Oct 25 ====== General ------- A major part of coarticulation seems to be dependent on how the listener comprehends speech, that they are "filling in the blanks" with more predictable knowledge. I assume this is possible because both listener and speaker communicate in the same language. To isolate the perception part of it, we make use of apparatus to analyse just the sound signals. But what if we could find an alternate way to transcribe speech into words without the recording device being aware of the language of the speech? Kühnert/Nolan:1997/1999 ----------------------- "English orthography has of course strayed somewhat from a phonemic analysis" -> "Somewhat" is a bit of an understatement! This does make me wonder if there are any orthographies that consistently incorporate coarticulation information the way we do for only a few words like "impossible". "Coarticulation at the phonological level is caused by rules of feature spreading along the lines discussed in recent developments in nonlinear phonology" -> Can we discuss how the difference between "linear" and "nonlinear" phonology is relevant for coarticulation? Can certain feature spreading only be accounted for if we assume a nonlinear phonological model? "In addition, doubts have been raised over the role of coordinative structures in speech production, and the assumption that the input to the physical speech mechanism is a contrastive phonological representation (see, for example, the discussions in Nolan, 1982; Kent, 1983, 1986; Shaffer, 1984)" -> Similarly, to the last question: what exactly does it mean for the phonological representation to be "contrastive" and why is this assumption important here? "Speaker-specific behaviour can not only be observed during the process of speech acquisition, but to a certain extent also in the coarticulation strategies of adult speakers." -> The fact that there is so much variation even among speakers makes me wonder if this perceivable. Does a person who consistently "coarticulates a lot" (e.g., by allowing features to spread farther) sound any different to a listener, even if they can't point out what sounds different? Repp argues that the various phenomena grouped under the term coarticulation have diverse origins and roles in speech development. What criteria could be used to distinguish them into more meaningful categories? Which categories would that be? It has been suggested that coarticulation enhances intelligibility. Given the coarticulatory variability identified in the literature, even within individual speakers, how can these two aspects be reconciled? In theory, one would expect that interindividual variability would hinder intelligibility because the listener may not know what to expect. Yet, this issue does not seem to arise. Why? The paper states that some studies suggest that children initially exhibit more coarticulation than adults, while others indicate that they are more segmental in their speech. What might explain these conflicting findings? Are there any differences regarding the experimental setup of the conducted studies? How can we describe speech as continuous while linguistic theory posits that phonemes are stored as discrete units in our minds? Is this contrast related to the physical constraints of the human vocal tract? If so, what can we say about the languages with many phonemes (more than 100), such as some African languages with click sounds, that use the full range of human vocal capabilities? In Section 3, they discuss if there is a variation of coarticulation between kids, with conflicted results about the question if those differences correlate with the speakers age. At what age is it suggested that a person has "worked out" his coarticulatory behaviour? Or is it a more dynamic thing, which can slightly change throughout your life? When Sievers (1876) mentions that every sound has a "static positional phase", does this apply to only the dominant articulator or the entire speech mechanism? So one of the fundamental differences between Coarticulation and Steuerung is that the former has to do with interaction between sounds using different articulators, whereas the later describes behaviour of the same articulator making two different sounds. Is that correct? Braun/Möbius:2023 ----------------- "Conversely, processing is impaired when the realization of a unit is inadequate for its context, i.e., either a reduced form in a non-predictive context or a non-reduced, or even hyperarticulated, form when it is highly predictable." -> I'm surprised that hyperarticulation can also result in impaired processing! This does make the situation nicely symmetrical, though. In section 5, it is mentioned that language processing is impaired if an utterance is hyperarticulated in a predictable context. Why is that? I.e., how can an overly precise articulation lead to processing difficulties? "Menzerath believed that experiments, not complicated instruments, are at the core of modern phonetics." Does the innovation and technological evolution of scientific instruments not play a key role in advancing the quality, accuracy and with that, the validity of results of experiments in such a scientific field? And how exactly did Menzeraths experiment designs differ from how other scientists in that era conducted their experiements? "Menzerath and de Lacerda inferred that no stable phase exists within speech sounds in continuous speech" (p22), what is meant here by "stable phase"? A duration of a certain length? A lack of other movement of the articulators? I'm also curious as to whether de Lacerda and Menzerath investigated any languages besides German and Portuguese as well as other sequences of sounds besides VCV. It can be inferred that coarticulation is not just a by-product or a negligible aspect of speech production, since speech would be intelligible without it. Considering Menzerath's demonstration of 'Lautabgrenzung', it was later established that this kind of allophone synthesis cannot produce intelligible speech. I wonder if it is so because of the mismatched units of speech, and that if a similar word synthesis demonstration was carried out with individually recorded speech units, it would sound more intelligible?