Enhancing accountability for the use of public sector resources ...
10
pages
English
Documents
Le téléchargement nécessite un accès à la bibliothèque YouScribe Tout savoir sur nos offres
10
pages
English
Documents
Le téléchargement nécessite un accès à la bibliothèque YouScribe Tout savoir sur nos offres
- cours - matière : economics
- exposé
- cours - matière potentielle : learning
- commonwealth countries
- pac performance
- sai of government-resource use
- public audit
- pac effectiveness
- financial accountability
- formal role
- executive
- government
Outline of today’s lecture (20/07/2004) I. Introduction to Prosody II. Dupoux, Pallier, Sebastian & Mehler (1997): “A Destressing Deafness in French?” III. What did we learn in this course? I. Introduction to Prosody 1. What is Stress? •phonetic prominence WAter aGAINFAther aBOUT •Different ways of indicating stress: [fár] ['far] [FAr] •Acoustic Correlates of Stress: (1)pitch (fundamental frequency; F0): High(er) (2)duration: Long(er) (3)amplitude (intensity; loudness): Great(er) •Unstressed syllables typically have reduced vowels (typically reduced to schwa) in English. •Consider the following words: raider radar •Both words have primary/main stress on the initial syllable; but inradar,there is secondary stress on the final syllable (that is, the final syllable inradaralso has some prominence compared to the final syllable ofraider, but it is not as prominent as the first syllables of these two words). For that reason, the vowel in the final syllable inradardoes not reduce to schwa. 2. Free Stress Languages vs. Fixed Stress Languages a. Free Stress Languages •no purely phonological constraints on where to put the stress; •depends on the morphological class or the type of suffix
1
•Note: “free” does not mean that stress assignment in random. That is, it is not the case that you can place stress wherever you want in “free” stress languages. Example: Russian múka : torment muká: flour oknó : window okná : window (gen. sg.) ókna : window (nom. pl) óknam : window (dat. pl.) b. Fixed Stress Languages •Stress always falls on a particular syllable. Example: Turkish Stress the last syllable of a word taní : to know tan -dík : acquaintance tan -d k-lár : acquaintances tan -d k-lar-ímmy acquaintances : tan -d k-lar- m-íz : our acquaintances tanõ-dõk-lar-õm-õfrom our acquaintancesz-dán : Other Examples: Stress the first syllable of a word: Czech, Hungarian, Scots Gaelic Stress the penultimate Syllable: Polish, Swahili English? convert (v) vs. convert (n) She converts word files into pdf files. They are converts. record (v) vs. record (n) We record quality music. The record company called. produce (v) vs. produce (n) Italy and France produce great wine. We sell organic produce. Germanic system (fixed initial stress) + Romance System (weight sensitive rules)
2
3. Syllable Weight Heavy:The syllable contains a complex rhyme (i.e., it consists of a long vowel/diphtong/(complex) coda). Light:The syllable contains a simplex rhyme (i.e., it has just a vowel (also known as an open syllable)). Predictable Cases: 4. English Stress: Predictable? a. Noun Stress Rule: Stress the penultimate syllable if heavy. Otherwise, stress the antepenult. agenda aroma Delaware catastrophe discipline [o:] b. Verb Stress Rule: Stress the final syllable if heavy. Otherwise, stress the penultimate syllable. abide obey hurry marry rally Note: /r/ inhurryandmarryand /l/ inrallyare ambisyllabic. That is, these consonants both belong to the coda of the first syllable (because in English, stress syllables must be strong) as well as the onset of the second syllable (due to Onset Maximization). c. Exceptions: Nouns: •Both of the following words have heavy penultimate syllable but onlyagendaobeys the Noun Stress Rule: agenda vs. calendar •Bothspaghettiandcatamaranhave light penultimate syllable. Despite that, the penultimate syllables are stressed in these words.
3
•Note spaghetti does not have a geminatetin English. It is pronounced as [spa.gE.Di] spaghetti catamaran •The following words, although all of them are nouns, are stressed on the final syllable. machine police report (n) balloon Verbs: The following verbs have initial stress although they have heavy final syllables. order linger reckon d. English Stress and Morphemes Stress Neutral Suffixes: -ise, -ly random random-ise atom atom-ise happy happy-ly Stress Attracting Suffixes: -ette, -ese, -ee: kitchenkitchen-ette mother mother-ese employ employ-ee Stress Retracting Suffixes -ic, -ity, -al atom atom-ic electricelectric-ity parent parent-al 5. Rhythmical differences and the foot: Stress-timed vs. Syllable-timed languages (1) (a)Which cats chase mice
4
(b)Many voters wanted freedom (c) Competent managers readily innovatePhonetic duration: a=b=c (English) Phonetic duration: c>b>a (if you read these like the way you speak French, Japanese, Turkish or Polish) (2) This is also the case in the following examples (thanks to Prof. Frans Plank): (a) Ken’s hat (b) Kenny’s hat (c) Kennedy’s hat Phonetic duration: a=b=c (English) Other examples: (d)cat in a hat (e)cat nip(f)the cat sat on the mat (g)as a bug in a ruga.k.a. Morse-code languages vs. Machine-gun languages Morse-code languages (stress-timed languages) devote the same amount of time to produce each foot: (a), (b), (c) have 4 feet; (d) and (e) 2; (f) and (g) 3.
Σ
σσσ…. A foot has at least one prominent (stressed) syllable. Depending on which part of the foot the stress falls, we give feet a different name (following Greek poetry tradition): Trochaic:
5
Σ
σσ wa ter mo ther Iambic:
Σ
σσ re cord (v) de mand II. Dupoux, Pallier, Sebastian & Mehler (1997): “A Destressing Deafness in French?” a. Statement of the Problem: •Learning a language does not only involve acquiring its sound pattern. •Foreign accents do not only reflect segmental problems but also prosodic ones. •“Why is that even proficient French speakers produce anomalous prosodic structures?” b. Aims and Research Questions: •Do speakers of different languages become sensitized to different prosodic properties (along with different phonetic contrasts)? •If yes, how would this affect the perception of non-native stress patterns? c. Languages and their stress properties: Spanish:Penultimate stress (default), but, like in Russian, many words may have stress in other positions.
6
Examples: bébe ‘drink’ (present tense) bebé ‘baby’ tópo ‘mole’ topó ‘met’ French:No such minimal pairs can be found. Primary stress falls on the last full vowel of lexical words. Assumption:Stress needs to be specified lexically in at least some of the cases in Spanish, whereas in French it could be completely left unspecified. Expectations:Unlike Spanish speakers, French speakers will have a hard time detecting stress differences and they should be able to ignore stress. Experiment 1: Methodology •ABX discrimination task •A, B, and X are produced by three different native speakers of Dutch. •A B are different with respect to the location of primary stress, X is either A or B. •The words conform to the pattern: CVCVCV •CVcvcv vs. cvCVcv (1st-2nd) Contrast was either: bópelo vs. bopélo cvCVcv vs. cvcvCV (2nd-3rd) bopélo vs. bopeló •Words were constructed such that the segments occur in both Spanish and French. •“Listen to words in a foreign language and decide whether the third one is identical to the first or the second.” bópelo ----500 ms.---bopélo---500 ms.---bopélo---4000 ms. (DECISION) --- 1000 ms---métilo ----500 ms.---metílo---500 ms.---métilo---4000 ms. (DECISION) --- 1000 ms---Possible outcomes: If the default stress locations are perceived without a problem: Spanish: both conditions should be OK. French: only the 2nd-3rd condition should be OK. Results:
7
•French subjects made significantly more errors than Spanish subjects. •French subjects were slover than Spanish subjects. •A-responses yielded more errors than B-responses. •A-responses were slower than B-responses. •For Spanish subjects both conditions were OK (confirming the above prediction). •For French, however, both conditions were problematic (disconfirming the above prediction). •Thus, the difficulty that the French have with stress must be general. Experiment 2 •It could be that the Dutch vowels and consonants were a bit different to the French. •Find a sitatuion that should be easy for the French and difficult for the Spanish: Ignore stress!! •If the difficulty lies in the fact that the French don’t pay attention to stress, then they should be able to ignore it. Again an ABX task. The subjects were told that the accent is irrelevant. A B X bópelo sopélo bópelo (accent congruent) bópelo sopélo bopélo (accent incongruent) B is always different from A in terms of phoneme and stress. X is identical to A either in terms of both accent and segment (accent congruent), or in terms of just the segment (accent incongruent). Results: •The incongruent conditions revealed more errors than the congruent ones. •Again, A responses generated more errors than B responses. •Spanish subjects made more errors than French subjects. •Spanish subjects had longer reaction times than French subjects. •Judging from the reaction times, ignoring accents was also very difficult for the Spanish in the congruent condition. •The French can ignore irrelevant variation in stress patterns, but Spanish subjects cannot.
8
Experiment 4 Research Questions: •At which level do French subjects have difficulty with stress? •Can it be that they lost their sensitivity to stress? •Can it be that the difficulty arises because the French subjects are unable to represent and store stress patterns in memory? Methodology •AX (same speaker) Different condition: fídape fidápe fídape lídape Same condition: fídape fídape fidápe fidápe lídape lídape lidápe lidápe •ISI: a pure 2KHz tone of either 200 ms. or 2200 ms was inserted in between. Expectations: If the French cannot perceive the acoustic correlates of stress, they should have difficulty in the accent condition with both the short and the long ISI. If the problem has to do with a phonological encoding problem that involves a later processing stage, they should only have problems with the accent condition with the long ISI. Results: •The long ISI condition yielded more errors in the accent condition than the phoneme condition. •French speakers are able to detect the acoustic correlates of stress. Overall Conclusions:
9
•Spanish subjects are essentially the same in terms of early processingFrench and of stress (presumably at the level of acoustic/phonetic processing); however, in order for the stress information to be retained, it must be recoded into a more abstract level (presumably at the level of phonology).
•In languages with fixed stress patterns, stress plays no lexical role, therefore, it is not represented at a more abstract level.
10
