Some languages have context hungry words. Let me define a context hungry word as a word that comes to life, that gains its meaning, that feeds on context.
One example of a context hungry word, I got from a lecture from C. Thi Nguyen on the gamification of public discourse, and why that’s bad for our values. (https://www.youtube.com/watch?v=1LpbGW3qLVg) In this lecture, he mentions the hide as a unit of measure meaning the amount of land needed to feed a family. In England where he gave his lecture, that could mean 20 to 25 hectares of land. In the nutrient rich soil of Central California, that could mean a 1/4 or a 1/5th of that. In the futuristic, hydroponic buildings of the Netherlands it could mean just one city block! “Hide” feeds on the capacity of the land to feed a family for its meaning.
Another example of a context hungry word, I read in Syrian Colloquial Arabic, page 171. The word is وقية (uqiya) and in this book it means 200g. But looking at the wiktionary entry it means many different units of measure. It also means a unit of currency in Mauretania and an ounce.
Depending on the city or country you are in, an uqiya can mean different things. Here is a list:
I can’t really find real sources for these measures, although there is a hadith that links it with being an ounce. Here is the hadith:
سألت عائشة زوج النبي
كم كان صداق رسول الله
قالت كان صداقه لأزواجه اثنتي عشرة أوقية ونشا قالت أتدري ما النش قال قلت لا قالت نصف أوقية فتلك خمس مائة درهم فهذا صداق رسول الله
(لأزواجه». (رواه مسلم
I can’t find the reference for this hadith but it is in the wikipedia page for uqiya.
Today, you can find the word used in Beirut as a substitute 200 grams. I am not sure where the 13.39 other grams cited in Wikipedia comes from. This is where adding citation needed tags is required.
A conversation between a shopper and a grocer might go like this:
بقديش الخيار؟
ب٥. الف ليرة بوقية او ٢٥٠ الف ليرة بكيلو
How much for cucumbers?
50,000 lira per uquiya or 250,000 lira per kilo.
By the way, these prices are about accurate given hyperinflation in Beirut.
This all leads to a question. Given a phrase “50,000 lira per uquiya,” how would a machine ever determine how many grams is meant? How would a machine ever determine how large England and Wales of 1450 was based on the phrase, “a demesne of a million hides.”
It might seem that merely applying the right machine learning models to the problem would be sufficient. And let’s for the moment assume that’s the case.
Here is an example.
Instead of a human shopper, we let a machine shop. This machine has been given a simple decision tree. Check the city they are in, and based on that use the appropriate uquiya. A machine might receive a purported uqiya of cucumbers (200 grams), but the machine looks at the wikipedia entry and see 213.39 grams. At this point, the transaction fails and hopefully the engineers accounted for this happening, and some fallback scenarios come into play: ask grocer to double check, and if grocer insists call person who made the order.
But this dispute might not ever be resolved…
Another dispute made famous in Italy is translating langostino into lobster on Italian menus. I couldn’t find this controversy online but when I was in Venice, this happened frequently while there for Carnivale so many years ago. The American will note how cheap it is, and then in horror recoil as a plate of something shrimp-like and small is offered instead. The issue would resolve itself with the Americans refusing to pay for something they didn’t order, and later, the wait staff happily gorging themselves on free squat lobster, and not spiny lobster — the lobster you and I know about, dear reader.
What about the sentence, “My crops failed last year, so my land was really half a hide, so I should be tithed on that.” The end goal of a machine learning tax collector might very well end up be maximizing taxes collected, so the machine might be more than happy to have the farmer whose crops failed starve. What machine learning model could apply to the calculation of hides?
Perhaps something crude as a k-means algorithm could determine the size of a particular hide given the inputs of access to water, average hours of sunlight per year, fecundity of the soil, and yearly mean temperature. The output could be bushels of wheat, and would only be applied to wheat growing regions.
Here we run into the problem that the more general the solution, the further away we are from the context for a hide. It’s really important to the farmer that the land be a hide: anything less, and the household goes hungry or even starves. Anything more, and that means maybe the land could be shared, or the surplus used for profit.
Is there a word so hungry for context, that it’s like a blackhole? Perhaps words like, “this,” “that,” and “here.”
Context hungry words seem meant to defeat mechanization and automation. They also seem to lead the way to those experiences in life that we can share with others but are impossible to articulate, the demesne of mysticism.
