Approximant

Semivowels

Some approximants resemble vowels in acoustic and articulatory properties and the terms semivowel and glide are often used for these non-syllabic vowel-like segments. The correlation between semivowels and vowels is strong enough that cross-language differences between semivowels correspond with the differences between their related vowels. Semivowels may be distinguished from the rest of approximants via a feature of prolongability; while all other approximants are prolongable, semivowels are momentary, as prolonging them results instead in the formation of their vocalic equivalents ( → ).

Vowels and their corresponding semivowels alternate in many languages depending on the phonological environment, or for grammatical reasons, as is the case with Indo-European ablaut. Similarly, languages often avoid configurations where a semivowel precedes its corresponding vowel. A number of phoneticians distinguish between semivowels and approximants by their location in a syllable. Although he uses the terms interchangeably, Montreuil (2004, p. 104) remarks that, for example, the final glides of English par and buy differ from French par ('through') and baille ('tub') in that in the French examples, the approximants appear in the syllable coda, whereas in the English ones, they appear in the syllable nucleus. This means that opaque (if not minimal) contrasts can occur in languages like Italian (with the i-like sound of piede 'foot', appearing in the nucleus: , and that of piano 'plane', appearing in the syllable onset: ) and Spanish (with a near minimal pair being abyecto 'abject' and abierto 'opened').

In a related sense, the terms semivowel and semiconsonant are often treated as synonyms in English linguistic tradition, though the latter may be regarded as somewhat antiquated. However, in other linguistic traditions, a distinction is made between the two terms. In these traditions, semivowels refer to non-syllabic vowel offglides (such as ) often found as partial components of syllable nuclei (such as in diphthongs), while semiconsonants refer to consonant onglides (such as ) often found in syllable onsets. An example demonstrating this difference can be observed in the interjections yay and wow .

Coordinate vowels

The following table demonstrates vowel correspondences with associated approximants. The § rhotic approximants listed here are not widely considered to be semivowels, but are provided as they do map to corresponding vowels.

Centralized correlates and symbolization

Although many languages have central vowels , which lie between back/velar and front/palatal , there are few cases of a corresponding approximant (or equivalently ). One is in the Korean diphthong or , though it is more frequently analyzed as velar (as in the table above), and Mapudungun may be another, with three high vowel sounds, , , and three corresponding consonants, , and , and a third one is often described as a non-labialized voiced velar fricative; some texts note a correspondence between this approximant and that is parallel to – and –. An example is liq (?) ('white').

It has been noted that the expected symbols for the approximant correlates of are ⟨⟩, ⟨⟩, or ⟨⟩ (as such, the correlate vowels may also be transcribed as ⟨⟩). An additional distinction of roundedness type may also be made between and : the former being compressed, and the latter protruded (the same distinction may be applied to and ). One might also distinguish and by slight degrees of fronting and backing (and by extension, and ), thus being referred to as post-palatal and pre-velar, respectively.

Alternation and glide-formation

In articulation and often diachronically, palatal approximants correspond to front vowels, velar approximants to back vowels, and labialized approximants to rounded vowels. In American English, the rhotic approximant corresponds to the rhotic vowel. This can create alternations (as shown in the above table).

In addition to alternations, glides can be inserted to the left or the right of their corresponding vowels when they occur next to a hiatus. For example, in Ukrainian, medial triggers the formation of an inserted that acts as a syllable onset so that when the affix is added to футбол ('football') to make футболіст 'football player', it is pronounced , but маоїст ('Maoist'), with the same affix, is pronounced with a glide. Dutch for many speakers has a similar process that extends to mid vowels:

• bioscoop → ('cinema')
• zee + en → ('seas')
• fluor → ('fluorine')
• reu + en → ('male dogs')
• Rwanda → ('Rwanda')
• Boaz → ('Boaz')

Similarly, vowels can be inserted next to their corresponding glide in certain phonetic environments. Sievers' law describes this behaviour for Germanic.

Non-high semivowels also occur. In colloquial Nepali speech, a process of glide-formation occurs, where one of two adjacent vowels becomes non-syllabic; the process includes mid vowels so that ('cause to wish') features a non-syllabic mid vowel. Spanish features a similar process and even nonsyllabic can occur so that ahorita ('right away') is pronounced . It is not often clear, however, whether such sequences involve a semivowel (a consonant) or a diphthong (a vowel), and in many cases, it may not be a meaningful distinction.

Contrast with fricatives

According to their standard definitions, approximants (such as ) contrast with fricatives (such as ) in that fricatives produce turbulence, but approximants do not. In addition to less turbulence, approximants also differ from fricatives in the precision required to produce them. When emphasized, approximants may be slightly fricated (that is, the airstream may become slightly turbulent), which is reminiscent of fricatives. However, such frication is generally slight and intermittent, unlike the strong turbulence of fricative consonants.

Spirants

Terminology and ambiguity

In discussing Spanish, Martínez-Celdrán (2004) suggested setting up a category of spirant approximants (such as ) contrasting with both semivowel approximants and fricatives, described as "non-rhotic central approximants". The term has since been used with regularity by Martínez-Celdrán and other authors, particularly when discussing the phonetics of Spanish and other Iberian languages such as Galician. Ball & Rahilly (2011) revived the term "frictionless continuant" – which historically referred more broadly to non-lateral approximants – to distinguish this same class, and continued to use the term in subsequent works.

However, as Ball (2026a) notes, both the spirant and frictionless continuant terminologies suffer from some level of ambiguity, as they both recall older (albeit now mostly disused) definitions with subtle differences from their current meaning. Other names that have been used for this class invoke fricatives, with prefix modifiers such as "weak", "approximated", and "lenited". No dominant and preferred term has yet emerged, but for simplicity, this page uses spirant approximants. The class includes all approximants that are both non-liquid (rhotic and lateral) and non-semivowel. They remain distinct from fricatives in having the characteristic non-turbulent (laminar) airflow of approximants.

Examples and related

Though spirant approximants are more constricted (having a lower F₂ amplitude), longer, and unspecified for rounding (viuda 'widow' vs. ayuda 'help'), the distributional overlap is limited. Spirant approximants can only appear in syllable onset (including word-initially, where semivowels never appear). The two overlap in distribution after and : enyesar ('to plaster') vs. aniego 'flood', and although there is dialectal and idiolectal variation, speakers may also exhibit other near-minimal pairs like abyecto 'abject' vs. abierto 'opened'. One potential minimal pair (depending on dialect) is ya visto 'already seen' vs. y ha visto 'and he has seen'.

The feature is not present in all dialects of Spanish. Other dialects differ either by merging the two and neutralizing the contrast, or conversely by enhancing the contrast through shifting the fricative to another place of articulation and potentially assibilating (such as , a feature called zheísmo), like in Rioplatense Spanish (some dialects of Rioplatense, such as the Buenos Aires dialect, even devoice to , a feature called sheísmo).

The syllabic fricatives found in some languages (sometimes referred to as "apical vowels", particularly in Sinological circles, as they behave like vowels phonologically) have also been reported to have similar features to spirant approximants, being approximant-like in their articulation.

Gutturals

For places of articulation further back in the mouth, from uvular to epiglottal (which may be grouped under the umbrella term guttural consonants), languages do not contrast voiced fricatives and approximants. Therefore, the assigned symbols in the IPA for the voiced fricatives of these places are typically also used for the approximants, and often without a lowering diacritic.

Occasionally, the glottal "fricatives" are described as approximants, since typically has no more frication than voiceless approximants (described in the section below), but they are often segmental phonations of the glottis without any accompanying manner or place of articulation.

Voicelessness

There has historically been some controversy over whether voiceless approximants can exist. Positions against the phonetic category have included the argument that voiceless airflow is necessarily turbulent and therefore by definition produces a fricative sound, while others have argued that the terminology is somewhat contradictory, as such sounds would be inaudible by definition of approximants (see § Disagreement over use of the term below). In contrast, more recent research distinguishes between turbulent (fricative-like) and laminar (vowel- or approximant-like) airflow in the vocal tract, with voiced fricatives characterized by more strongly turbulent airflow than voiceless fricatives due to tighter glottal constriction. Ball & Rahilly (1999) state that "the airflow for voiced approximants remains laminar (smooth), and does not become turbulent," while with voiceless approximants airflow is "usually somewhat turbulent"; because of this, the distinction between voiceless approximants and voiceless fricatives is not always clear. As such, voiceless approximants are rarely, if ever, distinguished phonemically from voiceless fricatives in the sound system of a language. Clark & Yallop (1995) discuss the issue and conclude "In practice, it is difficult to distinguish between a voiceless approximant and a voiceless fricative at the same place of articulation [...] there is no evidence that any language in the world makes such a distinction crucial."

Phonetic characteristics

Fricative consonants are generally said to be the result of turbulent airflow at a place of articulation in the vocal tract. However, an audible voiceless sound may be made without this turbulent airflow: Pike (1943) makes a distinction between "local friction" (as in or ) and "cavity friction" (as in voiceless vowels like and ). More recent research distinguishes between "turbulent" and "laminar" airflow in the vocal tract. It is not clear if it is possible to describe voiceless approximants categorically as having laminar airflow (or cavity friction in Pike's terms) as a way of distinguishing them from fricatives. Ball & Rahilly (1999) write that "the airflow for voiced approximants remains laminar (smooth), and does not become turbulent. Voiceless approximants are rare in the languages of the world, but when they do occur the airflow is usually somewhat turbulent." Audible voiceless sounds may also be produced by means of turbulent airflow at the glottis, as in ; in such a case, it is possible to produce an audible voiceless sound without the production of local friction at a supraglottal constriction, and thus lacking turbulence at the articulators. Catford (1982) [1977] defines both voiceless approximants and voiceless resonants, but suggests that both classes are necessarily turbulent in some form, stating that "[approximants] have non-turbulent flow when voiced; but the flow becomes turbulent when they are made voiceless [...] In voiceless resonants, such as , [...] there must always be turbulent flow through the glottis; if there were not, there would simply be silence."

Debated status

Voiceless approximants are treated as a phonetic category by (among others) Catford (1982) [1977], Ladefoged & Maddieson (1996), and Bickford & Floyd (2006). However, the term voiceless approximant is seen by some phoneticians as controversial. It has been pointed out that if approximant is defined as a speech sound that involves the articulators approaching each other but not narrowly enough to create turbulent airflow, then it is difficult to see how a voiceless approximant could be audible. As John C. Wells puts it in his blog, "voiceless approximants are by definition inaudible [...] If there's no friction and no voicing, there's nothing to hear." A similar point is made in relation to frictionless continuants by O'Connor (1973): "There are no voiceless frictionless continuants because this would imply silence; the voiceless counterpart of the frictionless continuant is the voiceless fricative." Ohala & Solé (2010) argue that the increased airflow arising from voicelessness alone makes a voiceless continuant a fricative, even if lacking a greater constriction in the oral cavity than a voiced approximant.

Ladefoged & Maddieson (1996) argue that Burmese and Standard Tibetan have voiceless lateral approximants and Navajo and Zulu voiceless lateral fricatives , but also say that "in other cases it is difficult to decide whether a voiceless lateral should be described as an approximant or a fricative". Asu, Nolan & Schötz (2015) compared voiceless laterals in Estonian Swedish, Icelandic, and Welsh and found that Welsh-speakers consistently used , that Icelandic-speakers consistently used , and that speakers of Estonian Swedish varied in their pronunciation. They conclude that there is "a range of variants within voiceless laterals, rather than a categorical split between lateral fricatives and voiceless approximant laterals".

Ball (2026a) reiterates that "it is a matter of debate as to whether voiceless equivalents of approximants can still be considered approximants or whether they should be termed fricatives". According to Ball, the arguments in favor of the recognition of voiceless approximants can be broken down into two primary positions: from a phonological standpoint, and from a phonetic one. The phonological position is one of "economical" explanations when addressing voiceless allophones of approximants in restricted environments. For example, it is simplest to describe only the devoicing of approximants within consonant clusters when preceded by voiceless consonants in English, such as in the words play and pray , rather than additionally describing changes in manner of articulation. The phonetic position focuses on the distinction of turbulent versus laminar airflow, as described above. However, it may also compound with the phonological position: as Ball notes, there is a clear auditory difference between the phonemic voiceless lateral in Welsh and the allophonic voiceless lateral in English , and demonstrates this difference with data from recordings of a bilingual Welsh–English speaker reading aloud the English word slay and the Welsh word lle ('place').

Rhotics

The phonetic features that define rhotic consonants are not well understood, despite many attempts to establish their unification. Instead, they are largely understood to be unified by their orthographic representation, such as with ⟨R r⟩ in Latin scripts, ⟨Ρ ρ⟩ in Greek scripts, and ⟨Р р⟩ in Cyrillic scripts. Similarly, within the IPA, rhotics are typically represented with a variety of symbols derived from the Latin letter ⟨R r⟩; in their unmarked forms (without diacritics), they represent a range of articulatory manners, including trills, taps, fricatives, and approximants (thus making [+continuant] one of their primary shared features). Nonetheless, there have been multiple studies showing their perceptual unity as a natural class, as well as to some extent their acoustic and phonological similarities. Furthermore, rhotics exhibit a degree of unity through contrast: they are distinctive enough for other consonant types to be defined as "non-rhotic" (i.e. [−rhotic] in featural phonology), such as the § spirant approximants described above. Studies have also shown sulcalization to be a frequent feature in rhotics.

The most widely recognized rhotic approximants are alveolar , retroflex , and uvular , along with their postalveolar and velar analogues. These approximants often exhibit free variation with one another, such as in English and in Dutch, though their target places of articulation may also be more stable in other languages. In languages where they are in free variation, they are grouped under a single phoneme often transcribed as simply ; this reflects the broader class of rhotics, typical orthography, and typographic ease rather than specifically the alveolar trill of the IPA (which may also be an allophone of this same phoneme). No language has been well-established to have a phonemic distinction between any two of these rhotic approximants.

Taps and trills

However, despite the widespread lack of a phonemic distinction between the typical rhotic approximants, many languages have a phonemic contrast between at least two general rhotic consonants; a well-known example is a contrast between a tap and a trill . In Spanish, these are typically distinguished in orthography with ⟨r⟩ for the tap and ⟨rr⟩ for the trill (though the trill may also appear in some words with a singular ⟨r⟩, and ⟨rr⟩ may be reduced to a tap in some environments). As described in the § spirants section above, Spanish is also well-known for its lenition (weakening) of plosive and fricative consonants to approximant ones (as are several other Romance languages), such as to . It has been reported by several authors that this lenition is extended to the tap and trill in some environments for multiple varieties of Spanish, where become , being referred to as an approximant tap and approximant trill, respectively. Thus, taking into account these allophones, a distinction between two rhotic approximants does exist in some languages.

Such sounds are distinct from typical taps and trills, as well as the reported tapped fricative (also usually transcribed as ) and fricative trill (usually transcribed as ) found in some languages, in that the approximant tap and trill lack the characteristic burst release of the aforementioned sounds, and maintain the characteristic formant structures of vowels and approximants. Martínez-Celdrán, Romera Barrios & Elvira-García (2026) demonstrate this distinction with a spectrogram analysis of two examples: a standard tap in pera 'pear' versus an approximant tap in dura 'hard', and a standard trill in burra 'female donkey' versus an approximant trill in adjudicar 'adjudicate'.

Nasals

Authors have occasionally used the terms "nasal approximants" and "nasalized approximants" to describe sounds with varying degrees of nasality. While they often are treated as synonyms, a distinction may be made between these two terms, particularly for usage in clinical linguistics: nasal approximants describe sounds that target nasal occlusives (such as ⟨⟩) but have insufficient articulator closure, while nasalized approximants describe sounds that target approximants (such as ⟨⟩) but have a lowering of the velum in their production or an insufficient closure of the velopharyngeal port. The distinction may be transcribed as a nasal occlusive with a downtack (such as ⟨⟩) for nasal approximants, and an approximant with a combining tilde above (such as ⟨⟩) for nasalized approximants.

Types

The following is an overview of the various types (subclasses) of approximants, aside from those already described in the sections above:

• Labial approximants: labial approximants do not involve the tongue in their articulation, with the constriction instead being primarily directed along the lips. They are considered to be the frontmost subset of § spirant approximants. For labial approximants that do involve articulation with the tongue, see § doubly articulated approximants below.

• Median approximants: median (or central) approximants involve some portion of the tongue in their articulation, with airflow directed along the midline. At their respective points of articulation, they contrast with lateral approximants (below). For § semivowels and § rhotic approximants, they may also be contrasted with median § spirant approximants of the same place.

• Lateral approximants: in lateral approximants, the center of the tongue is raised, and may make contact with the roof of the mouth. However, the defining location is the sides of the tongue, which only approach the teeth, allowing free passage of air. At their respective points of articulation, they contrast with median approximants (above).

• Lateral–median approximants: lateral–median approximants are a rare set of approximants that are produced with qualities between lateral and median approximants.

• Glottal approximants: glottal approximants may be best described as placeless with only laryngeal specification, as they take on the shapes of surrounding sounds in their articulation; see Phonation § Glottal consonants. Although similar in name, these are not the same as the glottalic approximants described below.

• Glottalic approximants: the typical airstream mechanism of approximants is pulmonic egressive, and their typical phonation state is modal voice. However, some languages also exhibit glottalic approximants, which may be realized with preglottalization (such as ), postglottalization (such as ), or laryngealization (such as ); see Glottalic sonorant. Although similar in name, these are not the same as the glottal approximants described above.

• Doubly articulated approximants: doubly articulated approximants feature two simultaneous places of articulation. For semivowels, these are typically with co-occurring labialization. For lateral approximants, these are typically with co-occurring velarization, uvularization, or pharyngealization.

Table

The following is a table of all attested approximant configurations, with two exceptions:

1. Regarding secondary articulation types, only doubly articulated approximants that have warranted the IPA to possess distinct letters for their combined qualities are included. Labialization is only shown for semivowels, as it aligns with the roundedness of their vocalic counterparts. Velarization (or uvularization or pharyngealization) is only shown for laterals ("dark l"). These types of secondary articulation, as well as palatalization, do occur with other subclasses of approximants (sometimes contrastively, such as in languages with hard–soft or broad–slender distinctions), but are not shown here.
2. For simplicity and to avoid diacritic stacking, § glottalic approximants are only shown with a preceding superscript, as in ⟨⟩. However, this should not be interpreted as preglottalization being the only realization of these sounds. Generally speaking, as with all glottalic sonorants, the timing of glottalization for approximants is fluid.

Additionally, the § labial and § glottal approximants are grouped with the § median approximants as they also apply to the § spirant subclass, though they are not actually median consonants as they do not involve the tongue as an articulator. The approximant tap and approximant trill § rhotics are grouped with their standard alveolar rhotic approximant counterpart, as no other configurations of these kinds have been reported.

For some configurations, letters shown with the lowered diacritic ⟨⟩ are often transcribed without it; the diacritic is used here to distinguish from their fricative counterparts.