Humanoid robot looking into the future (2023) by Aina Arbona GelabertFundación Antonio de Nebrija
What defines a language?
A language can be defined by its different components (pronunciation, grammar, vocabulary) and by its different social forms of use (fixed spelling, types of text, traditions of discourse...).
In addition to its linguistic and textual forms, a language has living use dimensions that are not so easy to formalize.
Language according to a machine (2023) by Aina Arbona GelabertFundación Antonio de Nebrija
We could say that a language has several dimensions.
A language is a lot more than just words.
Humanoid brain (2023) by Aina Arbona GelabertFundación Antonio de Nebrija
1 - The formal dimension
This tangible dimension comprises, for those who learn a language, managing grammar, vocabulary and, in written code, mastering spelling.
2 - The sound dimension
Intonation is so important! The intonation we give to our sentences has a huge impact on their meaning. Linguists call it "prosodic." Sounds (organized into phonemes) are a key factor in identifying our dialectal origin.
3 - The pragmatic dimension
This dimension includes uses such as courtesy, humor, the conception of time and space, the management of silences... When speaking, we use codes that are not always written, and that may depend on our specific field, such as certain cultural references.
In fact, if we talk to someone from another Spanish-speaking country or from another region to ours, there will probably be expressions and references that we will not be able to understand, because they are part of each person's mental construct.
Humanoid hand (2023) by Aina Arbona GelabertFundación Antonio de Nebrija
When you learn a language, you have to acquire a global mastery of all its dimensions.
Learning a language is, simultaneously, acquiring an increasing inventory of words and syntactic structures, handling them with competent pronunciation, and (which is more difficult) acquiring the social uses that condition the language.
Humanoid head (2023) by Aina Arbona GelabertFundación Antonio de Nebrija
If for us, as human beings learning human languages, this path is complicated, for machines it is even more so.
Now that we regularly use artificial intelligence, it is becoming increasingly clear that it is not enough for machines to master the tangible dimension of language.
Language as an epicenter in AI development
Having a conversation with an AI chatbot is one of the greatest achievements in the field of computer technology.
Presentation of humanoid robot (2023) by Aina Arbona GelabertFundación Antonio de Nebrija
The use of support operating systems such as Siri and the huge revolution that ChatGPT and other AIs have caused make the idea of being able to chat with a machine increasingly real. However, we realize in just under a minute that we are talking to a robot.
Why?
Because we have not yet been able to transmit the sound and philosophical dimensions to machines. From that desire, natural language processing (NLP) and language technology (LT) are emerging.
NLP and LT research how to teach a language to a machine in its entirety, so that it can not only interpret it, but also understand it.
What is the situation of Spanish in this new paradigm?
Spanish has suffered some difficulties when it comes to integrating into technology: lack of data, legal limitations in the access to data, misalignments in terms of the format of this data, lack of professionals, and the small size of companies in the sector
Humanoid robot looking into the future (2023) by Aina Arbona GelabertFundación Antonio de Nebrija
These barriers are being broken down thanks to the realization and awareness of the importance of this language in the world, not only in its cultural dimension but as an economic engine in different countries and sectors.
In order to compete in markets that need these technologies, Spanish must effectively develop its presence in current linguistic computing systems.
Why does a machine need to understand a language?
The most important instrument of communication for human beings is their voice. Our voice is the key element that helps us gain in-depth understanding of the intention, the emotions, and the meaning that underlies words.
Humanoid writing (2023) by Aina Arbona GelabertFundación Antonio de Nebrija
We constantly talk to AI systems. Voice chatbots are integrated into answering machines at institutions, banks, restaurants...
How many times have you called a number and had a robot answer you? Do you realize right away? What gives them away? Probably the lack of intonation when speaking and the fact that they do not understand the answers unless they fit with those programmed in their code.
The future of AI: being able to understand language
The greatest challenge facing computational linguistics is to be able to encode the sound and philosophical dimensions of languages, a challenge that LT and NLP specialists are trying to solve.
Humanoid reading a speech (2023) by Aina Arbona GelabertFundación Antonio de Nebrija
This remarkable technological breakthrough does not happen equally in all languages. Given the relevance of Spanish globally and the number of speakers, it is particularly important to develop the necessary technology to make this language understood by machines.
For this report we have consulted the study La tecnología del lenguaje: la inteligencia artificial centrada en el lenguaje.
Content curation and script: Aina Arbona
Graphic Editor: Carmen García
Consulting and review: Lola Pons (Universidad de Sevilla)
