Ask LL: parents' beliefs or infants' abilities?

« previous post | next post »

Andrew Clegg asks "Is this true?"

I'm more familiar with a different just-so story intended to explain the same alleged generalization: infants' phonetic abilities are initially limited, and this creates pressure to develop variants of words for caregivers (and other things infants are likely to want to name) that suit their preferences.

I don't know of any non-anecdotal studies of the generalization, much less of the relative strength of parental egoism and infant incapacity in explaining it; and I don't have time this morning to search. But maybe a reader can help Andrew out.


  1. Mark P said,

    October 29, 2009 @ 10:46 am

    My college roommate from Madras, India, told me their word for "mama" leaves off the initial m: "ama".

  2. Mr. B said,

    October 29, 2009 @ 11:04 am

    Call me crazy here (what do I know? I'm just a stinking English teacher), but the English word for "mother" is – wait for it – mother. (Well, insofar as there is a word for it; mama, ma, mom, etc. work, too. If the comic had been referring to a familiar term for a mother, then the statement would be closer to the truth.)

  3. Jacob said,

    October 29, 2009 @ 11:06 am

    In Finnish, mother is äiti and grandmother is mummu. Does that reflect the role of parents and grandparents?

  4. Christy said,

    October 29, 2009 @ 11:18 am

    I've heard the argument that the sound "mama" is common because it sounds like suckling. So babies wanting food might mimic eating and make some noise like "mama", which could be taken as asking for their mothers.

    No idea how accurate that could be, and it doesn't explain "papa"…

  5. Nick said,

    October 29, 2009 @ 11:26 am

    This essay (.pdf file) by Larry Trask was referenced in this Language Log post, and basically boils down to what T-Rex is saying (although it's worth reading in full).

    It also deals with why languages usually have more 'formal' parental terms, like Mr. B's comment touches on. Ryan (the cartoonist) should probably have made it more explicit that it's the 'mama' and 'papa' words that T-Rex is talking about, and not these formal words like 'mother' and 'father'.

  6. cs said,

    October 29, 2009 @ 11:28 am

    Are there languages where the terms are reversed? (Like, some variant on mama for father, or some variant on papa for mother.)

  7. Oskar said,

    October 29, 2009 @ 11:31 am

    I'm not so sure I buy it. Why is "pa" an easier phoneme to learn than "ta", "ka" or "da"? They're all stops, so they should be pretty much as easy to learn, and [t] and [k] are both voiceless just like [p]. And "na" should be as easy to learn as "ma", both of which are nasal stops.

    If the theory were true, it should be that almost all languages use simple phonemes form mother and father that babies can learn easily, but not necessarily the exact same phonemes.

  8. David Eddyshaw said,

    October 29, 2009 @ 11:38 am

    In Georgian, "mama" is "father" and "dede" is "mother".

    Just thought you'd like to know ….

    Actually quite a lot of languages have mama/dada/papa/nana type words for relatives other than parents as well (especially for grandmothers, I suspect, but uncles and aunts too.)

    Not surprising considering what a recent and localised development our "typical" modern Western nuclear family presumably is.

    This (of course) refutes the North hypothesis, as no grandmother is egotistical.

  9. Morgan said,

    October 29, 2009 @ 11:51 am

    Jacob, the native Finnish word for mother is "isoäiti". "Mummu" is a loanword from Swedish "Mommo" (approximately same pronunciation), a shortening of "Mormor", which is the actual Swedish word for grandmother.

    The Japanese words for father and mother are otoosan and okaasan. They don't really roll off the tongue, but I suppose the salient parts are "to" and "ka", "o" being a particle added before nouns for politeness? My limited acquaintance with Japanese was limited to loanwords (mama and papa) for casual use, and I am not acquainted with any older, native Japanese terms.

  10. Philip Spaelti said,

    October 29, 2009 @ 11:51 am

    @Oskar: have you ever actually seen/talked to a baby?

    Babies do indeed make the "ma" sound first and it is not a mystery why. The labial sounds "m" and "p" can be made most easily because they only require only opening/closing of the mouth, and do not require tongue control. "m" tends to come first because it doesn't even require that the baby control the oro/nasal opening. In effect the baby is not so much making an "m" as a leaky labial closure.

    Incidentally a sound of the "ti/di/(ki)" variety is usually not far behind. And thus many languages also have (baby) words of the titi varitey meaning either "father", or "breast/milk", etc. So: English "Mommy/Daddy". In (Old) Japanese "papa" was mother and "titi" was both father and breast, breastmilk.

  11. Sili said,

    October 29, 2009 @ 11:51 am

    Well, /m/ and /p(b)/ are bilabial – that somehow strikes me as 'easier' to do than anything involving the tongue.

    I have no plans about spawning, myself, but my friends' infant has just acquired /b/ as here second consonant (I believe the first was indeed /m/, but I'm not a regular in the household) – subsequently she needs to be wiped more often.

    I like the ego-explanation since it puts humans in the worst possible light – which rings true to me.

  12. Morgan said,

    October 29, 2009 @ 11:51 am

    Aha! And Philip has already answered my question! Thanks for the info. That's fascinating.

  13. Morgan said,

    October 29, 2009 @ 11:54 am

    Correction! I meant of course that "Isoäiti" is Finnish for GRANDmother (iso=large).

  14. Philip Spaelti said,

    October 29, 2009 @ 11:55 am

    As far as the English words "mother" and "father", the "-ther" reflects some original suffix also seen in "brother" and "sister". So the Indo-European nuclear-famaily terms also seem to reflect baby words "ma" and "pa".

  15. hsgudnason said,

    October 29, 2009 @ 11:55 am

    To use TRex's vocabulary: "It's a friggin' COMIC STRIP, people!" Don't be so touchy, or he'll come stomp on your house.

  16. John Cowan said,

    October 29, 2009 @ 12:01 pm

    Call me crazy here (what do I know? I'm just a stinking English teacher), but the English word for "mother" is – wait for it – mother.

    Yes, and if you look at the first syllable of mother (Latin mater shows it more clearly), you'll see the same ma babble-sound plus a suffix, the same one that shows up in father, sister, and brother in slightly different forms. Sometimes the baby talk words become the ordinary words, and then they are subject to ordinary sound-change until and unless they are replaced by fresh babble-sounds. This is happening in Italian today, where the inherited form padre (from an original pa babble, just like English father) is being replaced by the babble-sound babbo.

    In Finnish, mother is äiti

    That's a borrowing from East Germanic (in Gothic, for example, it's aiþei, where ei is how you write ii in Gothic spelling). The native Finno-Ugric word is emo.

    Why is "pa" an easier phoneme to learn than "ta", "ka" or "da"?

    Because it's easier to learn to open and close your lips, which a baby must do to suckle, than to master tongue positions.

  17. Axel said,

    October 29, 2009 @ 12:02 pm

    An informal survey of Japanese (one informant) reveals that some babies call their parents "totto" (father) and "kakka" (mother), but mama and papa is also common.
    Could it vary depending on what word the parents decide to give positive feedback to?

    But what do Finnish babies call their parents?

  18. JS Bangs said,

    October 29, 2009 @ 12:11 pm

    An anecdote: in my Romanian-speaking household our 18-month-old son has learned to say the word for "father" /tata/ before the word for "mother" /mama/, despite the fact that /m/ is supposedly easier for infants than /t/. In general, his babbling contains a wide variety of coronals and velars, but few labials.

    So whatever process favors the syllable /ma/ in matronymics, it's not universal to all infants, though of course it may still be statistically significant.

  19. Faldone said,

    October 29, 2009 @ 12:12 pm

    Here's an anonymous pdf from the University of Sussex Linguistic Department.

  20. Acilius said,

    October 29, 2009 @ 12:12 pm

    I'm sure you'll remember Roman Jakobson, as expounded by Larry Trask:

  21. Skullturf Q. Beavispants said,

    October 29, 2009 @ 12:13 pm

    I've sometimes seen baby talk represented in print as "goo goo ga ga". Is G typically one of the first consonants learned? (I also have a vague memory of reading that that consonant appears in most world languages but can no longer remember the source.)

  22. Peter M Reed said,

    October 29, 2009 @ 12:14 pm

    This is a link to a paper by Larry Trask (aimed at 1st year BA Linguistics students) [PDF], in which he discusses possible reasons for the crosslinguistic similarities in mama/papa words. There are also lots of data from different areal and genetic language groups. The paper references two Roman Jakobson papers (one is available on Google Books here).
    I hope these references are useful. FWIW, I don't think egoism is the whole story.

  23. Philip Spaelti said,

    October 29, 2009 @ 12:14 pm

    And finally we shouldn't forget Latin "mamma" which means breast(milk) and gives us the word mammal. So all languages perpetually rediscover these baby words, give them the same meanings, and incorporate them into the language as "real" words.

  24. Andrew said,

    October 29, 2009 @ 12:23 pm

    One thing that strikes me is that most of the languages mentioned in the strip are related, even if rather remotely – the only exceptions I can see are 'Chinese', Quechua and Cree. So it isn't so surprising that the word for 'mother' in those languages is similar.

  25. Faldone said,

    October 29, 2009 @ 12:25 pm

    The perpetual rediscovery is in the University of Sussex article that Acilius and I almost simultaneously linked to

  26. Andrew Clegg said,

    October 29, 2009 @ 12:48 pm


    No-one's being touchy! This is just the sort of well-considered and genial linguistic debate that a good comic strip is likely to inspire on Language Log. Are you a troll?

  27. Zubon said,

    October 29, 2009 @ 1:11 pm

    Did I learn an odd version of Japanese that I immediately thought of "haha" for "mother"? "Okaasan," yes, but "haha" is the less formal version, and babies rarely stand for formality. I don't know the proper linguistic term for how "ha" and "pa" are related in Japanese: the character is just the same, but ha° is pa (and ha" is ba). So "haha" is a version of "papa/baba," another win for T-Rex. See page 11 of the Trask PDF linked above for more.

  28. alex boulton said,

    October 29, 2009 @ 1:42 pm

    Gosh, this one got people going. On the ego front, I think we're probably mishearing mama & papa; a more likely explanation is that babies are saying "me me" all the time; except when it's "poo poo".

  29. Philip Spaelti said,

    October 29, 2009 @ 1:45 pm

    @Zubon: "Okaasan," yes, but "haha" is the less formal version, and babies rarely stand for formality.
    "Haha" is not the less formal version. In fact it is the more formal version. "Haha" is the way one talks about mothers, while "Okaasan" is the (recent traditional) form of address. [An older (formal) address form was "haha-ue", which is still heard in period-piece movies ("jidaigeki").] Of course the increasingly common form of address nowadays is "mama".

    The "h" in Japanese did indeed develop from a "p". When exactly is something of a debate, but the "p" sound is usually assigned to a Proto-Japanese period. In the historic period the sound is generally agreed to have been a labial fricative [ɸ], as is still used before [u]. In Old Japanese people could probably not distinguish p/ɸ, so borrowings from Chinese with [p] were reflected as [ɸ], and the ha-gyo characters were developped from Chinese syllables that had a [p]. The convention to add a small circle to distinguish [p] sounds was only developped late in the Edo or early Meiji period.

  30. Philip Spaelti said,

    October 29, 2009 @ 2:04 pm

    Just perhaps to clarify about formality. The words "haha" (mother) and "chichi" (father) are formal words in Japanese. But they are 'bare' with respect to politeness, so they cannot be used to address one's own mother/father or those of the person one is talking to. They can only be used in abstract discourse about mothers/fathers in general, or to talk about one's own mother/father. It is probably the later use which gave @Zubon the impression that they were less formal. Less formal ways of referring to one's own mother would be "fukuro" (literally 'the bag') or "kaachan" (without an "o" prefix).

  31. Craig Russell said,

    October 29, 2009 @ 2:32 pm

    For what it's worth, Ancient Greek also has both "mamme" and "mamma" as baby words either specifically meaning "mother" or used by infants crying to be fed.

    The comic playwright Aristophanes has a scene from his play "Clouds" (famous for including a parody of the philosopher Socrates written during his lifetime) in which a father Strepsiades is chewing out his son Phidippides for thinking that, now that he's gone to school, he's too good for his dad:

    You shameless boy! It was me that raised you;
    I figured what all your babbling meant:
    If you said "bru", I understood to bring you something to drink;
    When you said "mamma" I would bring you some food;
    Before you could say "kakka" I would pick you up, take you outside,
    And hold you out in front of me.

    In addition to the "mamma" word, it's pretty clear that "kakka" has a similar meaning across at least a few cultures as well.

  32. James C. said,

    October 29, 2009 @ 2:35 pm

    Some languages violate Jakobson’s Matronymic Generalization. Tlingit has in its poorly documented baby-talk vocabulary the form atlei (IPA /ʔatɬʰeː/) or atlee (IPA /ʔatɬʰiː/) for “mama”, which is derived from axh tlaa (IPA /ʔaχ tɬʰaː/) “my mother”. (I have no explanation for why the vowel changes.) Almost all Tlingit dialects lack labials, so *mama, *baba, *papa, and the like are impossible.

    I haven’t yet encountered the baby-talk form for father, which is axh éesh (IPA /ʔaχ ʔíːʃ/) normally. (I keep forgetting to investigate baby-talk and child-directed speech when I do fieldwork.)

  33. Pavel said,

    October 29, 2009 @ 3:03 pm

    Philip Spaelti said,

    Incidentally a sound of the "ti/di/(ki)" variety is usually not far behind. And thus many languages also have (baby) words of the titi varitey meaning either "father", or "breast/milk", etc. So: English "Mommy/Daddy". In (Old) Japanese "papa" was mother and "titi" was both father and breast, breastmilk.

    I'd never heard this before, but amazingly, in Malay, 'breast' is tetek! (I don't know if it's a coincidence, but the word for 'drip' is titik.) The more formal terms for mother and father are ibu and ayah, respectively, while the less formal terms are (e)mak and bapak.

  34. uberVU - social comments said,

    October 29, 2009 @ 3:11 pm

    Social comments and analytics for this post…

    This post was mentioned on Twitter by interests: Language Log: Ask LL: parents' beliefs or infants' abilities?

  35. J. W. Brewer said,

    October 29, 2009 @ 3:15 pm

    As in the Bangs household (but without any Romanian speakers present), my older daughter could say "dada" several months before she could say "mama," much to her mother's chagrin. For a brief period she actually seemed to have two distinguishable variants of "dada" (which I unfortunately failed to write down in IPA notation at the time), one of which was used only to refer to me and the other of which seemed to refer generically to adult human beings of either sex.

    The Jakobson/Trask hypothesis seems to assume that no one ever speaks to or in the presence of babies before the "babbling" stage begins, or at least that the babies lack the ability or interest to begin correlating the sounds they hear people making to particular referents. The first is obviously false, and I'm not sure what to make of the second one other than be dubious that either Trask or Jakobson had a basis for a scientifically-informed opinion.

  36. Nathan Myers said,

    October 29, 2009 @ 3:28 pm

    Craig: Hence "cacophony", right?

    Indonesian has "susu" for both breast and milk.

  37. Acilius said,

    October 29, 2009 @ 3:32 pm

    @JWB: I don't think that Jakobson and Trask presuppose that no one speaks to babies or that babies listen to no one. On the contrary, they assume that babies do hear the languages that the people around them speak. That assumption raises a question. Languages differ considerably in the sounds they use; yet babies the world over babble using the same few sounds. Jakobson and Trask contradict those who might try to argue that babies focus on those sounds because they are struggling to produce words they have heard, words which have meanings of particular importance to babies. Rather, babies produce sounds as they exercise their growing vocal tracts. That adults assign particular meanings to particular sounds babies produce may tell us something about those adults, but tells us rather less about the babies, and nothing at all about the sounds.

  38. Dan Milton said,

    October 29, 2009 @ 4:19 pm

    Another paper on the subject:
    Kin Tongue
    A Study of Kin Nursery Terms in Relation to Language Acquisition
    With a Historical and Evolutionary Perspective
    by Pierre J. Bancel * and Alain Matthey de l’Etang

  39. montgomery said,

    October 29, 2009 @ 4:22 pm

    @ Craig Russell: In addition to the "mamma" word, it's pretty clear that "kakka" has a similar meaning across at least a few cultures as well.

    FWIW, both "mama" and "kaka" refer to types of uncles in Nepali (mama on the mother's side, kaka on the father's). Mother is "aama", like in Hindi.

  40. jl_baldridge said,

    October 29, 2009 @ 4:31 pm

    More tangental, anecdotal fun:

    Prior to my birth, my parents were discussing what names I'd call them. They'd been thinking "Mama" and "Papa" or "Mommy" and "Daddy". But then one of my father's obnoxious friends inquired (and you must imagine this said in a most pronounced Boston accent) "What's the kid going to call you, Daaaaaddyyyyy?"

    My father drew himself up to his full height and pronounced, "The child will address me as Father Baldridge, SIR!"

    Somewhat later on, Mama advised him that they'd gotten me to say FatherSir, and that was probably good enough.

    My sister went through "saw-sir" and "faw-shir" before mastering the name.

  41. TB said,

    October 29, 2009 @ 5:02 pm

    Speaking of Japanese parent-words, I have always found it odd that the words for father (chichi) and breast/breastmilk (chichi) are homonyms in Japanese. Maybe T. Rex's theory has something to do with this as well.

  42. mingfrommongo said,

    October 29, 2009 @ 8:10 pm

    My eldest's first sound was probably something like 'blauggh', but the wife and I still argue over whether it was 'ma' or 'pa'.

  43. Mark F. said,

    October 29, 2009 @ 8:14 pm

    I don't think "parents' beliefs or infant abilities" is the right dichotomy. The limited speech abilities of infants is central to T-Rex's theory too. The question is what spin to put on parents' desire to hear the babies call them by name.

    What struck me about the "where to mama and papa come from" paper was how rarely the two meanings are reversed. I wouldn't have expected that.

  44. lhc said,

    October 29, 2009 @ 8:34 pm

    btw, although in Chinese cities people now say "mama" and "baba" for mother and father, the older words "niang" and "die" are still in very wide use in the countryside, and at least some people seem to think that mama and baba are loan words from european languages ( see discussion on baidu knowledge web here: Also, some mainlanders claim that the mama/baba terms are recent imports from hong kong and taiwan, where western influence was more widespread earlier in the twentieth century.

  45. lhc said,

    October 29, 2009 @ 8:40 pm

    oh and the babyish word for breast is mimi

  46. Filius Lunae said,

    October 29, 2009 @ 11:09 pm

    Mamá, Papá…

    In Spanish, the standard word for mother is madre (from Latin MATREM). Mamá would be the equivalent, in usage, to the English mom; and mami would then be mommy. Amá and ma are also heard, de……

  47. Aviatrix said,

    October 30, 2009 @ 12:04 am

    I'll repeat my advice to Ryan.

    "Ma" or "mom" is the sound you make when you have your mouth full of milk and your lips wrapped around someone's engorged nipple. Try it sometime! Tell her it's a linguistics experiment!

  48. Filius Lunae said,

    October 30, 2009 @ 12:45 am

    Being that Blogger doesn't support Trackbacks (sadly), I tried to use a third-party plug-in, but, alas, it didn't work. So I have to do a manual one.
    Mamá, Papá "[…] In Spanish, the standard word for mother is madre (from Latin MATREM). Mamá would be the equivalent, in usage, to the English mom; and mami would then be mommy. Amá and ma are also heard, depending on region and sociolect, though they are considered non-standard.

    The paradigm for father is very similar: padre is father (from PATREM); papá, dad (equivalent, again, in usage); papi, daddy; and, non-standard apá and pa (the first one of this pair, as above, often stigmatized). […]"

  49. vp said,

    October 30, 2009 @ 2:31 am

    For what it's worth, the first consonant sound my baby made was a voiced velar fricative.

  50. Kellen Parker said,

    October 30, 2009 @ 3:51 am

    A friend pointed out the other week that in Mandarin, all the titles of paternal extended family are repeated syllables while for the maternal extended family they're not, e.g. nainai vs waipo, yeye vs waigong. Himself a recent father had noticed his child had little trouble with repeating syllables but much more with things like "wai-po", and that it may be related to the fact that babies are often surrounded by paternal relatives and far less often by maternal.

    I buy it.

  51. Graeme said,

    October 30, 2009 @ 4:12 am

    While we're speculating our socks off, what do baby dinosaurs call their parents?

  52. Jo said,

    October 30, 2009 @ 5:52 am

    @John Cowan: I really wouldn't say "babbo" is replacing anything in Italian; aside from "Babbo Natale" – Santa Claus – it's really only used in Tuscany and has been for a long time (like, say, in Canto XXXII of the Inferno). Everywhere else, kids say papà or papi.

  53. Randy Hudson said,

    October 30, 2009 @ 8:27 am

    Dinosaur Comics always have a hover-over text, which is worth quoting:
    "Lots of languages have "ba" sounds for dads, too: "baba" in Persian, Swahili, Turkish and Bangla, Mandarin Chinese, "abba" in Aramaic and "ba" in !Kung. In other news, !Kung (the language AND people) is/are too awesome to just be mentioned in the title text here; their language uses CLICKS, that's what the "!" is!"

  54. Stephen Jones said,

    October 30, 2009 @ 8:34 am

    Sinhalese 'ama' for 'mama' and 'tata' for 'papa/daddy'.

  55. dwmacg said,

    October 30, 2009 @ 8:46 am

    Tossing another language into the ring (and in honor of Larry Trask): In Basque the word for mother is "ama" and the word for father is "aita". Grandmother and grandfather are reduplicatives of those words: "amama" and "aitite".

    With that, I've come close to exhausting my knowledge of Basque.

  56. Alon Lischinsky said,

    October 30, 2009 @ 9:10 am

    Guarani has sy (IPA [sɨ]) for mother, and túva (IPA ['tuʋa]) for father. Mapudungun has ñuke and chaw, respectively. I do not know whether there are any more intimate terms, but in any case I've never heard them. None of these seem too easy to master for the infant learner…

  57. Joe Fineman said,

    October 30, 2009 @ 1:10 pm

    Is it known what happened to the IE p@ter in Russian? Where did "otets" come from?

  58. Christian DiCanio said,

    October 30, 2009 @ 1:29 pm

    This must fall apart in Oto-Manguean languages where bilabials are often restricted (there was no Proto-OM bilabial stop).

    In Trique, for instance, 'mother' is /nni 3/ (but /nnãh 43/ for the vocative). 'Father' is /tʃe 3/, with no vocative/non-vocative distinction.

    However, there may be some bias in different cultures for how much they wish to interpret infants' vocalizations. Some cultures simply don't pay much attention to children's speech until they are intelligible. One wonders if cultures with more child directed speech are more likely to have names for "mother" with a bilabial.

  59. ChWatson said,

    October 30, 2009 @ 1:38 pm

    According to a Russian etymological dictionary, отец is a "common word of children's language" that replaced the IE word very early. It might be related to the word atta, that means "father" in Hettite, Gothic, Greek, Latin…

  60. Shannon said,

    October 30, 2009 @ 2:23 pm

    For another language point, in the Mayan K'iche' language, mother is naan and father is taat.

  61. Kent Scheidegger said,

    October 30, 2009 @ 5:45 pm

    The Thai word for father is "paw," just like Hoss called Ben. Mother is "ma" but the vowel is like the "a" in cat.

  62. marie-lucie said,

    October 30, 2009 @ 10:50 pm

    For what it's worth, the first consonant sound my baby made was a voiced velar fricative.

    Was she saying things like "gaga" or "googoo", and nothing else, or was it in response to something specific? Just because a baby produces what appears to be an identifiable sound or syllable does not mean that the baby means something by it. At 3 or 4 weeks of age my daughter said "la" when whe was crying. That does not mean she was attempting to talk. As someone else mentioned, babies go through a period of incoherent sounds ("babbling") before they produce something that sounds like an attempt at reproducing a word, especially if the parents have been saying "Say Mommy" or "Say Daddy".

    It is true that m or n for mothers and p or t for fathers (at least for "baby talk" or family address words) are extremely widespread, but they are by no mean universal as James C. shows with his Tlingit example. Another factors the "areal" one: words similar to the Tlingit one for 'mother' are found along the North Pacific Coast on both the American and the Asian side, in a variety of languages. This probably reflects intermarriage and/or slave-taking, which moves people across language boundaries, and puts in contact women from various origins with households where they are raising children (their own or those of their masters) and use some words of their own languages with the babies in their charge.

  63. Aaron F. said,

    October 31, 2009 @ 1:57 am

    In Georgian, "mama" is "father" and "dede" is "mother".

    Interesting! In Russian, "dyadya" (дядя) is "uncle."

  64. Kenny Easwaran said,

    October 31, 2009 @ 3:55 am

    Kellen Parker – that goes along with the data Trask mentions about mama/papa forms being common for all sorts of older relatives but not for younger relatives. If, in China, babies generally don't spend much time around maternal relatives until they're done babbling then babble-type words for those relatives won't replace existing words the way they will for paternal relatives.

    Hi Christian! I was hoping that Trask would mention in the article what tends to happen in languages that lack one or another of these relevant sounds. He seemed to suggest that babbling produces these sounds regardless of the language community that was present, but wasn't totally clear on whether this happens. If that's right, it would seem that either parents would be more likely to wait for babbles that fit the phonological patterns of the language before interpreting them as words, or that this might provide some pressure for the language to acquire phonemes that it lacks. (He seemed to suggest that something like this might be at play with Greek /b/ in "babbo", but it was unclear whether /b/ re-entered the language in other words first. I wonder if it also helps explain the fact that English has nasal vowels and phonemic glottal stops in words like "uhuh" and "nuh-uh" or however we write them, despite lacking these in other words.)

  65. 2009-10-30 Spike activity | Yuvablog said,

    October 31, 2009 @ 9:39 am

    […] Log discusses the hypothesis that words for mother and father (e.g. mama and papa) are so similar across languages because it's the first sounds children make and parents just assume their […]

  66. marie-lucie said,

    October 31, 2009 @ 11:41 am

    what tends to happen in languages that lack one or another of these relevant sounds

    Sometimes it is not that speakers are incapable of producing the relevant sounds, but that they associate them with baby talk, not with adult speech. As an example, it is well-known that French does not have the th sounds and that adult French speakers replace them with s, z (in France) or t, d (in Canada). Yet many francophone children are unable to produce s, z and replace them with the th sounds, which then are interpreted by adult French speakers as equivalent to s, z. When I first encountered the sound in English class, all the students burst out laughing: the teacher was trying to make us talk like babies!

  67. marie-lucie said,

    October 31, 2009 @ 11:42 am

    p.s. By "baby talk" here I mean the way small children talk, not the way adults talk to children.

  68. Timm! said,

    October 31, 2009 @ 12:14 pm

    My two cents:

    We actually discussed this in a linguistics class I took a few years ago. According to the prof, we're pretty certain that the easiest sound pattern for people to make is consonant-vowel-consonant-vowel (CVCV). We can surmise this because babies the world over babble in pretty rigid CVCV patterns and many "kiddie terms" for various things adhere more tightly to this pattern a more formal term (kitty, doggie, puppy, etc), many languages have strict CVCV patterns to their phonetics and many phonetic changes in speech are an attempt to get back to that CVCV pattern when it has been broken, particularly with addition and deletion. The prof pointed out how many people add an extra vowel sound to athlete (VCCVC) and realtor (CVCCVC) to end up with ath-a-lete (VCVCVC) and real-a-tor (CVCVCVC) and how the word west (CVCC) is oft pronounced as wes (CVC), particularly if the following word begins with a consonant to avoid a CCC cluster.

    Some consonants and some vowels are easier to produce than others. the /m/ and /a/ of mama and its cousins are particularly easy because they're pretty much all or nothing: to make the /m/ just close your lips all the way and vibrate your vocal cords. As long as your lips stay together it doesn't really matter where your tongue is or how fast you vibrate the cords; the resulting sound will essentially be /m/. For the /a/, keep on vibrating the vocal cords but open your mouth all the way and you've got it. There are no subtle distinctions between the two sounds – either the mouth is all the way closed, or it's all the way open, making it simple for even very small babies to make them.

    As for what caused mama and similar words to be so widespread across world languages, well that's kind of a chicken and egg problem. There are a lot of forces pulling in that direction, but it's probably impossible to say with any certainty what was the causal factor: humans are attracted to patterns, this is one of easiest sound patterns for humans to make, young kids need something to summon an adult, adults need a signal that's distinct to the know when they're being summoned as opposed to the kid just making noise. So, did adults just appropriate this really common pattern and assume it was meant for them, or do kids somehow know a pattern of sound will summon someone faster than random squawks? We'll probably never know.

  69. Jonathan said,

    October 31, 2009 @ 3:24 pm

    It may be worth mentioning that the reconstructed forms for PIE 'mother' and 'father' are not just different with respect to the initial consonants. The 'mother' word was *mV-H2te:r, that is, m + short vowel that's not /o/ + the kinship derivative suffix *-H2te:r (*H2 represents the so-called 'second laryngeal', some kind of dorsal fricative). I've seen the root vowel represented as /e/ or /a/; we can't tell because in every daughter language /e/ or /a/ plus the second laryngeal yields long /a:/ (in Proto-Germanic this later turned into /o:/, hence *mo:þe:r). If we reconstruct the root as *ma-, then at least the PIE form matches all those other languages typologically.

    The word for father, on the other hand, was *p-H2te:r, i.e. the root was the single consonant *p- (the first vowel in Proto-Germanic *faðe:r reflects the PG reflex of the laryngeal). What happened to the vowel, assuming that like the root *ma-, we are dealing with a 'nursery word' with the expected unmarked CV structure? A theory is that there was a vowel, but that it was not a full vowel like the /a/ in *ma, but an unstressed vowel that later disappeared from the daughter languages. The problem, of course, is that evidence for a PIE */ə/ is pretty thin: in many daughter languages, laryngeals turn into short vowels between consonants, yet this can be accounted for by epenthesis of a short vowel in the ancestors of the daughter languages, followed by loss of the laryngeal consonant. There is no need to project the epenthetic vowel back into PIE (in other daughters, laryngeals between consonants simply disappear without trace).

    Is it possible that PIE babies used the universally endearing nursery word 'ma!' for their mothers, but for their fathers the somewhat pejorative and possibly ptyalistic 'p!'?

  70. marie-lucie said,

    November 1, 2009 @ 2:04 pm

    Is it possible that PIE babies used the universally endearing nursery word 'ma!' for their mothers, but for their fathers the somewhat pejorative and possibly ptyalistic 'p!'?

    That seems to lay a lot of responsibility on the innocent shoulders of PIE babies and their hundreds of millions of descendants. "Ma" is far from universal for 'mother' as mentioned in several comments, and in some languages it is at least a component of a word for 'father'.
    And who says that the p in papa is pejorative? that early PIE babies wanted to spit on their fathers? After all, "baby words" are adults' interpretations of what babies seem to be saying. Would the fathers have wanted to reinforce a "ptyalistic" interpretation of their children's reaction to them?

  71. Jonathan said,

    November 1, 2009 @ 2:34 pm


    The question is: why does the PIE 'father' root have no vowel at all? It's not 'papa' or 'pa', it's 'p'! It's strange because the 'mother' root does have a vowel, which is cross-linguistically expected. There is no actual answer to this question, so I made a humorous attempt at one. I thought the humor was obvious, but I guess I was wrong.

  72. Bloix said,

    November 1, 2009 @ 3:06 pm

    In Hebrew, mama/mommy is "eema," mother is "ehm." Papa/daddy is "abba," father is "ahv."

  73. marie-lucie said,

    November 1, 2009 @ 4:15 pm


    You never know what people are going to take seriously, or what crazy explanations some people might invent! I am not a PIE specialist, so I can't comment on why the PIE word is reconstructed without a vowel, something which seems to need attention, but there might be other explanations for the divergence between languages as opposed to the unanimity with the 'mother' vowel.

    I wanted to emphasize the role of the parents in the transmission: if it was all left to pre-linguistic babies, parents might seize upon other syllables, since ma and pa are far from universal, and different families might have slightly different words. But parents know the words not from remembering what they themselves said at the age of 12 or 18 months but from their own knowledge of their language, transmitted through years of inteaction with other speakers. Even if babies can all produce labials, those are not usually their first sounds, but "gaga googoo" or the like do not remind parents of any actual words. It is only when parents think they recognize attempts at specific words that they start reinforcing those words, even if they have not already "primed" the baby with "Say Mommy", etc. In a family that uses "tata" or "daddy", "pa" from the baby will not be reinforced.

  74. kranky said,

    November 1, 2009 @ 11:40 pm

    I don't want to start a new thread (or branch) after so many posts, but no one seems to have noted that, in US English, at least, "mother" has been thoroughly replace by "mom," not just as an address form but referentially. The US world is now filled with "moms" of all kinds and "mothers" seem to have disappeared. Same with "dad" and "father."

    This has little or nothing to do with the child language issue. I do wish someone would do a serious study of the phenomenon though, or has someone?

  75. vp said,

    November 2, 2009 @ 1:37 am

    @marie lucie

    I have no idea whether my daughter meant anything by her voiced velar fricative. Her first word that had a clear (to us) referent was [kɐːkɯ] — "cookie". By this time she was already saying things like "mama" and "dada", but seemingly at random.

  76. vp said,

    November 2, 2009 @ 2:44 am


    I've always wondered whether *ph2ter for "father" could have been a zero-grade form of *peh2 "protect", thus meaning something like "protector". I admit I have no expertise to judge whether this little hypothesis could be correct :)

  77. John Cowan said,

    November 2, 2009 @ 6:54 pm

    My 16-month-old grandson Dorian, who lives at my house, has not yet said anything that I would call a word using fairly strict criteria — he babbles quite a bit, and still coos a lot too (to say nothing of squeals and other clearly non-linguistic ape-like noises), but he has no semantics for any of it. However, the phones in the babbles are quite complex and don't correspond to anything much he hears around him, with plenty of front rounded vowels and so on. (I sing him a German lullaby or two, but there's no evidence that he picks up his babbles from me or anyone else in particular).

    This suggests that Trask and m-l are right, and that babble sounds are explorations of the innate capabilities of the vocal tract rather than imitations of the surrounding speech. (He does, however, imitate the retroflex lateral clank, or whatever you call it — it starts out like a voiceless retroflex lateral, but the source of the sound is the tongue hitting the floor of the mouth — that I like to amuse him with.)

    "Dorian is currently an ape teaching himself to be a person." –me, to Dorian's mother, explaining Dennett's theory of the origin of individual consciousness

  78. Ariun said,

    November 3, 2009 @ 12:47 am

    >Nathan Myers said, Indonesian has "susu" for both breast and milk.

    Khalkh Mongolian has "suu" for milk, "me me" for breast in coochicoo(when adults talk to kids).

    Father is "aav", mother "eej". No Ms or Ps there! Unlike Mandarin Chinese, "baba" actually means "dirty" in Khalkh Mongolian coochicoo. Perhaps it's from the actual word "bohir".

    O linguists help me out here, what's the correct term for "coochicoo"?

  79. marie-lucie said,

    November 3, 2009 @ 12:57 am

    what's the correct term for "coochicoo"?

    Baby talk, or "motherese".

  80. ‘Mama’ and ‘Papa’ « Ken and Dot’s Allsorts said,

    November 3, 2009 @ 6:04 am

    […] Language Log, I came across this […]

  81. Nathan Myers said,

    November 4, 2009 @ 6:01 pm

    Hawaiian pidgin (and, I assume, whatever parent language loaned it) has "mimi" meaning the same as American slang "cooch", the vulva. It seems odd that American English avoids that precisely descriptive word in favor of the anatomically inaccurate "vagina", the which is not really ever visible to paparazzi.

  82. q said,

    November 5, 2009 @ 8:27 am

    @Hellen Parker:

    I think it has more to do with the fact that the "wai" in all those words is 外, meaning "outside" or "outer" – that is to say, the secondary relatives on the "outside" of the patriarchal family tree. This may tie into your statement that babies are more often surrounded by relatives from their father's family (not that I have personal knowledge of any such tendency) simply in that the words fit the ideology.

  83. Steve said,

    November 25, 2009 @ 2:58 am

    @cs ("Are there languages where the terms are reversed?"): In Uyghur, one of the only languages I know of that doesn't fit the pattern discussed above, one of the words for mother is /apa/. (the other common one is /ana/.) It's not totally reversed, though, as the words for father are /ata/ or /dada/.

RSS feed for comments on this post