In phonetics, rhotic consonants, or "R-like" sounds, are liquid consonants that are traditionally represented orthographically by symbols derived from the Greek letter rho, including ⟨R⟩, ⟨r⟩ in the Latin script and ⟨?⟩, ⟨p⟩ in the Cyrillic script. They are transcribed in the International Phonetic Alphabet by upper- or lower-case variants of Roman ⟨R⟩, ⟨r⟩: r, ?, ?, ?, ?, ?, ?, and ?.
This class of sounds is difficult to characterise phonetically; from a phonetic standpoint, there is no single articulatory correlate (manner or place) common to rhotic consonants. Rhotics have instead been found to carry out similar phonological functions or to have certain similar phonological features across different languages. Although some have been found to share certain acoustic peculiarities, such as a lowered third formant, further study has revealed that this does not hold true across different languages. For example, the acoustic quality of lowered third formants pertains almost exclusively to American varieties of English.
Being "R-like" is an elusive and ambiguous concept phonetically and the same sounds that function as rhotics in some systems may pattern with fricatives, semivowels or even stops in others--for example, the alveolar tap is a rhotic consonant in many languages; but in North American English it is an allophone of the stop phoneme /t/, as in water. It is likely that rhotics, then, are not a phonetically natural class, but a phonological one instead.
Some languages have rhotic and non-rhotic varieties, which differ in the incidence of rhotic consonants. In non-rhotic accents of English, /r/ is not pronounced unless it is followed directly by a vowel.
The most typical rhotic sounds found in the world's languages are the following:
In broad transcription rhotics are usually symbolised as /r/ unless there are two or more types of rhotic in the same language; for example, most Australian Aboriginal languages, which contrast approximant [?] and trill [r], use the symbols r and rr respectively. The IPA has a full set of different symbols which can be used whenever more phonetic precision is required: an r rotated 180° [?] for the alveolar approximant, a small capital R [?] for the uvular trill, and a flipped small capital R [?] for the voiced uvular fricative or approximant.
The fact that the sounds conventionally classified as "rhotics" vary greatly in both place and manner in terms of articulation, and also in their acoustic characteristics, has led several linguists to investigate what, if anything, they have in common that justifies grouping them together. One suggestion that has been made is that each member of the class of rhotics shares certain properties with other members of the class, but not necessarily the same properties with all; in this case, rhotics have a "family resemblance" with each other rather than a strict set of shared properties. Another suggestion is that rhotics are defined by their behaviour on the sonority hierarchy, namely, that a rhotic is any sound that patterns as being more sonorous than a lateral consonant but less sonorous than a vowel. The potential for variation within the class of rhotics makes them a popular area for research in sociolinguistics.
This section needs expansion with: an introduction explaining what "variable rhoticity" means in terms that summarize the rationale for binding all of its subsections under it, clarifying what they have in common. You can help by adding to it. (May 2020)
English has rhotic and non-rhotic accents. Rhotic speakers pronounce a historical /r/ in all instances, while non-rhotic speakers only pronounce /r/ at the beginning of a syllable.
The rhotic consonant is dropped or vocalized under similar conditions in other Germanic languages, notably German, Danish and Dutch from the eastern Netherlands (because of Low German influence) and southern Sweden (possibly because of its Danish history). In most varieties of German (with the notable exception of Swiss Standard German), /r/ in the syllable coda is frequently realized as a vowel or a semivowel, [?] or . In the traditional standard pronunciation, this happens only in the unstressed ending -er and after long vowels: for example besser ['b?s?], sehr [ze:]. In common speech, the vocalization is usual after short vowels as well, and additional contractions may occur: for example Dorn [dn] ~ [d?:n], hart [hat] ~ [ha:t]. Similarly, Danish /r/ after a vowel is, unless followed by a stressed vowel, either pronounced  (mor "mother" [mo], næring "nourishment" ['ne?]) or merged with the preceding vowel while usually influencing its vowel quality (/a(:)r/ and /?:r/ or /?r/ are realised as long vowels [a:] and [?:], and /?r/, /r?/ and /r?r/ are all pronounced [?]) (løber "runner" ['lø:b], Søren Kierkegaard (personal name) [?soe:?n 'k?i:?]).
In Asturian, word final /r/ is always lost in infinitives if they are followed by an enclitic pronoun, and this is reflected in the writing; e.g. The infinitive form dar [dar] plus the 3rd plural dative pronoun "-yos" da-yos [da'?os] (give to them) or the accusative form "los" dalos [da'los] (give them). This will happen even in southern dialects where the infinitive form will be "dare" [da're], and both the /r/ and the vowel will drop (da-yos, not *dáre-yos). However, most of the speakers also drop the rhotics in the infinitive before a lateral consonant of a different word, and this doesn't show in the writing. e.g. dar los dos [da: los ðos] (give the two [things]). This doesn't occur in the middle of words. e.g. the name Carlos [kar'los].
In some Catalan dialects, word final /r/ is lost in coda position not only in suffixes on nouns and adjectives denoting the masculine singular and plural (written as -r, -rs) but also in the "-ar, -er, -ir" suffixes of infinitives; e.g. forner [fur'ne] "(male) baker", forners [fur'nes], fer ['fe] "to do", lluir [?u'i] "to shine, to look good". However, rhotics are "recovered" when followed by the feminine suffix -a [?], and when infinitives have single or multiple enclitic pronouns (notice the two rhotics are neutralized in the coda, with a tap [?] occurring between vowels, and a trill [r] elsewhere); e.g. fornera [fur'ne] "(female) baker", fer-lo ['fer?u] "to do it (masc.)", fer-ho ['fe?u] "to do it/that/so", lluir-se [?u'ir.s?] "to excel, to show off".
Final R is generally not pronounced in words ending in -er. The R in parce que (because) is not pronounced in informal speech in French.
In Indonesian, which is a form of Malay, the final /r/ is pronounced, it has varying forms of Malay spoken on the Malay Peninsula. In Indonesia, it is usually a tap version, but for some Malaysian, it is a retroflex r.
Historical final /r/ has been lost from all Khmer dialects but Northern.
In some dialects of Brazilian Portuguese, /r/ is unpronounced or aspirated. This occurs most frequently with verbs in the infinitive, which is always indicated by a word-final /r/. In some states, however, it happens mostly with any /r/ when preceding a consonant. The "Carioca" accent (from the city of Rio de Janeiro) is notable for this.
Among the Spanish dialects, Andalusian Spanish, Caribbean Spanish (descended from and still very similar to Andalusian and Canarian Spanish), Castúo (the Spanish dialect of Extremadura), Northern Colombian Spanish (in cities like Cartagena, Montería, San Andrés and Santa Marta, but not Barranquilla, which is mostly rhotic) and the Argentine dialect spoken in the Tucumán province may have an unpronounced word-final /r/, especially in infinitives, which mirrors the situation in some dialects of Brazilian Portuguese. However, in Antillean Caribbean forms, word-final /r/ in infinitives and non-infinitives is often in free variation with word-final /l/ and may relax to the point of being articulated as /i/.
The native Thai rhotic is the alveolar trill. The English approximants /?/ and /l/ are used interchangeably in Thai. That is, Thai speakers generally replace an English-derived R(?) with an L(?) and when they hear L(?) they may write R (?).
In Istanbul Turkish, /r/ is always pronounced, with the exceptions in colloquial speech being the present continuous tense suffix yor as in gidiyor ('going') or yaz?yordum ('I was writing') and bir ('one') when used as an adjective/quantifier (but not other numbers containing this word, such as on bir ('eleven')). In these cases, the preceding vowel is not lengthened. The unfavorability of dropping /r/ can be explained with minimal pairs, such as çald? ('stole') versus çald?r (imperative 'ring').
In some parts of Turkey, e.g. Kastamonu, the syllable-final /r/ is almost never pronounced, e.g. "gidiya" instead of "gidiyor" (meaning "she/he is going"), "gide" instead of "gider" (meaning "she/he goes"). In "gide", the preceding vowel e is lengthened and pronounced somewhat between an e and a.
Among the Turkic languages, Uyghur displays more or less the same feature, as syllable-final /r/ is dropped, while the preceding vowel is lengthened: for example Uyghurlar [':la:] 'Uyghurs'. The /r/ may, however, sometimes be pronounced in unusually "careful" or "pedantic" speech; in such cases, it is often mistakenly inserted after long vowels even when there is no phonemic /r/ there.
Similarly in Yaqui, an indigenous language of northern Mexico, intervocalic or syllable-final /r/ is often dropped with lengthening of the previous vowel: pariseo becomes [pa:'seo], sewaro becomes [sewajo].
Lacid, whose exonyms in various literature include Lashi, Lachik, Lechi, and Leqi, is a Tibeto-Burman language spoken by the Lacid people. There are various reports of their population size ranging from 30,000 to 60,000 people. The majority are in Myanmar but there are also small groups located in China and Thailand. Noftz (2017) reports finding an example of a rhotic alveolar fricative in Lacid while doing phonological research at Payap University in Thailand in 2015. He was not able to continue his research and expressed the need for further examination of the segment to verify his results. It is postulated that the segment is a remnant of the rhotic fricative in Proto-Tibeto-Burman.
The Shekaki accent of the Kurmanji dialect of Kurdish is non-rhotic, that is the postvocalic flap "r" is not pronounced but the trill "R" is. When r is omitted, a "compensatory lengthening" of the preceding vowel takes place. For example:
Shekaki retains morphological syllables instead of phonological syllables in non-rhotic pronunciation.