Catalan phonology

The Catalan phonology has a certain degree of dialectal variation. Although there are two standard varieties, one based on Central Eastern dialect and another one based on South-Western or Valencian, this article deals with features of all or most dialects, as well as regional pronunciation differences.
Catalan is characterized by final-obstruent devoicing, lenition and fortition of plosives, voicing assimilation, and gemination ; a set of 7 to 8 phonemic vowels, vowel assimilations, many phonetic diphthongs, and vowel reduction, whose precise details differ between dialects.

Consonants

Phonetic notes:

, are laminal denti-alveolar,. After, they are laminal alveolar,.
, are velar but fronted to pre-velar position before front vowels. In some Majorcan dialects, the situation is reversed; the main realization is palatal,, but before liquids and rounded back vowels they are velar,.
,, are apical front alveolar,,, but the first two are laminal denti-alveolar, before,. In addition, is postalveolar or alveolo-palatal before,,,, velar before, and labiodental before,, where it merges with. It also merges with before,.
,, are apical back alveolar,,, also described as postalveolar.
, are apical alveolar,. They may be somewhat fronted, so that the stop component is laminal denti-alveolar, while the fricative component is apical post-dental. is rare and may not be phonemic.
, are laminal "front alveolo-palatal",.
There is some confusion in the literature about the precise phonetic characteristics of,,, and ; while and describe them as "back alveolo-palatal", implying that the characters would be more accurate, they use the characters for palato-alveolar affricates and fricatives while using for alveolo-palatal sounds in examples in other languages like Polish or Chinese. Otherwise, sources, like generally describe them as "postalveolar".

Obstruents

Obstruents assimilate to the voicing of the following consonant. Voiced obstruents undergo final-obstruent devoicing so that fred is pronounced with while fredes is pronounced with .
Table with minimal pairs:

Plosives

Voiced plosives become lenited to approximants in syllable onsets, after continuants: →, →, →.

Exceptions include after lateral consonants / and after labiodentals, e.g. bolígraf boníssim / .
In non-betacist dialects, remains unlenited / .
In some dialects, e.g. many Valencian accents, initial can be lenited: gat .
In the coda position, these sounds are always realized as stops; except in some Valencian dialects, where they might be lenited.

In Catalan and Balearic, labial and, and velar stops and may be geminated in intervocalic position before .
Intervocalic is dropped in regular speech in Valencian, with compensatory lengthening of vowel ; e.g. vesprada .
In Majorcan varieties, velar stops and become and word-finally and before front vowels, in some of these dialects, this has extended to all environments except before liquids and back vowels; e.g. sang .

The dorso-palatal may occur in complementary distribution with, only in Majorcan varieties that have dorso-palatals rather than the velars found in most dialects: guerra vs. sa guerra .

In the Valencian dialects final voiceless plosives may be lenited before a vowel: tot açò .

Affricates

The phonemic status of affricates is dubious; after other consonants, affricates are in free variation with fricatives, e.g. clenxa / and may be analyzed as either single phonemes or clusters of a stop and a fricative.

Alveolar affricates, and, occur the least of all affricates.
* only occurs intervocalically: metzines / .
** In Valencian, many instances of are deaffricated to : utilitzar .
* Instances of arise mostly from compounding; the few lexical instances arise from historical compounding. For instance, potser / comes from pot + ser. As such, does not occur word-initially; other than some rare words of foreign origin, but it may occur word-finally and quite often in cases of heteromorphemic plural endings: tots . Several linguists claim is not a phoneme on its own, but a simple combination of and, in the same way that the in English 'cats' is not phonemic.
The distribution of alveolo-palatal affricates, and, depends on dialect:
*In most of Valencian and southern Catalonia, most occurrences of correspond to the voiced fricative in Standard Eastern Catalan: gel .
* In Standard Eastern Catalan, word-initial is found only in a few words of foreign origin while being found freely intervocalically and word-finally: despatx / .
* Standard Eastern Catalan also only allows in intervocalic position. Phonemic analyses show word-final occurrences of /, but final devoicing eliminates this from the surface: raig .
* In various other dialects, including Valencian and its standard variety, occurs word-initially and after another consonant to the exclusion of . These instances of word-initial seem to correspond to in other dialects, including the standard on which the orthography is based: xinxa, pronounced in Standard Catalan, is in these varieties.

There is dialectal variation in regards to affricate length, with long affricates occurring in both Eastern and Western dialects such as in Majorca and few areas in Southern Valencia. Also, intervocalic affricates are predominantly long, especially those that are voiced or occurring immediately after a stressed syllable / . In Modern Valencian and have merged into, except in some parts of Southern Valencian.
In Aragonese Catalan and Central Valencian, voiced fricatives and affricates are missing.

Fricatives

The labiodental fricative occurs in Balearic, as well as in Alguerese, Standard Valencian and some areas in southern Catalonia. Everywhere else, it has merged with so that and occur in complementary distribution.

In Valencian, is realized as an approximant after continuants: avanç .
In Majorcan, and are in complementary distribution, with occurring before vowels.
In other varieties that have both sounds, they are in contrast before vowels, with neutralization in favor of before consonants.

In Majorcan and Minorcan, undergoes total assimilation to a following consonant : buf gros .
The dental fricative only appears in Ribagorçan and Lower Aragon, in contrast with. Spanish loanwords with this sound may be replaced by in both Catalan and Valencian.
The velar fricative is found in Spanish interferences, especially in Aragon and Southern Valencia.
The glottal fricative is found in loanwords and interjections, although is usually replaced by in loanwords.

Sonorants

Laterals

Laterals assimilate the place of articulation of the following consonant. The lateral may be geminated in careful speech. A geminated may also occur / .

While "dark l",, may be a positional allophone of in most dialects, is dark irrespective of position in Eastern dialects like Majorcan and standard Eastern Catalan. For simplicity dark l is not transcribed in this article.
In Aragonese Catalan, is palatalized to in consonant clusters; e.g. plou 'it rains'.
In Alguerese and Ribagorçan word-final is depalatized to : gall .

Nasals

Nasals assimilate the place of articulation of the following consonant. In careful speech, and may be geminated / 'unnecessary', emmagatzemar / .

In Alguerese and Ribagorçan word-final is depalatized to : any .

Rhotics

The distribution of the two rhotics and closely parallels that of Spanish. Wheeler analyzes intervocalic as the result of gemination of a single rhotic phoneme: serra / 'saw, mountains'.

Between vowels, the two contrast / 'myrrh' vs. mira /, but they are otherwise in complementary distribution. appears in the onset, except in word-initial position, after,, and, and in compounds, where is used.
* Majorcan contrasts and in word final position, e.g. xerr vs. mor . Note the minimal pair corr vs. cor shows no contrast as Majorcan tend to omit final r in most cases.
* In Majorcan final + can be assimilated to .
Different dialects vary in regards to rhotics in the coda, with Western Catalan generally featuring and Central Catalan dialects like those of Barcelona or Girona featuring a weakly trilled unless it precedes a vowel-initial word in the same prosodic unit, in which case appears, 'for', but. Note that non-silent final is transcribed as in both standard transcriptions of Catalan and Valencian, regardless of the next sounds or in absolute final position.
There is free variation in word-initially, after,, and, and in compounds, wherein is pronounced or, the latter being similar to English red: ruc .
* In Northern Catalonia and in some accents of Majorcan, a uvular trill or approximant can be heard instead of the alveolar trill; e.g. córrer .

Vowels

Phonetic notes:

The vowel is further back and open than the Castilian counterpart in North-Western and Central Catalan, it is slightly fronted and closed in Valencian and Ribagorçan, and further fronted and closed in Majorcan.
* Stressed can be further retracted to in contact with velar consonants, and fronted to in contact with palatals. This is not transcribed in the article.
** The palatal pronunciation of may merge with by some speakers.
The central vowel in stressed position is found in Majorcan and part of Minorcan, in the Balearic Islands.
* The realization of the reduced vowel varies from mid to near-open, with the latter variant being the most usual in the Barcelona metropolitan area, where the distinction between and is less pronounced than in other varieties that maintain the distinction.
** In some subvariants, it has a more back pronunciation, like.
The open-mid and are lower and in Majorcan, Minorcan and Valencian.
* is slightly more open and centralized before liquids and in monosyllabics.
* is most often a back vowel. In some dialects can be unrounded.
and can be realized as mid vowels in some cases. This occurs more often with.
* In Northern Catalan, Modern Alguerese and some places bordering the Aragonese and Spanish-speaking areas, open-mid and close-mid vowels may merge into mid vowels; and/or .
The close vowels are more open than in Castilian. Unstressed are centralized.
* In Valencian and most Balearic dialects are further open and centralized, especially in unstressed position.
Northern Catalan sometimes adds two loan rounded vowels, and, from French and Occitan.
* Similarly French and are mostly adapted with and, respectively.
Phonetic nasalization occurs for vowels occurring between nasal consonants or when preceding a syllable-final nasal: diumenge / .
Vowels can be lengthened in some contexts, e.g. coordinació / .

Stressed vowels

Most varieties of Catalan contrast seven stressed vowel phonemes. However, some Balearic dialects have an additional stressed vowel phoneme ; e.g. sec . The stressed schwa of these dialects corresponds to in Central Catalan and in Western Catalan varieties.
Contrasting series of the main Catalan dialects:

Reduced vowels

In Eastern Catalan, vowels in unstressed position reduce to three :,, ;,, ; remains unchanged. However there are some dialectal differences: Alguerese merges, and with ; and in most areas of Majorca, can appear in unstressed position.

Unstressed and appear only in some words, especially in learned terms from other languages and in the unstressed diphthong, which varies with in informal speech. In other cases, they merge with and .
In Eastern Catalan may be elided in contact with rhotics, this is partially reflected in the orthography: cargol vs. caragol . It is also observed in three combinations with the preposition per: per això → 'because of this', per aquí → 'around here' and per allà → 'around there'. This also occurs in per a, which in speech can be reduced to per, which can cause ambiguity. can also be omitted in rapid speech in the verb anar, in the word adeu and in contact with other vowels.
unstressed can rarely undergo univerbation if accompanied by another word and be pronounced as a single oxytone word, causing the reduction of to in Eastern dialects. Similar cases occur with being reduced to in unstressed syllables.

In Western Catalan, vowels in unstressed position reduce to five:, ;, ; remain unchanged. However, in some Western dialects reduced vowels tend to merge into different realizations in some cases:

Unstressed may merge with before a nasal or sibilant consonant, in some environments before any consonant, and in monosyllabic clitics. This sounds almost the same as the Barcelonan open schwa. Likewise, unstressed may merge into when in contact with palatal consonants.
Unstressed may merge with before a bilabial consonant, before a stressed syllable with a high vowel, in contact with palatal consonants, and in monosyllabic clitics.
Besides vowel harmony or vowel assimilation, unstressed and may be found sporadically in compounds like dèsset and dènou .

Term		Gloss
parla		'speech'
rere		'back'
lliri		'lily'
ferro		'iron'
mutu		'mutual'

Term		Gloss
parla		'speech'
rere		'back'
lliri		'lily'
ferro		'iron'
mutu		'mutual'

Vowel harmony

The harmony of Valencian is a clear example of harmony conditioned by a strong element: in some Valencian dialects, word-final post-tonic becomes and when the preceding syllable contains tonic open-mid vowels and ; that is, and propagate the palatal and labial features, respectively, to the final vowel, as the examples. The articulatory features extend from a phonologically privileged position—the stressed syllable—to a weak position—an unstressed syllable—a perceptual asymmetry emphasized by the fact that the harmony trigger belongs to the radical while the assimilated segment is normally an inflectional affix.

In the most widespread system of harmony, both open-mid vowels cause assimilation; in other systems, distributed over the harmonic territory quite randomly, only one of the vowels triggers the change. For example, in Cullera only the front vowel causes assimilation, while in Borriana the labial vowel is the only one that allows harmony. However, in both the broadest and the narrowest versions, and even in the sporadic cases of two-way harmony that are presented here, the pattern of strong → weak extension remains constant.
In the harmonic phenomenon just described, articulatory features spread from left to right. However, there is no shortage of assimilations in which the features spread to the left of the prominent position. This occurs in Majorcan when pretonic is closed to in words that contain a close tonic vowel, e.g. c'nill, c'sí, c''mú. The change involves the extension of the height feature again in the direction dictated by the strong → weak saliency relationship. Similar closures are documented in various Western languages; among these, Tortosan stands out, where the phenomenon, quite variable and often limited to the elderly, presents the peculiarity that height does not only propagate from tonic vowels, but can also do so from unstressed ones. In general terms, and in accordance with the data collected by and u...u b), respectively. The assimilation of mid vowels to a high vowel of a different point of articulation is possible, but in the sequence e...u it is reduced to some words c), and in the sequence o...i it is usually limited to fossilized cases, so that the disharmonious alternatives in d) reflect only copied pronunciations of the orthography.

Of the phenomena presented above, the most common and systematic is the change e...i → i...i. As in the examples, becomes when it precedes a stressed or unstressed. Closure can even affect a series of two pretonic vowels. Assimilation never affects stressed vowels and there is also no harmony when and do not occupy adjacent syllables.
With certain restrictions, the phenomenon can modify the final vowel of the first element of a compound and proclitic elements such as numerals or unstressed pronouns. In the last case, when the vowel of the pronoun is not strictly adjacent to the syllable that triggers the harmony, there is no assimilation; according to, the lack of spread is related to the fact that groups of pronouns generate a secondary accent, which would protect the original quality of the vowel.
also reports some examples of rightward assimilation between weak elements; that is, cases where an unstressed sequence i...e becomes i...i. Harmony to the right is documented only between vowels that are in pretonic position; therefore, the inflectional elements and the post-tonic vowels belonging to the radical are excluded from the change.
General Valencian is another variety in which the extension of features is limited to the main metric foot: in plain words, the final post-tonic, which is part of the main foot, is affected by harmony a); on the other hand, in proparoxytone words. In Valencian from the south of Alicante, the harmony affects an intermediate layer between the main metrical foot and the clitic group: the prosodic word .
Harmony in General Valencian:

Harmony from Southern Valencian :
In the harmony of Valencian, Majorcan and, mostly, Tortosan, the features extend from a strong element to a weak element. In the other possible model, on the other hand, the features are spread in the reverse direction, that is, from positions that are not prominent to positions that are stronger from the perceptual point of view. The trigger for change is in this case a weak element, with regular reduction in the unstressed syllable, that is, with the vowels, and in this position. Borrowings also have the peculiarity that they tend to block the neutralization of the unstressed middle vowels e and o, which are realized as and, respectively. In principle, these two trends should not be mutually exclusive; however, if the post-tonic sound is close-mid, the tonic mids are usually also realized as close, as shown by the plain words in example b), in which the levelling between the two vowels is almost universal. Therefore, the quality of the most prominent vowel is determined by the features of the following vowel, since the appearance of close-mid vowels in tonic position depends on the presence of vowels of the same pitch in the post-tonic syllable.

In proparoxytones there is greater variability. In the variety analyzed by esdrúixol words ; the syllabic adjacency between the two vowels is key to harmony, since words like Sòcrates or Hèrcules are usually presented without assimilation despite the presence of an unreduced post-tonic e. In the variety described by, on the other hand, post-tonic vowels do not condition the realization of the tonic vowel in esdrúixols b). On the other hand, and in accordance with the interpretation of the aforementioned authors, the adaptation of tonic vowels as open mids is compatible in all varieties with the appearance of unreduced mid vowels in pre-tonic syllables.

Example IPA Translation
Jespersen 'Jespersen'
Penèlope 'Penelope'
Sòfocles'' 'Sophocles'

Other harmony examples in Central Catalan:

Vowels in contact
One of the most unique features of Catalan and Valencian is the treatment of vowels that come into contact within the speech chain. When a word-final vowel meets an initial vowel there are two possible reactions: lengthening or weakening/elision of one of the vowels. In general terms, two consecutive vowels diphthongize more frequently in Valencian, North Western Catalan and Alguerese. Some examples :
;Lengthening
followed by, e.g. viu a Alacant .
followed by, e.g. he eixit ara .
followed by, e.g. platges i illots .
followed by, e.g. escrit o oral? .
followed by, e.g. una tribu urbana .
;Elision
Unstressed followed by stressed →, e.g. quina hora és? or .
Stressed followed by unstressed →, e.g. mà esquerra or .
Unstressed followed by unstressed →, e.g. agarra el gos .
Stressed followed by stressed →, e.g. què has fet? .
Unstressed followed by →, e.g. este home .
Stressed followed by unstressed →, e.g. no els volen .
Diphthongs and triphthongs
There are also a number of phonetic diphthongs and triphthongs, all of which begin and/or end in or.
In Standard Eastern Catalan, rising diphthongs are only possible in the following contexts:
in word-initial position, e.g. iogurt .
The semivowel occurs between vowels as in feia or diuen .
In the sequences or plus vowel, e.g. guant, quòrum, qüestió, pingüí ; these exceptional cases even lead some scholars to hypothesize the existence of rare labiovelar phonemes and.
Processes
There are certain instances of compensatory diphthongization in Majorcan so that troncs develops a compensating palatal glide and surfaces as . Diphthongization compensates for the loss of the palatal stop. There are other cases where diphthongization compensates for the loss of point of articulation features as in vs. .
The dialectal distribution of compensatory diphthongization is almost entirely dependent on the dorsal stop and the extent of consonant assimilation.
Voiced affricates are devoiced after stressed vowels in dialects like Eastern Catalan where there may be a correlation between devoicing and lengthening of voiced affricates: metge → . In Barcelona, voiced stops may be fortified.
In Catalan is known the historical process of erasing the vowel of unstressed final syllables. and, have proposed similar theories, in that one or more 'extrasyllabic' final consonants represent the opening of a syllable with null vowel or empty nucleus. However, in the case of Catalan, such a structure is fundamentally the one that has been proposed to trigger vowel epenthesis in cases such as the followings:
ampl →
segl →
aspr →
corr →
Assimilations
Catalan denti-alveolar stops can fully assimilate to the following consonant, producing gemination; this is particularly evident before nasal and lateral consonants: e.g. setmana, cotna, Betlem, rotllo. Learned words can alternate between featuring and not featuring such assimilation / 'atlas', sotmetre / .
There is dialectal variation regarding words with. While Central and North-Western Catalan tend to innovate with a palatalized pronunciation, Valencian and Balearic maintain the traditional pronunciation without palatalization, i.e., in most cases.
Valencian dialects features simple elision in many of these cases and learned words might not exhibit either assimilation or elision: atles and administrar.
Prosody
Stress
Stress most often occurs on any of the last three syllables of a word / 'compass', càstig 'punishment', pallús / .
Compound words and adverbs formed with may have a syllable with secondary stress 'willingly'; parallamps but every lexical word has just one syllable with main stress.
Phonotactics
The structure of the syllable shows a mandatory nucleus, and the optional presence of a margin before and/or behind the nucleus. Core and margin contrast articulatory due to the fact that the core is emitted with a higher degree of oral opening than the margin. In Catalan and Valencian, the core position is occupied by a vowel and the margin position by one or more consonants. Depending on the number of consonants that make up the margin and the location of the margin in relation to the core, syllabic structures of the type V, CV, VC, CVC, CCV, CCVC, CCVCC, VCC, VCCC, CVCC, CVCCC, etc. are possible. The margin that precedes the nucleus is called "explosive", and the one that follows it, "implosive", syllables can be classified as "open" or "closed" by virtue of the absence or presence of a margin postnuclear, respectively.
Any consonant, as well as and may be an onset. Clusters may consist of a consonant plus a semivowel or an obstruent plus a liquid. Some speakers may have one of these obstruent-plus-liquid clusters preceding a semivowel, e.g. síndria ; for other speakers, this is pronounced .
Word-medial codas are restricted to one consonant + / ). In the coda position, voice contrasts among obstruents are neutralized. Although there are exceptions, syllable-final rhotics are often lost before a word boundary or before the plural morpheme of most words: color / vs. coloraina / .
In Central Eastern, obstruents fail to surface word-finally when preceded by a homorganic consonant. Complex codas simplify only if the loss of the segment doesn't result in the loss of place specification.
When the suffix -erol is added to camp it makes, indicating that the underlying representation is, however when the copula is added it makes. The resulting generalization is that this underlying will only surface in a morphologically complex word. Despite this, word-final codas are not usually simplified in most of Balearic and Valencian.
Although the number vint is pronounced in Eastern Catalan due to the reduction of in, it regains its in numbers of its family. In the same dialects, the word sant regains its if it is followed by a vowel of a noun although there is hesitation among speakers. There is also hesitation about the words cent and vint when they are combined with other nouns. ésADir recommends pronouncing them with before the words anys, homes, and hores, usually silent in other cases, with equally permissible hesitations.
The preposition amb in Eastern Catalan is usually pronounced, but in careful speech when it is before a vowel sound or an is pronounced.
Word-initial clusters from Graeco-Latin learned words tend to drop the first phoneme: gnom, mnemotècnia /, pneumàtic /, pseudònim /, pterodàctil / .
Word-final obstruents are devoiced; however, they assimilate voicing of the following consonant, e.g. cuc de seda / . In regular and fast speech, stops often assimilate the place of articulation of the following consonant producing phonetic gemination: tot bé .
The fricatives and affricates,,, become voiced when they occur at the end of a word and are followed by a word beginning with a vowel / 'the' but els avis / 'the grandparents', baix / 'short, low' but Baix Ebre / .
The pronunciation of at the end of words is typically pronounced. If the word is followed by a voiced consonant or a vowel, it would theoretically be pronounced as, which would typically be pronounced. In some speakers the actual pronunciation would be an affricate. Also, among certain speakers of some variants, the same occurs with →, which may also vary with an affricate and assimilate voicing → or .
Dialectal variation
The differences in the vocalic systems outlined above are the main criteria used to differentiate between the major dialects:
Wheeler distinguishes two major dialect groups, western and eastern dialects; the latter of which only allow,, and to appear in unstressed syllables and include Northern Catalan, Central Catalan, Balearic, and Alguerese. Western dialects, which allow any vowel in unstressed syllables, include Valencian and North-Western Catalan.
Regarding consonants, betacism and fricative–affricate alternations are the most prominent differences between dialects.
Other dialectal features are:
;Vowels
In a number of dialects unstressed can merge with or according to the previous or following vowel. This merger is especially common in words with the prefix in- or im-.
In Southern Valencian subvarieties, especially in Alicante Valencian, the diphthong has become : bous .
In regular speech in both Eastern and Western Catalan dialects, word-initial unstressed — or —may be diphthongized to or : ofegar 'to drown, suffocate'.
;Consonants
Varying degrees of L-velarization among dialects: is dark irrespective of position in Balearic and Central Catalan and might tend to vocalization in some cases. In Western varieties like Valencian, this dark l contrasts with a clear l in intervocalic and word-initial position; while in other dialects, like Alguerese or Northern Catalan, is never velarized in any instance.
Iodització in regular speech in most of Majorcan, Northern Catalan and in the historic comarca of Vallès, Latin-derived words that had intervocalic + yod or velar + developed, rather than as in the majority of other regions. Accents with traditional ieisme use only in words where this sound developed from Latin initial - or intervocalic --.
In northern and transitional Valencian, word-initial and postconsonantal alternates with intervocalically; e.g. joc 'game', but pitjor 'worse', boja 'crazy'.
In northern Valencia and southern Catalonia has merged with realizations of after a high front vocoid; e.g. terrissa, insistisc vs. pixar, deixar . In these varieties is not found after other voiced consonants, and merges with after consonants; e.g. punxa .
In some Valencian dialects, and are auditorily similar such that neutralization may occur in the future. That is the case of Northern Valencian where is depalatalized to as in caixa. Central Valencian words like mig and lleig have been transcribed with rather than the expected, and Southern Valencian "has been reported to undergo depalatalization without merging with ", as in passets vs. passe'''ig'''
Historical development
Catalan shares features with neighboring Romance languages.
Marked contrast of the vowel pairs and, as in other Western Romance languages, except Spanish and Sardinian.
Lenition of voiced stops as in Galician and Spanish.
Lack of diphthongization of Latin short ĕ, ŏ, as in Galician, Sardinian and Portuguese, and unlike French, Spanish and Italian.
Abundance of diphthongs containing, as in Galician and Portuguese.
Abundance of and occurring at the end of words, as for instance and, unlike Spanish, Portuguese or Italian.
In contrast with many other Romance languages, Catalan has many monosyllabic words; and those ending in a wide variety of consonants and some consonant clusters. Also, Catalan has final obstruent devoicing, thus featuring many couplets like amic vs. amiga.

Example	IPA	Translation
Jespersen		'Jespersen'
Penèlope		'Penelope'
Sòfocles''		'Sophocles'