Why is the 'R' difficult?

There seems to be something special about the ‘r’ in language.

The pronunciation is always different across languages with the English sound that most of us here love and adore, the rolling ‘r’ that you get in Spanish and Hindi, the throat-clearing ‘r’ of French, the slightly off-key r/l/d sound of Japanese, and I feel like there are a few more variants that we transliterate to the ‘r’ letter when romanizing a word as well.

But then you also have the great difficulty of saying the ‘r’ sound, with rhotacism being an actual impairment, to things where 'r’s just sort of sneak into the pronunciation of words like “ruin” becoming “rurn” in some Southern dialects, the linking ‘r’ when connecting voweled words together, “lawer and order”, the disappearing ‘r’ at the end of words like “watuh”.

What’s so special about the ‘r’ family of sounds that humanity seems to have such a difficulty with it moving around, popping in, disappearing, and causing impairment?

In Thai, the five words ‘ra’, ‘la’, ‘pa’, ‘pra’, ‘pla’ are pronounced in five different ways among middle-class Bangkokians, with the ‘r’ often a rrrolling R.

But in lower-class rural Thai, the first two words above are pronounced the same as each other, and the three final words (‘pa’, ‘pra’, ‘pla’) are all pronounced as ‘pa’! Evidence: (1) When I inquire about the spelling of a name which begins with R or L, the rural Thai is often unable to tell me which letter. (2) Our late Aunt Pa sometimes had puns made on her name — puns based on the assumption that her name was the more common nickname Pla. (3) Sometimes a rural Thai trying to “put on airs” will “roll the R” in a word like ‘Pla’, which doesn’t have an R!

I don’t know why this is, beyond the general observation that sounds not learned in childhood are hard for an adult to produce. English speakers, for example, have difficulty pronouncing a word like ‘Ngo.’

Ladefoged and Maddieson’s authoritative book The Sounds of the World’s Languages devotes a chapter to rhotics — r-like sounds.

They describe how rhotics are unusual in several ways, mostly about acting partly as a vowel and partly as a consonant — they occur “close to the syllable nucleus,” and so “are likely to have syllabic variants, or to merge in various ways with contiguous vowels. Such processes, operating diachronically, are a particularly fertile source of differences between dialects of the same language.”

The OP mentioned rhotics often “disappear” — true, they sometimes do, but note there is a difference between “r-coloring” (no distinct sound is made, such as a trill or uvular aproximant, BUT the adjacent vowel changes quality — as in ‘non-rhotic’ Received British Pronunciation of English), and true disappearance (as in many African speakers of English as a second language, for whom “work” sounds to me like “wek”).

Finally, rhotics are uniquely heterogenous — there is no physical or acoustic property that unites them, as a class of sounds. So, within a language or dialect, they’ll tend to alternate freely with, or evolve into or from, sound that are perceived as other sounds in that language, but are in fact produced in a similar way (placement of the tongue, etc). Examples: “Taps, flaps, and trills all have similarities to stops because they all involve closure, and, indeed, often alternate with them. Fricative rhotics have obvious similarities to fricatives. And so on.”

Hindi has I think three different phonemic R sounds. One of them is a one-flap retroflex R. But they don’t consider it a type of R. They consider it a type of D, because its point of articulation is the same as their retroflex D. In fact the letter for this sound is a retroflex D with a dot under it — ड ड़

So the word for girl लड़की [ləɽki] I want to transliterate as larki, but most Indians want to transliterate as ladki.

One can add other languages as well. In Eastern North America, the “L” of the Lenni Lenape language has been described as a “liquid L” that seems to sound both like an “L” and an “R”, and has been transliterated as both (although usually as the “L,” as the name “Lenape” indicates.

Sadly, the last person who grew up speaking Lenape as her first tongue, Touching Leaves Woman, died a couple of decades ago, but not before she’d created a huge record of Lenape language and vocabulary.

In Urdu that same retroflex flap sound /ɽ/ is written with a modified r — they take ر‬ and make it ڑ.

So the Hindi orthography points to the etymology of the sound (historically a retroflex d between vowels), while the Urdu orthography points to its phonetics: لڑکی laṛkī

/r/ is an odd sound out. All the other consonants have defined points of articulation, where the tongue or other speech parts go *somewhere *and do something. But /r/ has nothing so definite. It just sort of awkwardly hangs out in empty space in the middle with its hands in its pockets hoping it doesn’t look foolish. It has no tangible means of support.


This thwead’s stawting to wub me the wong way!

My father started started school in the early 40s and had a problem saying R, which for a guy whose last name began with R was bit of an issue. My grandmother’s first and last name started with R, and for the purpose of this story I will call her “Rachel Ray”. A note was sent home to my grandparents, asking them in to discuss my father’s speech impediment with the teacher.

My Eastern European grandparents showed up at school, my grandmother allegedly stuck out her hand and said “Hello, I’m Wachel Way”. My father grew out of it, but I have both a niece and a 1st cousin with similar impediments.

So somewhat related to why Tony Soprano pronounces “Madonna” [madona] as “Muh-rone” [məron] because it’s supposed to be [maɾon] or Guy Fieri pronounces his name “Fee-yeddy” [fijɛdi] when it’s supposed to be [fieɾi]?

One of the defining characteristics (perhaps the defining characteristic) of a Catalan accent when speaking Spanish is precisely that the Ls are different in both languages; in most Catalan dialects, the L “moves to the ceiling” (it’s palatal vs the alveolar version of Spanish).

Thanks tothis person who thinks Spanish ortography is actually easier than it is, and who pointed out that both Ls are found in Portuguese, I was able to find the two IPA symbols: Spanish /l/, Catalan /ʎ/. For some reason the listings I find when searching for Catalan phonetics always say /l/, which is definitely not right.

Or, as another example, how Geddy Lee got his name. His given name was Gary, but his mother (a Polish Jew who survived several Nazi concentration camps as a teenager during WWII, and then immigrated to Canada) pronounced his name with what sounded like a “d” sound, leading Gary’s friends in school to think that she was calling him “Geddy.”

Yes. A similar kind of intervocalic effect like that occurs in at least some Southern Italian dialects. The retroflex ḍḍ in Sicilian that I’m familiar with occurs from a different intervocalic environment: bella > beḍḍa, cipolla > cipuḍḍa.

To transliterate the Sicilian into Devanagari, for clarity:
बेल्ला > बेड्डा, चीपोल्ला > चीपूड्डा

I know very very little Hindi, but when I read those pairs out loud, I feel like I’m saying something dirty.

Bella, beḍḍa just means beautiful. Cipolla, cipuḍḍa simply means onion.

Yes I knew the Italian words, but the flapped version written in Hindi just had a certain greasy feel to it.