Please post your comments here on Pater 2000/Kiparsky 2007.
3 thoughts on “English stress comments”
Presley
You argue against the idea that all exceptional stress is due to lexically indexed constraints by pointing out that this would miss the generalization that no sonorant-final syllables become stressed in a derived word. Could we do a chi-squared test or something like that to show that this is a strong enough generalization that we shouldn’t assume it’s due to chance?
Would it be logical to assume/is there proof that during acquisition the indexed constraints are re-ranked/created after the ranking of the “base” ones was established? I am thinking about the constraint such as *CLASH-HEAD-S2. Could it be a possible step for the children to produce [,ad][‘van]tage instead of ad[‘van]tage until this constraint is ranked properly?
It’s extremely interesting to see how English secondary stress has narrowly delimited bins for idiosyncratic variation, with idiosyncracy stemming from either lexical entries or “cyclic” effects. This complicated character of secondary stress, combined with the fact that secondary stress appears to map (almost) one-to-one with full vowels outside of main stress, is a temptation to try considering a segmental account of the facts, where it is assumed that there are metrical conditions on full and reduced vowels, but the idiosyncracy stems from the segmental representations of morphemes (similarly to treatments of vowel reduction in Dutch, see Van Oostendorp 1995, 1997).
In connection to this, there is an intriguing study that I just heard of today in Alexandra Jesse’s class on speech perception. When native speakers of English are presented with parts of words that share a segmental substring, but have different stress placement in that substring (for instance, MUsic versus muSEum; ADmiral vs. admiRAtion), they will not use the the stress information in “MU” versus “mu” and “ADmi” versus “admi” to disambiguate between the two words before the words start diverging segmentally. When Dutch L2 speakers are presented with the same stimuli, they will make use of the stress information, and will be able to disambiguate the words before they start diverging segmentally.
This study, although it’s not easy to interpret these results consistently, it least suggests that ‘mu’seum and ‘music, and ‘admiral and ‘admi’ration may both be underlyingly represented with a stress mark (to allow the full vowel to surface), so that native speakers of English would not be able to distinguish between the abstract representations of the shared substrings of the word pairs (‘mus- vs. ‘mus-; ‘admi- vs. ‘admi-). Dutch speakers, in contrast, would have a different system of computing secondary stress, one that may be more based on vowel quality rather than underlying stress marks, which may explain their sensitivity to the stress cues.
The reference for the study is:
Cooper, N., Cutler, A., & Wales, R. (2002). Constraints of lexical stress on lexical access in English: Evidence from native and non-native listeners. Language and Speech, 45(3), 207-228.
I haven’t actually read it yet, so I might be misrepresenting it.
You argue against the idea that all exceptional stress is due to lexically indexed constraints by pointing out that this would miss the generalization that no sonorant-final syllables become stressed in a derived word. Could we do a chi-squared test or something like that to show that this is a strong enough generalization that we shouldn’t assume it’s due to chance?
Would it be logical to assume/is there proof that during acquisition the indexed constraints are re-ranked/created after the ranking of the “base” ones was established? I am thinking about the constraint such as *CLASH-HEAD-S2. Could it be a possible step for the children to produce [,ad][‘van]tage instead of ad[‘van]tage until this constraint is ranked properly?
It’s extremely interesting to see how English secondary stress has narrowly delimited bins for idiosyncratic variation, with idiosyncracy stemming from either lexical entries or “cyclic” effects. This complicated character of secondary stress, combined with the fact that secondary stress appears to map (almost) one-to-one with full vowels outside of main stress, is a temptation to try considering a segmental account of the facts, where it is assumed that there are metrical conditions on full and reduced vowels, but the idiosyncracy stems from the segmental representations of morphemes (similarly to treatments of vowel reduction in Dutch, see Van Oostendorp 1995, 1997).
In connection to this, there is an intriguing study that I just heard of today in Alexandra Jesse’s class on speech perception. When native speakers of English are presented with parts of words that share a segmental substring, but have different stress placement in that substring (for instance, MUsic versus muSEum; ADmiral vs. admiRAtion), they will not use the the stress information in “MU” versus “mu” and “ADmi” versus “admi” to disambiguate between the two words before the words start diverging segmentally. When Dutch L2 speakers are presented with the same stimuli, they will make use of the stress information, and will be able to disambiguate the words before they start diverging segmentally.
This study, although it’s not easy to interpret these results consistently, it least suggests that ‘mu’seum and ‘music, and ‘admiral and ‘admi’ration may both be underlyingly represented with a stress mark (to allow the full vowel to surface), so that native speakers of English would not be able to distinguish between the abstract representations of the shared substrings of the word pairs (‘mus- vs. ‘mus-; ‘admi- vs. ‘admi-). Dutch speakers, in contrast, would have a different system of computing secondary stress, one that may be more based on vowel quality rather than underlying stress marks, which may explain their sensitivity to the stress cues.
The reference for the study is:
Cooper, N., Cutler, A., & Wales, R. (2002). Constraints of lexical stress on lexical access in English: Evidence from native and non-native listeners. Language and Speech, 45(3), 207-228.
I haven’t actually read it yet, so I might be misrepresenting it.