Embedded Information

I read all three readings by Professor Sumner. All of them discuss speech recognition’s dependence on speaker’s identity and voice. The reading from Cell Press claims that recognition of voice cues starts to develop from the earliest stage of our lives. It also interweaves this statement to the idea of “social weighting” or the biases activated when we hear a voice with certain characteristics.

The second reading I read is King’s and Sumner’s research on semantic association with variation in voice. This reading asserts that speaker-specific phonetic cues speech interpretation. A word spoken by two speakers whose identities are largely different generate different responses from participants in both offline and online tasks.

The third reading I read is Sumner’s and Kataoka’s, which examines soeech recognition with regards to variable productions of the words (accent). The study suggests that increased attention is related to priming and false recall rates. I find the false recalls most intriguing as the increased attention significantly decrease false recall rates.

In a nutshell, these three readings touch upon interpretations of words based on phonetic cues including speaker’s identity and voice.
I notice how this is related to last week’s reading and problem set. For the problem set, I answered that because of speech priming, I was able to guess that the speaker is African American, that it is summer, and that she is in her living room.

I also notice that English is a language whose words do not convey additional information, and it mostly depends on voice and speaker’s identity to embed information. Languages like Japanese or Korean, on the other hand, has specific pronouns for specific relationships. Without speech, the information about a person’s age, gender, or even socioeconomic class is conveyed just through the usage of pronouns whereas, in English, it takes more context to achieve the same purpose.