With many individuals worrying about privacy as well as the possible dangers that may arise with society’s reliance on computer technology, artificial intelligence (AI) is becoming particularly relevant. Artificial intelligence describes the ability that computers have developed to perform tasks that are commonly attributed to intelligent beings. From the Google search engine to the newest facial recognition software on cellphones, individuals interact daily with artificial intelligence. Hot topics in AI often revolve around the technology’s risks and benefits, yet artificial intelligence’s ability to intersect a vast number of fields is rarely discussed. Why is this? The answer possibly lies in the lack of understanding of AI’s true benefits as we focus too highly on its possible risks.
For example, the field of linguistics, which uses AI as part of its applications, is often overlooked in AI matters. While many associate linguists with polyglots, the study of linguistics encapsulates the scientific study of speech perception, sounds, grammatical structure, and meaning. UTM Linguistic Professor Barend Beekhuizen is a linguist who advocates for the use of computational methods in the field, and uses these methods in his research.
Professor Barend Beekhuizen specializes in computational linguistics research, a subfield of the linguistics department, where computational mechanisms are applied to linguistic data in order to extract particular information on language. “It has allowed us to look at linguistics from a perspective that just wasn’t possible before,” says Professor Beekhuizen, explaining that the fast generation of linguistic data across languages was not accessible until the introduction of computer science methods into the field.
Growing up in the Netherlands, Professor Beekhuizen’s love for prose influenced him to study Dutch literature at Leiden University. However, he quickly realized that he was more interested in language: “that’s where my passion for linguistics began,” Professor Beekhuizen says. While completing his master’s in linguistics at the same university, he discovered that his questions about how language works were not easy to test, so he began using computational techniques to build models that made linguistic predictions. He completed his Ph.D. at Leiden University, focusing his studies on computational models of language acquisition in children. Professor Beekhuizen is also interested in studying the variation of word categorization between languages and what those variations in linguistic discourse tell us about individuals’ representation of the world. More of Professor Beekhuizen’s research interests includes computational models in the use of colour term acquisition, as well as work in lexical ambiguities.
The University of Toronto has long offered courses with computational linguistics integrated into the computer science department; however, only recently did the university hire computational linguists, such as Professor Beekhuizen. Although not a new field study, using computation methods as part of the linguistic field is a novel approach, leaving some traditional linguists hesitant in its use. “Like any new method, there will be people who embrace it right away and people who are more skeptical,” explains Professor Beekhuizen. Fortunately, the subfield’s accessibility is drawing more attention to the study and exposing more individuals to its utility. “The field has embraced computational methods as part of what they want to give to future generations of students,” says Professor Beekhuizen.
The interaction between artificial intelligence and linguistics is more prominent and practical than many believe due to linguistics being central to everyday tools such as speech recognition and search engines. Additionally, the field is increasingly important in medical and healthcare-related language processes, where computational linguistic mechanisms can be used to extract information from patient files by using keywords. Moreover, computational linguistics techniques have been adapted for use by marketing companies to receive feedback from consumers and better understand consumer needs.
More students are also showing interest in the field. “I’ve noticed students on [both] undergraduate and graduate levels are interested to learn about computational methods to supplement their tool kit in other linguistic fields,” says Professor Beekhuizen. He’s had the opportunity to work with many of these students on research projects using artificial intelligence. Recently, a group of five undergraduate students completed a computational linguistics research project through U of T’s Jackman Scholars-in-Residence (SiR) program under the supervision of Professor Beekhuizen. The study investigated the context that certain translated words are used.
The English words “true,” “real,” “actual,” and “right” were analyzed in the subtitles of online TedTalks. The subtitles, translated into a dozen of languages, were analyzed through computational methods to understand the context of their use, as well as make deductions on how each word is translated in other languages. “We’re currently working on a write-up of this project with the students to send to an academic journal,” says Professor Beekhuizen. The initial research period was successfully conducted virtually last year.
To learn more about computational linguistics, students should consider taking Professor Beekhuizen’s course, LIN340: Language and Computers. LIN340 introduces the field by examining issues concerning computational linguistics and societies, such as equity concerns in speech recognition software. The course also provides insight on how artificial intelligence aids in answering specific linguistic questions, such as detecting sentence structures and grammatical relationships.
Computational linguistics offers an innovative perspective to AI that will hopefully help us understand its advantages. The linguistics’ subfield also raises awareness and develops tools for people requiring accessibility services, such as those needing aid in navigating disability or language barriers.