Catalan stress is iambic
Max W. Wheeler1
Paper delivered at the 12th Manchester Phonology Meeting, May 20-22, 2004
Abstract
Lexical stress in Catalan is free but is subject to important limitations. There is a three-syllable right-edge stress
window, and a word with a heavy penult may not have proparoxytone stress. Most recent work on Catalan has
claimed that its stress pattern is fundamentally trochaic (i.e. TROCHEE » IAMB). I argue that, if the basic stress
pattern of Catalan is taken as iambic (IAMB » TROCHEE), and if we acknowledge the prosodic category colon (κ)
standing above foot in the prosodic hierarchy (Green 1997) in the constraint ALIGNHEADκR, the abovementioned limitations fall out automatically. The salience in the lexicon of words of trochaic shape, such as
taules ‘tables’ or ferro ‘iron’ is a consequence of the fact that in all nouns and adjectives stress is required to
fall on the stem and is excluded from inflectional affixes such as –es (F.PL) or –o (M.SG). Four constraints are
shown to outrank faithfulness to limit the range of attested stress patterns: a prosodic markedness constraint
WHC (a version of the Weight-to-Stress principle), a morphology-prosody markedness constraint excluding
stress from inflectional suffixes, and two alignment constraints ALIGNHDκRT, and IAMB. In nominals involving
derivational affixation the prosodic constraint HEAD(PrWd) (3a), also ranked above faithfulness, together with
ENDRULE-R AND *LAPSE, which are ranked below faithfulness, ensures that only the rightmost input stress
surfaces.
1. Introduction
With some limitations to be described shortly, the position of word stress in Catalan is
contrastive. Thus the examples in (1) have the same vowels and consonants but differ in
which syllable carries primary stress.2 The examples in (2) are underived words showing
proparoxytone, paroxytone and oxytone stress patterns.
(1)
càrregues [karees] ‘loadN.PL’,
carregues [karees] ‘loadV.2SG.PRS.IND’
carregués [karees] ‘loadV.1SG.PST.SUBJ’
(2)
peŀlícula [plikul] ‘film’
manicomi [mnikmi] ‘mental hospital’
Benimodí [bnimui] (place-name)
Each prosodic word (PrWd) has one stress, which falls on the rightmost stressable element,
and every phonological word (= lexical word, except for items specified as clitics) is a
prosodic word. Compounds, which consist of more than one lexical word, are stressed as
sequences of prosodic words, that is, just as phrases are, with each lexical word retaining
stress, the rightmost element being most prominent. Lexical words may be free forms (bases
that may stand alone or in combination with inflectional affixes), or combinations of stems
with derivational affixes, such as suffixes, which carry their own lexical stress, or prefixes,
which are unstressed.
1
Department of Linguistics & English Language, University of Sussex; email [email protected]
These example words are given in NW Catalan pronunciation, where unstressed /a/ and /e/ are not affected by
vowel reduction. The corresponding forms in eastern Catalan are [kars], [krs], and [kres]
2
Max Wheeler Catalan stress is iambic
12mfm May 2004
2
The limitations on the position of word stress in Catalan words are of two kinds:
Firstly, the position of stress in all verb forms is determined by the inflectional morphology;
stress falls on the last stressable affix —essentially exponents of tense or aspect categories, or
of non-finiteness— or on the last syllable of the stem if there is no stressable affix. Personnumber affixes are stress-neutral. Stress in verbs is not considered further in this paper (see
Wheeler fc. §9.4).
Secondly, prosodic words of all kinds are subject to a ‘window’ constraint: stress is never
earlier in the word than antepenultimate (that is, there is a three-syllable stress window), and
if the penultimate syllable is heavy, stress cannot be antepenultimate.
Discussions of stress in Catalan phonology have generally assumed (rather than argued; see,
for example, Serra 1996, Bonet & Lloret 1998) that the basic pattern of the Catalan stress foot
is trochaic. Oliva (1992) however, takes the view that the basic pattern is iambic, and I argue
here that this view is correct, though the case for it is subtle rather than evident to inspection.
Non-lexical secondary stress is a central element of the rhythmic organization of
phonological phrases, but does not directly form part of the prosodic organization of
individual words, and is not considered further here.
2. Complex words with affixes
The stress pattern of complex words derived through affixation is straightforward. Each
prosodic word has only one stress (by HEAD(PrWd), (3a) below), which is that of the
rightmost stem or affix (by ENDRULE-R, (3b) below). Stems and derivational affixes have the
location of stress lexically marked, but the constraints HEAD(PrWd) and ENDRULE-R ensure
that only one is realized on the surface. (There is thus no way that lexical stress on a true
prefix could surface.) All derivational affixes are at least one syllable long, and nearly all of
them are ‘auto-stressed’. However, a few affixes, mostly deriving adjectives, such as -i, -ic,
and -id, are ‘pre-stressed’, that is, they require stress on the syllable preceding them (stress
which can, in turn, of course, be overridden by the stress of a following suffix). Nominal
inflectional affixes (gender and number) are unstressed and stress-neutral. Some relevant
constraints are spelt out in (3).
Max Wheeler Catalan stress is iambic
(3)
12mfm May 2004
3
a. HEAD(PrWd): Each prosodic word has a unique head (McCarthy 2002: 78)
b. ENDRULE-R: The head foot is not followed by another foot within the prosodic
word (McCarthy 2003: 111)3
c. FPH: Faithfulness to prosodic heads. Prosodic heads in the input are realized as
such in the output (after Beckman 1998, Alderete 1999, Kager 2000).
d. *LAPSE: *σ/ __σ: Assign one violation mark for each pair of adjacent unstressed
syllables (McCarthy 2003: 106)
In the examples below (4)-(6) the position of word stress in inputs is indicated with underline.
The footing of the candidates is in accordance with the constraints ALIGNL(PrWd, Ft),
PARSE-σ and FOOTBIN, though for the issues raised in this paper it would not matter if it were
otherwise; unlikely footings are not illustrated here. In the outputs central Catalan vowel
reduction is applied to unstressed syllables.
(4)
barcelonina ‘Barcelonese.F.SG’ [brslunin]: /barselon/ (cf. Barcelona) /+in/ /+a/
(F.SG.)
/barselon]St+in]Af +a]In/ HEAD(PrWd) FPH ENDRULE-R *LAPSE
a.
b.
c.
brslunin
(σ) (σ σ)(σ)σ
) brslunin
(σ)(σ)(σ σ) σ
brslunin
*!
*
*
*
*
**
*!
**
(σ) (σ σ)(σσ)
d.
brslunin
*!
**
*!
*
****
(σ) (σ σ)(σσ)
e.
brslunin
*
(σ)(σ)(σ σ) σ
Candidate (4a) retains the lexical prosodic heads of its stem and its affix, in accord with FPH,
but thereby violates HEAD(PrWd) which requires that there be only one head, and thus also
ENDRULE-R, since the two heads cannot both be the rightmost foot. The comparison of
candidates (4a) and (4b) demonstrates the ranking HEAD(PrWd) » FPH. The comparison of
(4b) and (4c) illustrates the role of ENDRULE-R; of input stresses, the rightmost is preferred.
Candidate (4d) violates HEAD(PrWd) by having no head at all. Candidate (4e) avoids a lapse
violation by having two primary stresses (in violation of HEAD(PrWd)), one of which is also
not faithful to the input (violating FPH).
3
ENDRULE-R is offered by McCarthy as a categorical constraint to replace the gradient RIGHTMOST:
ALIGN(HdFt, R, PrWd, R) ‘The head foot is rightmost in the prosodic word’ of Kager (1999: 167); *LAPSE
constraints are also part of McCarthy’s approach, after Kager (2002), to avoid gradient constraint evaluation.
Max Wheeler Catalan stress is iambic
(5)
12mfm May 2004
4
mariner [mine] ‘sailor’: /maR/ (cf. mar ‘sea’) /+in/ (adjective-forming suffix)
/+eR/ (agent suffix)
/maR]St +in]Af +eR]Af/ HEAD(PrWd) FPH ENDRULE-R *LAPSE
a.
mine
*!*
**
(σ)(σ)(σ)
b.
) mine
**
*
(σ)(σσ)
c.
min
**
*!
*
**
*!
*
(σσ)(σ)
d.
main
(σ)(σσ)
e.
min
*!
***
**
(σ)(σσ)
Candidate (5a) again retains all the input stresses, but incurs multiple fatal violations of
HEAD(PrWd). Candidate (5e) has no stress at all, also violating HEAD(PrWd). Candidates (5c)
and (5d) fatally retain underlying stresses other than the rightmost, while the winning
candidate (5b) has one head, the rightmost foot in the word.
Example (6) siŀlàbic ‘syllabic’ illustrates a pre-stressed suffix applied to a stem with marked
non-final stress. I ignore here the candidate with no stress.
(6)
siŀlàbic ‘syllabic’: /silab/ (cf. síŀlaba ‘syllable’) /+_ik/
/silab]St+_ik]Af / HEAD(PrWd) FPH ENDRULE-R *LAPSE
a.
silaik
*!*
**
(σ)(σ)(σ)
b.
) silaik
*
*
*
*
(σσ)(σ)
c.
silik
*!
(σ)(σσ)
d.
silik
**
*!
(σ)(σσ)
Candidate (6a) again is ruled out by HEAD(PrWd) through having multiple heads. Candidate
(6d) respects neither of the input stresses (two violations of FPH), though its head is
rightmost. Candidates (6b) and (6c) tie on ENDRULE-R, since in each case the head foot is
followed by another foot within the word; (6b) wins on *LAPSE.4
4
Though I follow McCarthy (2003) here in replacing gradient evaluation by RIGHTMOST with ENDRULE-R and
*LAPSE, the consequential account is, perhaps, not an entirely perspicuous way of expressing the fact that,
among underlying stresses, it is the rightmost that surfaces.
Max Wheeler Catalan stress is iambic
12mfm May 2004
5
Incidentally proparoxytone words like síŀlaba [sil] ‘syllable’ show that FPH outranks
ENDRULE-R and *LAPSE.
(7)
síŀlaba ‘syllable’ /silab+a/)
/silab]St+a]I/ HEAD(PrWd) FPH ENDRULE-R *LAPSE
a.
sila
*!
*
(σσ)(σ)
b.
) sil
*
*
(σ)(σσ)
c.
sila
*!
*
(σ)(σσ)
3. Prosodic word constraints
As mentioned in §1, in Catalan the location of stress —one per prosodic word— in non-verbs
is a lexical property. Not all conceivable stress locations are found, however, and the task of
this section is to account for the gaps.
Nominal inflectional affixes (such as -o-, -e-, masculine gender markers, -a- (-e- before -s):
feminine gender marker) are, not unusually, stress-neutral, or ‘extrametrical’. This restriction
might in principle reflect either correspondence, or markedness. Where the base exists as an
uninflected prosodic word, the stress-neutrality of affixes could be attributed to
correspondence constraints, along the lines suggested by Alderete (2001: 191-193). For
example, one might say that the stress profile of (8a) cunyades ‘sisters-in-law’ is constrained
to match that of the base (8b) cunyat [kuat] ‘brother-in-law’. But this would not provide an
account of the fact that the feminine gender marker /a/ ~ /e/ is never stressed, even in words
that have no correspondent without the agreement marker. Figura ‘figure’ (8c) is well
formed, but there is no form like *figura where -a is inflectional. (Catalan differs from
languages like Greek or Russian, in which nominal inflectional affixes may be stressed,
though they are often unstressed.)
(8)
a. cunyades [kuas] ‘sisters-in-law’ (/kuad/ /+e/ (F) /+z/ (PL))
b. cunyat [kuat] ‘brother-in-law’ (/kuad/)
c. figura ‘figure’ [fiu] /fiu+a/; *[fiua] */fiu+a/
Preferably, then, for Catalan, it is a markedness constraint like *ALIGNFTISUF (9) that ranks
above any faithfulness constraints concerning stress location.
(9)
*ALIGNFTISUF: Inflectional suffixes are not aligned with the head of a prosodic word.
Max Wheeler Catalan stress is iambic
12mfm May 2004
6
There are two other general stress constraints. The first is the three-syllable stress window at
the right edge, mentioned previously; see examples at (2). A polysyllable may have stress on
the antepenultimate syllable, for example, peŀlícula [plikul] ‘film’, or on the penultimate,
for example, manicomi [mnikmi] ‘mental hospital’, or on the last, for example, Benimodí
[bnimui] (place-name). But preproparoxytone stress is absolutely excluded.
The second prosodic constraint is one of quantity sensitivity: a word with a heavy
penultimate syllable cannot have antepenultimate stress. The weight of a final syllable
imposes no stress constraints, nor does the weight of an antepenultimate syllable. In (10)
examples are given of all the weight possibilities in trisyllabic words. In (10a) are all the
possibilities with heavy penults; in (10b) all the possibilities with light penults. As far as
possible monomorphemic words are chosen. The point is that the diversity of attested stress
locations within the three-syllable window is not determined by morphological structure.
(10)
a. Available stress patterns in trisyllabic words with heavy penults
LHL
LHL
LHL
LHH
LHH
LHH
HHL
HHL
HHL
HHH
HHH
HHH
molesquí ‘moleskin’, Novetlè (place-name)
sinopsi ‘synopsis’, enorme ‘enormous’
*
crisantem ‘chrysanthemum’, faristol ‘lectern’
adúlter ‘adulterous’, apòstol ‘apostle’
*
organdí ‘organdie’, almesquí ‘jonquil’
extracte ‘extract’, escombra ‘broom’
*
restaurant ‘restaurant’, exhaustiu ‘exhaustive’
escàndol ‘scandal’, extrínsec ‘extrinsic’
*
Max Wheeler Catalan stress is iambic
12mfm May 2004
7
b. Available stress patterns in trisyllabic words with light penults
LLL
LLL
LLL
LLH
LLH
LLH
HLL
HLL
HLL
HLH
HLH
HLH
comité ‘committee’, peroné ‘fibula’
bigoti ‘moustache’, figura ‘figure’
pàgina ‘page’, cèlebre ‘famous’
matalàs ‘mattress’, calamars ‘calamari’
capítol ‘chapter’, polígon ‘polygon’
Júpiter ‘Jupiter’, hàbitat ‘habitat’
Castellar [ksta] (place-name), bisturí ‘scalpel’
misteri ‘mystery’, llenguado ‘sole’
àlgebra ‘algebra’, ímpetu ‘impetus’
enrenou ‘uproar’, actual ‘present’
oxigen ‘oxygen’, ventríloc ‘ventriloquist’
pàncrees ‘pancreas’, làudanum ‘laudanum’
The gaps in (10) consist of all and only those forms with both antepenultimate stress and
heavy penults. To account for the three-syllable window together with the gaps in (10) I need
to mention some prosodic constraints proposed in recent work. A familiar constraint is
FOOTBINARITY (11).
(11)
FOOTBINARITY (FTBIN): Feet are binary under moraic or syllabic analysis (Kager
1999: 156).5
This constraint has two results: as regards the maximum size of feet, ternary and unbounded
feet are disfavoured. And the minimum size of a foot is set at two moras. (The minimal
aspect of FTBIN is, in fact, frequently violated in Catalan; for example in monosyllabic light
prosodic words like fe ‘faith’, pla ‘flat’.)
A well known parametric constraint is the constraint selecting either trochees or iambs as the
basic foot pattern. The active constraint in Catalan, as I shall argue, is IAMB. The formulation
in (12a) follows the model for trochee given by McCarthy (2002: 150). The formulation
spells out that IAMB is an alignment constraint.
(12)
a. IAMB: ALIGNR(Ft, Hd(Ft)). Every foot has its head on the right hand edge.
b. (LL) ≻ (LL), (LH) ≻ (LH), (HL) ≻ (HL), (HH) ≻ (HH)
c. (H) (L) (‘monosyllabic iambs’)
5
Hyde (2001: 58) gives good reasons for reassessing this constraint. The crucial part of it he takes to be the
limit on the maximum size of a foot. For him, the bimoraic minimality aspect falls out from other constraints.
With respect to the maximum foot size, however, he proposes the FootCap Condition:
FootCap Condition: Feet are maximally disyllabic.
The FootCap Condition is not a constraint (violable) but a condition on GEN; or, to put it another way, the
FootCap Condition forms part of the definition of ‘foot’.
Max Wheeler Catalan stress is iambic
12mfm May 2004
8
IAMB determines the harmonic preferences in (12b). Monosyllabic feet, namely (H) or (L)
(12c), do not violate IAMB as their heads are indeed aligned with the right edge of the foot (as
well as the left edge).
Another familiar prosodic constraint is the Weight-to-Stress Principle (WSP; Kager 1999:
155), expressing the fact that stress preferentially falls on heavy syllables. I adopt here (13)
Hyde’s formulation (2001: 74).
(13)
WHC (Weight-to-Head Constraint): Every heavy syllable must be designated as the
head of some foot.
Hyde’s Weight-to-Head Constraint (WHC) is in fact more general than the Weight-to-Stress
Principle, as there can be heads of feet which are not stressed, provided PARSE-σ is active.
Every stressed syllable must be the head of a foot, but not every head of a foot needs to be
stressed.
The last crucial constraint involves the prosodic category ‘colon’ (κ), a category that comes
in the prosodic hierarchy below the prosodic word and above the foot, and groups feet in a
manner analogous to the manner in which feet group syllables. The colon is argued for on the
basis of data from several languages by Green (1997: 98-133), following Hayes (1995: 119).6
Green (1997: 108) spells out the size of a colon with a COLONBINARITY constraint (14a)
(analogous to FTBIN). COLONBINARITY means that a colon may group two feet, or a foot and
a stray syllable (14b).
(14)
a. COLONBINARITY: Cola are binary (over feet and syllables).
b. Examples of cola: two feet {(σσ)(σσ)}, {(σσ)(σ)}
a foot and a stray syllable,{(σσ)σ} or {σ(σσ)}
Other colon constraints that I adopt from Green (1997: 108) are spelt out in (15).
(15)
a. PARSEFT: Feet are parsed into cola.
b. ALLκRT: ALIGN(κ, R, PrWd, R). Every colon is aligned with the right edge of
some prosodic word. (ALLκRT penalizes a word with more than one colon.)
c. ALIGNHDκRT: Every head colon is aligned right with some prosodic word; i.e.
‘prosodic words are right-headed’ (PRWDRIGHT in Green 1997)
In (15c) I rename Green’s PRWDRIGHT as ALIGNHDκRT (every head colon is aligned right
with some prosodic word) as a reminder that the right-headedness of words is being evaluated
by this constraint in terms not of feet, but of cola.
6
Seven of the analyses in Hayes (1995) make use of the colon; Green (1997) argues that this category is
required for an insightful analysis of stress location in Munster and E. Mayo Irish, and in Manx.
Max Wheeler Catalan stress is iambic
12mfm May 2004
9
I now turn to demonstrating how the combination (unranked with respect to one another) of
ALIGNHDκRT (15c), IAMB (12a), and WHC (13), ranked above faithfulness to input stress
(FPH at (3c)), rules out first preproparoxytones, and then proparoxytones with heavy penults.
In (16)-(17) are evaluated various parsings and stressings of hypothetical input
preproparoxytones, with H and L standing for any heavy or light syllable respectively. I
ignore here the possible influence of morphological alignment constraints such as
*ALIGNFTISUF (9), and do not mention HEAD(PrWd) (3a), and PARSEFT (15a), which are
assumed to be undominated. Heads of feet that are not stressed are indicated in bold.
Preproparoxytones LLLL excluded by ALIGNHDκRT, IAMB, WHC » FPH
(16)
a.
b.
c.
d.
e.
f.
g.
h.
i.
LLLL
ALIGNHDκRT IAMB WHC FPH
{(LL)(LL)}
**
{(LL)(LL)}
*
{(L)}{(LL)L}
*
) {(LL)(LL)}
*
{(L)}{(LL)L)}
*
*
) {(L)}{(LL)L}
*
{(LL)(LL)}
*
*
) {(LL)(LL)}
*
{(L)}{(LL)(L)}
*
*
Candidates (16a) and (16b) show two preproparoxytone parsings that are ruled out by IAMB.
Candidate (16c) avoids the IAMB violation, but at the cost of forcing the stressed syllable into
a non-final colon, violating ALIGNHDκRT. Candidates (16d), (16f), and (16h) violate
faithfulness only, but offer otherwise acceptable proparoxytone, paroxytone, and oxytone
alternatives respectively, of the types seen in (2). Candidates (16e), (16g), and (16i) are less
successful (trochaic) parses of the same stress patterns as (16d), (16f), and (16h).
Max Wheeler Catalan stress is iambic
12mfm May 2004
10
Preproparoxytones HLLL excluded by ALIGNHDκRT, IAMB, WHC » FPH
(17)
a.
b.
c.
d.
e.
f.
g.
h.
i.
j.
k.
l.
HLLL
ALIGNHDκRT IAMB WHC FPH
{(HL)(LL)}
**
{(HL)(LL)}
*
{(H)}{(LL)L}
*
{(HL)(LL)}
*
*
{(H)}{(LL)L)}
*
*
) {(H)}{(L)(LL)}
*
{(HL)(LL)}
*
*
*
{(HL)(LL)}
**
*
) {(H)}{(LL)L}
*
{(HL)(LL)}
*
*
{(H)}{(LL)(L)}
*
*
) {(H)}{(L)(LL)}
*
In (17) all the preproparoxytone candidates (17a)-(17c) violate either ALIGNHDκRT or IAMB.
They lose to (17f), (17i), or (17l), which are the best parsed proparoxytone, paroxytone, and
oxytone candidates respectively. These correspond to the attested types in (18).
(18)
aurèola ‘halo’
pastanaga ‘carrot’
Algemesí (place name)
Though I have only considered in (16)-(17) hypothetical inputs where the last three syllables
are light, it should be evident that additional Hs later on will do nothing to improve the
situation for preproparoxytone inputs, all of which are harmonically bound by ALIGNHDκRT,
IAMB and WHC.
I now turn to consider trisyllabic inputs with heavy penults, in order to explain the principled
gaps —those with heavy penults (*σHσ)— in the (10a) set. Here I take real examples, and
show in the tableaux that proparoxytone outputs are ruled out also by the constraints we have
been considering, rather than by faithfulness constraints. Tableau (19) evaluates candidates
for molesquí ‘moleskin’ considering various parses of the unfaithful proparoxytone, and the
faithful oxytone, candidates.
(19)
a.
b.
c.
d.
e.
f.
molesquí ‘moleskin’ ALIGNHDκRT IAMB WHC FPH ALLκRT
LHL
{(L)(HL)}
*!
*
{(L)(HL)}
*!
*
{(L)}{(H)L}
*!
*
{(L)(HL)}
*!
) {(LH)(L)}
{(L)}{(H)(L)}
*!
Max Wheeler Catalan stress is iambic
12mfm May 2004
11
In (19) each of the three proparoxytone candidates (19a-c) falls to one of the dominant
constraints being considered. Candidate (19d) provides a bad parse of the correctly stressed
form. Candidates (19e) and (19f) survive this evaluation. (19e) wins in the evaluation by
ALLκRT, mentioned at (15b), since (19f) contains a colon which is not right-aligned in the
word. (This is not crucial given that patterns (19e) and (19f) are indistinguishable in
pronunciation.)
Tableau (20) evaluates prosodic candidates for sinopsi ‘synopsis’; the first three (20a-c) are
identical to the failing proparoxytone candidates in (19).
(20)
a.
b.
c.
d.
e.
f.
g.
h.
sinopsi ‘synopsis’
LHL
{(L)(HL)}
{(L)(HL)}
{(L)}{(H)L}
{(L)(HL)}
) {(LH)L}
) {(LH)(L)}
) {(L)}{(H)L}
) {(L)}{(H)(L)}
ALIGNHDκRT IAMB WHC FPH ALLκRT
*!
*!
*!
*
*
*
*!
*
*
In (20e-h) four parses of LHL are considered; of which (20g-h) again equally violate
ALLκRT. In the following tableaux candidates with gratuitous violations of ALLκRT will not
be considered. The difference between candidates (20e) and (20f) is that the final light
syllable of (20e) {(LH)L} is not parsed into a foot, while that of (20f){(LH)(L)}contains a
violation of FTBIN (or whatever constraints produce the same effect). In the absence of
evidence to choose between these I arbitrarily pick the unfooted representation of a final
unstressed light syllable in the remaining tableaux, remembering that the alternative can
always be substituted for it.
In tableau (21) I consider the two LHH patterns together (LHH faristol ‘lectern’ and LHH
adúlter ‘adulterous’) since the unfaithful proparoxytone LHH candidates are common to
both.
Max Wheeler Catalan stress is iambic
(21)
a.
b.
c.
d.
e.
f.
g.
12mfm May 2004
12
faristol ‘lectern’, adúlter ‘adulterous’ ALIGNHDκRT IAMB WHC FPH
LHH
LHH
{(L)(HH)}
*!
*
*
{(L)(HH)}
*!
*
{(L)}{(H)(H)}
*!
*
faristol
{(L)(HH)}
*!
) {(LH)(H)}
adúlter
{(L)(HH)}
*!
*
) {(LH)(H)}
Candidates (21a-b, d, f) show that with IAMB and WHC constraints active there is no
effective way to combine a sequence of two heavy syllables into one foot. Footing them
separately, as in (21c) forces a preceding stressed syllable into a non-final colon, fatally, by
ALIGNHDκRT.
In (22) I examine the HHL patterns corresponding to organdí ‘organdie’ (HHL) and extracte
‘extract’ (HHL), repeating the model of comparison used in (21).
(22)
organdí ‘organdie’ extracte ‘extract’
HHL
HHL
a.
{(H)(HL)}
b.
{(H)(HL)}
c.
{(H)}{(H)L}
ALHDκRT
IAMB
WHC FPH
*!
ALLκRT
*
*!
*!
*
*
*
organdí
d.
{(H)(HL)}
e.
{(HH)(L)}
f.
{(HH)(L)}
g. ) {(H)(H)}{(L)}
*!
*!
*!
*
*
h. ) {(H)}{(H)(L)}
*
extracte
i.
{(H)(HL)}
j. ) {(H)}{(H)L}
*!
*
In (22a-b) a final HL sequence cannot, subject to IAMB and WHC, compose a single foot;
hence, as in (22c), a preceding stressed foot has to be in a separate, non-final, colon, violating
ALIGNHDκRT. The comparison between candidates (22c) and (22j), by the way, shows why I
reject the view that the trochee is the preferred foot pattern in Catalan. (HL) would form a
good trochee; therefore {(H) (HL)} would form a good bipedal head colon. In (22d-f)
Max Wheeler Catalan stress is iambic
12mfm May 2004
13
various disyllabic foot parses with an initial heavy syllable violate IAMB or WHC or both.
Candidates (22g-h), equally faithful to input stress, are evaluated by ALLκRT which, if
evaluation is categorical, does not choose between them —there is no difference anyway in
pronunciation. In (22i) it is shown again that a disyllabic foot with a heavy initial does not
pass IAMB.
Lastly in (23) consider HHH candidates of the types restaurant ‘restaurant’ and extrínsec
‘extrinsic’.
restaurant ‘restaurant’ extrínsec ‘extrinsic’ ALHDκ IAMB WHC FPH ALLκRT
HHH
HHH
RT
(23)
a.
{(H)(HH)}
b.
{(H)(HH)}
c.
{(H)}{(H)(H)}
*!
*
*
*!
*
*!
*
*
restaurant
d.
e.
f.
g. )
{(H)(HH)}
{(HH)(H)}
{(HH)(H)}
{(H)(H)}{(H)}
*!
*!
*!
*
*
h. ) {(H)}{(H)(H)}
*
extrínsec
i.
{(H)(HH)}
*!
*
j. ) {(H)}{(H)(H)}
*
The proparoxytone candidates (23a-c) are ruled out, as previously, by high-ranking
ALIGNHDκRT, IAMB, and WHC, and candidates (23g)/(23h) for restaurant, and (23j) for
extrínsec, win over the considered alternatives.
All remaining patterns of prosodic words are well formed in Catalan. Table (10b), repeated
here as (24), illustrated the range of trisyllables with light penults. Table (25) illustrates the
range of disyllables, and (26) the monosyllabic prosodic words. The foot and colon parses in
the right hand columns are those that pass the constraints ALIGNHDκRT, IAMB, and WHC that
have been illustrated previously.
Max Wheeler Catalan stress is iambic
(24)
(25)
14
Available stress patterns in trisyllables with light penults
LLL
comité ‘committee’, peroné ‘fibula’
LLL
LLL
LLH
bigoti ‘moustache’, figura ‘figure’
pàgina ‘page’, cèlebre ‘famous’
matalàs ‘mattress’, calamars ‘calamari’
LLH
LLH
HLL
HLL
HLL
HLH
HLH
HLH
capítol ‘chapter’, polígon ‘polygon’
Júpiter ‘Jupiter’, hàbitat ‘habitat’
Castellar [ksta] (place-name), bisturí ‘scalpel’
misteri ‘mystery’, llenguado ‘sole’
àlgebra ‘algebra’, ímpetu ‘impetus’
enrenou ‘uproar’, actual ‘present’
oxigen ‘oxygen’, ventríloc ‘ventriloquist’
pàncrees ‘pancreas’, làudanum ‘laudanum’
{(L)(LL)} or
{(LL)(L)}
{(LL)L}
{(L)(LL)}
{(L)(LH)} or
{(LL)(H)}
{(LL)(H)}
{(L)(LH)}
{(H)(LL)}
{(H)}{(L)L}
{(H)(LL)}
{(H)(LH)}
{(H)}{(L)(H)}
{(H)(LH)}
Available stress patterns in disyllabic words
LL
LL
LH
LH
HL
HL
HH
HH
(26)
12mfm May 2004
cafè ‘coffee’, menú ‘table d’hôte’
tribu ‘tribe’, mare ‘mother’
segon ‘second’, sotrac ‘jolt’
llapis ‘pencil’, pèsol ‘pea’
garbí ‘south west wind’, tauró ‘shark’
ardu ‘arduous’, taxi ‘taxi’
engany ‘deceit’, pinzell ‘paintbrush’
càstig ‘punishment’, príncep ‘prince’
{(LL)}
{(L)L}
{(LH)}
{(L)(H)}
{(H)(L)}
{(H)L}
{(H)(H)}
{(H)(H)}
Available stress patterns in monosyllabic words
H tros ‘piece’, verd ‘green’ {(H)}
L cru ‘raw’, re ‘anything’ {(L)}
4. Conclusion
In Catalan nominals, although a range of lexical stress patterns is attested, four constraints
outrank faithfulness to limit the range of patterns: a prosodic markedness constraint WHC
(13), a morphology-prosody markedness constraint *ALIGNFTISUF (9), and two alignment
constraints ALIGNHDκRT (15c) and IAMB (12a). In nominals involving derivational affixation
the prosodic constraint HEAD(PrWd) (3a), also ranked above FPH, together with ENDRULE-R
(3b) AND *LAPSE (3c), which are ranked below FPH, ensures that only the rightmost input
stress surfaces. The constraint ranking observed is
*ALIGNFTISUF, ALIGNHDκRT, IAMB, WHC, HEAD(PrWd) » FPH » ENDRULE-R,
*LAPSE
Max Wheeler Catalan stress is iambic
12mfm May 2004
15
REFERENCES
ROA = Rutgers Optimality Archive, http://roa.rutgers.edu/index.php3
Alderete, John (1999) Head dependence in stress-epenthesis interaction. In Hermans, Ben &
van Oostendorp, Marc (eds), The derivational residue in phonological Optimality
Theory, Amsterdam, Benjamins, 29-50, [= Faithfulness to prosodic heads. ROA 940000]
Alderete, John (2001) Morphologically governed accent in Optimality Theory, New York:
Routledge
Beckman, Jill N. (1998) Positional Faithfulness (PhD dissertation, U Massachusetts). ROA
234-1297
Bonet, Eulàlia & Lloret, Maria-Rosa (1998) Fonologia catalana, Barcelona, Ariel
Green, Antony Dubach (1997) The Prosodic Structure of Irish, Scots Gaelic, and Manx (PhD
Dissertation, Cornell University). ROA 196-0597
Hayes, Bruce (1995) Metrical stress theory: principles and case studies, Chicago, University
Press
Hyde, Brett (2001) Metrical and prosodic structure in Optimality Theory (PhD dissertation,
Rutgers University, New Brunswick, NJ). ROA 476-1101
Kager, René (1999) Optimality Theory, Cambridge, Cambridge University Press
Kager, René (2000) Stem stress and peak correspondence in Dutch. In Dekkers, Joost, van
der Leeuw, Frank & van de Weijer, Jeroen, (eds), Optimality theory: phonology,
syntax and acquisition, Oxford, Oxford University Press, 121-150
Kager, René (2002) Rhythmic directionality by positional licensing. ROA 514-0402
McCarthy, John J. (2002) A thematic guide to Optimality Theory, Cambridge, University
Press
McCarthy, John J. (2003) OT constraints are categorical. Phonology 20, 75-138
Oliva, Salvador (1992) La mètrica i el ritme de la prosa, Barcelona, Quaderns Crema
Serra i López, Josep (1996) La fonologia prosòdica del català, Tesi de doctorat, Departament
de Filologia i Filosofia, Universitat de Girona
Wheeler, Max W. fc. The Phonology of Catalan, Oxford, Oxford University Press