Notes on Gerfen & Hall (in press) Summary of class discussion: We discussed the relationship between incomplete neutralization as seen in the Gerfen and Hall 2001 data the phenomenon Labov 1994 identifies as near mergers. It was suggested that a good follow up to the study in the paper would be a perception study were speakers of Eastern Andalusian Spanish are asked to identify tokens of type kaspa vs kasta. Our hypothesis is that the difference would probably too minute for the hearer to perceive. Another question that was raized was the overall role spelling played in the phonetic distinction between the aspirated /s/ and the aspirated plosive of type /p/. Did a reading knowledge of Spanish facilitate the distinctive articulation? Would there be a notable difference between literate and illiterate speakers? Here Kathleen noted that if she were to repeat the analysis she would pay closer attention to spelling as well as lexical considerations instead of positing underlying forms for the relevant tokens. Another suggestion consisted in running another perception study in which speakers of another Romance language with phonemic geminates such as Italian would be asked to identify geminate segmets in tokens from Eastern Andalusian Spanish. Would they perceive the difference between the length of closure for /p/ as well as the degree of aspiration of the sibilant /s/? Would only clusters of type /pt/ count for them as a legitimate geminates on par with changes in this Spanish dialect (ex. kapta > kahtta) which more closely mirror Italian (CONCEPTU > concetto)? David Durian Gerfen and Hall (in submission) summary Gerfen and Hall's paper focuses on s-aspiration in word-internal codas in Eastern Andalusian Spanish. Via the paper, they present evidence from two controlled production experiments in which 11 native speakers, ages 30 to 38 were asked to read minimal pairs lists involving the words /kasta/:/kapta/ and /Testa/:/sekta/. Via their productions, subtle (but consistently variable) patterns in which the contrasting pairs of underlying /s/ and underlying voiceless stops (/p/ or /k/) are distinguished. Results of the experiments show that the aspiration of /s/ results in longer duration of aspiration, while the aspiration of /p/ or /k/ results in more medial consonant gemination, when the pairs are compared to each other, while both types exhibit each of these cues robustly when they are compared with words that do not contain medial coda obstruents. Based on their results, Gerfen and Hall argue that they inform four prominent issues in phonetic and phonological theory: a) The data add to the body of evidence suggesting that phonological theory needs to better take into account language-particular phonetics. b) They demonstrably show the limitations that are inherent when investigations rely solely on impressionistic transcriptions, as the differences they note here are quite subtle. c) Based on the evidence presented here, an adequate model understanding of the data requires a model of implementation that either a) incorporates "finely grained features" into the phonology; or b) follows the 'standard model" with the caveat that implementation is informed by URs. d) It is possible that they have located evidence of near merger in the data, and thus, they may add to our general understanding of near mergers. 1) On page 6, G & H note that "while a few tokens were in fact read without aspiration [...] speakers were in general highly comfortable with the task" while completing the first experiment task. This is used to argue that the speech used in the experiment approximates more closely the language they would use "while talking with friends and family at home" (6). What behaviors did they actually exhibit that lead them to conclude that the speakers were this comfortable? They note that the frame was used during the experiment mimicked "casual speech," but how the naturalness of the actual production elicited from the speaker gauged? In other words, were attempts made to capture examples of each speakers "causal speech" in actual casual conversations, so that use of a more casual register could be more explicitly verified, or was only the experimental data obtained? 2) Has the hypothesis that longer duration of aspiration in the /kasta/ context may result from "subtle timing differences related to the respective targets for glottal width, corresponding with underlying /s/ versus the underlying /p/" (28) been explicitly tested in experiments conducted since G & H (in submission)? Since it seems so promising for explaining the trends noted in the study, it would be wonderful to see if it actually "pans out" this way. 3) Have phoneticians devised ways to explicitly test if subjects are able to reference something like this production behavior: "the quality of the segment from which the general target of [-voice] originated [...] and adjust the glottal width accordingly" (31)? Since this is a core component in G & H's exploration of implications, it seems like this would be a really useful thing to get at, so the power of their argument is reinforced more strongly. 4) Has anyone tested the impact of literacy on the reinforcement of URs (cf 37) for the forms explored here? If they have, what did they find that literacy has this impact? 5) In the conclusion section, G & H discuss the subtle differences they note in production might indicate that we are observing near mergers here. During the experiments, was perception of the subtle differences in production tested, a la Labov (1994)? If so, was it found that the speakers do in fact produce, but do not perceive the differences? Steve Fondow Gerfen and Hall, 2001 The discussion involves differences in the aspiration/gemination of word- internal, coda final obstruents (/p,k/ vs. /s/) and elaborates on it through analysis of a study conducted on a group of speakers of Eastern Andalusian Spanish. It is shown that there are very slight phonetic differences which are consistent across the contrast indicated above. The statistics from the analyzed data indicate that of the various features under investigation (preceding V duration, aspiration duration and stop closure duration), the preceding vowel duration is not a determining factor in distinguishing coda /p/ from /s/, while, on the other hand, the duration of aspiration is normally longer with /s/ than with /p/. Likewise, there is a difference in the duration stop closure, namely, that the closure is longer with /p/ than with /s/. These two pieces of evidence are logical in that the first maintains the fricative nature of the /s/ and the second, the closure feature of the /p/. A second investigation was done to reconstruct the outcome of the first using the contrast of /s/ and /k/. Similar results were found. These results seem to indicate a finely defined contrast between these features that may or may not be perceived. It highlights the problems with transcription practices as lacking the necessary precision to indicate all important contrasts in a language. Further discussion suggests that there is an important interplay between the phonology of the system and its phonetics, in that it is necessary for one to have access to the either. Numerous types of evidence indicate the need for URs which maintain the contrasts /s/ and /p/ or /s/ and /k/ (morphology, borrowings, orthography?). Questions 1. Is the aspiration from glottal constriction or the approximation of articulators in a voiceless context? In the latter, is it possible that the articulators involved differed? (dental /s/ vs. bilabial /p/). This leads me to wonder why it should be glottal width targeting and aspiration/gemination duration targeting instead of simply a change in the way in which these sounds are produced and, therefore, represented in output. 2. Although there is an observable difference in the aspiration effects, is it perceived? That is important for the discussion of section 4 about discrete phonetic details in grammar. 3. If the URs do not contain /s/ and /p/, as hypothesized on pg. 32, why would careful speech permit speakers the possibility of realizing these sounds? Notes on Gerfen and Hall (2001): "Coda aspiration and incomplete neutralisation in Eastern Andalusian Spanish" Summary In this paper, we found an overall statistical trend across EAS speakers to show a phonetic difference in coda aspiration stemming from what seems to be an underlying /s/ and coda aspiration stemming from what seems to be an undelrying /p/ or /t/. We claimed that this tendency was evidence that either (1) phonology is not categorical and "aspiration" is not a binary process or (2) that phonology is categorical but phonetics does not feed directly from the output of phonology without being able to see any of the original input to the phonology. Important points 1. Underlying /s/ had more aspiration while underlying /t/ and /p/ had longer consonant duration in the phonetic output forms. However, the overall duration of aspiration + consonant closure remained constant. 2. Not all speakers showed the same pattern to the same extent. The genralizations made here are extrapolations about the population as a whole and do not apply to every individual speaker. 3. It is assumed that phonology has at least two levels, an underlying form and a surface representation, that can be distinguished from each other. Questions 1. Why did we divide it into two experiments? 2. Does the differing semantic information play a role in the distinction of the various forms? 3. We seem to assume that all aspirated forms must be treated the same way: why couldn't those that surface with an alternation, e.g. with word-final [d] except when a suffix is attached, have underlying /d/ while those that are aspirated always have underlying aspiration? 3. Does there have to be an UR at all? A new proposition: what if in the lexicon, we simply have finely-detailed phonetic forms (where could I possibly be going with this?). These are associated with various tags, like "this word means X." Hence for bidialectals, we have exemplars of both [kasta] and [kahtta] in the lexicon, but they are linked with the same semantic tagging. Then, when the speaker goes to produce the word, they discover that it is linked with this other word that has an [s]. Adopting Matt Goldrick's theory of a "cascading" system from phonology to articulation, it could then be simply the case that the associated [s] is realized as greater glottal width or aspiration when it is produced (Goldrick found that people for example produced the nonsense token [dhEz] with longer aspiration than they usually do for [d] when they are saying a tongue twister and are intending to say [thez]. I think this lets us know that words associated with produced words can have an effect on those produced words, even when there is no grounds whatsoever that there is some sort of mutual UR. 4. Issue of orthography -- how practical is this? Eunjong Summary : Gerfen & Hall Coda aspiration and incomplete neutralisation in Eastern Andalusian Spanish 1. The phonetic realizations of the aspiration of coda obstruents in EAS show the differences in the duration of aspiration and stop closure, depending on the underlying obstruents. 2. The production experiments indicates that two distinct underlying representations result in phonetic differences even in phonologically neutralized tokens: - Speakers know the underlying form of aspirated coda obstruents based on their knowledge of alternations. - Spellings may hint speakers? recognizing the underlying form differences of give tokens. 2-1)Question: What if phonological features of underlying forms are not (phonetically) corresponding to the phonetic realization (because underlying forms are so abstract)? It seems that the relation between fricatives (underlying forms)- aspiration (phonetic realization), stop (underlying forms) - closure (phonetic realization) is so direct and possible. 3. The results propose the problem of the split view of systems of Phonology and Phonetics: the source of phonetics is not restricted to the output from phonology but comes from underlying representation. 3-1) Question: What does ?input? mean in figure 13? Can we interpret it as ?underlying form? 4. Question: In measuring duration, did they consider the whole duration of word? (Relative portion of aspiration duration in whole word?..etc). Still, it is interesting that the duration of aspiration+closure in each token is not significantly different (p29). Fangfang This study examines the neutralization of syllable-final s-aspiration and stop-aspiration in EAS. Three acoustic parameters are selected: the duration of the voicing part of the vowel, the duration of the aspiration part of the vowel, and the duration of the closure. It has been found that the latter two are significantly different for s-aspiration vs. stop-aspiration with s-aspiration having consistently longer duration of aspiration and shorter duration of closure. It argues then the neutralization is not complete, and the fine-grained phonetic contrast still exists. Questions: 1. Even with the statistical significant result, we can still see some variation across speakers. How to account for the speaker variability then in the domain of synchronic phonology? 2. Does this paper argue that the phonetic difference results in different Urs? In other words, does the incomplete neutralization in the production results in the incomplete merge of the two underlying representations? David Alexander Gerfen and Hall 2001 examines neutralization of segments in syllable final word internal position in Eastern Andalusian Spanish spoken near the city of Granada. Such consonantal weakening in coda position is related to the aspiration of /s/ across Spanish dialects which typically affects the plural morpheme s: las mesas > lah mesah. Here the aspiration serves as a cue to plurality in lue of the fricative s. Similarly, the development of breathly voiced geminates in the Spanish varieted in question has been described as a way of maintaining a closed versus open syllable distinction. Take the minimal pair casta vs caspa: in this instance a voiceless obsruent (both /s/ and /p/) is aspirated in coda position and then undergoes a process of gemination giving: [kaht.ta]. Although this transcription appears to show a complete neutralization between the minimal pair in question (casta vs caspa) according to Gerfen and Hall 2001 one must consider the length of the vowel, the aspiration, and the geminate consonant in order to determine if the neutralization in coda position is in fact complete. In their account the underlying fricative /s/ in words of type casta is more aspirated while the underlying plosive /p/ in words like caspa is more geminated in nature. Therefore the neutralization is said to be incomplete. 1) Given the phonetic data that indicates a qualitative difference between syllable final fricatives like /s/ and plosives of type /p/ in what ways does this difference constitute a near merger in terms of Labov 1994? That is to say, do speakers actually perceive the difference (i.e the length of the aspiration vs the strength of the geminate segment)? 2) In what way can phonetic transcriptions be improved in order give an account of the articulatory differences as revealed in spectrogram analysis? In short, what level of accuracy can be expected from such transcriptions? 3) Diachronically speaking, how does the formation of geminates of type [kaht.ta] sean in EAS for casta and casta mirror similar processes in other languages such as Italian CONCEPTU > concetto, PERFECTU > perfetto?