittersweet feelings, for a number of reasons. But focusing on the sweet aspect, forging a new network of friends and like-minded linguists (or academics in general) reverberates until now even.

Lunch with the Language Development Lab.

After the talk I gave at HKU, with my friend Stephen Matthews.

My friends Aaron Chik and (by now) Dr. Rayne Yu!

Dr. Sevilla

My barber Anson

And of course Dr. Do! (Okay okay, Youngah and I are on a first-name basis 🤣)

I went back to Lamma Island, where I used to live. And somehow, I ended up at a friday quiz night!

On the last day I enjoyed some beers with (soon to be) Dr. Hwang and (currently) Dr. Yu, whom I also gave some Belgian chocolate:

Just to say that the network I built there was still alive and kicking. I have missed the people I met in Taiwan and Hong Kong a lot in the past few years, but I am also sure that our paths will keep crossing.

Conference in Košice, Slovakia (June 2025)

As you might surmise, June was a pretty busy month. It ended with the Word-Formation Theories VII & Typology and Universals in Word-Formation VI, where we had a lovely workshop on sound symbolism. The workshop was largely organized by Lívia Körtyelvessy (and Pius Akumbu!), who also recently was part of the editors for the Onomatopoeia in the world’s languages. I was fortunate enough to be invited to write the chapter on Mandarin in this handbook. But even more fortunate to keep on working with Lívia on a second project (that I did together with my student Ruiming Ma) and currently I’m organising a workshop on typology of sound symbolism and ideophones for next year’s SLE. You’re coming too, right?

Here are some pics from the workshop.

Sunsets and spreadsheets

During the summer, I worked quite hard on my FWO project and managed to identify a gap in the literature that has been occupying me for quite a while. I hadn’t really foreseen it in my grant application but I do have the “academic gut feeling” that this is something that’s needed in the field, so I’m working on that. But I’m only teasing it here because it’s still very much work in progress.

So instead, I’ll give you a few photos of fun events during the summer, without context.

Conference in Bordeaux, France (August 2025)

The next big event in our tale is the SLE 2025 conference in Bordeaux. Here I attended two very interesting workshops.

1. A workshop on the Principles of Isomorophism and Optionality, organized by Cameron Morin and Benoît Leclerq. We basically got to defend and later befriend the other camp (the organizers). It was really great to see some colleagues present in this workshop, such as (soon to be Dr.) Chiara Paolini, and my PhD student Ruiming Ma. It’s really great to see how much they have grown in the past few years.

2. But of course, my “main dish” was the workshop on iconicity, organized by Maria Flaksman and Chris Smith. Here I got to present for the first time my current FWO project in which I look at universal and borrowable patterns of iconic words in a sample of the world’s languages. All exciting stuff, keep an eye out for more. It’s also where the image in the header of this update comes from. Basically, I was able to plot the iconic lexicons of two languages (Japanese and Kichwa / Quechua) in terms of semantic features.

But again, your own work is only half the story at a conference. Getting to (re)connect and watch colleagues, friends, linguists in action, that’s what really makes it worth going there.

Like, here is Dr. Ellison Luk in action (together with Dr. Dana Louagie):

Or there’s a whole Taiwan crew (with Ian Joo and Shengfu Wang a.o.)

And some selfies with colleagues that I keep bumping into, such as Veronika Zikmundova and Pius Akumbu:

And sometimes it’s also just to share bad presentations or, when that doesn’t work, go to the beach:

But, should there be a junior researcher reading this, it often pays off to go to events like the conference dinner (this one was very bad) or the other social event stuff (receptions etc.). You might meet people that went to the same place on the same day “but how come we didn’t run into each other” like what happend to Dr. Niklas Erben Johansen and Prof. Aleksandra Bageshova.

Or meet HKU people like Dr. Hing Yuet Fung.

Or indavertedly end up being dressed as the French flag (with Anthe Sevenants and Alexander Van Herpe):

Conference: Santiago de Compostela (September 2025)

The final conference of this year’s conference season was the isLE conference, which is all about English linguistics. You know, I’m somewhat of an Anglicist myself, if not in training, then certainly in occupation. Here I was involved in two presentations again.

1. Together with Benedikt Szmrecsanyi (and Ruiming Ma, and Matt H. Gardner), we spread the word on the non-attraction of (un)filled pauses by grammatical optionality here as well. I can say that while the reception at SLE was courteous, it was also cautious from some people in the workshop. Very different story here in Compostela: most people at isLE actively love the work we’ve been doing, one of them being one of the keynotes, Alexandra D’Arcy. I found it very encouraging and am sure we can keep on investigating these things in the future.

I didn’t want to deny you guys this rather nice picture of Benedikt and myself:

2. Part of the fun but unforeseen past reverberations of conferences has been my work with modality specialist Dr. Alessandro Basile. Together, we’ve been working on quantifying some of the evolutions of modality markers in Singaporean English. Here, Alessandro did the hard work of presenting our work (the first picture is staged).

But the same as before, the dinner is often where bonds are forged. I was very lucky to be at a great table with some lovely linguists from Bamberg, Osnabrück, but also two KU Leuven colleagues, Juliette Kayenbergh and Hilke Ceuppens. I found both of their presentations really interesting and so I’m looking forward to see how their research develops in the next few years.

Finally, Santiago de Compostela was perhaps the most scenic city I got to visit in this conference round. We even got to go on the rooftop of the cathedral!!

The hospedería I stayed at is right behind me:

Finally, shoutout to Mathilde and Alessandro for taking me to eat Galician tapas (including the pulpo), which was delicioso.

Outro, languishing leaves

What people don’t always understand about this conference season, is that it’s fun yes, but it’s also very tiring. You (well, I) are typically always active, you’re sleeping less, you’re engaged more (even though I did play my zombie game on my phone during boring presentations – sometimes you walk in and after a minute you realize you made a mistake). And to make it really work, you need to bring your A game, I wouldn’t want to give my colleagues anything less nor expect anything less from them in return.

In between those intense periods, there’s a lot of boring work, full of spreadsheets and analysis, most of which will never see the light of day, all in the pursuit of knowledge. I keep finding it hard to resist the pressure of the LLM apologists whose research now solely revolves around probing the nature of “what the LLM thinks” (it doesn’t and it’s numbers with a lettered face), but I must say that occasional use is okay, but always intentional and after first exhausting other options.

Finally, to link the summer with this autumnal season in which I’m writing these words: after the summer I started preparing the classes I’m teaching this semester (first and second year intros to linguistics, colloquially known as ATW1 and ATW2 – yes I have become the Dirk Geeraerts this year, at least I try channeling him from time to time too, even though the course content and concomitant style has shifted towards typology JC Verstraete style and semantics/pragmatics Tim VD Cruys style). I’ve also spent some time (unfortunately unfruitfully) preparing for a job application and a major next grant. Let’s keep our fingers crossed for that one. And now it’s back to the trenches of manuscript writing. I invoked that ACADEMIA IS WAR framing as a segue to this goodbye picture: Guérnica by Pablo Picasso.

Until we read again!

Homebrew - terminal

I’m just assuming you are on a Mac of sorts and have access to the terminal. My favorite terminal implementation is iterm2. You can customize it by googling for a guide. Here’s a good one.

Make sure you also install Homebrew, the easy package manager. Just copy the install homebrew code in the terminal.

Three main functions: installing, updating, removing.

For installation, just copy the install function of a package. For this post’s packages those are qpdf and ocrmypdf.

brew install qpdf
brew install ocrmypdf

Most packages will be some sort of interface to the real package, those are called ‘casks’. For example, a lightweight pdf reader I like using is called skim. It’s install command is

brew install --cask skim

After a while you want to maybe update (every so often). That command is simply

brew upgrade

To remove packages, just

brew uninstall <packagename>
brew uninstall --cask <packagename>

Great, now we have qpdf, ocrmypdf and maybe skim. Next, step, navigate to the folder where your document is located (must be absolute path).

cd PATH/TO/FOLDER/JUST/COPY/THIS

You can check the files with the ls function.

ls

OCR-ing text

Very easy, the basic command is on the left. I often find I have to redo suboptimal previous ocr-ing, so that is where the force argument comes from.

ocrmypdft inputdocument.pdf outputdocument.pdf --force-ocr

Bam, presto!

Selecting pages from a file

You can clip page ranges, e.g., pages 1-3, or just single pages, e.g. 1, or a combination. Note: these pages are the actual pages of the document, not the internal numbering.

qpdf input.pdf --pages . 1-3 -- result.pdf

Combining files

To combine, we make use of similar syntax. Here we combine the first page from file1.pdf with the second-third and seventh page of file2.pdf

qpdf --empty --pages file1.pdf 1 file2.pdf 1-3,7 -- result.pdf

The AI con

It’s probably best I begin with a book that just came out: Emily Bender and Alex Hanna’s The AI Con: How to Fight Big Tech’s Hype and Create the Future We Want. The book, which I worked through in audiobook format, was ironically narrated by a human with the most robotic voice I have ever heard. Or maybe it was a robot, a deepfake; it’s hard to tell these days. Still, this is an important book and the authors’ stance is very clear: don’t believe the hype and don’t outsource human creative potential to a machine.

Now I think most of us have played with filters, in spite of the obvious privacy concerns and stuff like that. If the product is free, you probably are the product. And if you want to get more fun filters for your picture, you can pay to remain the product. It is in a way inevitable that asking people to prove they’re not robots by filling out captchas and identifying traffic lights in the end becomes monetized because of this late stage capitalism we find ourselves in.

The tech bros love making alluring promises of all the benefits we will be able to reap if we just put up with the sowing of AI seeds and the cultivation of the tools. A vision of the computer or Data in Star Trek - synthetic life that acts as any human would do or at least can completely understand and think independently (within well-defined and convenient borders like Asimov’s laws for rotobics). Or a while back (2016) Elon Musk promised neuralink chips that will lead us eventually to the singularity. It’s all smoke and mirrors, and not the good kind. And it’s also a global phenomenon, as these flags clearly illustrate (the provenance of this picture is discussed below).

But look at these pictures

Image generators like Dall-E and midjourney took the internet by storm. I think most of my friends played around with it, because it was such a novelty. You just write a prompt, get a few options (#PrincipleOfOptionality), choose one or tell it to go again, or how to adapt, and after some tries you hopefully land on something.

It wasn’t long before the complaints of artists and writers came that massive amounts of their work had been used to train these models. And if you don’t know any artists personally, it’s probably a far from your bed story that doesn’t concern you. But you probably should be, because if you’re on the internet, you’re probably also in it. Anyway, it’s not right that people’s work is stolen just like that. I hate that places where people could sell handcraft artisanal thingamajigs like Etsy are now overrun with generic and bland stuff. Part of this is also the Temus and AliExpresses of the world. I get that price is huge selling point (pun intended), but have you tried to be less consumerist?

In a number of online spaces I find myself in (e.g. OBOD, a fun patreon-linked discord), reactions to AI were first varied but increasingly unison. The novelty of being able to generate “art” wore off real quick with the aforementioned realization of massive nonconsensual scraping of people’s data. And still there were these guys, mesmerized by the hype, who kept defending it. All I could see was an influx of witchy books (favorite genre <3) that would be ghostwritten, if we can even assumer there is a ghost in the shell or homunculus at work within AI (probably not). This had all the astrological portents to be the fake witch books controversy of 2020 but on steroids. In the aforelinkedto discord server, other peeps clutched their pearls too, and I think soon an unspoken rule against AI generated stuff was put in place. Make no mistake though, we love memes and the sharing that is involved; they just need to be made by humans. (As I’m writing this a doom vision of AI generated memes suddenly pops up. Like what would even be the point.)

And I get it, recently we had the Ghibli filter fad, which is of course fun but completely antithetical to many of the exact themes treated in Ghibli movies. Perhaps it’s time we call upon Princess Mononoke to attack the AI servers with her pack of white wolves. Or take the less-enduring starter pack trend — in any case I saw less of this one — which sparked a small backlash with some #StarterPackNoAI resistance. The point is, the novelty of these ideas is really fun: what would this image look like in a Ghibli style? What would be a TVH starter pack? That’s a really creative human question. But the novelty wears off so fast, especially when you know the answers to these questions are generated by ripping off real artists’ work. There is a reason why we respect people that have honed their craft over time and have become masters at it. Take this beautifully drawing by Leen Sevens for example.

View this post on Instagram

A post shared by Leen Sevens (@cyberragdoll)

In the university

Time to turn to my workplace. There was about a half year window (early 2023) in which ChatGPT usage was novel and an influx of presentations included a quirky little section of “I asked ChatGPT what term X means” or something like that, and it still felt novel and edgy, the sort of feeling you may want to convey in a talk. But then, at least among my peers, that novelty wore off and it became kind of cringy.

For example, one presentation (June 2024) aimed at sharing best student supervision practices was rife with “so I asked ChatGPT, how do you …”, slide after slide. Not only did it make for a poor presentation structure, it just undermined the credibility of the speaker. Like what was even the point of inviting this person to share best practices if it was just gonna be synthetic text answers? Where was the expertise??

In KU Leuven’s latest issue of Sonar, pp. 8-13 the medical promises and potential of AI are briefly covered. And it’s not hard to imagine the idea of “putting big medical datasets into the computer and generalizing over them to arrive at more accurate diagnoses” as a benefit of AI. Certainly rings true, if indeed accurate. It is a vision of the future that Star Trek also prophesies. Yet, there is a small catch here. The technique that is being used for this is most likely based on neural networks, which strictly speaking is a subbranch of the catch-all term artificial intelligence.

In any case, the benefits of using this specialised usage of neural networks is evident, but I disagree the label it’s being served under. At the same time, don’t forget about the AI rat actually published (but now retracted) in Frontiers in Cell and Developmental Biology, see also coverage here.

Erm, how did Figure 1 get past a peer reviewer?! https://t.co/pAkRmfuVed H/T @aero_anna pic.twitter.com/iXpZ1FvM1G

— Dr Charlotte Houldcroft (@DrCJ_Houldcroft) February 15, 2024

By hopping on the AI train (with which I mean using “AI” as the main term, because it is in vogue) we risk losing the specificity of techniques, and the actual advancement of popular dissemination of scientific progress. ( waves hand vaguely while muttering “it’s all AI anyway” )

In the same issue of Sonar, I do like the current rector Luc Sels’s (the chancellor) admonition (p. 3) that we should really work on opening the black box that is AI (again, neural networks). The newly rector-elect Severine Vermeire also ran on increased literacy with regard to “(generative) AI”. I certainly hope the literacy starts with teaching students and colleagues to see through the hype and instead keep on developing the expertise necessary to judge these systems and outputs for what they’re worth. Let’s hope the Leuven.AI institute takes this issue seriously.

Yet I can’t get rid of this gnawing feeling in my chest. I am worried that the role of the expert will be confined to the affirmation or rejection of generated, synthetic artificial “intelligence” output. Experts of the future will have been exposed to increasingly larger proportions of AI generated slop, under the guise of training them to deal with it, risking throwing out the baby along with the bathwater. That is, training time is limited, and choices in the curriculum must be made. Let’s hope it’s the right ones.

At the same time, the literacy required to evaluate generative AI output (which, I want to reiterate, is not the same as the neural network medical case mentioned above; it’s just confusingly and erroneously all sold under the same name) is still lacking on all levels.

Let’s start with students. The stories of student papers that reek of GenAI synthetic text are too numerous to even link to one. From early-on lexical indicators (see here for a list, an issue which may be “solved” in more recent models, to leaving in sentences like “as a language model, I cannot …”, it’s all looking pretty grim. As far as I know, there are no reliable detectors for generative AI. But maybe that doesn’t matter. When a literary colleague of mine asked me if I knew any, I just replied: regardless of the process, if the outcome reads like a lot of bla bla and not a lot of boom boom (I hope this works in English, blabla and boemboem certainly works in Dutch), the thing is written badly and will receive a similar evaluation.

Why is that? An easy answer is that ChatGPT not so much hallucinates — this is perhaps the wrong term that has been applied to it — as much as it sells bullshit. There’s even an article “ChatGPT is bullshit”, published in Ethics and Information Techology (2024) which makes this exact point.

Alas, many people also lack a bullshit detector, no matter if the bullshit being presented is served on a written or spoken platter. And I get it, it’s part of our pragmatics that sometimes we just want to give face or get to a different issue, so we don’t critically engage with what’s being said. But if there’s one situation in which we must be cautious and critical, it’s when we are dealing with matters of the academy: it’s part of our moral code, e.g. the ALLEA code of conduct, which I just found out also has a vision on generative AI. After all, professionally, we should embody the advancement of society with factual expertise in our respective specialized domains. It’s why we consider plagiarism as a literal crime. Uncritical usage of generative AI comes pretty close to doing a plagiarismke.

And so it brings me no joy to report on the following issue, but I cannot help but mention it. Recently, we had rectorial elections at my institution. One of the candidates (notably the current vice-rector of internationalization) had this presentation on his faculty visit that was filled to the brim with AI generated images. We’re talking guys with six fingers, girls with anatomically impossible hand configurations, and flags of countries that don’t exist level here. This sparked immediate intense conversation with colleagues, not only because part of that candidate’s slogan was “commitment to people”. Maybe start with your own powerpoint, a very low bar to pass, because the cringe was real. In any case, I find it unacceptable that a vicerector of internationalization runs for rector using AI generated pictures featuring a.o. flags of countries that don’t exist.

I know this has been a long section, but I also want to quickly mention that I’ve left the whole green footprint out of the equation, but it’s certainly something that cannot be left out of the complete discussion. Thus far it seems that using, let’s say ChatGPT, is individually and per prompt not that taxing on the environment. But of course, it’s never a single prompt and a single individual. At scale, we probably should be wary if not worried. Let alone estimating the taxing burdens placed on the environment by not only training a single big ChatGPT model, but in fact training and retraining and overtraining multiple models until the next one is made “public”.

Research and writing

Is there then no use for generative AI?

This month, in the interest of writing this post and hoping to find some solutions to problems I was facing in my research, I decided to pay a month for ChatGPT. Note that this is because I didn’t feel entirely comfortable denouncing a technology without at least trying to evaluate what it can mean to me. I try to take the same approach to movies and books etc. and I kinda hate it when people have an opinion without having done the work of engagement with something, although I do recognize the value of hearsay.

One issue I struggled with was a pernicious python coding problem. Long time followers of this blog will know that I live in Rstudio. However, sometimes the big snake calls. I am not the best at writing python, mostly because I don’t do it often enough, but I can read (read: evaluate) it. So there was a python package that I found that could help me. And iterative reasoning with ChatGPT did help me figure out how to write the python code for my issue, and it did save a huge amount of time this time. At the same time, I didn’t hone my python skills here by visiting Stackoverflow until I somehow DIY’ed an ugly but bespoke working solution that can be used in the future. And that last part is exactly how you advance in programming languages like R and python — you try to add more stuff to your toolbox. While my problem was solved, no new instrument was added to that toolbox. (Oh gods, please don’t see ChatGPT as the only toolbox you’ll ever need.)

In another case study, a colleague of mine were talking about ChatGPT’s “deep research” feature. He engineered a prompt and I did one too, both on topics we are expert enough at. Basically, a collection of google searches woven into a synthetic text, produced in about 15 minutes, was spat out. The time it took was impressive. But the quality was rather shallow. Let’s just say that ChatGPT did not provide “PhD level research” as sometimes is claimed. The other alternative is that “PhD level” means something different for OpenAI than it does for us.

So I find it a difficult question. Obviously, using Gen AI as a search tool is playing a dangerous game but it can be a quick fix.

Indeed, when it first came out, a mate of mine said it was like “talking to a god” — in my experience that’s not how the gods communicate — but sure Jan 🙄🙄. My mom has recently discovered ChatGPT and uses it for personal tarot readings. While I find that pretty funny, she understands that behind it is just a number generating machine that randomly (I guess that is its oracular power ??) gives sequences of words that fit well together (you get the implicit biases for free).

Where I do see some benefit, or at least not fewer issues than competitors, is in checking grammar. I find it hard to see the difference between using ChatGPT as a glorified spellcheck or Grammarly (which also now touts use of AI as does every app in existence fol). Playing around with “make this text 10% less academic” was a nice 30 mins but I also noticed that some nuances in the original text just went missing. Yet maybe for learning purposes there is a place for Gen AI (if only to show the deficiency in the models), although again, the fear is that lack of AI literacy (or let’s face it, plagiarism literacy in general) does not lead to this takeaway…

Recently, we had two Gen AI writing workshops, presented by the team of the ILT institute, who have amazing expertise in teaching writing. The first one focused on charting the landscape of AI tools I was struck by just how many tools there are out there. Basically, we can outsource everything if we want: (i) subject selection, (ii) source identification, (iii) source evaluation, (iv) source handling, (v) development of experiments and materials, (vi) write and cite, (vii) check and publish.

Suddenly, I envisioned a future of AI tools reading papers, writing them, and the role of the researcher being confined to spending the hour it takes to upload materials to Editorial Manager, although that is probably also streamlinable in the future. Is this what triggered the Butlerian Jihad in the Dune series?

But then one of the presenters made a comment that shook me to its core because I hadn’t heard it before: “In this day and age of Gen AI, how dare students even present a text that hadn’t been ‘optimized’ by integrating AI feedback?!”. It’s been a few months, and I still don’t know how I feel about it. Of course, I want students (and colleagues too) to deliver the best texts they can, but I struggled to interpret what they meant with AI feedback.

In the second session that question was answered. The ILT team currently espouses a model called PAIRR (Sperber et al. 2025), involving the following steps in the student writing process: (i) draft, (ii) peer feedback, (iii) AI feedback, (iv) reflect, (v) revise. I guess theoretically his works out, as it stresses the iterative editing process that good writing requires. Practically, the worry that AI will do all steps in one go is definitely there. I get that the goal is to keep some control over the writing process and not to outsource everything, and also that using AI as a feedback tool (again, what is the difference between a ChatGPT giving feedback or a Grammarly doing the same) makes sense but maybe Pandora’s box can’t be closed anymore?

Where is the wabi sabi?!

I know this blog has been a meandering collection of thoughts and it’s totally fine if you left long before getting here. It’s a condensed reflection about (generative) AI and how it has invaded (probably the best term) my life. Every week there is a new doom story (I’m just linking here to Mark Dingemanse’s reporting on unsavory AI experiences with Oxford University Press or the increase of AI text on science blogs; sigh.). I can only reiterate: USE AI WITH CAUTION.

Final story time before the somewhat hopeful envoi. Recently, I found myself on LinkedIn (don’t know why), scrolling a bit absentmindedly through some updates of people. Before long I found this longish post by an acquaintance, talking about AI integration in the company and whatever. And then it suddenly hit me, I was reading generated text. I felt sick to the stomach. How dare this person inflict this upon me? It’s one thing when I go to ChatGPT to solve my problem, but it’s another if you post the output on a social medium without a single disclaimer.

The writing was bland and just a bit too clean. In processing this event, I finally worked through it by contemplating what I like about human text. And it’s actually simple. It’s the wabi sabi 侘び寂び of real texts. To put it bluntly, real texts are always a bit imperfect, because everybody speaks with their own voice. I love it when I can translate Dutch phrases literally into English or Chinese and still get their meaning across. Or when there’s little nuances that you can pick up by reading between the lines. Similarly, I want to read real imperfect scholarly works (theses, dissertations, articles) that at least hint at unique voices, not generic boilerplate outputs.

So let that be the final message for now, because clearly we can take a stance but that doesn’t make the problem of Big AI hype magically go away (believe me, I tried). Don’t just drink the koolAId but retain some of that wabi sabi, it’ll make your writing more palatable.

tl;dr

What do you mean “too long, didn’t read”?

Go read it, it’s not that long, not even in times of diminished/diminishing concentration due to cellphone addiction.

But okay okay, two papers of mine appeared on the same day in their online and #OpenAccess version. This deserves to be celebrated. Ecce this blogpost.

1. Van Hoey, Thomas. 2025. Waddling, Wandering and Waving: Literary Chinese Ideophones and the Motion Semantic Grid. Cognitive Semantics, 11, 1-29. https://doi.org/10.1163/23526416-bja10075 click this link

2. Thompson, Arthur Lewis, Van Hoey, Thomas, Chik, Aaron Wing Cheung and Do, Youngah. 2025. Iconic hand gestures from ideophones exhibit stability and emergent phonological properties: an iterated learning study. Cognitive Linguistics, 2025. https://doi.org/10.1515/cog-2024-0033 click this link

3. Ma, Ruiming, Thomas Van Hoey, and Benedikt Szmrecsanyi. 2025 online. Isomorphism-Inspired Theorising about Optionality and Variation: No Empirical Support from English Grammar. English Language and Linguistics. https://doi.org/10.1017/S1360674325000097 click this link

You can click those links to actually see what they are all about, and even read them, because both are open access.

Am I entering my acceptance era?

January 31. I wake up and, regretabbly out of a bad habit, scroll through all the apps I still allow notifications from. Among them, Outlook (you never know when the world happens to burn down during the night – an ever more likely scenario). Yet, this time it’s a great message:

Hi Thomas Congratulations on your high-quality paper submitted to COSE [Cognitive Semantics, tvh].
It will be the first OA paper accepted and published online in 2025. COSE will be turned to OA from 2025 onwards, sponsored by my universiy,Beihang, which means we have paid the OA fee for your article.

Best Thomas Li Editor-in-Chief

I am very excited because I kind of did that paper “between the soup and the potatoes” as we say in Dutch. It is linked to a special issue devoted to new applications of the Talmyan motion typology (Talmy 2000a, b) to the realm of ideophones. The editors for this special issue, Kiyoko Toratani and Kimi Akita, already had held a great online workshop about this topic a few months before, last summer. Alas, I couldn’t join then but I watched the recordings, and figured I could do something with (literary) Chinese ideophones. So I am very grateful to them for still inviting me and keeping me in the loop for this special issue.

Last February or so I did a bout of annotating and invoked the awen for creative writing. Add to that a few walks in the woods of Wuustwezel (#uilenbos) and whippity whoopi, the idea for a paper emerged. I made some vague promises in the abstract about what I predicted my findings to be. Okay, it must be said, after working on Chinese ideophones for almost a decade now, I have developed some intuitions of what the data is like, so it’s not like that came totally out of thin air.

Fast forward a few months, and I’m annotating the data between the FWO interview and the result to keep myself busy. I try a few clustering techniques. And lo and behold, one of my most favorite papers suddenly takes shape.

I am actually very proud of this one. It’s probably the most QLVL’y paper I’ve made to date. And the graphics came out pretty well.

For example, this 3D version of plotting first- and second-level components (it’s a bit technical, sowwy) was really cool to make.

I also know I had a good paper in my hands when I presented it at the 21st International Congress of Linguists (September 2024 in Poznań, Poland) and the feedback of the audience was really positive.

So yeah, am I entering my paper acceptance era??

Slow science

It certainly seems like this finally the paper acceptance era. For example, one key piece of research conducted during my postdoc at the University of Hong Kong also finally came through in the last few months. Note that this was not an easy birth: it took us quite a few years to cast our conference presentation into a coherent paper that the reviewers/gatekeepers (you know who you are Reviewer 2!) deemed acceptable. On top of that, the journal we submitted it to in the end, Cognitive Linguistics, had a switch of editor in chief, and that also delayed it, as far as I understand the history.

The concept of the paper is quite simple: we played the game of telephone. But with a twist: we would be looking at ideophones and gesture interplay + teaching some participants the wrong, non-iconic mappings.

This transmission experiment led to lots of interesting findings. But the ones that made it into the paper revolved around the gesture parameter of handshape, an aspect of sign languages and gesture in general that is generally viewed as not the most salient, yet important for conveying nuance. We found that we can break iconic mappings down into smaller features, like a phonological system. Sometimes iconicity “breaks” when the physiological naturalness of certain handshapes evolves into different configurations. Other times, iconicity acts as a strong enough force to resist those natural tendencies of handshape drift.

Cool stuff! Enjoy this picture.

There is another

Little did I know that on Friday afternoon another paper would be unleashed unto the masses online and open access.

In this paper, Ruiming Ma (it’s her first paper so, you go girl!), Benedikt Szmrecsanyi, and I devise a nifty three-step modeling procedure to investigate the issue we’re interested in: namely the non-effect of grammatical optionality on relative complexity, in spite of what some theories and principles may suggest (maybe this is a subtweet at Reviewer 2).

It’s short and sweet, and developed out of a nice MA thesis.

Click like, subscribe, and don’t forget to cite <3

Now go read them!!!

Van Hoey, Thomas. 2025. Waddling, Wandering and Waving: Literary Chinese Ideophones and the Motion Semantic Grid. Cognitive Semantics, 11:1-29.
https://doi.org/10.1163/23526416-bja10075 click this link

Thompson, Arthur Lewis, Van Hoey, Thomas, Chik, Aaron Wing Cheung and Do, Youngah. 2025 online. Iconic hand gestures from ideophones exhibit stability and emergent phonological properties: an iterated learning study. Cognitive Linguistics.
https://doi.org/10.1515/cog-2024-0033 click this link

Ma, Ruiming, Thomas Van Hoey, and Benedikt Szmrecsanyi. 2025 online. Isomorphism-Inspired Theorising about Optionality and Variation: No Empirical Support from English Grammar. English Language and Linguistics. https://doi.org/10.1017/S1360674325000097 click this link

You can click those links to actually see what they are all about, and even read them, because both are open access.

Who dis

So in a twist of logic, I probably don’t need to do an update, because most will have seen the wonderful places I’ve been able to go to in my time working on the complexity project with Benedikt Szmrecsanyi. For example, with the gracious support from Benedikt, I was able to travel to Oxford for the project start-up with Matt H. Gardner, to South Africa, Australia, Germany, Greece, Leuven (🫰), and soon the center of the world, aka Antwerp (🫰).

Formally, the time has come and gone that I was hired on that project. Of course, I am still involved and will remain to be so as long as I’m here. I feel I have finally gotten a grip on the issues within the variationist literature that we are talking about, and working with the SWITCHBOARD spoken corpus data definitely has helped with that. Besides, working with Benedikt is truly a pleasure. That is why we also have a shared PhD student, Ruiming Ma, who is looking at similar things as I was but in Chinese data, which is really exciting. You can read more about that project here.

Last February, I was also happy to receive my previous supervisor from HKU, Youngah Do, here in Leuven. It was a great opportunity to wrap up some projects we still had in the pipeline, but you know, after some reviewer time had passed. Academic schedules and scholarly time move at a different pace than most people realize.

In general, I have landed in a remarkably good research environment. Without exaggeration, it is hard to find days in which just a quick conversation with someone from the linguistics department says something that makes you go: oh, huh, interesting, lemme think about it, in particular the people from the QLVL. FunC research groups, Sometimes this concerns particular methods, or ways of looking at the data; other times, it’s just a topic or a turn in the conversation that will linger on. And bonus points for them for putting up with my stupid jokes at lunch time. It is appreciated.

As a group, we work well. I wouldn’t say it’s a hive mind culture, but more a concerned individuals focus group that discusses all sorts of issues. And we’ve also done some team building. For instance, here is the group pic of our QLVL teambuilding last summer.

New project

Now, one of the reasons why it has been a minute since I updated my blog is that *I’ve been busy.* Even though I got all the support I needed, after about half a year I realized that I couldn’t just rest on my laurels and diminish, go into the West and remain Galadriel. No, if I wanted all to love me and despair, to make this Galadriel metaphor even more convoluted, I would need to start applying to grants, professorships or postdoc fellowships.

While I originally had the idea to apply for an ERC Starting Grant, I was too late last year and my idea wasn’t full-fledged yet either. But luckily I was able to channel it into an application for the Senior Postdoctoral Fellowship of the Research Foundation of Flanders, our national funding agency. Fortunately, I didn’t have to sell my soul to Lucifer, so I was able to put off a Faustian deal, but just to be sure I went to this statue in Liège on the day of the interview after the event. #hustleculture

I’m grateful that Benedikt agreed to act as the supervisor the project. And I’ve also gotten my previous MA in Linguistics’ supervisor Jean-Christophe Verstraete on board. They are both a great match because the former specialized in variationist linguistics, and the latter in typology. And my project is situated at the crossroads between these two.

The full title of the project is “Borrowing iconic words: Loss, transfer, and reinvention of depiction”. To give you an idea of the written pitch, here are the first paragraph of my project.

The past two decades have shown that iconicity is a fundamental aspect of language. Words are iconic when their forms reveal something about their meanings. For example, woof sounds like the ‘sound of a dog barking’, and zigzag reveals something about ‘running while alternating from side to side’. Iconic words can be found in any natural language. Non-iconic words (prosaic words) can also display iconicity, such as size sound symbolism (/a/ is big, /i/ is small). This is especially true for core vocabulary, i.e., words that resist borrowing into other language families. Hence, there is the hypothesis that iconicity is negatively correlated with borrowability. Surprisingly, this hypothesis has not been tested on iconic words, making it hard to estimate the limits of the relationship between iconicity and borrowing.

Now, you can see that there is the impetus: there is this idea that is floating around. Clearly, I don’t entirely agree with it, for reasons my future papers will make clear. In any case, my proposed solution to this problem is to study the relationship between borrowability and iconicity starting from iconic words rather than from prosaic words that were (synchronically) rated in iconicity.

I’m very excited to start on this project. The first task consists of the creation of a typological database with lots of iconic lexical material, think onomatopoeia (meow, woof, bang!) and ideophones (zigzag, Jap. pikapika ‘flashing’). The want for such a database is not entirely new. Back in the covid times of 2020-2021, a few of us (Bonnie McLean, Ian Joo, Arthur Thompson, Mark Dingemanse, Jiyeon Park) had the idea to create some sort of “IdeoNet”. But we didn’t really have the funding or the space in our careers to do so. Having advanced a bit further in the career trajectory, I am happy we do have the funding and dedicated man-hour task (i.e., me and my project) to work on this. It is also rebranded as “Depicticon”. So keep an eye out and if you have data to share, please do so.

The rest of the project remains for a future update, shrouded in the mist that has been descending on an autumnal seasoned Leuven. Time to go watch some colored leaves fall down. Adieu.

Abstract

(skip this if it’s too technical)

Iconicity, or the resemblance between form and meaning, is often ascribed to a special status and contrasted with default assumptions of arbitrariness in spoken language. But does iconicity in spoken language have a special status when it comes to learnability? A simple way to gauge learnability is to see how well something is retrieved from memory. We can further contrast this with guessability, to see (1) whether the ease of guessing the meanings of ideophones outperforms the rate at which they are remembered; and (2) how willing participants’ are to reassess what they were taught in a prior task—a novel contribution of this study. We replicate prior guessing and memory tasks using ideophones and adjectives from Japanese, Korean, and Igbo. Our results show that although native Cantonese speakers guessed ideophone meanings above chance level, they memorized both ideophones and adjectives with comparable accuracy. However, response time data show that participants took significantly longer to respond correctly to adjective–meaning pairs—indicating a discrepancy in a cognitive effort that favored the recognition of ideophones. In a follow-up reassessment task, participants who were taught foil translations were more likely to choose the true translations for ideophones rather than adjectives. By comparing the findings from our guessing and memory tasks, we conclude that iconicity is more accessible if a task requires participants to actively seek out sound-meaning associations.

Background

As regular readers of this blog will know, ideophones are iconic words. That means that their meanings are somehow apparent in the sounds of those words. Or, there is a perceived resemblance between form and meaning.

We have known for a while now that people all over the world are sensitive so certain form-meaning mappings, especially in laboratory phonology studies, which use stimuli like kiki and bouba (go ask your grandma which one she thinks is ‘round and curvy’ and which one is ‘sharp and spiky’ — I bet she’ll make the correct choice [but also don’t come at me for your earnings]).

We have also known that infants and children pick up these words really early. Think of the time you had to learn the sounds of all animals. That’s easy and evocative because those onomatopoeic sounds have a strong resemblance in form and meaning.

Researchers have proposed that it’s exactly that form-meaning mapping that makes it easier to remember those onomatopoeic sounds. And so they’re easier to learn, because you need to less of an effort to make arbitrary connections, like you have to do for most words, e.g., the word wall doesn’t indicate anything about its meaning of WALL.

And if those words are easier to learn, they should also be easier to memorize right?

Guessing

But actually, learning is such a big and vague task, that it’s useful to separate how “guessable” these words are by virtue of their form. For instance, you can probably guess that miāo is a sound made by a cat (in Chinese), but can you guess what the word nuru-nuru in Japanese means? If I tell you that it means ‘slippery’ you might make an easy connection there between those /n/ and /r/ sounds, but one that may not be obvious just by looking at it. So that means that miāo is more guessable than nuru-nuru.

Okay, quick side note, another recent paper by McLean et al. (2023) talks about these issues as well, so go check that out when you’re done here.

There are quite a few studies that look the guessability of words, often using English speakers. In our study, we wanted to see if Cantonese speakers were good at guessing the meaning of ideophones. And we used ideophones from three languages (Igbo, Japanese, Korean).

This experiment, which replicated the methods of a study by Dingemanse et al. 2016 found that our Cantonese participants are very good at guessing the correct meaning when presented with two choice alternatives. As expected, especially for ideophones depicting sound meanings.

Memorizing

Okay, so we were interested in learning. After making sure that our participants were sensititive to these words, we were wondering how good they would be at remembering them.

So we conducted another experiment, inspired by Lockwood et al. (2016).

We made it quite tricky. Half of the form-meaning mappings that participants learned were correct (nuru-nuru meaning ‘slippery’) and half of them were wrong (nuru-nuru means ‘flashing light’). After being exposed to those words we asked them whether they had learned a particular mapping.

It’s kind of like the game of Memory: you get presented with a form and a meaning and you need to tell whether they fit together or not, based on what you learned. It turns out that, on average, it’s hit or miss. There is no particular memory benefit in the ideophones we tested with Cantonese speakers that leads to a good memorisation.

Also, importantly, we also conducted this experimental protocol with adjectives and the results were very similar. It’s all in the paper, but I’m just highlighting the ideophone part here.

Reassessing

Remember how participants sometimes had to learn the correct meaning for an ideophone and other times a wrong, coerced meaning? That comes back now.

After the memorization task, we basically told them: hey, we have taught you the wrong meaning for half of the words you had to learn. Now it’s your turn to think again about these words. And then we exposed them to another guessing task. But because it’s after teaching and after memorizing, it’s in fact a reassessment task.

The participants were pretty good at this task (details in paper).

What I’ll show here is whether participants decided to stick with the meaning they had learned or choose the alternative we presented to them. After all, if ideophones have iconic aspects to tap into, people should be reluctant to choose the alternative when given the choice to change what they had learned, and vice versa, should be eager to change when presented with the real meaning, abandoning the wrong meaning because it was not motivated in a form-meaning mapping.

I quickly mentioned that we also looked at adjectives. For those words, we didn’t have any special predictions, but it turns out they be have somewhat similar to ideophones, although the effect is much less strong when we ran models.

So here you can see (on the right side) how the yellow mean is low at the 20% mark. This means that for those items, which were taught in the right condition, participants stuck to the original meaning they had learned. For coerced meanings (right side, blue), we get a mean of 57% or so, indicating very easy choices for the alternatives.

Looking at adjectives the true form-meaning mapping was also low, but the coerced one much closer to the halfway mark.

For the other languages we found similar results (in the paper), though maybe less pronounced. The models we ran indicate that there are significant effects of language (Japanese > Korean > Igbo) and of course of type (ideophone > adjective), as well as some important interactions with the condition boosting or curbing the effects.

But you’ll have to read the whole paper for that!

Main take away

Ideophones are, once again, proven to be quite guessable. However, the guessability aspects do not necessarily correlate with memory benefits (veridical recall). This may be because not all tasks invite people to think of form-meaning mappings as driven by iconicity. In fact, once given the chance to rethink and reassess the meanings of certain words, participants may tap into iconicity and choose the correct meaning after all, even if they didn’t remember whether they had learned a particular form-meaning pair.

Not all tasks investigating iconicity are equally good at drawing out iconicity. And the learning benefit of iconic words should at least be split up into guessability and memorization, if not more.

Where can I find that paper

Van Hoey, Thomas, Arthur L. Thompson, Youngah Do & Mark Dingemanse. 2023. Iconicity in Ideophones: Guessing, memorizing, and reassessing. Cognitive Science 47(4). e13268. https://doi.org/10.1111/cogs.13268.

You can find the paper version by clicking this link.

There is an online repository with all data, code and figures, which can be found here.

It also hosts the presentation we gave at the Deutsche Gesellschaft für Sprachwissenschaft (DGfS 2023), which I twote about here:

Happy to have presented at #DGfS2023

"Iconicity in ideophones: Guessing, memorizing, and reassessing"

Work with Arthur Thompson, Youngah Do and Mark Dingemanse

Missed it? Don't worry - soon in Cognitive Science #cogsci @cogsci_soc pic.twitter.com/uvZxp7xOVN

— Thomas Van Hoey 司馬智 | ABB - collocate-ideophone (@Simazhi) March 8, 2023

Have reading and have fun with ideophones!

ABB, not ABBA

Imagine.

The year is 2019. It’s September. You’re eating Chicago deep dish “pizza” in Taipei. Suddenly you hear people nearby play a fun game where they have to take turns saying a words of a certain form. If they’re too slow or say a wrong one, they’re out.

This actually happened to, and I couldn’t believe my ears. The words in question all had the syllabic template of ABB.

lǜ-yóuyóu
yìng-bāngbāng
hēi-xūxū

These, respectively, mean “very verdant green”, “super hard” and “deep black”. And as most speakers of Chinese languages will know, the BB part of them will add a certain vivid quality to the A part. The greenness of lǜ 綠 is elaborated in a vivid manner by the slipperiness of yóuyóu, which helps because yóuyóu 油油 means ‘oil oil’. On the other hand, the hardness of yìng 硬 ‘hard’ is elaborated by the hardness depicted by bāngbāng 邦邦. And the blackness of hēi 黑 by the BB part xūxū 魆魆.

These are one of the most fun parts of learning Chinese (honestly, all kinds of vivid words are fun to a certain degree, see Dingemanse & Thompson 2020) and absolutely necessary to become a fluent speaker of Mandarin, but also Cantonese or other varieties, and they have special predictive properties as well (see Thompson et al. 2022).

ABB, sound symbolism and ideophones

Already in the 1970s we find scholars like Benjamin K. T’sou (1978) stating that ABB forms are a particular locus for identifying sound symbolism in Chinese. Even if you don’t know what sound symbolism is, you may have hear about words like kiki and bouba. If I told you that one of these words sounds round and curvy, while the other is spiky and angular, there’s a very high chance that you would say kiki is the spiky angular one. There’s a whole field of linguistics that tries to identify patterns where the formal aspect of language (sound, but also writing) in some way reveal something about the meaning. That field is called iconicity.

Now, strictly speaking, not all statistical formal patterns that tell us something about the meaning are iconic, but that’s not the topic of this post. This post is also not about identifying particular mappings in ABB words.

What this post is about, is how ABB words - after that great observation by T’sou - were basically only studied from formal aspects and how they hang together with other permutations of patterns in which occur, think juggling letters like ABAB, ABB, BBA etc.

That led me to the belief (in my dissertation, Van Hoey 2020) that we should really be looking at these words from a more function point-of-view. If ABB words are so vivid, then that must mean there is at least the possibility that there is ideophony at play in them.

Ideophones can be cross-linguistically defined as “marked words that depict sensory imagery and which belong to an open lexical class” (Dingemanse 2019). But you probably know them as onomatopoeias. The only meaningful difference (for the majority of people in the field) is that onomatopoeia is seen as depiction of sound, while ideophones are much broader, covering senses like visual patterns, inner feelings, textures etc., which are not really in the scope of the term onomatopoeia.

Where am I going with this

Okay, so if ABB words potentially involve ideophone-like qualities, then most likely it will be in the BB part, as reduplication is a highly frequent trait of ideophones across languages. The A part, then, is somewhat of a fixed collocate of this ideophone. How do I know it’s fixed? Well, if you can play a game that’s not just random syllables of the form ABB, then there is a certain degree of idiomaticity at play as well.

So it stands to reason to think of ABB words as collocate-ideophone constructions.

But ABB words are not the only party in town (a reviewer in my article that I’m gonna plug in a second did not like that phrasing so I’m recycling it here). There are many other collocates we can look at, in varying length.

In my article that recently came out in Cognitive Linguistics (currently ahead of print) I basically regale the story of ABB words, how they were studied traditionally, and what it would mean to treat them as collocate-ideophone constructions.

Of course, if they are collocate-ideophone constructions, they are probably “special” among those constructions, and I do find that they are prototypical in a sense. So finally I look at all the data from a number of perspectives, such as type frequency, token frequency, cue validity, dispersion. Finally, I put these single sources of information together in “tuples”.

Anyways, I hope you’ll check out the article.

Van Hoey, Thomas. 2023. ABB, a salient prototype of collocate–ideophone constructions in Mandarin Chinese. Cognitive Linguistics aop. https://doi.org/10.1515/cog-2022-0031.

And if you don’t have access to it, I’m more than happy to share it with you.

I don’t wanna read, I just want to hear your voice

Okay, for these people I also have a solution. I presented this in a talk last summer.

Van Hoey, Thomas. 2022. ABB, a salient prototype of collocate-ideophone constructions in Mandarin Chinese. In. Toronto; Nagoya: York University; Nagoya University.

You can find the talk on my YouTube channel, which I’m linking to here but also embedding below.

Goodbye gatherings

One of the most salient hallmarks of my imminent departure were all the gatherings with friends and colleagues. I have enjoyed all of them, whether it was pizza party, home dinner, walk around the block or gathering on or under a roof.

Also the idea of liminality was felt most clearly at the following exact time: 7 December 2022 at around 21 o’clock. At this particular time I met up with my friend Rita on the beach – itself not quite land and not quite sea, between the wild nature of Lamma Island and the steampunk inducing Power Station, under the full moon. I had already officially left HKU, but not quite yet Hong Kong. But at the same time the liminal is a time and space where some of the most transformational things can happen, like me reflecting on and reappreciating the times past and the times ahead. Oh and we made a bonfire, which is always super fun.

I cannot stress enough how thankful I am that I met these people. Even though I left, I feel good knowing that I’ve taken a piece of them with me, in return for offering up a piece of myself to them. Thanks for being guiding stars in my life.

The final week

Apart from the stress induced by dealing with all the administration (in a broad term, including taxes, rental contracts, clearing out my apartment, avoiding mr Toxic), I kept thinking of what I wanted to do these final week. Other people might want to go hiking or whatever, but I just mostly wanted to revisit places I had enjoyed before. And that includes museums. Here are some pictures.

The last day

I went to Madame Tussaud’s, where I was alone. So of course I perfected my selfie skills and took *many* selfies memories I will treasure forever.

That time I got crowned to Miss Universe

That time I got kidnapped by the Hulk

That time I got to sing on stage with JJ Lin

That time Gaga wanted to take a pic with me

That time GEM and I were recording our new hit song

That time Jay Chou requested to work with me

That time Super Junior Siwon and I were on the red carpet

That time I got into a fight with Bruce Lee. Be like water my friends.

That time Marilyn and I stood on top of a vent

That time I visited Yayoi Kusuma in her natural habitat

That time I taught the Queen (RIP) how to take a selfie

That time I was sparkling in the sun with ~Edward Culllen~ Robert Pattinson

That time I got breakfast at Tiffany’s

Then I said goodbye to HKU

And then I left…

To Taiwan

That’s right, I left Hong Kong to go to Taiwan for a week, to meet up with CJ, aka Baetje.

We also went to a museum lol, where they had an exhibition about the crazy 80s.

I also met up with some of my Tatung University students.

And I got to revisit NTU and meet up with some of my NTU peeps!!

And we did karaoke (which I had done before leaving HK as well)

Alas, all good things come to an end. Time to say goodbye to CJ and head home. Still in the liminal zone. At least I got *a whole row* on the plane for myself.

Goodbye CJ <3

Renaissance picture with my dad and my luggage

Whoopi longlingly awaiting my arrival

A citizen of Wuustwezel (Loenhoutttttt) once more. I’m sorry but we have the most beautiful town hall.

Oh yeah it became 2023 and I started work at KU Leuven, with Benedikt Szmrecsanyi (more on that in another update). Here is me on my first day, going from Leuven’s city hall to the Erasmushuis, which is where the Faculty of Arts is housed.

魯汶風雅，於斯爲盛。 “Literary pursuits at Leuven, may I flourish here.”

This calligraphy hangs right across from my office; it’s as if they knew I was coming here.

To Oxford

Starting the new project in Leuven required that I pay a visit to Matt Gardner in Oxford. So I did, and in passing visited some scenic views of Oxford, as well as the musical Wicked. A trip to Oxford, in this day and age, also requires compatible reading. I chose Babel, Or the Necessity of Violence: An Arcane History of the Oxford Translators’ Revolution by RF Kuang. It was a great book.

By the way, I saw a student use ChatGPT for writing a paper while on the train.

My road of good intentions led where such roads always lead, i.e., the musical Wicked.

Can we also talk about the bookstores of London, urgh.

Including the Atlantis bookshop <3

And Treadwells, where I trod well.

Finally, I also met up with Franzi and Antonio.

Leaving liminality

Imbolc came and February passed by as well.

I went for a fancy birthday dinner for my mom and we took my aunt as well. Looking good ladies!

And I was invited for another birthday dinner (Chiara, in yellow).

I can say that the feeling is much different from when I got to Hong Kong. If you recall (or if you want to reread), I felt quite uprooted in Hong Kong. While I eventually grew some roots there, the feeling here has been much different. I can’t figure out if it’s the familiarity of coming back home, back to an alma mater, back to a number of friends (although meeting up with some of them turns out to be more arduous than it should be :s :s ), or feeling relieved about the absence of a certain coworker. In any case, I’m back, transformed after 7 years abroad and ready to face the future.

Leaving liminality, but, you know what, I’m Leuven it.

Spring commences 立春 4 Feb

I guess I’ve always been into cyclical systems like calendars. For instance, learning about how the traditional Chinese calendar worked in my second year of college was super fascinating. Actually, I should be calling them calendars, since it’s both a solar and a lunar calendar that at times needed to be put in sync again. Anyways, here is a picture of me looking at the so-called Wheel of the Year, which contains a cycle of eight seasonal festivals. I like the concept very much, but I didn’t feel like it was particularly applicable to a Hong Kongtext, where it is cold for like 3 weeks in February and December.

Here’s a picture of a beautiful tree on the HKU campus. I guess I’m a tree guy.

Rain water 雨水 19 Feb

While it wasn’t really cold, being covered by a blanket gives a sense of comfort and coziness that is second nature to Belgians. And apparently also to Polish and Siberian people, as evidenced by this picture of me, Rita and Dasha.

This was the period in which omicron finally made its way into Hong Kong, and all the barbershops closed. Luckily, we had an outside barber on Lamma.

Beautiful island with its beautiful sunsets and sunset views. I miss it already. But I am not quite sure if I miss the island or the people and their vibe. Possibly both.

Insects awaken 驚蟄 6 Mar

The insects awaken and with them the blue skies returned.

But, like, not every day. Sometimes it was still foggy. And on such a day I decided to surprise my friends Tomo and Jinyoung by awaiting them on the Peak.

Spring equinox 春分 21 Mar

Night and day are equally long and friends that had been held up (holed up?) abroad started making their way back to Hong Kong. Like Marcelo, who wanted to have a big bite at Big Bite.

I also found this replica of Taiwan’s bent post office boxes.

Bright and clear 清明 (aka my birthday) 5 Apr

While my 30th birthday was an absolute disaster, at least initially, “due to circumstances”, my 31st was exactly what I wanted. My friends that came over (shoutout to Tomo, Jinyoung, Marcelo and Yingheng) really managed to surprise me and the atmosphere of the evening was just right. Forever grateful.

The highlight of the evening was this beautiful smörgåsbord of Cheesecake Factory cakes.

Excitingly, I landed an interview for a faculty position at another university in Hong Kong. Disappointingly, the interviewers were late for the interview, which led me to panic, in an icecold manner, whether or not I was in the correct zoom room for about 4 minutes. Unmannerly, I did not receive any replies on whether I was still in the long/short list after inquiring after a month had passed. I mean, do I really want to work for a department that does not respect potential colleagues? Here is a picture of me with the mixed feelings after the interview. It went well, but they were late… 0/10 would not recommend.

Corn rain 穀雨 20 Apr

Some of the friends made along the way are actually routes, if that makes sense. There is a sense of comfort in walking along pathways you know well and know better with every iteration. For me that would be the many walks I took along Victoria Harbour (HK side), from HKU to the pier. It didn’t hurt that the weather was amazing and the views appetizing.

Summer commences 立夏 6 May

May came around and that means two things: 1) Birthday of CJ, which I couldn’t celebrate with him :( . 2) Eurovision, which I managed to watch all night long with Rita. Great edition, at least for the two of us.

My supervisor Youngah came back to Hong Kong and I had actually missed her while she was gone for about two months, even though we saw each other regularly on zoom.

I was also saying goodbye to some friends in preparation for my summer trip: I would be presenting a paper at the Iconicity in Language and Literature conference in Paris at the end of May. Like this fun group of HKU.

Or my dear friend Dasha. See you in Europe soon!!

Corn forms 小滿 21 May

Wow I’m back home! And I’m super in love with the way home looks at the cusp of summer.

I also went to Paris! Let’s face it, there are too many pictures from that trip, I am only including the ones with friends here (blame the IG poll).

And also some of the obligatory sightseeing. By the way, I started watching Emily in Paris while it was Thomas in Paris and it’s both cringey and fun. Also, I look like this self-portrait by Van Gogh, but I am Van Hoey.

Corn on ear 芒種 6 Jun

Back home and even more in love with the fields that surround my home. And taking long walks with the dog.

Time to meet some friends: ultra fans of this blog and friends since highschool: Ruben and Cedric

Friends from sinology days: Lai Man, Silke, Joke, Yoana, Anneleen and me

Friends from my student organization: Ann-Sofie and Elisabeth

Summer solstice 夏至 21 Jun

In case it wasn’t clear, the weather was very good. 10/10.

In this period, I went to two very important places. The first is Leuven, where I finally went to visit my friends from the East Asian Library again (Arjan and Benedicte).

And I also met up with fellow R aficionada and Cognitive linguist Mariana. This turned out to be a very amusing afternoon, and also a very fruitful one, as it led me to my new job, which I will start very soon.

The second trip was visiting Nijmegen (Netherlands), where I spent the afternoon and them some with my other supervisor Mark. Very enjoyable as well and beautiful campus.

As it turns out, one of my friends from my NTU days is now studying in Nijmegen, so I also met up with Yeh Chih, which was the perfect end to a perfect day.

And I also went to Lier with my dad

And met with my friends Margaux, Ruben and Kristof

And almost went to Bonnie’s wedding (sorry I couldn’t make it but happy we saw each other in Brussels!!!)

Moderate heat 小暑 7 Jul

Even thought his solar term is called moderate heat I didn’t think it was very moderate; it was quite hot. And it only got hotter when my Spanish friend Antonio came to visit me. It had been a few years since we had seen each other but some friendships can definitely overcome this.

And then, one early morning – mind you I was still working during this time and often had meetings starting at 7 AM with a team that was working in HK time – I woke up and my mom told me she had phone problems. Grumbling and not yet quite awake I went to have a look and she showed me a screen of an airplane route, casually saying: “So, we need to pick CJ up this afternoon.” Thinking I was still dreaming, I splashed some water on my face and reconfirmed.

You see, a devious and cunning scheme had been going on behind my back, and nobody had told me that this was happening. So there I was, later that day, picking up my dearest baetje at the train station. It felt really good to see him again after 11 months.

Major heat 大暑 23 Jul

So this summer back home turned out quite great I would say. We went to Brussels for a walk in the Sonian Forest and what I was selling to my friends as “the Stonehedge of Belgium” but is actually a monument to forestry that’s only about a century old (sorry not sorry).

We met up with family at a summer.

We met donkeys under an amazing sky.

And CJ’s brother TJ also popped in for a visit.

And, we crossed something of my bucket list, the Vlooyberg tower after visiting my aunt Jenny!

Autumn commences 立秋 8 Aug

So it turns out CJ absolutely destroyed me at mini golf. At least the food with the sino friends was good and comforting.

I developed a curious interest in labyrinths. Here’s one in Mechelen.

Here’s us celebrating our favourite K-pop band VIXX. #Starlight forever, but also room for new (more active) bands.

Here is one of their best songs.

All good songs must come to an end. On CJ’s last day here we went to Leiden. One place we stopped at was the Tolkien store (nerd alert), where I was attacked by an Uruk-Hai.

And then, a goodbye kiss at the windmill. Luckily, this time, it didn’t take a year before we met up again.

On one of my last days back home I got to meet up with Ann from Taiwann.

Then, a rainbow as I was leaving.

And an originally 7 days of quarantine that was shortened to 3 days right before I got back. Here’s the HK skyline. By the way, New York does not deserve the epithet concrete jungle as much as Hong Kong.

End of heat 處暑 23 Aug

The end of heat is only apt for Belgium I suppose, in any case not for Hong Kong. Luckily, Jinyoung was staying at my place during the first few days so I didn’t feel too alone.

We also got new members in the Language Development Lab.

And I got to embrace Changhe (I forced him into my pun).

And I got to – finally – participate in a psycholinguistic experiment. The cap is not comfortable.

One of the things that brought me great joy in the last few months was the participation in pub quizzes with Marcelo and friends. Even though we never won, I had great fun.

White dew 白露 8 Sep

Spontaneous encounters are the best. Here’s one with the Polish girls Anna and Rita. Thanks for being so fabulous, even on the 8 AM ferry.

It was at this point that work became really busy as well. And that lasted all the way until the end. # No rest for the wicked (this was the alternative title I had in mind for this update).

Autumn equinox 秋分 23 Sep

Luckily, the more work found itself to me, the more I was able to balance it with outings with friends. And, oooff, I really needed that counter balance. A collection of moments.

Drinking cuba libre with Rita because of Cuba’s progressiveness.

Mo also stopped by.

Aaron’s first shake shack.

Saizeriya with Xiaoyu and Bingzi.

Frites with Jinyoung and Tomo.

Somehow ending up at the Hermès show with Xiaoyu.

My hair stylist Anson Law (like the mirror guy) I’m really gonna miss <3

Oh, and I beheld a beautiful broken branch. So I’m sharing it with you here.

The sun must be greeted too.

Cold dew 寒露 8 Oct

This solar term’s title is not applicable to Hong Kong, error 404. This can be evidenced by the warm October nights we spent at the power station beach.

Also one inside picture with Mona, Ka, Kayi and Mattew

Frost 霜降 23 Oct

In writing this blog, I have come to realize that the solar terms’ titles reflect my mood rather the Hong Kongnese environment. It started to dawn on me that time spent in the Fragrant Harbour was stretching thin. One thing I got to cross of the bucket list was Cantonese opera (I think it might have been Teochew opera but I’m not sure). Marcelo and Ksenia took me to Farewell my concubine.

The next natsukashii thing is of course Halloween (samhain), one of the highlights of Lamma life. It started with an eerily purple hued sky, and lasted a good amount of time, which I spent watching people with friends.

Zack Sparrow.

This beautiful Green Fairy is my Canadian friend Zoë.

Ewen <3

I got to wear this sick helmet.

Jon the axe killer (pen killer)? There was also a girl with pompoms and a shirt that said “Go ceiling” – she was a ceiling fan.

Mi amigo Carlos Cuarto.

Beginning of winter 立冬 7 Nov

I MET AN INFLUENCER – rope skipping legend Pakhung Jai.

And I was visited by more and more people from HKU: Aaron, Samuel and Wayne. Fond memories were created.

Oh, and maybe you don’t know this about me (hi nice to meet you) but quite often girls will fall asleep on my shoulders. Proof.

This has been a lot

To end with a sort of cliffhanger, I think I will end here, and talk about the transitory period of my leaving Hong Kong, going to Taiwan, arriving in Belgium and beginning my new job for the next update (early February?). That way, we can do the last five solar terms and be done with it.

As always, thanks for reading, and thanks for enriching my life this year, whether we met IRL or just interacted online. 2022 was a pretty great year, despite having to deal with some major unpleasant moments that can only be epitomized by #Free Britney Spears’s hit single Toxic. Luckily, I got to lean on a select bunch of you to get through it and I’m very grateful for that.

Let me know if you are interested in reading particular things on this blog.

’ello ’ello hola bonjour and ciao!

Productivity

In terms of productivity, I’m glad to say that we already have one publication out. We in this case refers to Youngah Do and Arthur Thompson. I’ve written about the paper here if you feel a desire to learn more about correlations between articulatory gestures and semantic features.

Here is a picture of us celebrating our acceptance in Cognitive Linguistics.

I’ve also done quite a lot of reviews for other papers, namely for 8 articles with multiple rounds for most of them. It has been interesting to “be on the other side” of the author-reviewer interface. I’m not convinced that they are two sides of the same coin. However, I can share that one of the authors (or team of authors?) sent a very kind message that they really appreciated the way the reviewers had done their job. So while I’ve had to face a learning curve in reviewing, I think my style has become better over the year. Sorry for the one paper I rejected but congrats to the ones that got major revisions or even better. I’m somewhat proud to already have seen some of the articles in print.

And I must say, I actually enjoy reviewing – there is a certain excitement or arousal when confronted with new ideas. So I hope for more review opportunities this year (but not too much).

Here is a picture of my Reviewer 1 face.

There’s of course also lots of projects in the work. And if often feels like juggling many balls. In other words, it can feel a bit overwhelming or like drowning in a bath of project balls. You will see the picture this metaphor leads to in a bit. Or scroll down now and find out now.

But I the point I want to make is that there are ways to stay on top of different projects. I find myself most productive working deeply on one or two at the same time rather than multitasking all the time. To help me do that, I use a combination of my bullet journal (let me know if you want me to make a blogpost about that), google calendar and a simple kanban system that involves post-it notes but that can also be implemented in systems like notion. Once again, let me know if you want me to talk about that, I love talking about academic workflow tips and tricks, and I know I’m not the only one. For instance, my kōhai 後輩 from Japanology Ann-Sofie is building a business out of giving tips for effective planning, check out Plan A here.

Here is the picture of me metaphorically drowning in a pool of multitasking balls. I smile because I have dealt with it.

Network

Especially in the latter half of 2021 I have made quite a few new friends, and that has been a very good web of support for events that were to come.

Halloween

For instance Halloween, which I spent with my Korean friend Jinyoung. She came to Lamma where it is apparently a big deal to celebrate it. I really loved seeing all the costumes. We didn’t really dress up, but here are some pictures that will give you an impression.

Christmas and New Year

For Christmas and New Year I hosted friends at my place. The deal was simple: I would make dessert and they would cook all the rest. And an even better part of the deal was that we had different cuisines from across Asia: China, Korea and Japan (in no particular order I love you all equally).

For New Year my Japanese friend Tomo came over again and this time I was treated to a full-blown Japanese dinner. This was after we had eaten pizza so of course we couldn’t finish it. But on the upside, we had food for a few days after the event.

But it wasn’t the only Christmas party I went to. In fact, I had a wonderful Christmas Eve: I was on a boat, with the CJ highschool crew. I had a good time.

And I also went over to my supervisor Youngah’s place. Jinyoung had told me to memorize the phrase chodaehae jusyeoseo gamsahabnidaang 초대해 주셔서 감사합니다앙, which means “thanks for inviting me”. Luckily it maps well on the meter of Jingle Bells so I was able to do so. The food was also really good.

And speaking of good food: also with other friends from school we had some really nice dinners. I didn’t take a picture of the food at Christmas but at Thanksgiving – which was the first time I celebrated that, like ever – we had a big turkey. With the magic of perspective it looks even bigger.

And at the end of 2021 I also found myself part of a pubquiz team, which has been pretty fun as well. Our team name is “I thought it was a disco”, which is a welcome change from the typical “Jezus Kwistus” I get back in Belgium. Here’s a picture of us being victorious.

Chinese New Year

In the start of 2022 we also said goodbye to Szeto and Ian, two members of our short-but-interesting-lived TypoClub. So Ian, Szeto, Gabor and I had a goodbye lunch for Szeto (going to Italy) and Ian (going to Sweden). Given the current stringent covid measures in Hong Kong I don’t know when I will see them again.

I also said goodbye to one of my office mates, Jackie, who has been very fun to work across from. Luckily we still have Gabe and Anna to talk to. Well, after this covid wave ends. Here is a picture of the four musketeers on Lamma. Good memories and better times.

For Chinese New Year, finally, I also had the same nationalities over. Ying Heng, Jinyoung, and Tomo made delicious food. Behold!

We also celebrated Jinyoung’s birthday.

And in general we wish you happy and prosperous Year of the Tiger!

Other meetups

I also appreciate the other friends that have made my life richer. For instance, shoutout to Maggie and Sasha (🇺🇦️) for being so full of wisdom. Maybe it helped that we were also full of Flammkuchen.

And of course Dasha, Rita and Anna. The occasional hangout has meant a lot to me. I love how we can vent about life to each other. And also enjoy the simple comfort that a blanket brings. #blanketlyfe (It was really cold in Hong Kong the past few weeks.)

Mood pictures

Are you still here? If so, here are just some miscellaneous pictures from the past few months.

I learned a lot. For instance, while the hermetic law of as above so below has a truth to it, the law of as within so without is something I no longer find true in certain contexts. Go ponder on that one reader, and let me know if you want to discuss it further.

Oh yeah, the staycation!

I got these coupons last year from the hotel I did my 21 day quarantine at, and they were expiring. So I did a staycation in a fancy room on the 78th floor in an executive suite. Honestly, I just came for the bathtub and the view, both of which were excellent. Here is the view.

And here is what the view was seeing.

There was also Dongzhi:

A meet-up with a student from Taiwan, Tungle (rhymes with jungle).

A PhD defense, congratulations Tingyu! I should say, Dr. Huang! Here is a picture of an almost complete Language and Development lab.

Fogginess

But what I’m really looking forward to is clear skies.

Wouldn’t that be nice? And it seems the Weather Gods have answered the call. On this 1yearversary the weather is clear and the temperature agreeable. I think a small party is in order. Here’s to whatever the future may bring. 🐯️ 🐯️ 🐯️ 🐯️ 🐯️

Have a nice day and thanks for stopping by!
*tries to look seductively at the camera*

Abstract

Iconic words are supposed to exhibit imitative relationships between their linguistic forms and their referents. Many studies have worked to pinpoint sound-to-meaning correspondences for ideophones from different languages. The correspondence patterns show similarities across languages, but what makes such language-specific correspondences universal, as iconicity claims to be, remains unclear. This could be due to a lack of consensus on how to describe and test the perceptuo-motor affordances that make an iconic word feel imitative to speakers.

We created and analysed a database of 1,860 ideophones across 13 languages, and found that seven articulatory features, physiologically accessible to all spoken language users, pattern according to semantic features of ideophones. Our findings pave the way for future research to utilize articulatory properties as a means to test and explain how iconicity is encoded in spoken language. The perspective taken here fits in with ongoing research of embodiment, motivation, and iconicity research, three major strands of research within Cognitive Linguistics. The results support that there is a degree of unity between the concepts of imitative communication and the spoken forms through cross-domain mappings, which involve physical articulatory movement.

I want to read the paper:

Because it was the Editor’s choice for the issue of Cognitive Linguistics it was published in, the paper has been made open access, free of charge for us and of course, free of charge for readers.

Click here to read the paper.

I want to watch a video instead:

For the 10th installment of the monthly iconicity seminar, creatively entitled IcoSem (hosted by Ian Joo) I presented the main points of our paper. Watch the video here.

Thanks everybody who was at the talk and especially for everyone that engaged in the Q&A session!

I want to look at the data:

Sure enough, here’s the OSF directory with the data.

I just want to cite you guys:

We’ve got you covered for that as well:

Thompson, Arthur Lewis, Thomas Van Hoey & Youngah Do. 2021. Articulatory features of phonemes pattern to iconic meanings: Evidence from cross-linguistic ideophones. Cognitive Linguistics 32(4). 563–608. https://doi.org/10.1515/cog-2020-0055.

Fang Shangyang

Fang’s debut collection Burying the mountain is available since the end of October of 2021, and a copy of it currently finds itself in my shopping cart, patiently awaiting the moment I decide to pull the trigger and press the button which will send it on its way to me. So far, I haven’t but the day is definitely drawing closer.

In any case, you can find more about the poet from his twitter or the collection from the publisher. Currently, that page features him reading some of his work in a video, which I will embed here first.

While there is something in the way he’s reading poetry that I find slightly off-putting, I can’t help but respect how he appears to be visibly captivated by his own writing. Admirable.

Thunder only happens when it’s raining

However, Fang is maybe not why you’re here! You’re perhaps wondering why I am writing about him and his work on this blog. When I saw in the aforementioned interview that he had written a poem entitled 轰隆隆 is the sound of thunder, I knew I had to see what that was all about. After all, I have a greater-than-average interest in Chinese onomatopoeias and ideophones. And I also happen to have written about iconicity in meteorological expressions before!

In my paper “Does the thunder roll? Mandarin Chinese meteorological expressions and their iconicity” (Van Hoey 2018, doi: 10.1163/23526416-00402003) (Contact me if you want to read a copy but you don’t have access), I talk about the two kinds of motivation that we can observe in weather-related ideophones.

A first kind, let’s call them colloquial ideophones, are formally marked in their phonology, mostly through different types of reduplication (full and partial). The way iconicity or sound symbolism works in these words is that when you try to map aspects of the sound on the meaning of the expressions and notice a resemblance. For instance, the back vowels and nasals /oŋ/ in hōnglónglóng perhaps evoke the rumbling (notice the back vowel and nasal in rumbling) of thunder. Or in ablaut ideophones like xīlī huālā or dīdī dādā you get that alternating “over here (high vowel /i,i/) … over there (low vowel /a,a/)” effect. Combined with the fricatives of /x/ and /h/, it’s easy to imagine a kind of sluicing rainfall in xīlī huālā. And then l- reduplication is well-known in Chinese as a marker of continuation. Compare this to the unaspirated alveolar /d/’s of dīdī dādā that indicate drips and drops.

Just so we’re clear, I’m not saying these words are compositional in the way Hamano has argued for Japanese, but conversely that I can see how structural mappings can be identified or imagined.

• hōnglónglóng (轟)隆 隆 ‘sound of thunder’
• xīlī huālā 唏 哩 嘩 啦 ‘streaming and sluicing, heavy rainfall’
• dīdī dādā 滴 滴 答 答 ‘lighter rainfall, just beginning or about to end’

The second kind, literary ideophones by lack of a better term also typically display formal phonological markedness (full or partial reduplication) and also sometimes some phonesthemic-like elements. However, what sets them apart is that they can rely more on radical support. Radical support is the idea that the graphical form of a (Chinese) word contains a semantic element that is expressed, often repeated, that indicates something about the meaning of that word. Most people know that onomatopoeia (sound ideophones) in Mandarin often have a 口 MOUTH element, e.g., jījī zhāzhā 嘰嘰喳喳 ‘sound of birds chirping; sound of people chattering’. But literary ideophones typically have other radicals. In the examples below WATER 氵 and RAIN ⻗. They are also typically older and not as in use in spoken language as the colloquial ones; but of course, they often turn up in song lyrics, sung in karaoke bars. So I wouldn’t flat out say they don’t belong to the contemporary Chinese, just that there is a genre effect.

• línlín 淋淋 ‘streaming and sluicing; dash and splatter’
• línlí 淋漓 ‘light drizzle’
• màimù 霢霂 ‘soaked and steeped’
• fēifēi 霏霏 ‘heavy snowfall or rainfall’
• pīlì 霹靂 ‘thunderclap, thunderbolt’

Hōnglónglóng is the sound of thunder

Anyways, I can go on and on about these words but it’s now time to turn back to Fang Shangyang and his 轰隆隆 is the sound of thunder. I will present the poem below, which, in the spirit of fair use and commentary on (please Shangyang Fang or publishing house, I do this out of love and don’t come back to sue me, I hope you sell many copies), I have augmented with pinyin in red so as to show how to pronounce these words and so that you can enjoy the poem to its fullest.

Also, pay attention to what kind of ideophones are used here. They virtually all belong to the colloquial stratum, which tells you that indeed, we should be reading the poem out loud to make most sense of it.

Finally, I hope the lay-out matches that of the original poem but I can’t be sure. Just know that it’s supposed to have a few lines per paragraph usually.

Okay, enjoy!

轰隆隆 hōnglōnglōng Is the Sound of Thunder

By Shāngyáng Fāng 方商羊

is the rain stitching a bell jar, or the clock remembering its days
as a clepsydra.扑通 pūtōng is the sound of: man stumbling
＆ 哈哈哈哈 hāhāhāhā is the laughter that chases after
玲玲琅琅 línglíng lángláng is the water sword whacking the 哐当锵锵 kuāngdāng qiāng qiāng jade pendants
of cold rocks. 飒飒瑟瑟 sàsà sèsè is the west wind & 扑扑簌簌 pūpū sùsù
is the falling leaves that 萧萧翩翩 xiāoxiāo piānpiān with an autumnmoon

In fact, 簌簌 sùsù is the sound of all soft things falling
The tears 簌簌 sùsù on the face. The snow 簌簌 sùsù.The begonia 簌簌 sùsù
on the mahogany bone of a folding fan. 簌簌 sùsù the brushstroke
paints the rice paper into a landscape of black mountains
泠泠淙淙 línglíng cóngcóng the rivers flow where the inked mountain is not depicted
flow in the uncharted whiteness

The bold strings 嘈嘈 cáo cáo like blizzard, the fine strings 切切 qièqiè like murmur一
the melody of a girl playing 琵琶 pípá on the orchid boat
back in the Tang dynasty which we still hear through 白居易 Bái Jūyì’s poetry

If the river could speak, would the mountain understand its utterance
The sound of a frog is 呱呱呱 guāguāguā, which is annoying
If you cannot stand the 呱呱呱 guāguāguā, please put the frog into a boiler
The hard part: try not to mourn the frog

The sound of the autumnmoon is 寻寻觅觅 xúnxún mìmì,，冷冷清清 lěnglěng qīngqīng
凄凄惨惨戚戚 qīqī cǎncǎn qīqī,，which is also the 点点滴滴 diǎndiǎn dīdī sorrow of 李清照 Lǐ Qīngzhào

Because to mourn the frog one must speak in the language of frog
which is to 呱呱呱 guāguāguā

咿咿呀呀 yīyī yāyā is a青衣 Qīng Yī singing with 盈盈脉脉 yíngyíng mòmò lips of red beans
飘飘冉冉 piāopiāo rǎnrǎn is their furbelow 依依霏霏 yīyī fēifēi among the willows & catkins
式微 shìwēi is the evanescent light of a lantern. Wrong. 式微 shìwēi is the swansdown
evening, a girl calling her beloved: why don’t you come back? Come back

In truth, 式微 shìwēi is 王维 Wáng Wéi writing a poem to the shadow of his leaving friend
who promised to come back with a sprig of magnolia to light up the yard
His poem has the sound of clouds spread low on the moving water

The lamplight threads a dim song across the voice of trees
that can only be heard by insects & birds

关关喈喈雍雍嘤嘤嗷嗷萋萋肃肃呖呖 guānguān jiējiē yōngyōng yīngyīng áo’áo qīqī sùsù lìlì
are the sounds of birds
薨薨喓喓嘒嘒趯趯啁啾了了 hōnghōng yāoyāo huìhuì tìtì zhōujiū liǎoliǎo
are the sounds of insects
噼里啪啦 pīlipālā are the firecrackers bumping into the air
叽里咕噜 jīlīgūlū is a mouse stealing oil & falling from the chair
呦呦 yōuyōu the deer sings.萧萧 xiāoxiāo the horse chases the wind
淅淅沥沥 xīxī lìlì the knuckles of rain knocking windows of a departed train
知了 zhīliǎo : is the cicada chirping its name: 知了知了 zhīliǎo zhīliǎo
布谷 bùgǔ is the cuckoo calling itself: 布谷 bùgǔ cuckoo cuckoo 布谷 bùgǔ

For us, the petals fall with no sound. For the fire ants, the petals hit
the ground with a sound as loud as thunder, which is 轰隆隆 hōnglōnglōng

But what is the ant’s onomatopoeia for the thunder & falling flowers
What is the petal’s onomatopoeia for the wind that brings forth its own
falling
Still, we keep hearing, hearing, hearing the hiss of white, yellow
& blue petals scattered in a bowl bitten by the ants’ bitter teeth

沥 lì 淅沥沥 xīlìlì the rain
轰隆隆 hōnglōnglōng 轰隆隆 hōnglōnglōng 轰隆隆 hōnglōnglōng轰隆隆 hōnglōnglōng
the thunder

Outro

Pretty cool huh?!

Buy his book if you want to support him or read all of his poems.

Tidymodels: first thoughts

• Just like with the tidyverse, I love the modularity and relatively clear sequence of steps that can guide you through “any” analysis. Although, we’re not quite there yet in terms of straightforward applicability.
• I love the tutorials and the “tidytuesday” series on YouTube. Those case studies really provide excellent showcases for how to do common things.
• The idea of making the interfaces to different models more uniform so you can be more efficient.

Let’s talk metaphor for a second:

• The conceptual metaphor of DATA IS MONEY, expressed in phrasings like You need to spend your data budget wisely are helpful. I generally don’t see studies in linguistics that split the data first in a training and testing data set, although we probably should be doing that.
• The conceptual metaphor of PREPPING DATA IS FOLLOWING A RECIPE (the recipes package ) is highly amusing and makes sense.
• The evocative names of the other packages in the core tidymodels set are good too: rsample for resampling, parsnip for modeling (this is a parsnip, in Dutch it’s called pastinaak), workflows for defining workflows, tune for tuning, yardstick for measing performance of your models, broom for cleaning the output of models…

What I don’t super like about tidymodels, is that as of yet, sometimes it’s very hard to see what some tidymodels equivalents to super common modeling operations are.

• For instance, I don’t really know how to do an Multiple Correspondence Analysis (MCA) within tidymodels, even though the Principle Components Analysis (PCA) is very well represented in the tutorials and examples. I’ve tried some recipes::step_dummy() to get to similar results as I would have gotten with FactoMineR or ca, but I find the results not similar enough to surrender my dimension reductions for qualitative variables completely to tidymodels. (Julia Silge and friends, if you are reading this, please figure this out for me.)
• I kind of hate that most analyses just stop at the “Oh I collected the metrics, time to turn off my computer” moment. I don’t think the analysis is finished after calculating a model’s performance; this interpretation is often lacking, and I don’t know if that’s a general quirk of data scientists or just specific to the showcases of these packages, but I wish it didn’t stop there.
• Finally, we come to the topic of this post: how do I decide whether or not to keep interaction effects. In “normal” modeling, this is quite straightforward:

Step 1. Make a model with multiple predictors, no interaction.

Step 2. Make another model with the supposed interaction.

Step 3. anova(model1, model2).
If the p value is significant, 
the more complex model (the one with the interaction) is what you want.

But what about the tidymodels approach to anovas?

Googling “tidymodels anova” will bring you to a lot of pages. Unfortunately, they are not super useful in answering our question:

RQ. How do I do a simple comparison of models to decide if the interaction is valid?

This page takes you to a broom function; this one to the excellent Tidy modeling with R book, more specifically, the chapter on comparing models with resampling. And I agree, resampling seems a very good way to compare models, and I’ve tried to apply that that page to the case study I will present below (nothing too crazy, just the iris dataset), but if I follow the guides there, I end up with no difference between the model with and without interactions, which is not what our anova will say. So yeah. 🤷️

Let’s load packages

library(tidymodels)   # general tidymodels packages
library(skimr)        # fancy way of inspecting data, not necessary
library(ggplot2)

Let’s look at the data

iris %>% head()

  Sepal.Length Sepal.Width Petal.Length Petal.Width Species
1          5.1         3.5          1.4         0.2  setosa
2          4.9         3.0          1.4         0.2  setosa
3          4.7         3.2          1.3         0.2  setosa
4          4.6         3.1          1.5         0.2  setosa
5          5.0         3.6          1.4         0.2  setosa
6          5.4         3.9          1.7         0.4  setosa

skimr::skim(iris)

What I will be investigating, is if Petal.Length can be predicted by Petal.Width and the Species. But we want to know if there is an interaction between Petal.Width and Species. In other words, our main model will look something like this:

Petal.Length ~ Petal.Width + (or *) Species

Let’s plot the data:

iris %>% 
  ggplot(aes(Petal.Width, Petal.Length)) +
  geom_point(aes(color = Species)) +
  geom_smooth(method = lm, color = "orange",
              formula = 'y~x') + 
  geom_smooth(method = lm, aes(color = Species),
              formula = 'y~x') +
  theme_minimal()

We can see three nice groups of Species. Our general linear model smooth (orange) doesn’t seem too bad, but it doesn’t take into account that there may be an interaction between the Species and the Petal.Width, as evidenced by the three different slopes per Species.

The “normal” way of testing this

The code will be quite short:

mod1 <- lm(Petal.Length ~ Petal.Width * Species, data = iris)
# summary(mod1)

mod2 <- lm(Petal.Length ~ Petal.Width + Species, data = iris)
# summary(mod2)

anova(mod1, mod2) %>% tidy()

# A tibble: 2 × 7
  term                          df.residual   rss    df sumsq statistic  p.value
  <chr>                               <dbl> <dbl> <dbl> <dbl>     <dbl>    <dbl>
1 Petal.Length ~ Petal.Width *…         144  18.8    NA NA        NA    NA      
2 Petal.Length ~ Petal.Width +…         146  20.8    -2 -2.02      7.72  6.53e-4

We see a very small p value (0.00065) so we know that there is a difference between mod1 (no interaction) and mod2 (with interaction), so we have to choose the one with interaction.

Done! Or not?

Tidymodels

Tidymodels is a lot more verbose than just these three lines of code, but that’s actually a good thing: you can are more conscious of the steps involved, more explicit and can easily add more models. However, here it will feel a bit redundant, but bear with me.

set.seed(1234)
iris_split <- initial_split(iris, strata = Species, prop = 0.8)
iris_train <- training(iris_split)
iris_test  <- testing(iris_split)

# the model without interaction
rec_normal <- 
  recipe(Petal.Length ~ Petal.Width + Species, 
         data = iris_train) %>%  # we train on the training set
  step_dummy(all_nominal_predictors()) %>% 
  step_center(all_numeric_predictors())

# the model with interaction
rec_interaction <-
  rec_normal %>% 
  step_interact(~ Petal.Width:starts_with("Species"))

iris_model <-
  linear_reg() %>% 
  set_engine("lm") %>%     # if you want different engines, this is where you would do that
  set_mode("regression")

# normal workflow
iris_wf <-
  workflow() %>% 
  add_model(iris_model) %>% 
  add_recipe(rec_normal)

# interaction workflow
iris_wf_interaction <-
  iris_wf %>% 
  update_recipe(rec_interaction)

iris_normal_lf <-
  last_fit(iris_wf, 
           split = iris_split)

iris_inter_lf <-
  last_fit(iris_wf_interaction, 
           split = iris_split)

normalmodel <- iris_normal_lf %>% extract_fit_engine()
intermodel  <- iris_inter_lf %>% extract_fit_engine()

anova(normalmodel, intermodel) %>% tidy()

# A tibble: 2 × 7
  term                          df.residual   rss    df sumsq statistic  p.value
  <chr>                               <dbl> <dbl> <dbl> <dbl>     <dbl>    <dbl>
1 ..y ~ Petal.Width + Species_…         116  17.8    NA NA        NA    NA      
2 ..y ~ Petal.Width + Species_…         114  16.1     2  1.72      6.10  0.00304

iris_inter_lf %>% collect_metrics()

# A tibble: 2 × 4
  .metric .estimator .estimate .config             
  <chr>   <chr>          <dbl> <chr>               
1 rmse    standard       0.318 Preprocessor1_Model1
2 rsq     standard       0.968 Preprocessor1_Model1

mod2 %>% glance()

# A tibble: 1 × 12
  r.squared adj.r.squared sigma statistic  p.value    df logLik   AIC   BIC
      <dbl>         <dbl> <dbl>     <dbl>    <dbl> <dbl>  <dbl> <dbl> <dbl>
1     0.955         0.954 0.378     1036. 3.70e-98     3  -64.8  140.  155.
# ℹ 3 more variables: deviance <dbl>, df.residual <int>, nobs <int>

intermodel

Call:
stats::lm(formula = ..y ~ ., data = data)

Coefficients:
                     (Intercept)                       Petal.Width  
                          3.7894                            1.0309  
              Species_versicolor                 Species_virginica  
                          1.8314                            3.0006  
Petal.Width_x_Species_versicolor   Petal.Width_x_Species_virginica  
                          1.0536                           -0.2426  

intermodel %>% tidy()

# A tibble: 6 × 5
  term                             estimate std.error statistic  p.value
  <chr>                               <dbl>     <dbl>     <dbl>    <dbl>
1 (Intercept)                         3.79      0.197    19.3   1.11e-37
2 Petal.Width                         1.03      0.230     4.49  1.73e- 5
3 Species_versicolor                  1.83      0.557     3.29  1.34e- 3
4 Species_virginica                   3.00      0.585     5.13  1.22e- 6
5 Petal.Width_x_Species_versicolor    1.05      0.652     1.62  1.09e- 1
6 Petal.Width_x_Species_virginica    -0.243     0.621    -0.391 6.97e- 1

intermodel %>% glance()

# A tibble: 1 × 12
  r.squared adj.r.squared sigma statistic  p.value    df logLik   AIC   BIC
      <dbl>         <dbl> <dbl>     <dbl>    <dbl> <dbl>  <dbl> <dbl> <dbl>
1     0.958         0.956 0.375      516. 1.55e-76     5  -49.6  113.  133.
# ℹ 3 more variables: deviance <dbl>, df.residual <int>, nobs <int>

iris_inter_lf %>% 
  collect_predictions() %>% 
  ggplot(aes(.pred, Petal.Length)) +
  geom_point() +
  geom_abline(intercept = 0, slope = 1, color = "orange") +
  labs(x = "Predicted Petal.Length",
       y = "Observed Petal.Length",
       title = "R² plot") +
  theme_minimal()

Disclaimer on the iris data set

In recent years, it has become more public knowledge that the ubiquitous iris data set was first published in the Annals of Eugenics in 1936 by Ronald Fisher. As this tweet and this post point out, it’s perhaps not the best thing that this data set is so readily used in data examples. Some proposals for other data sets can be found here.

Obviously, eugenics is bad – think of how important this issue was to the projected future in Star Trek with Khan and friends – and I agree that iris is kind of boring. But, just like the Annals of Eugenics rebranded itself to the Annals of Human Genetics, distancing themselves from that terrible phase in science (and we still see the beast rear its head once in a while), I can’t help but think that this silly description of flowers is quite innocent. Perhaps, death of the scholar does exist? After all, if we have to throw away iris because of Fisher’s bad personal views (once again, not good), do we also have to throw out the stats techniques he developed? I know that I’ve used the Fisher-Yates Exact Test to calculate mutual attraction.

And let’s also talk about Karl Pearson, who was the first editor of the Annals of Eugenics. Reading up on his wiki bio was not pleasant either. So should we throw out the Pearson correlation? Or even worse: the p-value (which was first formally introduced in the Pearson’s chi-square test)? Gone with Principled Components Analysis! Histograms? History you mean!

The point is that it is necessary to treat the data and work as separate from their personal life. That means that I agree with the efforts to rename buildings that were named after Fisher or Pearson at UCL, but that at the same time we should still be okay with using iris or statistical techniques developed by these people. The best two arguments for not using iris are that it’s boring and that there exists a penguins dataset (found here).

Food and coffee

Since we basically had to cram a year of not seeing each other in 43 days, we had a lot of good food and coffee. Here are pictures (insert thousands of words).

Places we visited

Of course, we also went places.

Opulence, you own everything

First up is the Monopoly exhibition on the Peak. I had always suspected it was a tourist trap and my suspicions were confirmed. But I’m happy CJ was willing to indulge in my wish to go there. It was expensive, but they had a fun money grabbing game in the installation which kinda made it worth it. Also the history of the game by itself. If anybody ever wants to play Monopoly with me, I’m down~

DB

We also went to Discovery Bay, which is probably best described as Club Med in Hong Kong or Center Parcs HK. It’s where CJ grew up and for kids that looks like a super safe and nice community. We visited his old primary school and also the house he lived in the longest there. I love doing house tours. Anyways, pictured here are the central plaza with the Pacific Coffee CJ used to get LAN internet from and the DB beach.

Big Buddha

One day, we also went to the Big Buddha. It was not the firs time I had been there, but it would be the first time taking the cable car – last time we had taken a bus. The first picture is us in our cable car. In the second one you can spot him on the left.

But to our astonishment, the Buddha was under construction! Luckily it did not dampen our fun. We took pictures with the free-roaming cows and statues of the 12 arhats, and still climbed up to his bodhiness.

There was also the Po Lin monastery which was pretty beautiful (and I have seen many Chinese temples so I have a frame of reference). Adding to the athmosphere was the praying of the monks that could be heard from outside of their hall which we weren’t allowed to enter.

Chi Lin Nunnery

Chi Lin Nunnery and the adjacent Nanlian Garden, which I have visited a few times before (just look at previous blogs), also had the honour of being frequented by us. And we had the honour of witnessing the golden pavilion.

The nunnery itself is also nice; it’s a kind of fake old temple which is an acquired aesthetic.

I have friends

One of the shows we watched together was Crazy Ex-Girlfriend. Well, I had watched seasons 1-3 and we watched season 4 together. One of the best things in that show is the crazy amount of fun songs the cast have given us. Here is I have friends.

Friends… songs… that brings us to the great karaoke moments we got to spend with CJ high school friends!

But of course we also hung out away from the microphone.

Yes, they have a Ponyta in their living room. It’s a thing.

Lamma Lyfe

Some final moments of our life on Lamma. First there was the ghost month. People will leave out offerings to the ghosts in exchange for luck. But, as my hair stylist warned me, it’s best not to touch the offerings or even sing while going up during ghost month. Very dangerous.

I remember in Taipei they also stayed away from bodies of water, lest the ghosts would snatch you in and turn you into a ghost. Ghastly.

We also saw our first snake on Lamma! A little copperhead race.

For an island that has a snake path, which I often tread on, I can’t help it but be disappointed that so far I’ve only seen this little copperhead racer.

WHERE ARE THE PYTHONS?

WHERE ARE THE COBRAS?

WHERE ARE THE BAMBOO PIT VIPERS?

Sshhhhttt

Soothing moon-lit views of the island to calm you down Thomas.

Cute pictures of CJ with an oversized dog that should not be living a tropical clime to calm you down Thomas.

Tranquil sunset views to calm you down Thomas.

Summer and its subsequent separation

From my facebook message:

I’ve learnt in the past few years that after wonderful summers together, the next season is always heralded by separation September. It’s weird how it doesn’t really matter where but just being with you somehow always feels natural. 💕️

This was the caption for the picture that we took right before CJ went in for customs and to catch his flight. Once again, we were separated by barriers I couldn’t pass.

I can only reiterate that, in the words of Vance Joy, whose music you really need to be in the mood for because it’s basically all variations of the same song but that one song is so good, in this iteration, you’re the missing puzzle piece, and I can’t wait until we’re back together again.

Because when you’re in a room with me that missing piece is found. You know when you’re by my side darling nothing can bring us down.

CJ & Thomas out!

Turning 30

There’s no way around it; this year I got my ticket to get on tram 3, the way we colloquially talk about reaching this milestone. I know that so many people did not get to spend their birthdays the way they wanted in the past year or so, let alone their 30th. So I’ve been quite fortunate that neither last year nor this year a lockdown was in effect at the moment of reaching that age. Yet, I still had a number of thoughts surrounding that age (huh, but age is just a number?! yeah right…).

I’m proud of what I’ve been able to accomplish so far in life, given the chances afforded to me. But in Chinese there is this Confucian saying sānshí ér lì 三十而立 “At thirty, I stood firm”, meaning that you’re independent and can take care of yourself and your situation at the age of thirty. And I guess that’s kind of true in my situation, but it just didn’t feel like that.

On the contrary, I felt quite uprooted: in a new place, no strong network of friends. One might even say, I felt like I was “only thirty” (sānshí ér yǐ 三十而已), or maybe just felt like a new kid on the block (sānshí ér yǐ 三十兒立). In any case, due to circumstances, I spent the b-day alone and ended up just sitting on the beach (sānshí ér zuò 三十而坐) wallowed in existential despair. Here are two pictures of that beach that night:

I truly longed for the warmth of that campfire and those people that night, but could not bring myself to go over and join their party.

But I was more than pleasantly surprised that I actually have more friends than I thought here. Throughout the following two weeks or so, I met up with multiple people, and they really pulled me out of my sullen slump. And I am so grateful for that. In this time of covid, we need live interactions more than ever.

So thanks Darryl and family for the birthday cake.

And thanks Ka, Mona, Kayi and Matthew for the dinner and the never ending supply of mussels.

And thanks Youngah for the lunch. And thanks Marcelo and Arthur for the dinner. And thanks to all the friends I’ve made or reconnected with in the past few months: Ian, Gabor, Szeto, Jinyoung, Pavel, Hengheng, Jonah, Tomo and so on and so on. You’ve really given me the sense of growing some roots here, helping me ground.

It has taken me a while but I’m in a better space now, and I am also extremely stoked that CJ is nearing the end of his quarantine. So August is going to be awesome!

Walking in the mountains

One of the things I have been enjoying since moving to Hong Kong is leisurely strolling around my place and the place I work at. I don’t think I’m a full-on hiker, honestly, I come from Belgium and every slope is like a mountain to me okay. But I do enjoy some nice brisk walks.

So here I found myself walking up to Victoria Peak. (I don’t like the work hike in English, because in my mental vision that involves full on hiking gear with ropes and pickaxes and what have you. I don’t think we even have the meaning of ‘leisurely walking amidst the mountains’ in the Dutch equivalent bergbeklimmen ‘mountain-climbing’. Idk. Random thoughts.)

The view from the Peak was quite nice and a good reward for the upwards trek.

Along the way I encountered Pinewood battery, the remnants of an air defense (built by the British). At first I was heaving and huffing from the climb, but then I regained my composture and reflected a bit about the moment. Much later I wrote a poem about it (I was a bit bored on the ferry ha).

Pinewood Battery

Covered by leaves, I hiked the Peak and reached a verdant clearing, pinetrees all around a central circle — batt’ries that were breached when Eastern foes took flight to higher ground.

The remnants ruins of that air defense once used to shoot down planes from azure sky. They witnessed fragrant bloodshed, times were tense, yet tranquil on a tourist trail they lie.

A ghastly whisper beckoned, contemplate the lives cut off too soon for Empires lost. Who mourns here now the dead, their restless fate, except for hikers; was it worth the cost?

This all in mind I ask two aunties dear, please take a picture and make sure it’s clear.

I’ve been also been walking around Lamma. Here are some pics from my walk to the windmill.

Oh, and did I tell you about the sunsets. So many beautiful sunsets. It’s truly incredible.

And it’s not just me who is slightly obsessed with sunsets. Every day I see lots of Hong Kongers take pictures of it, enjoying the daily dying of the light. It’s almost as if if nobody caught it on film, we couldn’t be sure that the next day the sun would make the effert to come back and shine upon us all.

The moons I’ve witnessed on Lamma are also nice. It’s almost as if Lamma is a manifestation of Mount Penglai or maybe even Avalon. The latter is more probable, as there are many British elements present there (but no apple trees).

I envision myself writing a poem or two about the place that is Lamma in the future. But here is a poem that I thought of when coming across this renovation site in my village (not written by me but the wonderful Doreen Valiente).

The Redevelopment.

The old house vanished in a dusty cloud, its secrets scattered to the winds of air. Whether its dwellers humble were or proud, they are forgotten, and its walls stand bare.

Soon the last fragment’s gone. The raw earth gapes. The loud-voiced, sun-bronzed builders have moved in. Now by their hands arise new-fangled shapes, new bricks, fresh concrete, motors roaring din.

A new room grows, where once in the same place those others dwelt, who lived and loved and died. Who came there young, grew old, perhaps found grace with new hope now who comes here to reside?

So turns the wheel of time; the cycle runs for houses, years and galaxies of suns.

This poem came to me the instant I saw that ruin. I linked it to my own uprooted state that had slowly been trying to undo that uprootedness. And I feel like I succeeded through a budding network of friends and colleagues, but also by becoming more independent. In a way I was able to undergo some internal redevelopment. What helped me was seeking out the beauty expressed by other people. In other words, I went to artsy places, trying to “art” myself – this is play on words, to ground oneself in Dutch is aarden. Naturally, grounding onself through art is then to art oneself.

Arting myself

I kept bumping into artsy spaces, so I took it as some kind of sign that I should do more of that. For instance, I decided on a whim to go to Tai Kwun, and they had this really cool exhibition about the moon (also a Taschen bookstore, so nice books).

Then I also happened to get free tickets to Art Central. Here are some really nice paintings.

Then there were also some triptychs with contemporary elements, painted by Japanese artists, which were very interesting.

Then some manga/anime art:

And some funny things:

Some cool things:

Even the performance art was interesting.

And here I am with some art by Lousy Lousy.

The only thing I did not like at all was this artwork by an artist who took pie charts to a whole new level. Like, him choosing that to be interesting shows that he hasn’t read up on how bad it is to use pie charts to compare groups. Do not do it people.

A while later, for French may, which weirdly comprises May and June, I went to the Hong Kong Museum of Art, which was hosting an exhibition Mythologies: Surrealism and Beyond. Here are pictures of famous works:

But there were also some other really nice exhibitions.

And lastly, in the category Arting myself, I rediscovered some art made in highschool:

Time to sail away

So there you have it. I’m doing well, and increasingly so. The communities I’m lucky to be(come) a part of are becoming more solidified, both professionally and personally – although this update has decidedly focused more on the personal part.

Now it’s time to take my junk and sail away into the sunset. Cheers.

WTF is PsychoPy

PsychoPy (Peirce et al. 2019 doi: 10.3758/s13428-018-01193-y) is an application for the creation of experiments in behavioral science (psychology, neuroscience, linguistics, etc.) with precise spatial control and timing of stimuli. There are two ways of creating an experiment: either by writing raw python code, or by using the Builder to get a more GUI experience (to which some code can be appended).

For someone with a more corpus-oriented background (le moi), it was looking like a daunting task to create this kind of experiments for which it was important that the reaction times etc. were controlled for. Note that other software like Qualtrics, or even Google Forms, is still great for surveys for which such control is not super relevant; but it makes sense to think that in psycholinguistics we ideally would want to capture such information in a realiable manner. There’s also a budding psycholinguistic community that is based on the Shiny framework for R, like ShinyPsych, and I know the aforementioned Bonnie is also scripting experiments based on R, although I don’t know if its ShinyPsych.

So, because PsychoPy is supposedly easy to use, open source, and because it can also be deployed on its linked platform Pavlovia to actually run the experiments, it seemed like a good choice to base our experimental work on, especially in times of Covid. However, for the preparation of stimuli and the post-hoc analysis, I still turn to R.

Builder

This is what a general PsychoPy experiment in the Builder view looks like.

In the lower box, you can see the different “routines” that make up the experiment. For instance, here you have the instructions ‘instructPractice’, followed by the ‘trial’ and ‘feedback’, which are repeated in a loop. Then you have some new instructions ‘instruct’, followed by the “real experiment” ‘trial’ and a loop again. Finally, a few words of thanks for participating.

In the upper box, you have the uniquely named components of a routine, in this case ‘trial’. You can see that there is a fixation cross present. At some point, a word stimulus will appear. People will have to press a key to respond and then the experiment will go to the next trial in the loop.

So the general buildup is pretty explanatory, and you find heaps of videos on youtube that show the basic anatomy. On of my favorites is the following, on creating a word based reaction time study (watch at your own pace, or as the Japanese say mai peesu de マイペースで):

Following along with such instructional videos can get you up and running pretty fast, although it felt like a struggle because I had to start from scratch. Below I’ll list some things that were not super obvious to me in the beginning. But first, deploying the experiment online.

Deploying and running on Pavlovia

One of the selling points of PsychoPy is the supposed seamless integration with a platform called Pavlovia. You can upload your experiment to this platform, pilot the experiment and finally run it. You do have to pay for running it, but the credits are not very expensive, and you only have to pay for the experiments that are completed; so not for participants that somewhere along the way leave.

Since it was not super intuitive of how to actually deploy the experiment, I once again turned to youtube. Luckily, there were some good demo videos for this aspect as well. My favorite, which helped me the most, is this series by a certain Jordan Gallant:

• Intro
• Pavlovia account
• Uploading an experiment
• (Sharing an experiment)
• (Basic git commands)

And now after running the first experiment, I can see say that it works and is accurate enough.

Community

Another aspect that I haven’t touched on yet, is the community. Most people familiar with R or python or what have you know that this aspect is actually quite important. From my mentioning of youtube you can already gauge that there is some vibrancy concerning PsychoPy, but it’s not until you head to https://discourse.psychopy.org/ that you can really find out how thriving it actually is. People will ask questions about errors they face, much like stack overflow, and some of the issues I encountered were indeed answered there (yet not all, spoiler). But so this makes for another compelling reason of why PsychoPy may be a good vehicle for your experiments.

Not very obvious aspects

Here is a series of issues I faced thus far.

Other handy resources

Luckily the community is quite good, so there are some documents that help you set up the experiment and “translate” it to Pavlovia.

Two must-reads are:

• the 2020 cribsheet
• the updated 2021 cribsheet

Conclusion

So all in all I’m quite happy we are using PsychoPy. Having become familiar with the basic functions has enabled me to set up experiments, in such a way that analysis afterwards is clear and reproducible. There is a learning curve, but there is also a supportive community to help out.

Flying from BE to HK

As I detailed in this post, I spent the latter half and the first month of 2021 or so in Belgium, back home, and loved it. But I was also getting excited for moving to Hong Kong for my new job at HKU. This was not an easy getaway. Not only did I have to apply for the work visa and all other necessary documents; but in times of covid I would also need to have tests taken at the right time within the right window to even be allowed to use the plane ticket that I also paid for.

So there I was, within my 72 hour window before my flight, lining up outside, between the cars in the drive-through covid test site in Brasschaat. The cold made the depth reached by the unnecessary long stick up my nose hurt even worse and I felt a phantom pain as I was making my way back.

Fast forward two days. Hooray, a negative test, now off to Brussels Airport, where I had to take a fast test to enter the Netherlands, which was unnecessary and stupid, but required. 40 minutes later this also gave me a negative test result and I could finally board the plane from Brussels to Amsterdam. Going up. Going down.

Then I had to wait for quite a while and finally got ready to board. It didn’t look like it was going to be a full flight, but maybe a lot of people were late? Anyway, we had to show yet once again all the million test results and right documents, so we could board the plane.

As it turned out, there were only four passengers, and 10 crew members. Unfortunately, one of the four of us had decided to pay for an upgrade to business class while right before boarding and that meant that he got to enjoy all the perks that came with that, while the remaining three of us had to stay in economy. Later, on of the flight attendants told me that if he hadn’t done that, we would all have been upgraded for free. Yet, all things considered, it was a joyous flight. Here is a picture of me expressing my joy.

When we arrived at the airport, it felt like we had just entered a ghost town. A long guided trail led us from station to station. We had to show our test results, show that we had the necessary documents to enter, show that our tests were not just photoshopped, show this, show that …and finally got tested again (within my 72-hour window!!). Then we were sat at tables to await the test results and fitted with fashionable wristbands that we could not take off for the following 21 days of quarantine.

Caged bird

After being fitted with the trendy covid fitbit, I was taken to the hotel I had booked for 21 consecutive nights. I chose L’Hotel Island South – now rebranded as the Nina Hotel Island South – and honestly, it was a good experience, given the circumstances. People following my instagram (@simazhi y’all) could see daily updates on the food + ratings. Here’s a small selection.

My room had two beds, but luckily also wifi, a bathtub, tv, a minifridge and a nice city view.

Oh yeah, I should probably also mention that by this point it had been 2.5 months since I had had a haircut, due to the restrictions in Belgium and also quarantine. So expect to see some wild hairstyles in this post.

I was surprised a number of times during my stay at the hotel, two of which had to do with the wonderful care packages from my current boss Dr. Youngah Do. Words cannot express the happiness I felt at receiving this.

Emotionally, though, the lack of real world contact was quite hard. It’s one thing to see other people on a computer screen and interact with them, but I was so happy every time I caught a rare glimpse of a hotel staff member delivering food. And the tests on day 17 and 20 were also kind of enjoyable, since I got to interact with real humans, who talked back to me, in real life and proximity. There was also a fun little gimmick in the hotel: a delivery from a robot.

I mean, the robot delivery was a fun activity but real interaction is still best. And it didn’t help that I could see people from my window, all the way down (I was on the 23th floor) in the distance. I tried to still my feelings about this aspect of life that had suddenly been taken away from me by not focusing on the world outside, even though the weather was so good those days. My cage was a physical containment, but the mind can wander and I developed a meditation practice, but also watched like half of Netflix.

Bird set free

Finally, the day came that it was my final day. When the clock struck 00:00, I did not lose my glass shoe but instead took off my covid fitbit tracker, deleted the Stay Home Safe app, and felt a huge weight lifte off my back.

Yet, the next day, as I prepared my luggage for leaving, I felt a bit 捨不得 ‘reluctant to say goodbye’. I once again took pictures of every corner of my room, and to this day still think of the gingo leaf painting that was my only fixed view for 21 days. But then it was time to prepare the bags and take a picture with the hotel once I was out.

Outside, I was being picked up by my friend and colleague Dr. Arthur Thompson, who kindly offered me a place to sleep for the unforeseeable future, while I looked for a place to live. So we went to eat ShakeShack (first post-quarantine food) and then took the ferry to Lamma island, and then took a sunset pic on the beach.

Don’t worry, it’s not a nuclear power plant.

Nesting at HKU

The day after my release on the 28th of February, we went to the University of Hong Kong, where I got to meet my new boss in real life again. It was very joyful and bound to lead to a good cooperation. Here is a picture of the three of us in our small iconicity and ideophones team. As well as a view from the grad students office on the 9th floor where I would take a seat for a while.

I’m planning to periodically write about the research that we’re doing so stay tuned for that.

I got a haircut at some point, which was perfect to take sunset pictures with.

Migratory bird

The homefront has told me they want to see more pictures so in this section I’m just showing a few places I went to visit in these first two months. Perhaps it is also useful to link to my first visit to Hong Kong in 2013. If you can read Dutch then this is really a gem of a blogpost. It also lays the foundation my tourist impression of Hong Kong and explains why I visited Chi Lin Nunnery 志蓮淨苑 and the adjoining Nanlian Garden 南蓮園池 within the first few weeks of being here.

For my travel companions and myself in 2013, this garden really was stunning and a well-deserved cover of the contemporary Lonely Planet guide. I find that strolling in the garden really enables me to put my mind at ease. So I’ve returned to it with CJ and Arthur in subsequent visits.

Let me know if you want to see more pictures of this idyllic place on earth. But Hong Kong is not only Chi Lin Nunnery, lol; there’s also lots of temples. For instance, there is the Man Mo Temple 文武廟, dedicated to the god of writing and literacy Man Tai 文帝 and the marial Mo Tai 武帝. Fun fact, tai in Cantonese, or dì in Mandarin means ‘emperor’ as well and there are lots of emperors throughout history who posthumously are referred to as Wéndì 文帝 and Wǔdì 武帝.

When I first visited the Man Mo in 2013 I was a bit disappointed, not a lot was going on and the nice spiral incense cones we were promised in the guidebook weren’t lit at all. Things were different this time around. The day I happened to visit the temple it was super busy, and I also offered some incense to the deer statue. Deer are an interesting animal and I’ve come to appreciate them increasingly in the past few years, for instance on my trip to Japan in 2019 (for which I did not write a blogpost??).

Also, here I am with the Choi Hung “Rainbow” station!

Bird-watching

I also love going to see some art (mostly painting or sculpture). And discovering that HKU has a university museum with exhibitions was kind of nice. I got to see Ng Lung Wai’s Painting across the threshold, where he uses a folding paint technique. Very cool effect.

There was also an exhibition on Congolese artists, “Colours of Congo: Patterns, Symbols and Narratives in 20th-Century Congolese Paintings”, which was very beautiful and interesting. I learned a lot from the selection of paintings they had, and the pointillism style really appealed to me.

If you’re in Hong Kong, do go check these exhibtions out!

Roosting on Lamma

It became clear quite early on that I was interested in settling on Lamma Island. And even though this island has snakes, bugs and wild boar, it’s also quite vibrant and fun to escape the city on a daily basis. I still love the ferry commute, although it’s quite a journey to go to work, but I love the view of seeing Hong Kong in all its vertical concrete glory rise up as I approach it.

So I hope this was a fun read for you. More adventures in the future and stay tuned for April and May in June! Baibai from the rooftop of the IFC, looking at Kowloon!