On the Benefits and Dangers of AI: The Role of Sociolinguistics in Understanding Speech Technologies

By Prof. Nancy Niedzielski (Department of Linguistics, Rice University, Houston, Texas)

May 23rd, 2024

All lectures will be held in RIGA 717, 14.15hrs

While the incredible benefits of speech applications to technology (i.e., speech recognition and synthesis, or artificial intelligence) in areas such as medicine, the legal arena, and human-computer interaction cannot be denied, there have unfortunately been (and continue to be) astounding failures. Often these failures are the result of gross misunderstandings about how language works. Many of these technologies fail to recognize and account for the simple indisputable fact that languages change and languages vary. First, this disregard can lead to the exclusion of entire groups of speakers (especially speakers of non-standard varieties, dialects, or groups with accents differing from educated norms) from using these technologies. Second, this also leads to the creation of an alarming amount of "junk science" (and therefore, "junk technology"). And finally, it creates unnecessary fear of what these technologies are actually capable of, and of what they are not. 

In this lecture series, we will explore speech technologies and examine several ways in which the increased inclusion of sociolinguistic understanding of language change and variation can – and should – play a much bigger role in the advancement of speech technologies.

All are very welcome.

For further info, contact nils.langer-TextEinschliesslichBindestricheBitteEntfernen-@uni-flensburg.de

By Prof. Nancy Niedzielski (Department of Linguistics, Rice University, Houston, Texas)

May 30, 2024

All lectures will be held in RIGA 717, 14.15hrs

While the incredible benefits of speech applications to technology (i.e., speech recognition and synthesis, or artificial intelligence) in areas such as medicine, the legal arena, and human-computer interaction cannot be denied, there have unfortunately been (and continue to be) astounding failures. Often these failures are the result of gross misunderstandings about how language works. Many of these technologies fail to recognize and account for the simple indisputable fact that languages change and languages vary. First, this disregard can lead to the exclusion of entire groups of speakers (especially speakers of non-standard varieties, dialects, or groups with accents differing from educated norms) from using these technologies. Second, this also leads to the creation of an alarming amount of "junk science" (and therefore, "junk technology"). And finally, it creates unnecessary fear of what these technologies are actually capable of, and of what they are not. 

In this lecture series, we will explore speech technologies and examine several ways in which the increased inclusion of sociolinguistic understanding of language change and variation can – and should – play a much bigger role in the advancement of speech technologies.

All are very welcome.

For further info, contact nils.langer-TextEinschliesslichBindestricheBitteEntfernen-@uni-flensburg.de

By Prof. Nancy Niedzielski (Department of Linguistics, Rice University, Houston, Texas)

June 6th, 2024

All lectures will be held in RIGA 717, 14.15hrs

While the incredible benefits of speech applications to technology (i.e., speech recognition and synthesis, or artificial intelligence) in areas such as medicine, the legal arena, and human-computer interaction cannot be denied, there have unfortunately been (and continue to be) astounding failures. Often these failures are the result of gross misunderstandings about how language works. Many of these technologies fail to recognize and account for the simple indisputable fact that languages change and languages vary. First, this disregard can lead to the exclusion of entire groups of speakers (especially speakers of non-standard varieties, dialects, or groups with accents differing from educated norms) from using these technologies. Second, this also leads to the creation of an alarming amount of "junk science" (and therefore, "junk technology"). And finally, it creates unnecessary fear of what these technologies are actually capable of, and of what they are not. 

In this lecture series, we will explore speech technologies and examine several ways in which the increased inclusion of sociolinguistic understanding of language change and variation can – and should – play a much bigger role in the advancement of speech technologies.

All are very welcome.

For further info, contact nils.langer-TextEinschliesslichBindestricheBitteEntfernen-@uni-flensburg.de