It is fairly common in my family. I had speech therapy when I was a kid and it is long gone.
I can do both the Same-lips-different-tongue approach and the different-lips-same-tongue way. It depends on if the “r” sound is being said like “roh/ruh” or like “er/ar”. Try both…see how the lips are different?
I had it. Speech therapy from grade 2 to grade 7 eventually cured(?) it. But now I have no detectable accent at all unless you are trained to hear it. From East TN and I sound completely different from parents and siblings. But, my r’s are all good.
Speaking of cruel, this must be a common enough problem that somebody came up with a joke about it. You’ve probably heard it…the kid with the problem is told to recite,
Robert gave Richard a rap in the ribs for roasting the rabbit so rare
Which the student repeats as:
Bob gave Dick a poke in the side for not cooking the bunny enough.
I had this problem as a child and speech therapy took care of it. I was pretty young, about four and a half, but I do remember some of the methods. My therapist taught me substitute sounds for problem words. For example, I couldn’t say “red” but I could say letter “R”. So she had me say “R” and “Ed” (name of favorite uncle). If you say it quickly it comes out as “red.”
There were several other approximations I learned and by the time I started school at age six the problem was no longer noticeable. No one in my family, myself included, knows when I stopped saying “R-Ed” and started saying “red.”
LunaV, this is a long shot, but if your state has an Office of Vocational Rehabilitation, they might be able to pay for some therapy. It’s probably worth checking out.
I took a brief course at Vandy in January and February. The professor had a slight speech difference. I couldn’t catch exactly what it was, but it was somehow becoming to her and became part of her personality.
“Rhotacism” is such a retarded name for this condition, for one thing it also means over-pronouncing /r/. What genius thought it would be a good idea to use the same word for two contrary opposites? The inability to pronounce /r/ should logically be called arhotacism (but maybe the person who first named it didn’t know any Greek). The word arhotacism would also remove the cruelty for someone whose trouble was with initial r- but can pronounce it intervocalically, as Tess and her uncle Ed found out.
I had it, and now I’m fine. Had a bit of speech therapy and hated it. I think I just grew out of the problem. I do recall the therapist saying R over and over again, and I’d try to say it, and the therapist would say “Can you hear the difference?” and honestly, I could not hear the difference between what I was saying and how I was supposed to be saying it.
So maybe if your son could record himself and listen back through headphones, that might help. Or find another therapist. Actually, the school should provide this for you.
Hah! I’ll grrowl at them, too! Arrrrrrr. Avast, ya landlubberrr. Sometimes, when I’m tired, my teeth clack together on an R. 
/r/ has got to be the hardest of all consonants to get a fix on because it lacks a clear crisp point of articulation. Especially our American alveolar approximant /ɹ/. It just sort of hangs there in the middle of space with no obvious means of support.
This was mentioned to me once in high school. Thanks, I’d forgotten about it - I’ll check into it.
The Altaic languages (like Turkish, Mongolian, and Manchu), and also Tamil, are characterized by the absence of initial r-. No native words in these languages start with r, but they use r’s a lot at the middle and end of words. This is different from languages like Hawaiian where /r/ just doesn’t exist at all.
The initial r’s that Altaic languages and Tamil have nowadays were picked up as loanwords. Often when they did, to make them more easy to pronounce they used the same technique as Tess of the Derbyville described: insert a vowel in front, called a prothetic vowel. This explains why the Hungarian word for Russian is Orosz. It was a loanword into Old Hungarian from the Old Turkish form Orus, at a time when the early Turks hadn’t yet gotten used to pronouncing initial r-. But they got used to it over the centuries and in modern Turkish they just say Rus.
In Spanish we refer to that as “un defecto de frenillo”, the frenillo being the little muscle that links the tongue to the bottom of the mouth, perpendicular to and in front of the tongue’s base. It may or may not be really a physical defect, but if (when) it is, I don’t know if there’s much to be done.
One of my teachers was Doctor Irurre. Or, as he said it, “Doctod, Id-dude” (yes, he not only couldn’t pronounce the r sounds, but he mixed the RR and R). His son had inherited the problem too, poor guy.
That’s it, I’m moving.
I went to college with a Puerto Rican woman whose first two names were Ana Mary. She wrote out the approximate pronunciation for gringos: “Ana Moddy.”
How about the classic L/R confusian in Asia? The old joke about “flied lice.” Although Thai does have an L and an R, there are some Thais who honestly can’t tell the difference between the two, and whether it comes out L or R is just a matter of chance. I remember one person back when I was up North who really did say “flied lice” no matter how much a fellow American and I tried to teach her otherwise. We told her what “lice” and “rice” meant in English, which only produced a lot of giggles. She never could get it. But this does not seem all THAT prevalent, so it’s feels strange to me that it exists at all.
Let’s make for LunaV a list of the world’s r-free languages. There are several in the New World, including some of the big ones like Cherokee, Lakota, Navajo.
Korean is another language with no initial r-, and is often connected with the Altaic family too. Korean has one phoneme represented by the letter ᄅ (riŭl) which only becomes /r/ in the middle of a word between vowels, but is /l/ at the beginning or end of a word.
When they romanize it as initial r- that’s only a phonemic transliteration in the McCune-Reischhauer romanization system, that isn’t meant to indicate actual pronunciation which varies depending on its position in a word. [l] and [r] are the allophones of the Korean phoneme /r/. The square brackets around a symbol mean it’s a phonetic sound, the actual pronunciation. The slashes around a symbol mean it’s a phoneme, which is a slot or position that can be occupied by one or more allophones, which are all classified by the speaker as the same thing even when pronounced differently.
Almost all the Turkic languages have shifted the r’s in their protolanguage to /z/. This has happened a little in English, French, and Latin too. Old English ceoren became Modern English choose. Old French chaire>Modern French chaise. The only exception to this loss of /r/ among Turkic languages is Chuvash, spoken in a subnational republic in Russia. In Chuvash the original /r/ has survived, compare Turkish kız ‘girl’, Uzbek qiz with Chuvash khĕr. So Chuvash is called “r-Turkic” and classified in a group by itself. Another explanation might be that the ancestral form of r-Turkic was a sister language of Proto-Turkic, not a daughter language like the rest. Chuvash words tend to resemble their Altaic cognates in Mongolian more than Turkic.
When the Magyar tribes lived on the Pontic steppes in the 8th-9th centuries before migrating to Hungary, they were in league with Turkic tribes and their language got a lot of Old Turkic loanwords. From these we can tell it was an r-Turkic language, for example compare Turkish yaz- ‘write’, Chuvash śyr, Hungarian ír.
What’s more, when I listen to modern Turks speaking, I hear their r’s often becoming voiced sibilants, essentially a palatalized [z]. It sounds like the ancient shift is still continuing and if it keeps up future Turks may start writing <z> in place of <r>.
What the existence of all these r-impaired languages indicates to me is that rhotacism-sufferers should know they have lots and lots of company–such that many languages have reduced or modified this troublesome sound. Myself, after studying a lot of Altaic languages and Tamil, have begun to feel that r doesn’t belong in the initial position and now it annoys me a little even though it’s plentiful in my own language. If I lived in the Göktürk realm of 7th century Central Asia, I would never need to ever hear an initial r- (unless some Indo-European-speaking Tajiks or Tocharians were in town).
My 7-year-old has this, along with a number of other speech problems. Her other problems have all been corrected, or are well on their way to being corrected (she still pronounces her “th” sound as “f” if she gets in a hurry). Her speech therapist told her they’d work on her “r” next year.
My BIL had a lot of trouble with this sound when he was younger, and a doctor found out his jaw was set too far back. My MIL declined to have the jaw repaired. He’s in his 50s and still can’t pronounce his r sounds.
My point is, in addition to seeing a speech therapist, you might try having a doctor examine him to make sure there’s no physical problem (if you haven’t already).
Heyyy! JK, I know we Aussies have “r- at- the- end” difficulties. 
Johanna, that’s quite a bit of knowledge there. Thanks.