|
|
Jessica Boynton |
|
|
Coda Simplification in Non-Standard EnglishI grew up translating English into English for my mother. I grew up watching others assume she was ignorant because she couldn't understand what they were saying, and because she talked differently. Some people have even asked her which country she's from. It wasn't until I started studying phonology that I even picked up on her accent. Then, in learning about optimality theory, I realized that one constraint in particular seemed to characterize her speech - *COMPLEX(coda). I remember trying to teach my mother how to pronounce labyrinth (it was our favorite movie) and hearing her stumble over complex words until she'd just say "oh, you know what I mean." Most people do that, but in trying to pronounce the word she'd do weird things to it. Like make it longer. So, when choosing my topic for my phonology project this semester, I decided to tackle this aspect of her accent. I started out proposing that she simplifies codal consonant clusters by epenthesizing schwas. Then things exploded. But before we get into that, let me introduce you to my mother. She's a registered nurse at the hospital back in Big Rapids, although she'll have to give up that career soon as her hearing loss progresses beyond what hearing aids can overcome. She grew up in northern Michigan, living mostly in the wooded Newaygo area interacting primarily with her 100+ cousins. My mother had normal hearing until she was about a year and a half old, but then she had an inner ear infection that resulted in considerable hearing loss (60%-75%). Since then, she's had a gradual increase in hearing loss, so that now she can barely interact over the phone (even with hearing aids) and relies primarily on reading lips. From age 6 till 11 she was explicitly taught how to pronounce words and how to read lips. It's really amazing that she mimics standard English so well considering how long she's been without hearing. She also reads lips expertly, unless the speaker mumbles or has a mustache. Another factor that probably influences her speech is an unusually high palate. This is but one of many flaws in my research design. To research this project, I made a list of words with complex codas for my mother to read. Then, to simulate natural speech, I wrote them into an enchanting story about a klutzy nymph, a barking dwarf, and a melodramatic elf. Then I talked my mother into reading the story into a recorder (I left the room for this so she would be more comfortable). I transcribed her speech and analyzed it to the best of my abilities, comparing her pronunciations to the standard ones provided by Michael Hammond in his book The Phonology of English. However noble my efforts, there were some weaknesses to my research design. First off, this was my first attempt at transcribing speech, so I was dependent upon data supplied by a woefully unpracticed transcriptionist (assuming, of course, that transcriptionist is a word). Also, though whenever I looked at all 98 transcribed words it seemed insurmountable, I didn't have enough data to do more than generalize. They represent to broad a spectrum for an in depth analysis. This data set served well as a starting point, but ideally I would have used this to hone in my next set of target words. Another problem with the data is that it's all 'reading aloud' speech. A set of 'list' speech and a set of 'natural' speech data would provide a nice balance for an analysis. I originally doubted Michael Hammond's accuracy on the standard pronunciation of some words, but apparently I mispronounce those specific words like my mother does. That detail notwithstanding, his transcriptions don't reflect some lenition (like unreleased word final obstruents) nor do they reflect [p] epenthesis which usually occurs m__obstruent if I'm not mistaken. I don't blame him leaving out these details, as they weren't relevant to his aims. They are, however, to mine. These minor difficulties aside, I found some really neat patterns. There are some unconditioned changes - or at least changes for which I cannot identify a condition - some conditioned changes, and then some generalizable exceptions to these patterns. Then there are some data that confuse me greatly. As for the unconditioned changes, it seems like my mom likes to centralize sounds. [s] palatalizes to some degree; it's not quite a [_] but it's close. Also, [_] always becomes [_] which, in case you don't have your IPA chart handy, is the voiced dental implosive (although, to be totally honest, I'm not sure that it is implosive). Also, vowels are different. I'm not exactly sure how so, but they're just different enough that I feel like a liar when I write the vowel used in standard English as the vowel she uses. [_] (which, because of some technical difficulties I'm having, is my way of writing the alveolar lateral approximate)seems prone to lenition, although I can't characterize it. I didn't know how to reflect degrees of lenition in the transcription, but, in prose, the following situations occur: I think this happens in standard English too, but Hammond doesn't address it.
As for the conditioned changes, [d] and [t], are deleted at the end and middle of codal clusters. There are numerous examples of [d] and [t] deletion at the end of codal clusters, for instance: There are fewer examples of the medial deletion, such as: Also, in places where speakers of standard English would insert a [p], she doesn't. She fails to follow a rule that would, essentially, complexify a coda. For example: Some data seem to resist the generalizations, but these data can actually be generalized themselves. First off, [r] seems to protect the consonant that immediately follows it. For example: Also, sometimes word final d and t are not deleted, but this only happens when the next word starts with a vowel and the offending phoneme is actually picked up as an onset to that word. In other words, deletion of these units is not necessary to simplify the coda because these units have actually been adopted as onsets. For instance: Some other data don't resist other rules as much as they seem to have created their own. This can usually be blamed on unfamiliarity and awkward orthography _ especially since these labels are so subjective. For example: Finally, there are some oddities that I simply cannot characterize. There are some vague patterns, but there's not enough regularity to make generalizations. Sometimes weird voicing patterns occur. For instance: S + voicless plosive + obstruent clusters, like, explode: And other random oddities surface, like: Overall, my original hypothesis was pretty accurate _ my mother simplifies codal consonantal clusters. In her attempts to do this, she will always neglect to epenthesize consonants where standard English would, with the possible exception of (37) 'glimpsed' [g_Imp] where the [p] actually appears in the orthography. Some sounds, namely [d] and [t] are more expendable than others. If possible, she'll use left capture to remove these from coda position. But this only works if the word following the phoneme begins with a vowel. So her next option is to simply delete the sound. This, however, is only an option for [d] and [t]. Sometimes she'll epenthesize a schwa into particularly confusing constructions, which effectively makes codas into onsets. Other times she'll use metathesis to at least simplify the coda, even if it doesn't actually remove any part of it. [r], for some reason, offers protection to the consonant immediately following it. And some data follow rules I can't figure out. From an optimality theoretic standpoint, it seems like whatever constraint keeps words separate (I'll call in WORD because I'm not even sure if it's a markedness constraint or a faithfulness constraint) would be very lowly ranked, to allow left capture to do what in can to satisfy the highly ranked *COMPLEX(coda). If left capture isn't an option, MaxIO can be violated in some cases to delete part of the codal cluster. In extreme situations, DepIO can be violated to make a codal unit into an onset. Sometimes LINEARITY can be violated to rearrange phonemes in the cluster into a more pronounceable order. I suspect that there are feature-based conditions at work here that I have yet to identify. However, an extremely tentative ranking would look like: *COMPLEX(coda ) _ LINEARITY _ DepIO _ MaxIO _ WORD A generative approach would have rules that look something like: But I can't even begin to define the epenthesis or metathesis because I'm not sure if anything actually conditions it (perhaps they're just coincidentally similar idiosyncracies?). Both of these attempts at analyzing the data would be greatly improved with additional data. Minimal pairs would be nice too. I would love to collect more data, using my current generalizations to focus on specific phenomena one by one, and better explain the patterns. With a better analysis, I could then look more deeply into the implications of the constraint rankings and rule orderings. Until then, I'll just have to sit around pondering. Word Story(1 - 25) Once1 upon a time there lived2 a dwarf3, an elf4 and a nymph5 of the woods6. These three unlikely friends7 formed8 the Eighth9 Distinct10 Triumvirate of the Marshland11 Realm12. This mostly13 means14 that they waged15 war against16 werewolves17, bronzed18 gods19 and warped20 giraffes21 that had absorbed22 tainted water amidst23 the Streams of Substandard24 Charm25. (26 - 41)One day, while she searched26 for emeralds27, the nymph28 (being something of a klutz29) fell into the Darned30 Caves31 of Triumph32 Dreamt33. She cowered34, bruised35 and alarmed36, and glimpsed37 into the dimly lit labyrinth38 of quartz39, wishing more dearly than ever before that she were one of the winged40 air-nymphs41. (42 - 53)Her pulse42 quickened43 when she heard44 a distant45 belch46. Then she saw him _ her friend47, the dwarf48. "Foolish nymph49!" he barked50 succinctly51 as he made his way to her. "Always neglects52 to call before she drops53 in." (54 - 63)"You moved54 here from the Gorge55 of Eclipsed56 Depth57?" she asked58. "Yeah, I think59 it was two months60 agoÅc" the dwarf61 was interrupted when the elf62 waltzed63 in. (64 - 91)"Behold64 the wealth65 of the cavernous66 realms67! Glimpse68 the caress of Nature as she carves69 the dismal landscape70! Sculpts71 the hearths72 of lords73 forgotten! in the most74 ancient75 of crafts76!" He clasped77 the nymph's78 hand79. "Father Time warps80 not the charms81 forged82 through her tools83. The silk84 art85 shall be divorced86 from the gnarled87 course88 of age that rasps89 away at Youth's90 tender countenance91." (92 - 98)"Melodramatic elves92," the dwarf93 breathed94. The elf95, shocked96 at the dwarf's97 vulgarity, skulked98 off. Transcriptions
1.once [w_ns]
|