BBN-ANG-243 Advanced Phonology: Phonological Analysis Lecture 7: - - PowerPoint PPT Presentation

Page 1 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

BBN-ANG-243 Advanced Phonology: Phonological Analysis

Lecture 7: Word Stress part 1 Annotated version: see the Yellow Pages after sets of slides

Kiss Zoltán / Starcevic Attila / Szigetvári Péter / Törkenczy Miklós

Dept of English Linguistics, Eötvös Loránd University

(1) Phonetic properties of stress : suprasegmental & relative compare [coronal] j H r * * * * * grid representation of prominence kiss kisses syllable prominence due to loudness, pitch and/or length

Page 2 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

THESE NOTES (THE YELLOW PAGES) ARE ANNOTATIONS EXPLAINING (LIKE THE ACTUAL LECTURE WOULD HAVE DONE VERBALLY) THE SLIDES THAT PRECEDE. THE NUMBERS IN BRACKETS ( ) REFER TO THE SECTION NUMBERS OF THE SLIDES. These two lectures are about word stress, i.e. the stressing of words spoken in isolation as if each word were a “one-word sentence” (not when words are combined into phrases or sentences – phrase stress will be discussed in the set text on phrase stress/intonation). (1) Stress is a suprasegmental feature. Unlike the features like [coronal] or [voice] it not realised

• n a single segment, but it extends over more than one segment: it is associated with a syllable.

Furthermore stress is not an absolute property: it is the relative prominence of syllables. In contrast to features like [voice] or [coronal], whose value is determinable independently of the environment of the segment, it is not possible to tell whether a particular syllable is stressed

• r unstressed without comparing it to other (neighbouring) syllables.

Phonetically, stress (prominence) does not have a single unique phonetic correlate but is associated with (a combination of) more than one measurable phonetic property: increased loudness (amplitude, the amount of air pushed out of the lungs), pitch (fundamental frequency, the rate of vibration of the vocal cords) and increased length.

Page 3 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2) Word stress (2.1) Degrees (1 stress vs more than 1 stress in a word) ! Hungarian * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * matek matekos matekozik matekozhatnak matekozhatnának átmatekozhatnának ! English * * * * * * * * * * * * * * * * * * * * atom atomic atomistic excommunicate `s?l ?sNlHj `s?lHrsHj Djrj?li|vmHjDis Notation spelling átom atómic àtomístic èxcommúnicate transcription !`s?l ?!sNlHj $`s?!lHrsHj $Djrj?l!i|vmHjDis átcm ct 3 ]mwk àtcmístwk 4 ekskcmj 3 œwnwkejt

Page 4 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2.2) Predictability (fully predictable vs. fully lexical) Hungarian: leftmost syllable ! Macedonian: antepenultimate syllable FULLY PREDICTABLE & FIXED Polish: penultimate syllable ! classical Arabic: FULLY PREDICTABLE & NOT FIXED ! Russian, Italian, Spanish NOT (FULLY) PREDICTABLE & NOT FIXED (partially) LEXICAL

Page 5 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2.1) Some languages have just one stress in a word (i.e. two degrees of stress: stress versus no stress). Hungarian is an example where each word has an initial stressed syllable – otherwise the word is “flat” stress-wise independently of how long it is. English is different because more than one syllable can be stressed in a word, there are degrees of prominence. Notation: stress in English can be indicated before the first segment of the syllable (in transcription, superscript=1ry stress, subscript=2ry stress) or on top of the stressed vowel (in spelling and transcription, acute accent=1ry stress, grave accent=2ry stress). (2.2) In some languages stress is predictable (from some property of the of the stressed syllable, e.g. its position in the word). In others it is not predictable, it must be specified (memorised by the speaker) for every word (or some/most words). Unpredictable stress is called lexical stress. In some languages stress always falls on the same syllable of a word: fixed stress (e.g.

• Hungarian. In other languages (e.g. English) any syllable of a word may be stressed (not

fixed). If stress if fixed, it is predictable, but it may or may not be predictable if not fixed.

Page 6 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2.3) Stress & quantity/syllable weight (quantity sensitivity) (i) syllable weight: heavy, superheavy, light (VL=long V, D=diphthong, VS=short V, "." = syllable boundary;

• pen syllable ends is V, closed syllable ends in C)

HEAVY SYLLABLE: H LIGHT SYLLABLE: L

(contains a long V or diphthong or is closed) (contains a short V and is open) VL(C). or D(C). or VSC(C). V.

SUPERHEAVY (HSH)

HL vi.sa uHi-y? HH aw.ful n9-e?k HH ban.quet a`M-jvHs HSHH an.gel Dim-cY?k HSH flaunt ekn9ms HSH act @js LL A.bba `-a? LH a.crid `-jqHc (ii) Two interpretations of diphthong a. D = VL ne.on mHi - ?m HH b. D = VSC ne.on mH - i?m LH

Page 7 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(iii) Quantity-sensitive system, example: Classical Arabic

• a. !j` s` <a`>

'he wrote' b. jh !s`9 <atm> 'book (nom sg)' L L L H !a` k` í` <stm> 'date (nom sg)' it !R`9 qh <jt> 'he participates' L L L L H L l` m`9 !ch9 <kt> 'kingdom (nom sg)' L H H Ignore last syllable; stress rightmost H if there is one, otherwise stress leftmost (L) syllable (iv) Quantity-insensitive system, example: Hungarian paripa !L L L Barbara !H L L karácsony !L H L Melinda !L H L Abigél !L L H Aladár !L L H (not the same as the pattern of Hungarian metrical verse: taka!rá el a !bérci te!tőt)

Page 8 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2.3) Metrification (i.e. the assignment of stress) is sensitive to the distinction between heavy and light syllables in some languages. These stress systems are quantity-sensitive. (i) a syllable is light (L) if it is open (=ends in a vowel) and the vowel is short (=it is not a long vowel or a diphthong a syllable is heavy (H) otherwise, i.e. if the syllable is closed (=ends in a consonant) and/or its vowel is a long vowel or a diphthong (see the table in (i)) A subtype of a heavy syllable is superheavy (HSH). These are syllables which are not minimally heavy, i.e. they are heavy even without their final consonant L see the table (2.3) i. (ii) Diphthongs may be analysed in two ways: either as a long vowel or as a short vowel followed by a consonant (a glide). This has a consequence for the weight of the syllable they are in: under the first interpretation it is heavy (since it counts as a long vowel), under the second it is light (since it is an open syllable with a short vowel (because the consonant (the glide) syllabifies into the next syllable due to Onset Maximisation, see Lectures 4-5 on syllable structure). We will follow the first interpretation Lsee the syllabification and the syllable weights in the box with the red border (iii) In quantity sensitive systems heavy syllables tend to attract stress, Lsee the Classical Arabic examples (iv) Hungarian is quantity-insensitive: syllable weight has no role in metrification. Lsee the Hungarian examples Hungarian metrical verse is different: it is quantity sensitive. Lsee the Petőfi quotation

Page 9 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2.4) Stress & the domain of stress assignment (extrametricality) In a stress system with extrametricality some peripheral parts of the word are systematically outside the domain

• f metrification, e.g. last syllable in Classical Arabic:
• a. !j` s` <a`>

'he wrote' b. jh !s`9 <atm> 'book (nom sg)' L L L H !a` k` í` <stm> 'date (nom sg)' it !R`9 qh <jt> 'he participates' L L L L H L l` m`9 !ch9 <kt> 'kingdom (nom sg)' L H H

Page 10 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2.4) Metrification (i.e. the assignment of stress) can systematically ignore some part of a word (typically at the end). This part is called extrametrical and it is “invisible” to metrification. LAs you can see in the Classical Arabic example, the final syllable is extrametrical in that it cannot get stress and does not influence the place of stress within the word.

Page 11 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2.5) Stress & vowel quality (vowel reduction: full vs. reduced vowels) In a stress system with Vowel Reduction unstressed syllables have a limited set of vowels (called ‘reduced vowels’), i.e. fewer vowel contrasts, compared to stressed syllables (e.g. Catalan, Russian).

Page 12 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2.5.1) Inventory: Trad BrE (‘old school’) vs. Current BrE Lexical set Trad BrE CuBE NEAR

H? H9

SQUARE

d? D9

START/PALM/BATH

@9 @9

NURSE

29 ?9

FORCE/NORTH/THOUGHT

N9 n9

CURE [palatal] __

T? 79

POOR

n9

Lexical set Trad BrE CuBE KIT

H H

DRESS

d D

TRAP

z `

STRUT

U U

LOT

P N

FOOT

T 7

Lexical set Trad BrE CuBE FLEECE

h9 Hi

FACE

dH Di

PRICE

`H @i

CHOICE

NH ni

MOUTH

`T `v

GOAT

?T ?v

GOOSE

t9 |v

Page 13 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2.5.2) Vowel reduction in English In English vowel quality and stress are interdependent, there are two sets of vowels: full vs. reduced. (i) stressed syllables have full vowels and (ii) there is a strong tendency for reduced vowels to occur in unstressed syllables (= vowel reduction). parámeter Jàpanése Zo?!q`l?s?\ Z$cY`o?!mHiy\ Therefore: a reduced vowel can alternate with any full one á t o m ~ a t ó m i c a c á d e m y ~ à c a d é m i c é d i t ~ e d í t io n | | | | | | | | | | | | | | ` ? ? N ? ` ? ` ? D D ? H H F R R F R F R F R F F R R F . However: while stressed syllables must have full vowels, unstressed syllables can have full vowels too. October ambassador róbot órgasm Nj!s?va? `l!a`r?c? !q?vaNs !n9f`y?l variation direct c@i!qDjs % cH!qDjs % c?!qDjs

Page 14 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2.5.3) FULL vs. REDUCED: Overlap between full and reduced vowels

FACE PRICE CHOICE MOUTH GOAT NEAR START CURE FLEECE

/

HAPPY SQUARE NURSE NORTH GOOSE FOOT COMMA

/

LETTER DRESS TRAP STRUT LOT

Page 15 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2.5.4) A possible alternative analysis of the vowel inventory If we analyse the vowel of STRUT ZU\ and the vowel of COMMA/LETTER Z?\ as allophones of the same vowel phoneme, then the reduced vowels are a subset of the full ones

FACE PRICE CHOICE MOUTH GOAT NEAR START CURE FLEECE

/

HAPPY SQUARE NURSE NORTH STRUT

/

COMMA

/

LETTER DRESS TRAP GOOSE FOOT LOT

Page 16 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2.5) In a language with vowel reduction there is a connection between stress and vowel quality:

• nly a subset of the inventory of vowels (see the TradBrE and the CuBE inventories in

(2.5.1), each vowel is given with the name of the lexical set of words in which it occurs) can

• r tend to occur in unstressed syllables. There is no connection between stress and vowel

quality in languages without vowel reduction: any vowel of the inventory can occur in any syllable (e.g. Hungarian). (2.5.2) English has vowel reduction: there is a strong tendency for a subset of vowels, called reduced vowels to occur in unstressed syllables (as opposed to the full vowels that tend to

• ccur in stressed syllables.

LYou can see how the vowels change as the stress moves in the examples in the middle of p.13 In English vowel reduction is a tendency: while stressed syllables must have full vowels, unstressed syllables can have full vowels too, in some cases there’s even variation. LSee the vowels highlighted in red in the examples at the bottom of p.13 (2.5.3) A further complication in the English system is that there is an overlap between full and reduced vowels: some vowels belong to both the reduced and the full set (GOAT,

FLEECE/HAPPY, GOOSE, FOOT) while some a full only and one is reduced only

(COMMA/LETTER). LSee the consequence of this overlap in (2.6.2) (2.5.3) In an alternative analysis of English vowels where ZU\ and Z?\ are analysed as allophones

• f the same phoneme, STRUT and COMMA/LETTER would have the same vowel phoneme and

thus there would be no reduced only vowels. We will not pursue this analysis in this lecture.

Page 17 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2.6) Unstressed syllables with full vowels: “TERTIARY stress” 2 0 1 0 3 (2.6.1) Notation èxcommúnicate $Djrj?!li|vmHjDis "3ry" stress is not indicated explicitly in transcription: full vowel without stress mark (2.6.2) Indeterminacy Since there is an overlap between full vowels & reduced vowels, sometimes it is not possible to tell if a syllable has a full vowel or a reduced vowel, i.e. whether it is “3ry stressed” or unstressed vomit Z!uNlHs\ 1 0 or 1 3 ambush Z!`la7R\ 1 0 or 1 3 (format Z!en9l`s\ 1 3) F ? F ? F F (2.6.3) Other analyses Some other analyses do not distinguish 3ry stress. For them full V = stress. See later in (2.6.5)

Page 18 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2.6.4) The predictability of tertiary stress (i) If you know that a syllable is unstressed, how can you tell if its vowel will be reduced or full (=3ry)? You cannot: it is mainly (mainly) lexical (a) lexical/arbitrary fórmat Z!en9l`s\ clímate Z!jk@il?s\ páradise Z!o`q?c@ir\ cháracter Z!j`q?js?\ (b) morphological gráduateV Z!fq`cY|vDis\ gráduateN Z!fqzcYT?s\ (c) phonological Closed Syllable Tendency

• pen initial syllable

closed initial syllable A.mérica )`-!lD, ?-!lD, Mon.tána lNm-!s`, a.trócious )`-!sq?v, ?-!sq?v, Oc.tóber Nj-!s?v,  No.vémber m?v-!uD, m?-!uD, con.trástV )jNm-!sq@9rs j?m-!sq@9rs

Page 19 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(ii) If you know that a syllable is has a full vowel, how can you tell if it is stressed or unstressed (1ry/2ry vs. 3ry)? a stressed syllable in a word may be a rhythmically stressed/strong syllable in a sentence (and may also lose it's rhythmical prominence under certain conditions due to the Rhythm Rule). In the sentence a stress can be reduced to 3ry as a result of the Rhythm Rule (Rhythmic Stress Deletion, Stress Shift), but 3ry stress cannot be upgraded to become stressed for rhythmic reasons. (a) àcadémic She's an academic [AcadEmic] [2 1] $`j?!cDlHj – – Rhythm Rule It's an academic year [Academic] yEAr [2 3] 1 ávalanche It's an avalanche [Avalanche] [1 3] !`u?k@9mR It's an avalanche year [Avalanche] yEAr [2 3] 1 It's the first avalanche today [Avalanche] todAY 2 3 1 – *Rhythm Rule *[AvalAnche] todAY *2 2 1

Page 20 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(b) èverlásting The rain is everlasting [EverlAsting] [2 1] $Du?!k@9rsHM – – Rhythm Rule She shivered in the everlasting rain [Everlasting] rAIn [2 3] 1 Octóber It rains in October [octOber] [3 1] Nj!s?va? She shivered in the October rain [octOber] rAIn [3 2] – – *Rhythm Rule *[October] rAIn *[2 3]

Page 21 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2.6.5) Distribution: 3ry stress vs. 2ry stress (i) Rightmost stress in a word = primary (1ry) non-rightmost stress(es)= secondary (2ry) Let us assume we know where the last stress is. (ii) Distribution (a) 3ry & 2ry stress: complementary distribution: 2 σ before 1ry stress & after 1ry stress position 2 σ before 1ry stress after 1ry stress 2ry (=major) stress àcadémic ! 3ry stress ! róbot

Page 22 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(b) 3ry & 2ry stress: 1 σ before 1ry stress position 2 σ before 1ry stress 1 σ before 1ry stress after 1ry stress 2ry (=major) stress èverlásting àcadémic ? ! 3ry stress ! Octóber róbot Rhythm Rule sàrdíne $r@9!cHim < sàrdine sándwich $r@9cHim !r`mvHcY Rhythm Rule T Octóber Nj!s?va? < Octòber ráin Nj$s?va? !qDim Rhythm Rule X )$Njs?va? !qDim (c) position 2 σ before 1ry stress 1 σ before 1ry stress after 1ry stress 2ry stress àcadémic sàrdíne ! “3ry stress” = unstressed with full V ! Octóber róbot

Page 23 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

Two-syllable words # F STR # Rhythm Rule T F (full V) = 2ry sàrdíne $$r@9!cHim sàrdine sándwich $r@9cHim !r`mvHcY dìréct $c@i!qDjs dìrect débit $c@iqDjs !cDaHs cartóon $j@9!s|vm càrtoon nétwork $j@9s|vm !mDsv29j Three-syllable words # F STR 0 # Rhythm Rule X F = 3ry Octóber Nj!s?va? Octòber ráin Nj$s?va? !qDim Titánic s@i!s`mHj Titànic's bánd s@i$s`mHjr !a`mc salvátion r`k!uDiR?m Salvàtion Ármy r`k$uDiR?m !@9lh (d) position 2 σ before 1ry stress 1 σ before 1ry stress after 1ry stress # F STR # # F STR 0 # 2ry stress àcadémic sàrdíne ! ! “3ry stress” = unstressed with full V ! ! Octóber róbot (The complementary distribution of 2ry stress & 3ry stress can be an argument for not distinguishing between them: we will not pursue this here and continue to refer to a full vowel that does not have 1ry/2ry stress as having 3ry stress)

Page 24 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2.7) Difference between stresses: 1ry vs. 2ry – pitch change 2 1 Academic. 2 2 1 Peter's an academic. 2 2 2 1 Peter's an academic for good. The distinction between 1ry & 2ry stress is a predictable phrase/sentence level distinction. At the level of words 1/2ry stress is the same kind of stress (as opposed to 3ry & 0 syllables, which are unstressed). At phrase/sentence level the rightmost major = 1ry

Page 25 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2.6) We will use “3ry stress” as shorthand syllable that has a full vowel but no stress (2.6.2) Since some vowels are both in the full and the reduced subset, if there is such a vowel ina syllable without stress, it is not possible to tell if it has “3ry stress” or no (zero) stress. L In the first two examples you cannot tell, but in fórmat you can because [a] is full only so the final syllable must have “3ry” stress (2.6.4) i. Although there is a strong tendency for unstressed syllables to have reduced vowels, it is generally not predictable if vowel reduction will apply in an unstressed syllable or not L Look at the syllable highlighted in red in (a), in similar environments sometimes reduction

• ccurs, sometimes it does not

Sometimes there is a morphological clue: in words longer than two syllables -ate reduces in nouns, but not in verbs L Look at the examples in (b). Sometimes there some phonological motivation: in word-initial closed syllables immediately preceding the 1ry stress vowel reduction may be suspended, but in open syllables in the same position there is a strong tendency for it to apply. This is called the closed-syllable tendency for

• trad. 3ry stress.

L Look at the examples in (c). The first two rows illustrate the tendency, however the last one shows that it does not always apply. (2.6.4) ii. If a syllable has a full vowel, it may be stressed (1ry or 2ry) or unstressed (“3ry stress”). How can you tell the difference? An optional phrase/sentence level rule called the Rhythm Rule can make the distinction: this is a rule that downgrades a stress to “3ry stress” (so that its syllable loses its rhythmic prominence, but its vowel does not become reduced) if it is too close to another stress L This is what happens to the penultimate stress of académic in (a) and everlásting in (b) when another stressed word follows. This cannot happen to a word like Octóber, which looks very similar superficially, having a full-vowelled syllable followed by a more prominent full-vowelled syllable Nj!s?va?. If we put this word in a shifting context where the Rhythm Rule could apply, we find that it does not apply and the prominence relations do not change. The second syllable always remains more prominent that the first. We can explain this by claiming that this is due to a difference in stress: the first syllable of everlasting has (2ry) stress, but the first syllable of October has “3ry stress” (=is unstressed but has a full vowel..

Page 26 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(2.6.5) Assuming that we know where the last stress (called 1ry stress) is in a word, what is the distribution

• f “3ry” stress compared to other stresses (the 2ry stresses) preceding the 1ry?

(2.6.5) ii. “3ry” and 2ry are in complementary distribution two syllables before 1ry stress (3ry does not

• ccur here) & anywhere after 1ry stress (2ry does not occur here) = the difference is predictable

L see the table in (a) Both occur (and therefore seem to contrast) one syllable before 1ry stress L see the table in (b) – it is clear that the first syllable of sàrdíne has 2ry and not “3ry” stress since the Rhythm Rule applies to it. However, of these words two-syllable words of the pattern #FullV+STRESS# behave like sàrdíne (their first syllable has 2ry stress) and three syllable words of the pattern #FullV+STRESS+0stress# behave like Octóber (their first syllable has 3ry stress) = the difference is predictable here too. L see the table and the examples in (c) remember: # means word-boundary Thus 2ry and “3ry” stresses are in complementary distribution. L This is summarised in table (d) The complementary distribution of 2ry stress & 3ry stress can be an argument for not distinguishing between them: we will not pursue this here and continue to refer to a full vowel that does not have 1ry/2ry stress as having 3ry stress. (2.7) The difference between 1ry stress and 2ry stress is a sentence/phrase level distinction rather than a word-level one. If we say a word in isolation, we say it as a sentence, with the grammatically/attitudinally appropriate tone on the tonic syllable of the sentence (e.g. falling if it is a statement). If the same word occurs in a sentence consisting of more than one word, it may

• ccur in such a position that pitch change does not occur on any of its syllables, i.e. the place of 1ry

stress in a word is the place of potential pitch change: 2 1 2 2 2 2 1 Cigarette. I quit smoking cigarettes for good. Therefore at the word level we only need to know the location of stresses because the difference between 1ry and 2ry, the actual place of pitch change (on the 1ry) is determined at sentence/phrase

• level. The location of the 1ry stressed syllable is also predictable: in a neutral sentence it is the

rightmost major-stressed syllable. Since we are looking at isolated words here, all we need to know is the location of major stresses – the rightmost one will be the potential bearer of pitch change in a sentence, i.e. the rightmost one is the 1ry stress. We use the terms 1ry and 2ry stress as shorthand for rightmost and non-rightmost major stress in a word, respectively. See the set text on phrase stress/intonation

Page 27 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(3) Predictability of stresses in a word (3.1) Non-rightmost stress is predictable (‘2ry’) when the place of the rightmost stress (‘1ry’) is known (3.2) The phonotactics of ‘2ry’ stress in English words (constraints)

NO 2RY AFTER 1RY:

*#...1...2...# (by definition, see (3.1)) gráduateV Z!fq`cY|vDHs\ > 1 0 3 NOT *1 0 2 a.

NO STRESS CLASH:

* #...STR STR...# (= *#...21...#; *#...22...#) very few “exceptions” sàrdíne Rhythm Rule: sàrdine tín *sardìne tín prìncéss Prìncess Ánn *Princèss Ánn b.

EARLY STRESS:

*#00; *#30; *#03 exceptionless

Page 28 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(3.3) Relationship between the stressing of derived words and their bases

STRESS PRESERVATION

The derived word has to preserve the place of the stresses of its base. base derived word

STR STR

T {chá racter} {chà racter} + ístic STR

STR

T {o rí ginal} {o rì ginál} + ity STR

STR

X {Ja pán} {Jà pan} + ése

STRESS PRESERVATION is an inherently violable constraint: it applies as long as NO STRESS CLASH and EARLY STRESS are not violated = it is more important to obey NO STRESS CLASH and EARLY STRESS than STRESS PRESERVATION:

a. ranking NO STRESS CLASH, EARLY STRESS >>

STRESS PRESERVATION

! Long underived words with a stress later tan the 3rd from the beginning of the word: 2ry stress is unpredictable: either

• ne or two syllables are skipped

Àpalàchicóla (*Apàlachicóla) but Wìnnipesáukee (*Winnìpesáukee) 1σ 1σ 2σ

Page 29 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

(3) The predictability of stresses (1ry, i.e. rightmost stress & 2ry stress, i.e. non-rightmost stress), are they predictable or lexical? (3.1) Let us assume that we know the place of the rightmost stress (1ry stress). (3.2) The place of stresses preceding the rightmost stress (i.e. the 2ry stresses) are predictable and is due to the interaction of constraints, i.e. restrictions on the occurrence of 2ry stress, some

• f which are violable. These constraints are not equally important, they are ranked, i.e. some

are more important to obey than others. These constraints are: (a) NO STRESS CLASH: There should be no adjacent stresses (2ry or 1ry) Thus these stress patterns are not ok *#...21...#; *#...22...# Two-syllable words with final stress are a systematic exception (see (2.6.5) ii. above L see the examples in (3.2) a. and (2.6.5) ii. c. above (b) Early Stress There must be a stress (2ry or 1ry) on one of the two syllables at the beginning of a word Thus these stress patterns are not ok: *#00; *#30; *#03 NO STRESS CLASH and EARLY STRESS together uniquely determine the place of secondary stress when there is a stress (2ry or 1ry) on the third syllable (counting from the beginning of the word): in this case there must be a 2ry stress in the first syllable too (e.g. Pàtagónia $o`s?!fnvmHi?) because otherwise either NO STRESS CLASH or EARLY STRESS would be violated (e.g. *Patàgónia )o?$s`!fnvmHi? (e.g. *Patagónia )o?s?!fnvmHi?) (3.3) What if the known stress (2ry or 1ry) is later than the third syllable of the word, say the fourth? In this case NO STRESS CLASH and EARLY STRESS together require that there must be a 2ry stress either on the first or the second syllable: the patterns *#xxxX and *#xxXX are out, but both #XxxX or #xXxX are ok since they equally satisfy No Stress Clash and Early Stress. These words tend to be derived words and here another constraint comes into play STRESS PRESERVATION The derived word has to preserve the place of the stresses of its base.

Page 30 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

If there is a choice between two possible patterns of 2ry stress placement, it is STRESS

PRESERVATION that determines which one will win.

L Look at the examples on page 28. The reason why there is a difference between the placement of 2ry stress in the derived words chàracterístic and orìginálity is that their bases (cháracter and oríginal) have their 1ry stresses on different syllables, and the derived words both preserve the place

• f the 1ry stress of their bases.

Notice that it is the immediate base that counts: the base of orìginálity is oríginal not órigin. However, stress preservation is not always possible. Stress Preservation is an inherently violable constraint: it applies as long as No Stress Clash and Early Stress are not violated. It is not possible to preserve the major stress of Japán in the derived form Jàpanése, because the hypothetical form that would preserve it (*Japànése) would violate NO STRESS CLASH. It is more important to obey NO STRESS CLASH and EARLY STRESS than STRESS PRESERVATION. L This is why we have the ranking in (3.3) a. What happens in underived words in which 1ry stress falls later than the third syllable from the beginning of the word? In these words EARLY STRESS and NO STRESS CLASH permit 2ry stress placement either on the first or the second syllable of the word, but STRESS PRESERVATION cannot decide between these two candidates since there is no base whose stress should be preserved. In this case the place of 2ry stress is unpredictable (and therefore lexical). L Look at the examples at the bottom of page 28.

Page 31 of 31

Phonological Analysis BBN-ANG-241_stress1 <EngPhonBA_243_stress1_2020_annotated.wpd>

2ry stress is

! calculated from the location of 1ry stress ! weight insensitive ! iterative (a word can have more than one) ! stress preserving (if possible) ! partially unpredictable: long underived words