Questions & explanations
1. Why does the Rhythm Rule not apply in 'thirteen candles'?
In 'thirteen candles', the word 'candles' has stress on the first syllable 'can-', which is not directly next to the stressed '-teen' because there is a weak syllable in between? Actually, 'candles' is two syllables: 'can-' (stressed) and '-dles' (weak). The stressed '-teen' and 'can-' are adjacent? Wait, 'thirteen' ends with a stressed syllable, and 'candles' begins with a stressed syllable, so they are adjacent. The rule should apply. But if we consider that 'candles' has initial stress, the clash exists, so the rule would apply, shifting stress to 'thir-'. However, some speakers might not shift if the following word has a weak first syllable? Actually, 'candles' has a strong first syllable, so clash occurs. The question might be tricky: perhaps the rule doesn't apply because the following word has a weak first syllable? No. Let me correct: In 'thirteen candles', both stressed syllables are adjacent, so the rule applies. But maybe the answer is that it does apply? I need to be accurate. Actually, the Rhythm Rule typically applies when the following word has initial stress. So 'cand
2. Suppose we have constraints A, B, C. The current ranking is A >> B >> C. The observed winner violates B and satisfies A, C; a loser satisfies B but violates A. What does the algorithm do?
The algorithm checks the highest constraint where the winner and loser differ. The winner violates B, loser satisfies B. But first it looks at higher constraints: both satisfy A, so no difference. Then at B: winner violates, loser satisfies. So B is the constraint to demote. The algorithm demotes B to just below C (the lowest constraint that the loser violates? Actually, the loser violates A, which is higher than B. But the algorithm demotes B to below the highest constraint that the loser violates? Standard algorithm: demote the constraint to just below the lowest constraint that the loser violates? Wait, correct: demote to below the highest constraint that the loser violates? Actually, the standard demotion: find the highest constraint that the winner violates and the loser satisfies; demote it to just below the lowest constraint that the loser violates? Let's be precise. The algorithm: find the highest-ranked constraint that the winner violates and the loser satisfies; demote that constraint to immediately below the lowest-ranked constraint that the loser violates. Here, the loser
3. Give an example where OO-correspondence is more important than a markedness constraint that would simplify the form.
In English, the word 'bomb' has a silent /b/ in the singular, but in 'bombard' the /b/ is pronounced. OO-correspondence between 'bomb' and 'bombard' would want them to be similar, but they are not. Actually, a better example: in some languages, a base form might have a complex cluster that is simplified in isolation, but in derived forms the cluster reappears to be faithful to the base. For instance, in French, 'petit' [pəti] (small) and 'petite' [pətit] (feminine) show OO-correspondence: the feminine form keeps the final /t/ to be similar to the masculine? Wait, the masculine loses the /t/ due to a constraint, but the feminine keeps it because of OO-correspondence with the masculine? Actually, the masculine [pəti] is the base, and the feminine [pətit] is derived by adding a vowel, but the /t/ appears. OO-correspondence between the masculine and feminine would require the feminine to have the same segments as the masculine, but it adds a /t/. So OO-correspondence is violated. Hmm. Better example: in English, 'sign' [saɪn] and 'signal' [sɪgnəl] show OO-correspondence: the /g/ appears
4. Compare the grouping you would perceive for a sequence with loud-soft vs. soft-loud patterns.
For a loud-soft pattern, you perceive trochaic groups: each loud sound starts a group. For a soft-loud pattern, you perceive iambic groups: each loud sound ends a group. So loud-soft is grouped as (loud-soft), while soft-loud is grouped as (soft-loud). The law predicts that loudness differences lead to trochaic grouping, but if the pattern is soft-loud, the loud sound is at the end, so it becomes iambic? Actually, the law says that loudness differences favor trochaic grouping, meaning the louder sound is perceived as the beginning. So soft-loud would still be trochaic? Let me clarify: The law states that a sequence of alternating loudness is grouped trochaically, with the louder element first. So soft-loud would be perceived as (soft-loud) with the loud at the end? That would be iambic. But the law says loudness leads to trochaic, so the louder should be first. So if the pattern is soft-loud, the louder is second, so the grouping might be ambiguous. Actually, experimental findings show that listeners tend to group loud-soft as trochaic and soft-loud as iambic? I need to be accurate.
5. Give an example of two candidates where one is harmonically bounded by the other.
Consider constraints NOCODA and MAX. Candidate A: 'pat' (violates NOCODA once). Candidate B: 'pa' (violates MAX once). Candidate C: 'p' (violates MAX twice). Candidate C is harmonically bounded by A and B because it has more violations of MAX and also violates NOCODA if 'p' is a syllable? Actually, 'p' violates MAX twice and NOCODA? Let's think: 'p' as a syllable violates ONSET? Better example: For input /pat/, candidate 'pat' violates NOCODA; candidate 'pa' violates MAX; candidate 'p' violates MAX twice and NOCODA? No. Simpler: For input /pat/, candidate 'pat' violates NOCODA; candidate 'pa' violates MAX; candidate 'p' violates MAX twice and ONSET? Actually, 'p' as a syllable violates ONSET and MAX twice. So 'p' is harmonically bounded by 'pa' because 'pa' has fewer MAX violations and no ONSET violation? But 'pa' also violates ONSET? Let's use a clear example: constraints ONSET and DEP. Input /ata/. Candidate A: 'ata' (violates ONSET twice). Candidate B: 'ʔata' (violates DEP once, ONSET once). Candidate C: 'ʔaʔta' (violates DEP twice, ONSET once). Candidate C is harmonically bounded
6. Compare the foot structure of the words 'banana' and 'America'. How are they different?
The word 'banana' has three syllables: ba-NA-na. It can be analyzed as one iambic foot (ba-NA) plus an extra unstressed syllable 'na' that is 'unfooted' or part of a degenerate foot. The stress is on the second syllable. 'America' has four syllables: A-ME-ri-ca. It has a dactylic foot (stressed-unstressed-unstressed) on 'A-ME-ri' and then an unstressed 'ca'? Actually, 'America' has main stress on ME, so it might be a trochaic foot 'A-ME' (stressed-unstressed?) No, A is unstressed. Better: 'America' is often analyzed as having an iambic foot (a-ME) and then a trochaic foot (ri-ca)? The key difference is that 'banana' has a simple iambic pattern, while 'America' has a more complex structure with secondary stress on 'ri'? Actually, 'America' has no secondary stress? Let's say: 'banana' has one stressed syllable (NA), while 'America' has one main stress (ME) and possibly a secondary on 'A'? In many analyses, 'America' has primary stress on ME and secondary on A? The foot structure differs in the number of feet and the type of feet.
7. Compare the onsets of 'Stein' (stone) and 'Schtein' (non-word). Why is 'Schtein' not a possible German word?
'Stein' starts with /ʃt/ (scht), which is a valid onset. 'Schtein' would start with /ʃt/ plus /aɪn/? Actually 'Schtein' would be /ʃtaɪn/ which is fine? Wait, 'Schtein' is not a word, but /ʃt/ is allowed. The issue is that German does not allow /ʃt/ followed by a consonant? Actually 'Stein' is /ʃtaɪn/ with /ʃt/ onset. 'Schtein' would be /ʃt aɪn/ same? No, 'Schtein' would be /ʃt aɪn/ but the spelling 'scht' is unusual. The point: German allows /ʃt/ but not /ʃt/ plus another consonant? I need to correct: German allows /ʃt/ and /ʃp/ as onsets, but not /ʃt/ followed by another consonant like /ʃtr/ is allowed in 'Straße' (street) with /ʃtʁ/. So 'Schtein' might be possible? Actually 'Stein' is spelled with 'St', not 'Scht'. The rule: after short vowels, 'st' is pronounced /ʃt/; after long vowels, it's /st/. So 'Stein' has long vowel? Not sure. Let's give a clearer example: 'Pflicht' (duty) has onset /pfl/ which is complex. Compare 'Pflaume' (plum) vs 'Flaume' (non-word). The point is that German allows many consonant clusters.
8. If two rules apply to the same input, how does rule ordering affect the output? Give an example.
Rule ordering matters when rules can apply in sequence. For example, suppose we have Rule A: [-sonorant, -voice] → [+voice] / V __ V (intervocalic voicing), and Rule B: [+voice, -sonorant] → [-continuant] / # __ (word-initial stop hardening). If we apply A first, /saba/ (with /s/ voiced to [z]?) Actually, consider /apa/: A applies to /p/ → [b], then B does not apply because [b] is not word-initial. If we reverse order, B applies first but /p/ is voiceless, so no change; then A applies to /p/ → [b]. Same output? But if Rule A creates a voiced stop that then triggers another rule, ordering can differ. For a clear case: Rule A: V → [+nasal] / _ N (vowel nasalization before nasal consonant), Rule B: N → ∅ / _ # (nasal deletion word-finally). If A applies first, /pan/ → [pãn] → then B deletes /n/ → [pã]. If B applies first, /pan/ → [pa] (nasal deleted), then A cannot apply because no nasal consonant left, so output [pa]. Thus, ordering produces different outputs.
9. Compare how Government Phonology and traditional feature theory represent the feature [voice] in obstruents. Which approach is more restrictive?
In traditional feature theory, [voice] is a binary feature: [+voice] for voiced, [−voice] for voiceless. In Government Phonology, voicing is represented by the element |L| (low tone, slack vocal cords) for voiced obstruents, and |H| (high tone, stiff vocal cords) for voiceless obstruents. However, in Government Phonology, these elements are typically used for tone languages; for obstruents, voicing is often represented by the absence of |H| (voiceless) or presence of |L| (voiced). Government Phonology is more restrictive because it claims that only elements that are phonologically active in a language can be used. For example, if a language has no tone, |L| and |H| may not be available for voicing contrasts. This predicts that voicing contrasts are limited in tone languages, which matches some typological observations. Traditional feature theory allows [voice] universally, which is less restrictive.
10. Where is the stress in the Latin word 'imperator' (commander) using the weight rule?
The word 'imperator' has syllables: im-pe-ra-tor. The second-to-last syllable is 'ra' (short a, no coda) – light. So stress falls on the third-to-last syllable, which is 'pe' (short e, no coda) – also light? Wait, check: 'pe' is light. But the rule says stress the third-to-last if the second-to-last is light. So stress on 'pe'? Actually, 'imperator' has stress on 'pe'? No, correct stress is on 'ra'? Let's re-evaluate: In Latin, the penultimate (second-to-last) is 'to'? Syllables: im-pe-ra-tor. Penultimate is 'ra'? Actually, 'ra' is the third from end? Let's count: 'tor' is last, 'ra' is second-to-last, 'pe' is third-to-last. 'ra' has short a and no coda? 'ra' is open syllable, light. So stress on third-to-last 'pe'. But the actual Latin stress is on 'pe'? I think 'imperator' is stressed on 'pe' (im-PE-ra-tor). Yes, that matches the rule: penultimate light, so stress on antepenultimate.
11. Give an example where O-Contig is violated.
If the input is 'pat' and the output is 'pʔat' (inserting a glottal stop between /p/ and /a/), then the output segments /p/ and /ʔ/ are adjacent, but their input correspondents /p/ and /a/ are not adjacent? Actually, /p/ and /a/ are adjacent in the input, but /p/ and /ʔ/ are not. O-Contig requires that if two output segments are adjacent, their input correspondents must be adjacent. Here, /p/ and /ʔ/ are adjacent in output but their input correspondents /p/ and (nothing) are not adjacent because /ʔ/ has no input correspondent? Better: Insertion creates a segment with no input correspondent, so O-Contig is violated because the inserted segment breaks adjacency. For example, input 'pat' output 'pʔat': output segments /p/ and /ʔ/ are adjacent, but their input correspondents are /p/ and (none), so adjacency is broken. O-Contig penalizes internal insertion.
12. Compare how Clements' and Sagey's models represent the feature [spread glottis] for aspirated stops. Which model allows for independent specification of this feature across different places of articulation?
In Clements' model, [spread glottis] is under the Laryngeal node, which is a single node for all laryngeal features. This means that if a language has aspirated stops at different places, the feature [spread glottis] is specified once under the Laryngeal node, and it applies to all stops that share that node. In Sagey's model, the Laryngeal node can be specified independently for each place of articulation because the geometry allows for multiple Laryngeal nodes attached to different place nodes. For example, in Hindi, which contrasts aspirated and unaspirated stops at labial, dental, retroflex, and velar places, Sagey's model can represent each stop's aspiration independently, while Clements' would require a separate root node for each. Thus, Sagey's model offers more flexibility for languages with complex laryngeal contrasts.