Je + Mettre ≠ Bilingualism
If I showed you a picture with clearly labeled pictures and explanations versus a conversion table of the basic grammar rules of a language, which do you think would help you understand a language easier?
Like, what? Do you remember when you first learnt your native language? Surely you did not learn it by memorizing ridiculous tables like these… If this was not how you became fluent in your native tongue, why aren’t you taught this way with second languages as well?
Most of us have probably attempted to learn another language in one way or another. A majority of the time, especially in the U.S. public school system, we are taught like a grammar machine. Starting from the “present tense” and their “conjugations”, and then moving on to the past, future, and conditional tenses. Blah, blah, grammar talk that most people find exhausting and often find themselves not only unable to apply these rules, but even giving up on languages as a whole.
This isn’t how languages work, and this isn’t how humans should learn them either. As many of us may already know, languages don’t intricately follow grammar rules. In fact, English grammar probably has more exceptions to rules than rules themselves. We actually aren’t meant to acquire languages with such rules, however, it just so happens that machines do. What scientists have found is that through Natural Language Processing (NLP), computers can actually be programmed to learn every single intricate grammar rule in order to understand human, “natural” language.
Languages are very useful with communication, however, sometimes it can also be a barrier for some people. Which is why technology companies like Google and Amazon are using Machine Learning to teach computers how to communicate with human language systems to further convenience our lives. This is why when you ask Google Assistant (or Siri for Apple users 🙄) on your phone to translate a sentence, not only can it translate it, but it can also comprehend and carry out your request. Thank God for NLP!
Natural Language Processing
Typically, computers are programmed and can be expressed information to through programming languages such as Python, Java, or C#. This is to be differentiated from Natural Language, which is the language at which humans communicate through complex grammar and vocabulary. This also has several outliers which we have adapted to more efficiently communicate at our comfort. This form of language is rather complex and is harder for computers to comprehend as there is often no exact translations or specific computations like in programming languages.

To get past these complications, scientists first dissected our languages and broke it down into basic grammar structures and parts of speech that we are taught at school. The most efficient method so far is by using Parse Trees. These are webs of parts of speech that categorize and separate parts of sentences, then go further into detail as you look closer at the phrases. This way computers can more easily differentiate between words that can be used as either nouns or verbs, such as “leaves”.
Well, how does the machine understand our voices in the first place? These verbal sentences can often be understood through Speech Recognition, which is using the wavelengths of our speech and analyzing them through Deep Learning and Fast Fourier Transform. Fast Fourier Transform analyzes spectrogram data, that uses those same wavelengths, and maps their frequencies in Hertz. Through machine learning, computers can learn tremendous sets of data and understand the frequency patterns for certain words and Phonemes, which are specific sounds that distinguish syllables in different words. /u:/ is a Phoneme for the sound ‘oo’ like those in the words moon, soup, and duke, despite having different spellings. Once the word is retained, the computer can connect them to their meanings.

Once a sentence is comprehended, the computer will provide feedback. Through Knowledge Graphs, computers will take the question phrase and the noun and connect them within this massive graph of networks and database of information. Say you ask, “When was Da Vinci born?” By connecting the phrases “when was ___ born” and “Da Vinci”, computers can easily search through the data base and answer the question. This is how Siri and Google typically work.
Human Language Acquisition
Humans, however, are different from machines. Shocker, I know, but this is important. The methods of language acquisition for us is a lot simpler than memorizing charts, grammar structures, and vocabulary. Languages were created to communicate ideas to other people, they’re a form of expression. We shouldn’t be learning languages like a machine; we didn’t create them like a machine!
”Acquisition requires meaningful interaction in the target language — natural communication — in which speakers are concerned not with the form of their utterances but with the messages they are conveying and understanding.” Stephen Krashen
Stephen Krashen is a linguistics professor and researcher who emphasizes this concept of language acquisition: that language should be taught by meaning and not by rules and memorization. He had a hypothesis for Language Acquisition and Learning. That Language Acquisition is based on learning a knowledge and subconsciously accepting it into your brain versus Language Learning which is learning about languages and their grammar structures. One will lead to bilingualism, while the other just leads to forgotten notes.

Biologically, language comprehension actually takes place in the temporal lobe in an area called “Wernicke’s Area”, named after a German scientist. While language production takes place in an area called “Broca’s Area”, named after a French physician. These areas were originally discovered by brain lesions, damage to brain tissue. Certain patients would have damage to these areas, and scientists infer that this damage caused their particular aphasias.
Despite these being the biological areas that give us the capacity to comprehend and produce language, this doesn’t actually explain why or how language originated. This is because language isn’t something that we were created with. It is something we adapted to use for our advantage. Historically, the Incan empire didn’t even have any written language system, despite having been the largest empire in the Americas until defeated by the Spanish in 1572. They simply communicated through the tying of rope, which was their language.
This emphasizes the concept that language is simply just a form of communication between people. This is also why it is constantly changing around us. Think of the new lingo that teens use everyday that never existed before. These words are just as much English as any other word in the language.
As new jokes, technologies, and concepts are added to our culture and everyday lives, we adapt and change how we communicate to include these new ideas. This is another barrier that AI would have with Natural Language Processing, but this is also the advantage that us humans have. The ability to advance our capacities of communication and advance our societies through this. Think of how sad our conversations would be if we couldn’t adapt to the new and changing world around us.
It is also very important to stress that language is how we communicate with ourselves as well. As our language can expand exponentially, so can our thoughts and methods of expression. This freedom of expression is essential to the advancement of our society and ourselves as individuals. Any attempt to prevent or silence that is an insult to the phenomenal gift that is our brain.
Hi! My name is Alyssa Gould. I’m super passionate about languages and biology of all kinds.
Feel free to contact me at anytime at alyssa.gould@outlook.com and add me on LinkedIn!
