The architecture of the english lexicon

A.2 Presentation of the data

The transcriptional conventions used here are designed to simplify the presentation of the data. Data appearing in brackets represents the phonetic representation found in the Celex corpus. This representation, similar to traditional dictionary pronunciations, will be replaced in the tables by a quasi-phonemic representation for a number of reasons. One is the contrast between what are certainly corresponding phonemes (e.g., /ær/: [¼:]Ü[¨]) found in British and American pronunciations. Another is the presentational simplification of presenting vowels and diphthongs that participate in the vowel quantity alternations that this investigation focuses upon with the same symbols, distinguished solely by length markings. The system used is as follows :

(A.1)

/a/ = Br. [¾, a], Am. [¾] // = [eÆ]

/e/ = [ƒ] /«/ = [iÆ]

/i/ = [í] /ö/ = [aÆ]

/o/ = [Ì] /¯/ = Br. [æ·], Am. [o·]

/u/ = [ó] /¬/ = [Æu·]

/ô/ = [ô] /u·/ = [u·]

/æ/ = [æ] /a·/ = [a·]

/or/ = Br. [¿:], Am. [¿r] /o·/ = Br. [¿:], Am. [¿·]

/ær/ = Br. [¼:], Am. [¨] /oÆ/ = [¿Æ]

/ar/ = Br. [Ì:], Am. [ar]

The vowels seen in the left column will be generally described as "short" while those on the right will be described as "long". The variation between British [¾] and [Ì] was considered to be for the most part conditioned by the following consonant, and these are merged in the notation as /a/, which behaves phonologically like a short vowel in all cases despite the phonetic length of the British back variant, e.g. p‡st as [p̲:st]. The convention /¬/ for the triphthong [Æu·] has been chosen for its stressed instantiation despite the fact that this phoneme does not always appear as (or behave phonologically as) a long vowel, and when reduced surfaces as [Ææ]. The unstressed form of this triphthong will be written as /u/. In general, /æ/ shows up in the quasi-phonemic notation to indicate a reduced vowel. Syllabic sonorants are also indicated with a leading /æ/, e.g., /æl/, /ær/, /æn/, /æm/. Primary and secondary stresses are marked with the acute and breve as usual.

Morpheme boundaries will be noted inside the quasi-phonemic notation with hyphens, e.g., dnot‡tion /deø-n¯t-a³°-æn/. Stress patterns will be described using a bracketed list, e.g., [2,0,1-0]. Only primary, secondary and zero stress will be indicated in this descriptive formalism, which in this case indicates a word with penultimate stress and initial secondary stress. The hyphen will be used to indicate the suffix boundary, as many patterns within stems remain consistent, regardless of the weight of the suffix. In most tables, example forms are given simply with orthography, long vowels being marked with the macron, and stresses marked with the breve and acute, as in the main chapters. The same conventions outlined above apply to the marking of long vowels in orthography. The remainder of the appendix is a presentation of the distribution of stress patterns in the major suffix classes amongst the Latinate or "level one" suffixes. The classifications are based on a combination of stress placement and morphological criteria such as the behavior of prefixes and roots in combination with the suffixes in question. Information on the distribution of stress patterns in unsuffixed words has been presented in chapter three of the main text.
A.3 Suffixes like /-al/

The suffixes grouped here include the monosyllabic suffixes /-al/, / or/, / ive/, /-ous/, / us/, /-um/, /-ar/, and /-an/. All are vowel-initial, and the suffix vowel is always reduced. The underlying vocalism of some of these suffixes may be inferred from the following forms which show further suffixation:

(A.2) ton‡lity /t¯n-‡l-iti/ form‡lity /form-‡l-iti/

tut—rial /t¬t-oør-i-æl/ dit—rial /dit-o³r-i-æl/

lmin—sity /luømin-—s-iti/ pomp—sity /pomp-—s-iti/

ˆdject’val /ˆïikt-´v-æl/ sbstant’val /sbstænt-´v-æl/¹³⁶

prit‡nical /puørit-‡n-ik-æl/
Forms showing a monosyllabic stem are universally stressed with the pattern [1-0] and are about evenly divided between heavy (CVC-) and long-vowel (CVV-) roots. A minority shows a light stem (CV-):¹³⁷
(A.3) CVV- 43% t—nal /to³n-æl/ l—cal /lo³c-æl/

r‡zor /ra³z-ær/ m—tor /mo³t-ær/

f‡mous /fa³m-æs/ hŽinous /ha³n-æs/

pl—sive /plo³s-iv/ m—tive /mo³t-iv/

l—cus /lo³k-æs/ m’nus /m´n-æs/

sŽrum /se³r-æm/ d‡tum /da³t-æm/

r—man /ro³m-æn/ p‡gan /pa³g-æn/

CVC- 44% f—rmal /f—rm-æl/ fr‡ctal /fr‡kt-æl/

vŽndor /vŽnd-ær/ c‡ptor /k‡pt-ær/

nŽrvous /næ²rv-ær/ v’scous /v’sk-æs/

‡ctive /‡kt-iv/ c‡ptive /k‡pt-iv/

d’scus /d’sk-æs/ c‡mpus /k‡mp-æs/

d’ctum /d’kt-æm/ qu‡ntum /kw—nt-æm/

gŽrman /ïæ²rm-æn/ œrban /æ²rb-æn/

CV- 13% m‡mmal /m‡m-æl/ Žrror /Žr-ær/

zŽalous /zŽl-æs/ fŽrrous /fŽr-æs/

p‡ssive /p‡s-iv/ m‡ssive /m‡s-iv/

ph‡llus /f‡l-æs/ st‡tus (alt.) /st‡t-æs/

vŽllum /vŽl-æm/ ‡lum /‡l-æm/
About 13% of the forms show a short light syllable in the stressed foot, contrary to expectations about foot structure. A number of these forms have long variants, e.g., st‡tus /st‡t-æs/Ü/sta³t-æs/, l—cus /l—k-æs/Ü/lo³k-æs/, d‡tum /d‡t-æm/Ü /da³t æm/.

Words showing two syllables preceding the suffix show two stress patterns with the following distribution:

(A.4) [1,0-0] 45%

[0,1-0] 55%

[1,0-0]

CVV-CV- 14% sŽasonal /se³zæn-æl/ fœneral /fu³nær-æl/

‡uditor /¿²·dit-ær/ c—unselor /k‡unsæl-ær/

d‡ngerous /da³nïær-æs/ lœminous /lu³min-æs/

sŽcretive /se³kræt-iv/ nœtritive /nu³trit-iv/

œterus /u³tær-æs/ tœmulus /tu³mÆæl-æs/

st‡dium /sta³di-æm/ —pium /o³pi-æm/

pœritan /pu³rit-æn/ lœtheran /lu³Jær-æn/

CVC-CV- 27% —rbital /—rbit-æl/ c‡rdinal /k‡rdin-æl/

c—nqueror /k—nkær-ær/ Žmperor /Žmpær-ær/

c‡ncerous /k‡nsær-æs/ dŽxterous /dŽkstær-æs/

pœrposive /pæ²rpæs-iv/ sŽnsitive /sŽnsit-iv/

c‡lculus /k‡lkÆæl-æs/ ’ncubus /’nkÆæb-æs/

m‡ximum /m‡ksim-æm/ pŽndulum /pŽndÆæl-æm/

mŽxican /mŽksik-æn/ p‡rtisan /p‡rtiz-æn/

CV-CV- 59% p’votal /p’væt-æl/ n—minal /n—min-æl/

Žditor /Ždit-ær/ m—nitor /m—nit-ær/

vŽnomous /vŽnæm-æs/ —minous /—min-æs/

p—sitive /p—zit-iv/ ‡dditive /‡dit-iv/

st’mulus /st’mÆæl-æs/ tŽtanus /tŽtæn-æs/

pl‡tinum /pl‡tin-æm/ b‡rium /m—nit-ær/

‡frican /‡frik-æn/ vŽteran /vŽtær-æn/

There are also a few words (less than 1%) with unexpectedly heavy but unstressed penults:
(A.5) CVC-CVC- sœbstantive /sœbstænt-iv/

CV-CVC- g—vernor /gœværn-ær/

‡djective /‡ïækt-iv/

c‡vernous /k‡værn-æs/

Of the second, slightly more numerous pattern with penultimate stress, a large majority of these forms show heavy penultimates. These stems are either bisyllabic morphemes with final heavy syllables, or prefixed monosyllables:
(A.6)

[0,1-0]

s-CVV- 41% cathe³dral propo³sal equa³tor adv´sor explo³sive des´rous cohe³sive progna³thous

papy³rus detr´tus facto³tum ango³lan

s-CVC- 53% tröœmphal revŽrsal precœrsor detŽctor

momŽntous horrŽndous

vind’ctive defŽnsive oly²mpus prospŽctus ug‡ndan subœrban momŽntum addŽndum

s-CV- 16% col—ssal rebœttal profŽssor abŽttor

oprŽssive sucŽssive coll—ssus narc’ssus

flagŽllum err‡tum tibŽtan mor—ccan

The general stress distribution noted above is paralleled by the longer forms. Again, words with light penults show the final stress pattern [1,0-0] while words with heavy penults show [0,1-0]:
(A.7) Trisyllabic stem forms:

[0,1,0-0] 52%

s-CVV-CV- lev´athan

s-CVC-CV- impŽrsonal prep—sterous impŽrvious

s-CV-CV- or’ginal med’cinal amb‡ssador

compŽtitor carn’vorous const’tutive

compŽtitive emŽritus es—phagus alœminum curr’culum amŽrican

[2,0,1,0] 37%

s`-s-CVV- ˆnecdo³tal “ntertr´bal reøappra³isal loøcomo³tive “ndec´sive “gnora³mus ˆrbore³tum pr˜tozo³an

s`-s-CVC- bnef‡ctor (alt.) “ntercŽptor ˆmbidŽxtrous smic—nscious “nexpŽnsive ucaly²ptus “nterrŽgnum rferŽndum mˆnuf‡cture

s`-s-CV- rtrogrŽssive strptoc—ccus

[1,0,2,0] 11%

s²-s-CVV- su³pervØsor ‡rchetyøpal

s²-s-CVC- ch´roprˆctor bŽnefˆctor

s²-s-CV- prŽdecssor

[1,0,0,0] 0 %

s²-s-CV- sp’ritual
Four syllables:

[2,0,1,0,0] 91%

s`-s-CVV-CV- sbcuta³neous

s`-s-CVC-CV- “nterpŽrsonal ph˜tosŽnsitive

s`-s-CV-CV- ˆrithmŽtical ˆbor’ginal “nsect’vorous

sperc’lious “nterl—cutor pr“mogŽniture t“ntinn‡bulum h“ppop—tamus c˜smop—litan

[0,2,0,1,0] 5%

s-s`-CVV- insctic´dal tyrˆnnos‡urus el“zabe³than

s-s`-CV- des“der‡tum

[2,0,0,1,0] 4% tmperamŽntal rˆdio‡ctive smicondœctor

pritonŽum

[1,0,0,2,0] 0% sŽmicondctor (alt.)

Five syllables: [2,0,0,1,0,0] “nfinitŽsimal hŽteroge³neous

ˆntedilu³vian vterina³rian

[0,2,0,1,0,0] eccleøsi‡stical

[2,0,2,0,1,0] “nterc˜ntinŽntal

As the number of syllables in the word increases, instances of penultimate stress drop dramatically. Most examples are actually cases of multiple suffixation such as / ent al/ or / ment al/, such as env“ronmŽntal, “nterc˜ntinŽntal. Secondary stresses usually appear on alternating syllables to the left of the main stress. Words which fail to conform to this generalization are infrequent and tend to be morphologically complex.

A.4 Suffixes like /-ent/

The members of this group, which includes /-ent/, /-ant/, / ence/, / ance/, /-ency/ and /-ancy/, show a stress pattern that is identical with that of the preceding group. That is, words with heavy penults receive penultimate stress, while words with light penults show antepenultimate stress. The underlying segmental forms of the suffixes, which tend to surface as / ænt/ or / æns(i)/, are revealed by further suffixation, e.g., c˜ntinŽntal, prsidŽntial. The principal difference between the two sets, which suggests a separate grouping, is morphological. Whereas in the / al/ grouping monosyllabic long vowel stems retain their form when prefixed, e.g., rev’ve Ü rev’val, many stems in the / ent/ group, with apparently similar structure, fail to show the stressed long vowel seen in the unsuffixed form, and instead display a bisyllabic foot with short vowels, e.g., res’de Ü rŽsident. This alternation is not seen with heavy stems, e.g., resp—nd Ü resp—ndent.

Forms showing a monosyllabic stem are universally stressed with the pattern [1-0] and are divided between heavy and long-vowel forms, as well as a sizable minority of light, short stem forms:

(A.8) CVV- 35% cla³imant m´grant stu³dent po³tent

gu´dance cre³dence re³gency va³cancy

CVC- 46% sŽrvant s—lvent sŽntence prŽgnancy

CV- 18% Žrrant plŽasant prŽsent pŽnance

Žssence clŽmency cœrrency

Words showing two syllables preceding the suffix show two stress patterns with the following distribution:

[1,0-0] 41%

[0,1-0] 58%

Of the first set, both prefixed and unprefixed stems are seen, in contrast to the pattern seen in the / al/ group. Most stem forms show a final light syllable:

(A.9)

[1,0-0]

CVV-CV- 7% ve³hemence c—untenance pre³cedence (Br.)

CVC-CV- 33% ‡spirant ‡bstinent s—mnolence

—rdinance mŽndicant tœrbulent

c—mpetency c—rpulency tœrbulency

CV-CV- 59% ’nnocent —ccupant hŽsitance

rŽsident ‡ssonance rŽticence

hŽsitancy v’rulancy rŽsidency

Of the second pattern with penultimate stress, the vast majority of these forms show heavy penults. These stems are either prefixed heavy monosyllables or, much less commonly, bisyllabic stems with heavy second syllables:

(A.10)

[0,1-0]

s-CVV- 29% obe³isance compla³inant impe³dence

compo³nent clairv—yance appl´ance

malfe³asance exc´tant acc—untancy infre³quency cohe³rency

s-CVC- 50% tröœmphant defŽndant abœndance

transcŽndent convŽrgence astr’ngency

cont’ngency ins—lvency ascŽndancy

mal’gnancy

s-CV- 21 % adm’ttance abh—rrent fluorŽscent

propŽllant recœrrence inclŽmency

ex’gency transp‡rency (alt.) discrŽpancy

There are a number of forms here, such as compl‡inant, which are obviously prefixed long-vowel monosyllables and thus violate the generalization applied above to the / ent/ group. Many of these show unusual allomorphy in their paradigms, for example:
(A.11) compo³seÜcompo³nent surve³yÜsurve³illance obe³yÜobe³isant
A few words allow for both realizations, e.g., sœbsident, subs´dent, ‡spirant, asp´rant. Many others are words with open syllables in the penult, such as convŽyance, appl’ance. Of the forms that show the unexpectedly stressed light penult, most prefixed cases correlate with verb forms that show this light stressed stem, e.g., recœr, recœrrent, adm’t, adm’ttance, propŽl, propŽllant.

In the longer words, forms with light penults usually show the stem stress pattern [1,0-] while words with heavy penults show [0,1-]. Multiply prefixed stems show the same tendency to reduce long-vowel root monosyllables; many [0,1,0-0] forms are prefixed bisyllabic stems:

(A.12)

Trisyllabic forms:

[0,1,0,0] 52% protu³berance commu³nicant co’ncidence

rec—gnizance inc—mpetent amb’valent (alt.)

bell’gerent omn’potent sign’ficance

bell’gerency omn’potency irrŽlevancy

inc—mpetancy res’liency

[2,0,1,0] 48% ˆntece³dence “ncohe³rence d“sinfŽctant

c˜rresp—ndent quid’stant jrispru³dence

˜mniprŽsent “ncandŽscence ˆmbiva³lent (alt.)

“nterm’ttent c˜rresp—ndency “neff’ciency

Four syllables:

[2,0,1,0,0] 44% d“sinhŽritance “nsign’ficance

[2,0,0,1,0] 56% sperabœndant sperintŽndent
The high percentage of forms with the pattern [2,0,0,1,0] is due to the presence of the prefix /super /, which takes stress on its initial syllable regardless of the stress seen in the stem.
A.5 Suffixes like /-ic/

This group is of central interest to a study about vowel length alternation in English, because the suffixes to be listed here are intimately involved in two of the principal length-based alternations: vowel shortening (Myers 1987) and CiV lengthening (Chomsky & Halle 1968). The suffixes grouped here include the monosyllabic suffixes /-ic/, / ion/ and /-ule/, and the bisyllabic suffixes / ity/, / ible/, /-ify/. To these may be added the i-extended suffixes , /-i-al/, / i or/, / i ous/, /-i-us/, / i um/, /-i-ar/, / i an/, / i ent/, / i ence/, /-i-ant/, / i ance/, and the combination /-ul-ar/. All are vowel-initial. Words in this group almost all take their main stress on syllable preceding the suffix. In most cases this places stress on the antepenult, except where the suffix is monosyllabic and the stress then falls on the penult. The initial suffix syllable is always light, making the surface stress pattern of these words similar to that of the above groups.

A.5.1 Shortening suffixes

The "shortening" suffixes placed in this group are /-ic/, /-ity/, /-id/, / ish/, / ible/, /-ify/, / ule/, and its extension /-ul-ar/. All begin with high vowels. Words belonging to this set overwhelmingly show stress on the syllable preceding the suffix. Among bisyllabic stems, a very small amount, around 1% of the words involved, show antepenultimate stress; most of these are /-ic/ forms :

(A.13)

c‡tholic ‡rabic ch—leric b’shopric rhŽtoric

lœnatic hŽretic ‡rsenic p—litic tœrmeric

Žligible nŽgligible intŽlligible c—rrigible imp—verish

m—lecule rŽticule m’nuscule vŽstibule r’dicule

ar’thmetic (noun) ‡ntarctic (Br. alt.) Ho³meric (Br.)

Most of these stems tend to show regular stress patterns when they are further suffixed, e.g., pol’tical, herŽtical, cholŽrical, rhet—rical, ˆrithmŽtical, molŽcular, rid’culous. None of the longer stems show this exceptional pattern.

Forms showing a monosyllabic stem are universally stressed with the pattern [1-0(0)] and are divided between heavy, long-vowel and short-vowel forms:

(A.14)

CVV- 19% sce³nic se´smic pho³bic pro³bity (alt.)

de³ity cru³dity fe³asible ‡udible

r´sible cru³cible u³nify cru³cify

no³tify co³dify (alt.) glo³rify tu³bular

CVC- 33% cŽntric t—xic Žthnic d’gnity

dŽnsity l‡xity sŽnsible flŽxible

t‡ngible f—rtify m‡gnify f‡lsify

fœrnish v‡nquish ‡mpule c‡psule

v’rgule pœstule ’nsular c’rcular

CV- 48% c—nic t—nic m’mic qu‡lity

v‡nity s‡nity crŽdible v’sible

Ždible m—dify v’lify ty²pify

cl‡rify chŽrish v‡nish gl—bule

n—dule gr‡nule p—pular rŽgular

t‡bular t’tular

Here, while many words with long vowel stems in their "bare" forms do show a short vowel, e.g., c—neÜc—nic, t—neÜt—nic, a sizable minority of about 20% retain their long vowel, e.g., b‡se /ba³s/Ü b‡sic /ba³s-ik/, scŽne /se³n/Ü scŽnic /se³n ik/, although some show regularized alternants like /sŽn-ik/, cy²clic /s´kl ik/ Ü /s’kl-ik/.

Words showing two syllables preceding the suffix overwhelmingly show the identical pre-suffixal stress pattern, apart from the few exceptions noted above.

(A.15)

[0,1-0(0)]

s-CVV- 8% apha³sic phone³mic ac—ustic strate³gic

gratu³ity obe³sity credu³lity ame³nity (alt.)

impu³nity defe³asible dedu³cible redu³cible

s-CVC- 38% rom‡ntic i‡mbic sem‡ntic eclŽctic

indŽmnity divŽrsity etŽrnity modŽrnity

rotœndity profœnity divŽrsify ödŽntify

exŽmplify elŽctrify indŽmnify contŽmptible

convŽrtible conv’ncible submŽrsible acc—mplish

ext’nguish avœncular

s-CV- 54% dem—nic rob—tic err‡tic dupl’city

necŽssity longŽvity son—rity doc’lity

pol‡rity fat‡lity pomp—sity demy²stify

be‡tify hum’dify pers—nify transm—grify

sol’dify syll‡bify adm’ssible comp‡tible

illŽgible div’sible ast—nish dem—lish

bin—cular spect‡cular vern‡cular

Many of the forms with long penults show hiatus, e.g., prosa³ic, trocha³ic, but a few are notable because their unsuffixed stems did not show a long vowel, for example, strate³gic /stræte³g-ik/ Ü str‡tegy /str‡tæïi/, syste³mic /siste³m-ik/ Ü sy²stem /s’stæm/.

For the trisyllabic stem forms, the distribution is even more skewed towards light syllable stems:

(A.16)

[2,0,1-0(0)]

s`-s-CVV- 9% ˆnalge³sic pˆraple³gic ˜rthope³dic pr˜miscu³ity

˜pportu³nity sperflu³ity ˆmbigu³ity sp˜ntane³ity

“nsecu³rity c˜ntinu³ity “ndefe³asible “nexh‡ustible

s`-s-CVC- 18% hØer‡rchic d“alŽctic hyøpodŽrmic uønivŽrsity

tˆcitœrnity uønif—rmity c˜nfratŽrnity

n˜nconf—rmity “ndestrœctible c˜mprehŽnsible

s`-s-CV- 73% ˆutom‡tic pˆtri—tic con—mic quan’mity s˜lid‡rity last’city v˜lat’lity srend’pity

“nfidŽlity ˆnony²mity “ndiv’sible c˜nsangu’nity

d“sest‡blish “nexprŽssible “ncomp‡tible prpend’cular
Words of more syllables all show penultimate stress, but vary in the placement of the secondary stress. Many are multiply suffixed:
(A.17) Four syllables:

[2,0,0,1,0(0)] 37% ˆgorapho³bic pripatŽtic p˜lythe’stic

chˆracter’stic h˜mogene³ity s“multane³ity

d“scontinu³ity vˆriab’lity mˆnageab’lity

sparab’lity nˆvigab’lity prˆcticab’lity

“nstrument‡lity nivers‡lity sp“ritu‡lity

c˜ntradist’nguish

[0,2,0,1,0(0)] 63% enthuøsi‡stic kaleøidosc—pic hermˆphrod’tic

advØsab’lity excØtab’lity accptab’lity

pred“ctab’lity or“gin‡lity convntion‡lity

fam“li‡rity part“cul‡rity convrtib’lity

inflxib’lity perfctab’lity inc˜ntrovŽrtible

inc˜mprehŽnsible
For multiply suffixed forms, the difference in stress appears to often be morphologically influenced, with stress falling on the initial syllable of multisyllabic prefixes, e.g., /pri-pætŽt-ik/, /p˜li-J«-’st-ik/ vs. /en Jsi ‡st ik/, /d«-trmin-’st-ik/, /pætæ`rn-æl-’st-ik/, although this does not always hold, e.g., mon—polistic /mon˜-pol-’st-ik/.

The few words with five syllables preceding the suffix are multiply suffixed forms in /-ity/:

(A.18)

[2,0,0,0,1,0,0] :

/-ive/+/-ity/: rˆdioact’vity

[0,2,0,0,1,0,0]:

/-able/+/-ity/: inducab’lity inc˜mparab’lity invØolab’lity

insparab’lity

/-ible/+/-ity/: inc˜rrigib’lity inligib’lity

[2,0,2,0,1,0,0]:

/-able/ + /-ity/: “nterchaøngeab’lity “nappl“cab’lity

/-al/+/-ity/: h˜mosxu‡lity c˜nfidnti‡lity

/-ible/ + /-ity/: c˜mprehnsib’lity “ndiv“sib’lity “rresp˜nsib’lity
There are a few longer words, such as inc˜nsequnti‡lity and hterosxu‡lity.
A.5.2 /-ion/ and the CiV lengthening suffixes

Words of this group also take stress on the syllable preceding the suffix, and are further characterized by hiatus within the suffixal material. The frequent suffix /-ion/ shares this categorization with a set of suffixes typified by a glide /-i-/ extended by /-al/-type suffixes. Consequently, the environment for Chomsky & Halle’s (1968) rule of CiV lengthening is created, and most low and mid vowels in open syllables do in fact appear long in this position. The suffix /-ion/ rarely appears unextended, (e.g., rŽgion, med‡llion), and the vast majority of these forms are nominal abstracts in /-t-ion/ or / s-ion/, formed from the set of Latinate verbs which made their adjectives in /-ive/, which was often similarly extended to / t ive/ or /-s ive/. Only a few verbs, like op’ne and rebŽl, form their verbal abstract by directly suffixing / ion/ to form op’nion and rebŽllion. Forms in / at ion/ are treated separately (¤ A.8.1).

Forms showing a monosyllabic stem are universally stressed with the pattern [1-0] and are divided between heavy , long-vowel and short-vowel forms:

(A.19)

CVV- 22% rŽgion n‡tion m—tion œnion ‡uction

CVC- 52% sc‡nsion sœction ‡ction s‡nction

CV- 26% p’nion st‡llion p‡ssion r‡tion (alt.)
All words showing two syllables preceding the suffix show the identical penultimate stress pattern. Among the bisyllabics all three syllable types are found in the stressed penult. Most stem forms are prefixed monosyllables:

(A.20)

s-CVV- 36% cohe³sion emo³tion comple³tion

infla³tion perhe³lion ero³sion

s-CVC- 48% projŽction dimŽnsion compœnction

concŽption recœrsion

s-CV- 28% bat‡llion v—lition aud’tion contr’tion op’nion

comp‡nion fru’tion regrŽssion rebŽllion

For trisyllabic stems, all forms show penultimate stress with the pattern [2,0,1,0]:

(A.21)

s`-s-CVV- 48% d“sillu³sion d“stribu³tion rsolu³tion

s`-s-CVC- 22% mˆled’ction gnuflŽction c“rcumscr’ption

rtrospŽction sˆtisf‡ction rprodœction

s`-s-CV- 29% ˆbol’tion ˆdmon’tion dcomprŽssion

rtrogrŽssion c“rcumc’sion
Words of more syllables all show penultimate stress. They vary in the placement of the secondary stress:

(A.22)

Four syllables:

[2,0,0,1,0] c“rcumlocu³tion “ntercommu³nion rdistribu³tion

c˜ntradist’nction ˆnimadvŽrsion “nterpos’tion

[0,2,0,1,0] inc˜mprehŽnsion elctrocu³tion derquis’tion

irrsolu³tion
Most of these forms can be characterized as prefixed forms, and the difference in stress appears to often be morphologically influenced, multisyllabic prefixes taking stress on their initial syllables. Some words show alternants with both kinds of stress, e.g., d“ssatisf‡ction, dissˆtisf‡ction. The words in /-i-/ plus /-al/-type suffix show parallel stress patterns.