Why humans need to speak to machines on their terms

As AI becomes an integral part of our lives, many new words, expressions will gain currency



By Shalini Verma

Published: Mon 23 Sep 2019, 9:00 PM

Last updated: Mon 23 Sep 2019, 11:27 PM

Language is an emotional subject. The recent language debate in India has stirred up reactions of all emotional shades and shapes.
We just have to look at Bangladesh of the 1950s, then East Pakistan, to understand the strong emotional underpinnings of language. Did you know that Unesco honors February 21 as the International Mother Language Day? This commemorates the 1952 tragic killing of activists and students, who were part of the Bengali Language Movement in Bangladesh. The demand to reinstate Bangla as the official language was part of the mass cultural unrest to reclaim the Bangla identity in East Pakistan.
Mother tongue can easily become synonymous with culture because it carries in its grammar and vocabulary its people's history, ethos, and tradition. There is so much encoded in how and what we say that 'lost in translation' becomes extremely perceptible as we switch between languages.
In a 2013 study by Efrosini Kokaliari, Gia Catanzarite, and Joan Berzoff, called It Is Called a Mother Tongue for a Reason, the researchers found that their bilingual clients defaulted to their mother tongue when 'expressing strong affects, in their dreams, or when dealing with death or trauma.' We often express distinct values and identities in our mother tongue vis-à-vis our secondary language. For example, it is common knowledge that we Indians, regardless of our fluency in English, tend to speak to our pet dogs in English because we picked up the hobby from the British a few generations ago. Subliminally, we assume that our foreign bred dogs understand English.
Language is deeply personal. So, forcing people to speak a language, be it their own or someone else's can feel like oppression. However, there is nothing permanent about language. British linguist David Crystal famously said that the only languages that do not change are the dead ones.
Language evolves in an unconscious, organic way - with time and geography. It is no secret that languages like Arabic, Spanish, and Hindi have picked up local accents and vocabulary across their geographic expanse. When I was learning Arabic, I became aware of the extent to which the language has adapted to the many rich influences. My Arabic teacher would smile sagaciously and say, "You can write it like this or like this. No problem."
Language borrows from the language of conquerors, rulers, and traders. Teachers making a case for learning English assert that it is the business language in many countries. And it has also changed significantly over time. Most of us would constantly need the dictionary when trying to decode Chaucer's Middle English, which borrowed heavily from the French, Latin, and the Old Norse.
The infiltration of LOLs and ROTFLs in our languages was the result of our pragmatic texting needs. But before that a 'hello' took on a whole new meaning when we started to use telephones. 'Hello' diversified into many accented sounds in different cultures. Our grandparents were experts at communicating through telegrams, which developed a lingo of its own. But texting has had a more pervasive impact on our language because it is so entrenched in our lives. Now a simple LOL is a viable proxy for a guffaw or a chuckle. The Grammar Nazi within us often recoils at the way our languages are getting refashioned by the texting multitude. But this is not the first time, nor will it be the last.
As artificial intelligence (AI) becomes an integral part of our daily lives, many new words or expressions will gain currency. Now digital assistants are becoming more versatile as they adapt to new languages and industry domains. It is possible to chat with a digital assistant in Arabic, to follow up on the muffins you ordered along with your coffee. But then you realise how versatile our languages are when you train a digital assistant to converse. Just consider the many ways in which you can order a cup of coffee.
But the big question is, will our language adapt to the limitations of AI or will machines pick up our slang, colloquial innuendos, and bilingual expressions? My sense is that there will be a bit of give and take.
When our conversations with machines will inevitably become more frequent and nuanced, we will see the real impact of AI on our language. We will need to dress up our thoughts in a language that machines can quickly understand. For example, we are likely to be more direct in how we command a digital assistant, and thus lose some of the social niceties and emotions. When we command Alexa or Google Assistant to play a song, they obey when we address them on their terms.
Natural language processing will also evolve to help machines become more eloquent in addressing our everyday needs. The training data set will be large enough to train AI models in a general way. It will continue to learn from different fields of linguistics such as pragmatics that addresses social uses of language. Our emotional bond with our language will remain, but when it comes to conversations with AI, humans and machines will meet halfway. 
Shalini Verma is the CEO of PIVOT technologies


More news from OPINION