So as I haven’t been able to speak the past short while, one thing I have noticed is that it is harder to communicate with others. I know what you are thinking: “Wow, who could have possibly guessed? It’s harder to converse when you can’t speak?”. Indeed, I didn’t expect it either.

But how much harder is it to communicate?

One proxy you can use is the classic typing metric, words per minute (wpm). So I spend some time looking at various forms of communication and how they differ between one another.

For most tests, i used https://www.typingtom.com/english/typing-test/30s

So I list below the forms of communication I have tried and how slow they are.

Here are the rough tiers that I found:

Ultra-slow-speed tier

(~10-20wpm) Shaping out non-standard letters with my hands

This is obviously the worst method of communication. Most people don’t know sign language, but can pretty intuitively learn how to infer most-but-not-all letters without needing to use a table. With people I have spend more time with, they have managed to learn it moderately well, but probably they should just learn sign language.

Then with some words, even with the word spelled out, people sometimes struggle to understand the written word spelled out in letters to translate that into the normal way they understand words.

That being said, sometimes people can use their context skills to infer what is wanted by just the first letter or two, so it’s not completely useless. And often it can be easiest since no materials are needed.

Pretty-slow tier

(~40wpm) Drawing on a whiteboard

(~40wpm) Typing with one hand on my phone

(~45wmp) Typing on my laptop with one hand

I find it slightly surprising how close these end up converging together.

For the most part. Writing on a whiteboard has the added benefit of being much easier to share in some contexts, while writing on a device has the benefit of being able to use Text-To-Speech (TTS). But I find both kinda inadequate in their own ways.

(But you see, there aren’t that many situations where typing with one hand comes up, so perhaps I just haven’t had that much practice with it? unclear)

Respectable tier

(~60-70wpm) Typing on my phone with two hands

(~80-90wpm) Typing on my laptop

Yeah I was somewhat surprised when typing on my phone with two hands, that it was not actually as much slower than typing on my laptop is. However, I guess this doesn’t factor into account that when typing on my phone, I might be outside in the cold or rain and simultaneously trying to walk, which combine to make typing on the phone feel much worse.

And yeah, I do wish I was faster at typing on my laptop, but I guess I never got around to it. But it makes sense that using two hands you get roughly double speed than you do with one hand.

Actually-fast tier

(~180-200) speaking at a normal pace

I asked a few people to do a speaking speed test at a comfortable talking speed when reading, and found that it is much faster than typing by a significant margin, about double again. And this is effortless.

Speech also includes tone-of-voice and such, in a way that is only implicitly captured when typing and using a real-time TTS model. (my partner still sometimes doesn’t quite decouple that the tone of the “OK” on the TTS is not the tone with which I actually mean it).

Very-fast tier

(~260-340wpm) Speaking at a rushed pace

I then subjected my same friends to the torture of reading the same passage as fast as they could. And they managed to achieve another ~1.5x in speed compared to normal speaking speed. It goes to show how language is quite optimized for speaking.

What have we learned?

One update from doing all of this, is “wow, maybe when I get my voice back, I should just consider improving my Speech-to-Text game” (~10h maybe?), since the input is just so much faster than typing. (2-4x faster!). I used to be a big STT hater, so this is a moderately big update for me.

Some notes though:

One thing, is that the wpm of most of the methods are slightly higher than one might expect based on the naive number. When I do end up typing some sentence out, people can often infer what I am trying to say before I am finished typing. (I usually do end up still typing out the whole sentence anyway though). So one could potentially optimize for this somehow.

Another note, is that when speaking, I very rarely make verbal typos, and when I do, they are quite phonetically similar. When typing however, I make typos more often typographically similar, but when they are passed to a TTS model, the result is often catastrophic and illegible to people who want to understand what I just said.

This list also excludes some possible communication methods that I did not put in the effort to learn. ASL can reach speeds comparable to speaking if you learn all the vocab fluently. If one spends a year or two learning stenography, one can achieve 200-300wpm by typing as well. But I never learned either of these.

Overall, I remain bullish on speaking, more than ever, so I will try see what I can do in the future with this information.