Tag: languages

Vai Script

The Vai script of Liberia was created from scratch in about 1834 by 8 completely illiterate men who wrote in ink made from crushed berries. Because of its isolation, and the way it has continued to develop up until the present day, we thought it might tell us something important about how writing evolves over short spaces of time. The 8 Vai creators set out to design symbols for each of their language’s syllables, inspired by a dream. Their chosen symbols represented physical things like a pregnant woman, water, and bullets, as well as more abstract traditional emblems. Over the first 171 years of its history, the Vai script did become increasingly compressed. The simplification occurred over generations of users; symbols with the highest complexity were simplified the most. These changes are far from random. Languages pass a kind of natural selection process via memory and learning, where the hardest to recall features do not survive. As the letters became less complex, they also became more uniform. This is despite the language never having been adopted for mass production or for bureaucratic needs. These uses are what seemed to help standardize other languages – for example, Mesopotamia’s writing standardization coincided with the implementation of state-wide systems.

1.5 ma Language?

Everett examines the culture of the first known human species, Homo erectus, focusing especially on their physical and cultural evolution such as tools, travel, and settlements. He then makes the case that these accomplishments are best explained by the invention of language. Language in turn is shown to be the transfer of information by symbols, where other components of language, such as grammar, play roles in support of symbolic communication. Concrete evidence for symbols among erectus populations is found in their tool construction and “dialectal” tool distinctions.

Indian Legalese

The modern form of Babu English turns up most frequently in the language of India’s legal system. Here’s a single sentence from an order from the Himachal Pradesh state high court issued in 2016: “However, the learned counsel appearing for the tenant/JD/petitioner herein cannot derive the fullest succour from the aforesaid acquiescence occurring in the testification of the GPA of the decree holder/landlord, given its sinew suffering partial dissipation from an imminent display occurring in the impugned pronouncement hereat wherewithin unravelments are held qua the rendition recorded by the learned Rent Controller in Rent Petition No. 1-2/1996 standing assailed before the learned Appellate Authority by the tenant/JD by the latter preferring an appeal therebefore whereat he under an application constituted under Section 5 of the Limitation Act sought extension of time for depositing his statutory liability qua the arrears of rent determined by the learned Rent Controller in a pronouncement made by the latter on 6.11.1999, wherefrom an inference spurs of the JD acquiescing qua his not making the relevant deposit qua his liability towards arrears of rent within the statutorily prescribed period, application whereof suffered the ill fate of its dismissal by the learned appellate Authority under the latter’s order recorded on 16.12 2000.” When the matter came up in appeal before the Supreme Court, the baffled judge sent it back to the high court, observing, “We will have to set it aside because one cannot understand this.” “It seems that some judges have unrealized literary dreams. Maybe it’s a colonial hangover, or the feeling that obfuscation is a sign of merit… It can then become a 300-page judgment, just pontificating” In October, Subhash Vijayran filed a public interest litigation in the Supreme Court, which is in the process of hearing his petition requesting that legal writing be simplified. “The writing of most lawyers is: (1) wordy, (2) unclear, (3) pompous and (4) dull,” his petition states. “We use 8 words to say what can be said in 2. We use arcane phrases to express commonplace ideas.”

See also the elements of bureaucratic style:

What became clear to me in this exchange is that the passive voice is itself unsuited for the lexical landscape of United’s email, which itself is part of a larger world we now find ourselves in, where corporate and government bureaucracies rely heavily on language to shape our perception. Munoz’s email relies heavily on the passive voice to evade culpability, but he also employs a host of other rhetorical moves that collude to put the blame on the man who was assaulted and carried out on a stretcher. Like a well-trained bureaucrat, Munoz used an array of syntactical choices in a predictable, quantifiable, and deliberate manner, and it’s time we recognize it for what it is.

Lox Hasn’t Changed in 8 ka

The pronunciation in the Proto-Indo-European was probably ‘lox,’ and that’s exactly how it is pronounced in modern English. Then, it meant salmon, and now it specifically means ‘smoked salmon.’ It’s really cool that that word hasn’t changed its pronunciation at all in 8 ka and still refers to a particular fish.

On GPT-3

GPT-3 is scary because it’s a tiny model compared to what’s possible, with a simple uniform architecture trained in the dumbest way possible (prediction of next text token) on a single impoverished modality (random Internet text dumps) on tiny data (fits on a laptop), and yet, the first version already manifests crazy runtime meta-learning—and the scaling curves still are not bending! The samples are also better than ever, whether it’s GPT-3 inventing new dick jokes or writing (mostly working) JavaScript tutorials about rotating arrays. Does it set SOTA on every task? No, of course not. But the question is not whether we can lawyerly find any way in which it might not work, but whether there is any way which it might work.