Allophones, (Nearly) Homophones, and Voice Recognition

In these two pieces, Gussenhoven and Kenstowicz discuss many of the important base topics of phonology. Gussenhoven focuses mainly on the physical anatomy of the vocal tract and how different body parts such as the larynx, glottis, and vocal cords can alter and contribute to speech. He identifies a multitude of different phonetic terms, such as voiceless, devoicing, and aspiration, using examples from a variety of languages to corroborate his points. Kenstowicz builds upon the information presented by Gussenhoven, using the gamut of phonetic terms to outline a series of governing rules. Specifically, he focuses on the varied forms of the letter “t” in the English language to describe the properties of phonemes and their allophones. The selection from Gussenhoven’s work essentially serves as a dictionary of phonetic terms for Kenstowicz’s denser writing.

The most interesting point I garnered from Kenstowicz’s piece is that most native English speakers do not realize they are following a set of defined phonetic rules. Somehow, with repetition and time, we become familiar with rather odd patterns in the English spoken language. For example, comparing the pronunciations of “writer” and “rider” as well as “seated” and “seeded” reveal that these pairs of words sound nearly identical (due to the flapped allophone of “t” in these cases). Yet, due to slight variation in the vowel length, native speakers can identify the intended word and distinguish between these nearly homophonous terms. 

Though I understand why the pronunciations of “writer” and “seated” contain more of a [D] sound than a regular [t] in everyday conversation, I am curious to see whether these rules of pronunciation always apply. That is, if a person spoke slowly and deliberately, emphasizing each syllable, couldn’t the word “writer” result in two explicitly divided parts – with the latter half starting with the written [t] instead of the usual [D] sound? Interestingly enough, Google Translate’s audio button pronounces “writer” with a clearer [t] sound, while Siri says the word with the flapped [D]. Siri also struggles to determine the correct word, unless given clear context and additional information; for example, it is easier for Siri to understand one is saying “writer” when “book” is mentioned in the same statement (another example – “horse” and “rider”). For Siri at least, these words seem to be pretty much homophonous. Though Kenstowicz stated that these words in fact had distinct traits, I believe that without any other context, even native English speakers would struggle to identify the correct choice given these pairs of words.

I still have some lingering thoughts about the reading that hopefully someone can explain in more detail. Specifically, after reading Kenstowicz’s statements about the multiple levels of phonological representation, I understand that there are some aspects of language that cannot be clarified simply through explicit phonetic rules. However, I had some trouble parsing through the dense examples and would appreciate any concise explanations of the “second level.” Is his main point just that it is impossible to apply black-and-white rules to linguistics?