Spanish phonology

This article is about the phonology and phonetics of the Spanish language. Unless otherwise noted, statements refer to Castilian Spanish, the standard dialect used in Spain on radio and television. For historical development of the sound system see History of Spanish. For details of geographical variation see Spanish dialects and varieties.

Spanish has many allophones, so it is important here to distinguish phonemes (written between slashes / /) and corresponding allophones (written between brackets [ ]).

Consonants
The phonemes, , and are realized as approximants (namely , hereafter represented without the undertack) or fricatives in all places except after a pause, after a nasal consonant, or—in the case of —after a lateral consonant; in such contexts they are realized as voiced stops. Some examples would be: hacia Bogotá [aθja βo̞ɣo̞ˈta] ('towards Bogota'), el búho [e̞l ˈβu.o̞] ('the owl'), and el delfín [e̞l de̞lˈfin] ('the dolphin').

The phoneme is realized as an approximant in all contexts except after a pause, a nasal, or a lateral. In these environments, it may be realized as an affricate. The approximant allophone differs from non-syllabic in a number of ways; it has a lower F2 amplitude, is longer, can only appear in the syllable onset (including word-initially, where non-syllabic  normally never appears), is a palatal fricative in emphatic pronunciations, and is unspecified for rounding (e.g. viuda  'widow' vs ayuda  'help'). The two also overlap in distribution after and : enyesar  ('to plaster') aniego  ('flood'). Although there is dialectal and ideolectal variation, speakers may also exhibit other near-minimal pairs like abyecto ('abject') vs abierto ('opened'). There are some alternations between the two, prompting scholars like to postulate an archiphoneme, so that ley  would be transcribed phonemically as  and leyes  as.

In a number of varieties, including some American ones, a process parallel to the one distinguishing non-syllabic from consonantal  occurs for non-syllabic  and a rare consonantal. Near-minimal pairs include deshuesar ('to bone') vs. desuello  ('skinning'), son huevos  ('they are eggs') vs son nuevos  ('they are new'), and huaca  ('Indian grave') vs u oca  ('or goose').

The phoneme (as distinct from ) is found in some areas in Spain (mostly northern and rural) and some areas of South America (mostly highlands).

Most speakers in Spain (except for Western Andalusia and all Canary Islands), including the variety prevalent on radio and television, have both and  (distinción). However, speakers in Latin America and those parts of southern Spain have only (seseo). Some speakers in southernmost Spain (especially coastal Andalusia) have only (a consonant similar to ) and not  (ceceo). This "ceceo" is not entirely unknown in the Americas, especially in coastal Peru. The phoneme has three different pronunciations ("laminal s", "apical s" or "apical dental s") depending on dialect.

The phonemes and  are laminal denti-alveolars. The phoneme becomes dental  before denti-alveolar consonants, while  remains interdental  in all contexts.

According to some authors, is post-velar or uvular in the Spanish of northern and central Spain. Others describe as velar in European Spanish, a uvular allophone  appearing before  (including when  is in the syllable onset as ).

A common pronunciation of in nonstandard speech is the voiceless bilabial fricative, so that fuera is pronounced  rather than. In some Extremaduran, western Andalusian, and American varieties, this softened realization of /f/, when it occurs before /w/, is subject to merger with /x/; in some areas the homophony of fuego/juego is resolved by replacing fuego with lumbre or candela.

is a marginal phoneme that occurs only in loanwords; many speakers have difficulty with this sound, tending to replace it with or. In a number of dialects (most notably, Northern Mexican Spanish, informal Chilean Spanish, and some Caribbean and Andalusian accents) occurs, as a deaffricated.

Consonant neutralizations
Some of the phonemic contrasts between consonants in Spanish are lost in certain phonological environments, and especially in syllable-final position. In these cases the phonemic contrast is said to be neutralized.

Sonorants
The three nasal phonemes—,, and —maintain their contrast when in syllable-initial position (e.g. cama 'bed', cana 'grey hair', caña 'sugar cane'). In syllable-final position, this three-way contrast is lost as nasals assimilate to the place of articulation of the following consonant —even across a word boundary; or, if a nasal is followed by a pause rather than a consonant, it is realized for most speakers as alveolar (though in Caribbean varieties this may instead be  or an omitted nasal with nasalization of the preceding vowel). Thus is realized as  before labial consonants, and as  before velar ones.
 * Nasals and laterals

Similarly, assimilates to the place of articulation of a following coronal consonant, i.e. a consonant that is interdental, dental, alveolar, or palatal.

Assimilatory nasal and lateral allophones are shown in the following table:

The alveolar trill and the alveolar tap  are in phonemic contrast word-internally between vowels (as in carro 'car' vs caro 'expensive'), but are otherwise in complementary distribution. Only the trill can occur after, , or  (e.g. alrededor, enriquecer, Israel), and word-initially (e.g. rey 'king'). After a stop or fricative consonant, only the tap can occur (e.g. tres 'three', frío 'cold').
 * Rhotics

In syllable-final position, inside a word, the tap is more frequent, but the trill can also occur (especially in emphatic or oratorical style) with no semantic difference—thus arma ('weapon') may be either (tap) or  (trill).

In word-final position the rhotic will usually be; The tap/trill alternation has prompted a number of authors to postulate a single underlying rhotic; the intervocalic contrast then results from gemination (e.g. tierra >  'earth').
 * either a trill or a tap when followed by a consonant or a pause, as in amo paterno 'paternal love'),
 * a tap when the followed by a vowel-initial word, as in amo eterno 'eternal love').

Obstruents
The phonemes, , and  become voiced before voiced consonants as in jazmín ('Jasmine') , rasgo ('feature') , and Afganistán ('Afghanistan'). There is a certain amount of free variation in this so that jazmín can be pronounced or.

Both in casual and in formal speech, there is no phonemic contrast between voiced and voiceless consonants placed in syllable-final position. The merged phoneme is typically pronounced as a relaxed, voiced fricative or approximant, although a variety of other realizations are also possible. So the clusters -bt- and -pt- in the words obtener and optimista are pronounced exactly the same way: Similarly, the spellings -dm- and -tm- are merged in pronunciation, as well as -gd- and -cd-:
 * obtener
 * optimista
 * adminículo
 * atmosférico
 * amígdala
 * anécdota

Vowels
Spanish has five vowels, , , and. Each occurs in both stressed and unstressed syllables:

Nevertheless, there are some distributional gaps or rarities. For instance, an unstressed high vowel in the final syllable of a word is rare.

Phonetic nasalization occurs for vowels occurring between nasal consonants or when preceding a syllable-final nasal, e.g. cinco ('five').
 * Allophones

Arguably, Eastern Andalusian and Murcian Spanish have ten phonemic vowels, with each of the above vowels paired by a lowered (or fronted) and lengthened version, e.g. la madre 'the mother' vs. las madres  'the mothers'. However, these are more commonly analyzed as allophones triggered by an underlying that is subsequently deleted.

Diphthongs and triphthongs
Spanish has six falling diphthongs and eight rising diphthongs. While many diphthongs are historically the result of a recategorization of vowel sequences (hiatus) as diphthongs, there is still lexical contrast between diphthongs and hiatus. There are also some lexical items that vary amongst speakers and dialects between hiatus and diphthong: words like biólogo ('biologist') with a potential diphthong in the first syllable and words like diálogo with a stressed or pretonic sequence of and a vowel vary between a diphthong and hiatus. hypothesize that this is because vocalic sequences are longer in these positions.

In addition to synalepha across word boundaries, sequences of vowels in hiatus become diphthongs in fast speech; when this happens, one vowel becomes non-syllabic (unless they are the same vowel, in which case they fuse together) as in poeta ('poet') and maestro  ('teacher'). Similarly, the relatively rare diphthong may be reduced to  in certain unstressed contexts, as in Eufemia,. In the case of verbs like aliviar ('relieve'), diphthongs result from the suffixation of normal verbal morphology onto a stem-final (that is, aliviar would be || + ||). This contrasts with verbs like ampliar ('to extend') which, by their verbal morphology, seem to have stems ending in. Spanish also possesses triphthongs like and, in dialects that use a second person plural conjugation,, , and  (e.g. buey, 'ox'; cambiáis, 'you change'; cambiéis, '(that) you may change'; and averiguáis, 'you ascertain').

Non-syllabic and  can be reduced to, , as in beatitud  ('beatitude') and poetisa  ('poetess'), respectively; similarly, non-syllabic  can be completely elided, as in (e.g. ahorita  'right away'). The frequency (though not the presence) of this phenomenon differs amongst dialects, with a number having it occur rarely and others exhibiting it always.

Prosody
Spanish is usually considered a syllable-timed language. Even so, stressed syllables can be up to 50% longer in duration than non-stressed syllables. Although pitch, duration, and loudness contribute to the perception of stress, pitch is the most important in isolation.

Primary stress occurs on the penultima (the next-to-last syllable) 80% of the time. The other 20% of the time, stress falls on the ultima and antepenultima (third-to-last syllable).

Nonverbs are generally stressed on the penultimate syllable for vowel-final words and on the final syllable of consonant-final words. Exceptions are marked orthographically (see below), whereas regular words are underlyingly phonologically marked with a stress feature [+stress].

In addition to exceptions to these tendencies, particularly learned words from Greek and Latin that feature antepenultimate stress, there are numerous minimal pairs which contrast solely on stress such as sábana ('sheet') and sabana ('savannah'), as well as límite ('boundary'), limite ('[that] he/she limit') and limité ('I limited').

Lexical stress may be marked orthographically with an acute accent (ácido, distinción, etc.). This is done according to the mandatory stress rules of Spanish orthography, which are similar to the tendencies above (differing with words like distinción) and are defined so as to unequivocally indicate where the stress lies in a given written word. An acute accent may also be used to differentiate homophones, such as mi (my), and mí (me). In such cases, the accent is used on the homophone that normally receives greater stress when used in a sentence.

Lexical stress patterns are different between words carrying verbal and nominal inflection: in addition to the occurrence of verbal affixes with stress (something absent in nominal inflection), underlying stress also differs in that it falls on the last syllable of the inflectional stem in verbal words while those of nominal words may have ultimate or penultimate stress. In addition, amongst sequences of clitics suffixed to a verb, the rightmost clitic may receive secondary stress, e.g. búscalo ('look for it').

Alternations
A number of alternations exist in Spanish that reflect diachronic changes in the language and arguably reflect morphophonological processes rather than strictly phonological ones. For instance, a number of words alternate between and  or  and, with the latter in each pair appearing before a front vowel:

Note that the conjugation of most verbs with a stem ending in or  does not show this alternation; these segments do not turn into  or  before a front vowel:

There are also alternations between unstressed and  and stressed  and  respectively:

Likewise, in a very small number of words, alternations occur between the palatal sonorants and their corresponding alveolar sonorants  (doncella/doncel 'maiden'/'youth', desdeñar/desdén 'to scorn'/'scorn'). This alternation does not appear in verbal or nominal inflection (that is, the plural of doncel is donceles, not *doncelles). This is the result of geminated and  of Vulgar Latin (the origin of  and, respectively) degeminating and then depalatalizing in coda position. Words without any palatal-alveolar allomorphy are the result of historical borrowings.

Other alternations include ~  (anexo vs anejo),  ~  (nocturno vs noche). Here the forms with and  are historical borrowings and the forms with  and  forms are inherited from Vulgar Latin.

There are also pairs that show antepenultimate stress in nouns and adjectives but penultimate stress in synonymous verbs (vómito 'vomit' vs. vomito 'I vomit').

Phonotactics
Spanish syllable structure can be summarized as follows, in which parentheses enclose optional components:
 * (C1 (C2)) (S1) V (S2) (C3 (C4))

Spanish syllable structure consists of an optional syllable onset, consisting of one or two consonants; an obligatory syllable nucleus, consisting of a vowel optionally preceded by and/or followed by a semivowel; and an optional syllable coda, consisting of one or two consonants. The following restrictions apply:


 * Onset
 * First consonant (C1): Can be any consonant, including a liquid.
 * Second consonant (C2): If and only if the first consonant is a stop or a voiceless labiodental fricative, then a second consonant—which can only be a liquid —is permitted. Although the onsets  and  do occur, they are not native to Spanish.
 * Nucleus
 * Semivowel (S1)
 * Vowel (V)
 * Semivowel (S2)
 * Coda
 * First consonant (C3): Can be any consonant
 * Second consonant (C4): Must be . A coda combination of two consonants only appears in loanwords (mainly from Classical Latin), never in words inherited from Vulgar Latin.
 * Medial codas assimilate place features of the following onsets and are often stressed.

Examples of maximal onsets: transporte, flaco , clave

Examples of maximal nuclei: buey, Uruguay

Examples of maximal codas: instalar, perspectiva

In many dialects, a coda can only be one consonant (one of n, r, l or s) in informal speech. So, realizations like, , are very common, and in many cases they are considered legitimate even in formal speech.

Because of these phonotactic constraints, an epenthetic is inserted before word-initial cluster beginning with  (e.g. escribir 'to write') but not word-internally (transcribir 'to transcribe'), thereby moving the initial  to a separate syllable. This epenthetic is pronounced even when it is not reflected in spelling (e.g. the surname of Carlos Slim is pronounced ). While Spanish words undergo word-initial epenthesis, cognates in Latin and Italian do not:


 * Lat. status ('state') ~ It. stato  ~ Sp. estado
 * Lat. splendidus ('splendid') ~ It. splendido  ~ Sp. espléndido
 * Fr. slave ('Slav') ~ It. slavo  ~ Sp. eslavo

Spanish syllable structure is phrasal, resulting in syllables consisting of phonemes from neighboring words in combination, sometimes even resulting in elision. This phenomenon is known in Spanish as enlace. For a brief discussion contrasting Spanish and English syllable structure, see.

Phonology
Phonological development varies greatly by individual, both those developing regularly and those with delays. However, a general pattern of acquisition of phonemes can be inferred by the level of complexity of their features, i.e. by sound classes. A hierarchy may be constructed, and if a child is capable of producing a discrimination on one level, he/she will also be capable of making the discriminations of all prior levels.


 * The first level consists of stops (without a voicing distinction), nasals,, and optionally, a non-lateral approximant. This includes a labial/coronal place difference (for example, vs  and  vs ).


 * The second level includes voicing distinction for oral stops and a coronal/dorsal place difference. This allows for distinction between, , and , along with their voiced counterparts, as well as distinction between and the approximant.


 * The third level includes fricatives and/or affricates.


 * The fourth level introduces liquids other than, and . It also introduces.


 * The fifth level introduces the trill.

This hierarchy is based on production only, and is a representation of a child’s capacity to produce a sound, whether that sound is the correct target in adult speech or not. Thus, it may contain some sounds that are not included in the adult phonology, but produced as a result of error.

Spanish-speaking children will accurately produce most segments at a relatively early age. By around three-and-a-half years, they will no longer productively use phonological processes the majority of the time. Some common error patterns (found 10% or more of the time) are cluster reduction, liquid simplification, and stopping. Less common patterns (evidenced less than 10% of the time) include palatal fronting, assimilation, and final consonant deletion.

Typical phonological analyses of Spanish consider the consonants, , and the underlying phonemes and their corresponding approximants , , and  allophonic and derivable by phonological rules. However, approximants may be the more basic form because monolingual Spanish-learning children learn to produce the continuant contrast between and  before they do the lead voicing contrast between  and. (In comparison, English-learning children are able to produce adult-like voicing contrasts for these stops well before age three.) The allophonic distribution of and  produced in adult speech is not learned until after age two and not fully mastered even at age four.

The alveolar trill is one of the most difficult sounds to be produced in Spanish and as a result is acquired later in development. Research suggests that the alveolar trill is acquired and developed between the ages of three and six years. Some children acquire an adult-like trill within this period and some fail to properly acquire the trill. The attempted trill sound of the poor trillers is often perceived as a series of taps owing to hyperactive tongue movement during production.

Codas
One research study found that children acquire medial codas before final codas, and stressed codas before unstressed codas. Since medial codas are often stressed and must undergo place assimilation, greater importance is accorded to their acquisition. Liquid and nasal codas occur word medially and at the ends of frequently-used function words, so they are often acquired first.

Prosody
Research suggests that children overgeneralize stress rules when they are reproducing novel Spanish words and that they have a tendency to stress the penultimate syllables of antepenultimately stressed words, to avoid a violation of nonverb stress rules that they have acquired. Many of the most frequent words heard by children have irregular stress patterns or are verbs, which violate nonverb stress rules. This complicates stress rules until ages three to four, when stress acquisition is essentially complete, and children begin to apply these rules to novel irregular situations.

Dialectal variation
Some features, such as the pronunciation of voiceless stops, have no dialectal variation. However, there are numerous other features of pronunciation that differ from dialect to dialect.

Yeísmo
One notable dialectal feature is the merging of and  into one phoneme (yeísmo); this was traditional in many peninsular (mainly southern) dialects. In recent times, it has spread to metropolitan areas in other parts of the Iberian Peninsula (such as Santander and Valladolid), where simply loses its laterality, and in some South American countries, where the phoneme resulting from the merger is realized as. In Buenos Aires, the sound has recently been devoiced to among the younger population, and the change is spreading throughout Argentina.

Seseo, ceceo and distinción


Speakers in northern and central Spain, including the variety prevalent on radio and television, have both and  (distinción, 'distinction'). However, speakers in Latin America, Canary Islands and some parts of southern Spain have only (seseo), which in southernmost Spain is pronounced  and not  (ceceo).

Realization of
The phoneme has three different pronunciations depending on the dialect area:
 * 1) An  alveolar retracted fricative (or "apico-alveolar" fricative), sounding to some ears a bit like English . This is characteristic of the northern and central parts of Spain and is also used by many speakers in Colombia's Antioquia department.
 * 2) A  alveolar grooved fricative, much like the most common pronunciation of English . This is characteristic of western Andalusia (e.g. Málaga, Seville, and Cádiz), Canary Islands, and Latin America.
 * 3) An  dental grooved fricative  (an ad-hoc symbol), which has a lisping quality and sounds something like a cross between English  and  (but is not the same as the  occurring in ceceo dialects). It occurs in eastern Andalusia, for example in Granada, Huelva, Córdoba, Jaén and Almería.

Obaid describes the apico-alveolar sound as follows:
 * "There is a Castilian s, which is a voiceless, concave, apicoalveolar fricative: the tip of the tongue turned upward forms a narrow opening against the alveoli of the upper incisors. It resembles a faint and is found throughout much of the northern half of Spain."

Dalbor describes the apico-dental sound as follows:
 * " is a voiceless, corono-dentoalveolar groove fricative, the so-called s coronal or s plana because of the relatively flat shape of the tongue body.... To this writer, the coronal, heard throughout Andalusia, should be characterized by such terms as "soft," "fuzzy," or "imprecise," which, as we shall see, brings it quite close to one variety of  ....  Canfield has referred, quite correctly, in our opinion, to this  as "the lisping coronal-dental," and Amado Alonso remarks how close it is to the post-dental , suggesting a combined symbol  to represent it."

In some dialects, may become the approximant  in the syllable coda (e.g. doscientos  'two hundred'). In many places it debuccalizes to in final position (e.g. niños  'children'), or before another consonant (e.g. fósforo  'match')—in other words, the change occurs in the coda position in a syllable.

From an autosegmental point of view, the phoneme in Madrid is defined only by its voiceless and fricative features. This means that the point of articulation is not defined and is determined from the sounds following it in the word or sentence. Thus in Madrid the following realizations are found: >  and  >. In parts of southern Spain, the only feature defined for appears to be voiceless; it may lose its oral articulation entirely to become, or even a geminate with the following consonant ( or  from  'same'). In Eastern Andalusian and Murcian Spanish, word-final, and  (phonetically ) regularly weaken, and the preceding vowel is lowered and lengthened:
 * > e.g. mis  ('my' pl)
 * > e.g. mes  ('month')
 * > e.g. más  ('plus')
 * > e.g. tos  ('cough')
 * > e.g. tus  ('your' pl)

A subsequent process of vowel harmony takes place so that lejos ('far') is, tenéis ('you all have') is and tréboles ('clovers') is  or.

Coda simplification
Southern European Spanish (i.e. Andalusian Spanish, Murcian Spanish, etc.) and several lowland dialects in Latin America (such as those from the Caribbean, Panama, and the Atlantic coast of Colombia) exhibit more extreme forms of simplification of coda consonants: These dropped consonants do appear when additional suffixation occurs (e.g. compases 'beats', venían  'they were coming', comeremos  'we will eat'). Similarly, a number of coda assimilations occur:
 * word-final dropping of (e.g. compás  'musical beat' or 'compass')
 * word-final dropping of nasals with nasalization of the preceding vowel (e.g. ven 'come')
 * in the infinitival morpheme (e.g. comer 'to eat')
 * the occasional dropping of coda consonants word-internally (e.g. doctor 'doctor').
 * and may neutralize to  (e.g. Cibaeño Dominican celda/cerda  'cell'/'bristle'), to  (e.g. Caribbean Spanish alma/arma  'soul'/'weapon', Andalusian Spanish sartén ), to  (e.g. Andalusian Spanish alma/arma ) or, by complete regressive assimilation, to a copy of the following consonant (e.g. pulga/purga  'flea'/'purge', carne  'meat').
 * ,, (and in southern Peninsular Spanish) and  may be debuccalized or elided in the coda (e.g. los amigos  'the friends').
 * Stops and nasals may be realized as velar (e.g. Cuban and Venezuelan étnico 'ethnic', himno ).

Notice that final dropping (e.g. mitad  'half') is general in most dialects of Spanish, even in formal speech.

These deletions and neutralizations show variability in their occurrence, even with the same speaker in the same utterance, implying that nondeleted forms exist in the underlying structure. This doesn't mean that these dialects are on the path to eliminating coda consonants, since these processes have existed for more than four centuries in these dialects. argues that this is the result of speakers acquiring multiple phonological systems with uneven control similar to that of second language learners.

In Standard European Spanish, voiced obstruents are devoiced before a pause as in ('thirst').

Loan sounds
The fricative may also appear in borrowings from other languages like Nahuatl and English. In addition, the affricates and  also occur in Nahuatl borrowings.