Moloch and its countless congeners: the efflorescence of triliteralism

« previous post | next post »

Quoting Wikipedia, Barbara Phillips Long writes:

…[S]ince 1935, scholars have debated whether or not the term refers to a type of sacrifice on the basis of a similar term, also spelled mlk, which means "sacrifice" in the Punic language. This second position has grown increasingly popular, but it remains contested.
Barbara was inspired to look this up by Gary Wills' article on the subject in The New York Review (12/15/12), which surfaced in some commentary she had read about the most recent school massacre. In the essay, Wills wrote "The gun is our Moloch." Leaving aside her opinions on guns and public safety in the U.S., here is the link if you are curious.
Barbara's question is whether there has there been any resolution of the debate about the origins, evolution, or meaning of the word Moloch. The Wiktionary entry did not clarify things for Barbara, since there's no reference to Punic, but a reference to Ammonite:

New Latin, from Μολόχ (Molókh), Greek rendition of Hebrew מולך (mólekh, Moloch), borrowed from Ammonite  (mlk), an Ammonite god mentioned in the Pentateuch, worshipped by Canaanites and Phoenicians, said to have demanded child-sacrifice.

The Encyclopedia Britannica has a different take:
Moloch, also spelled Molech, a Canaanite deity associated in biblical sources with the practice of child sacrifice. The name derives from combining the consonants of the Hebrew melech (“king”) with the vowels of boshet (“shame”), the latter often being used in the Old Testament as a variant name for the popular god Baal (“Lord”).
Barbara concludes:
Also, I got the impression that the word Moloch or its predecessors, used in various scriptural translations, did not appear in secular texts, but I don't actually know if that is the case. Are there inscriptions on monuments or other written sources?

I can't answer her question directly, but when I started poking around for useful leads, I was astonished by how many guises and languages in which cognates of Moloch show up.  They all start from the simple little sequence of three consonants — √mlk — that constitute one of the typically productive Proto-Semitic roots.

Examples of words with the root mlk: Mameluke, Melchizedek, Melkite, Moloch.

West Semitic, to rule, dominate, possess, own; Common Semitic noun *malk-, ruler, king.

1. Melkite, from Aramaic malkāye, plural of malkāy, royal, royalist, from malkā, king.
2. Melchizedek, from Hebrew malkî-ṣedeq, my king (is) righteousness, from malk, presuffixal form of melek, king + , my.
3. Moloch, from Hebrew mōlek, from Canaanite *mulk, perhaps variant of Canaanite *malk, *milk, king.
4. Mameluke, from Arabic mamlūk, owned, slave, Mameluke, passive participle of malaka, to own, possess.

(American Heritage Dictionary of the English Language, 5th ed.)

Thus, √mlk is productive in different ways in many branches of Semitic and, in these various languages, its derivatives mean everything from "ruler" and "righteous king" to "slave" and a demonic deity demanding costly sacrifice (e.g., human children).  When √mlk jumps beyond Semitic, it becomes all the more prolific and varied in meaning.

For example, Melchior (from Hebrew מלכיאור‎ [king of light], from מלך‎ [king] + אור‎ [light]) during medieval times comes in European languages to be the name of one of the three Magi, a rather rare male given name from Old Persian, and even to signify a very large wine bottle with the capacity of about 18 liters, equivalent to 24 standard bottles.  (source)

I have known persons with the surname or name Malik.  Since the ones I knew were not themselves of Semitic descent, and I met them mostly in Central Asia and South Asia, but also in America, the origins and development of the spread of the term reveal themselves to be extraordinarily complex.

Malik, Melik, Malka, Malek, Maleek, Malick, or Melekh (Phoenician: ; Arabic: ملك; Hebrew: מֶלֶךְ) is the Semitic term translating to "king", recorded in East Semitic and Arabic, and as mlk in Northwest Semitic during the Late Bronze Age (e.g. Aramaic, Canaanite, Hebrew).

Although the early forms of the name were to be found among the pre-Arab and pre-Islamic Semites of the Levant, Canaan, and Mesopotamia, it has since been adopted in various other, mainly but not exclusively Islamized or Arabized non-Semitic Asian languages for their ruling princes and to render kings elsewhere.Malik is also the elder or chief of all tribes and Khans in Pashtuns like Malik Ahmad Khan in Yousafzai. It is also sometimes used in derived meanings.

The female version of Malik is Malikah (Arabic: ملكة; or its various spellings such as Malekeh or Melike), meaning "queen".

The name Malik was originally found among various pre-Arab and non-Muslim Semitic peoples such as the indigenous ethnic Assyrians of Iraq, Amorites, Jews, Arameans, Mandeans, Syriacs, and pre-Islamic Arabs. It has since been spread among various predominantly Muslim and non-Semitic peoples in Central Asia, the Middle East, and South Asia. Malik is also an angel in the Quran, who never smiled since the day the hellfire was created.

The last name "Malik" also refers to people belonging to the Punjab, Uttar Pradesh and Haryana region in India and Pakistan.

The earliest form of the name Maloka was used to denote a prince or chieftain in the East Semitic Akkadian language of the Mesopotamian states of Akkad, Assyria, Babylonia and Chaldea.[2][full citation needed] The Northwest Semitic mlk was the title of the rulers of the primarily Amorite, Sutean, Canaanite, Phoenician and Aramean city-states of the Levant and Canaan from the Late Bronze Age. Eventual derivatives include the Aramaic, Neo-Assyrian, Mandic and Arabic forms: Malik, Malek, Mallick, Malkha, Malka, Malkai and the Hebrew form Melek.

Moloch has traditionally been interpreted as the epithet of a god, known as "the king" like Baal was an epithet "the master" and Adon an epithet "the lord", but in the case of Moloch purposely mispronounced as Moleḵ instead of Meleḵ using the vowels of Hebrew bosheth "shame".



A note on Sanskrit mleccha

One of the most mysterious words in Sanskrit is mleccha:

mlechchha, also spelled mleccha, people of foreign extraction in ancient India. A Sanskrit term, mlechchha was used by the Vedic peoples much as the ancient Greeks used barbaros, originally to indicate the uncouth and incomprehensible speech of foreigners and then extended to their unfamiliar behaviour. Mlechchhas were found in northwestern India, and there is reason to believe that the people known in Akkadian as Mlakkha were the original mlechchhas. As a mlechchha, any foreigner stood completely outside the caste system and the ritual ambience. Thus, historically, contact with them was viewed by the caste Hindu as polluting.


By providing this information about mleccha, I do not mean to imply that there is a necessary connection with Semitic √mlk.  My only intention, considering the lack of an accepted etymology in Indo-Iranian, much less Indo-European, is to call its attention to scholars versed in Middle Eastern languages who may take mleccha into consideration for future research.

The Sanskrit word mleccha does not have a standard Indo-European etymology and has no counterpart in Iranian languages. However, it has cognates in Middle Indo-Aryan languages: Pali milakkha, and Prakrit mliccha, from the latter of which originate Sindhi milis, Punjabi milech, Kashmiri brichun (weep or lament), Western Pahari melech (dirty). The Sanskrit word occurs as a verb mlecchati for the first time in the latic Vedic text Śathapatha‐Brāhmana dated to around 700 BCE. It is taken to mean "to speak indistinctly or barbarously". Brahmins are prohibited from speaking in this fashion.

As mleccha does not have an Indo-European etymology, scholars infer that it must have been a self-designation of a non-Aryan people within India. Based on the geographic references to the Mleccha deśa (Mleccha country) to the west, the term is identified with the Indus people, whose land is known from the Sumerian texts as Meluḫḫa. Asko Parpola has proposed a Dravidian derivation for "Meluḫḫa", as mel-akam ("high country", a possible reference to the Balochistan high lands). Franklin Southworth suggests that mleccha comes from mizi meaning 'speak', or 'one's speech' derived from Proto-Dravidian for language.

Pali, the older Prakrit used by Theravada Buddhism, uses the term milakkha. It also employs milakkhu, a borrowing from a Dramatic Prakrit.



Closing questions

I've been dabbling in Semitic languages for six decades or so.  There's something about them that fascinated me from the very beginning and still captivates me today.  Namely, the virtual sacrality and near exclusivity of the triliteral roots from which they are formed.  I realize that quadriliteral roots do exist, but they are relatively few in number.  I was originally under the impression that all Semitic roots were triliteral, but when I asked Devin Stewart whether this were true, he replied:

Quadriliteral roots go back to proto-Semitic, they are in Hebrew, Arabic, and every other Semitic language. I have attached an article about proto-Semitic. [VHM:  omitted here] Clive Holes wrote an article about quadriliterals in Gulf Arabic, and someone else about Kuwaiti Arabic. I wrote something about the formation of quadriliterals in Egyptian Arabic in graduate school but I never published it. They are more frequent in modern Hebrew than they are in Biblical Hebrew, and more frequent in modern Egyptian Arabic than they are in classical Arabic.  [VHM:  significant that their number seems to be increasing with time] So, overall, I would say the basic answer is no, they are not very few in number, and in addition, their number seems to be growing. In fact, they are where a lot of the innovation of new forms happens, as in the creation of the verbal noun avwara in Egyptian, and the verb yi'avwar, etc., from ōvar "over" meaning exaggerated, excessive, or going beyond what is acceptable.

[VHM:  The general impression this gives is that the triliteral principle of Semitic morphology of Semitic languages is ever so slowly and gradually breaking down.]

I treasure my Bohtlingk & Roth and Monier-Williams Sanskrit dictionaries, my Liddell & Scott Greek-English Lexicon, and my Oxford Latin Dictionary (all arranged according to alphabetical principles), but Edward William Lane's Arabic-English Lexicon (8 vols., 2nd half of the 19th c., incomplete) is a wonder unto itself.  All of the thousands of words in this massive dictionary (more than three thousand pages) are arranged according to their triliteral roots.  From the moment I first beheld Lane's Lexicon, I was absolutely dumbfounded by this stark fact.  How could all the words of Classical Arabic be organized and locatable by their roots consisting solely of three consonants, etyma that go back to the beginnings of the Semitic language family?  Compared to Indo-European roots, which are of varying lengths and mixture of vowels and consonants, the consistently triliteral principle adhered to in Proto-Semitic and its daughter languages boggles the mind.  Just take a gander at the appendices of Indo-European and Semitic roots in The American Heritage Dictionary of the English Language and you'll see what I mean.

I do not know about the etyma of other proto-languages (would love to hear about them), but I am eager to learn from Language Log readers what they deem to be the philosophical and linguistic premises that led to the formation and maintenance of the triliteral regularity of the Semitic lexicon.


Selected readings


  1. Coby L said,

    May 29, 2022 @ 10:28 am

    Malik can also be a Slavic diminutive of a common word meaning "small"; for example, the Soviet (Ukraine-born) diplomat Yakov Malik.

  2. Peter B. Golden said,

    May 29, 2022 @ 11:16 am

    Sir Gerard Clauson's classic "An Etymological Dictionary of Pre-Thirteenth-Century Turkish" (Oxford: Clarendon Press, 1972) places all entries under a biliteral or trilaiteral heading, e.g. EMG has: emig "nipple, teat, breast, udder;" imig "(of the weather) 'mild, warm'," emge:k "pain, agony;" emgek "the fontanel, the gap in the crown of an infant's skull before the bones join up;" ömge:n "an anatomical term" in Qarakhanid Turkic ömge:n "the jugular vein; Chaghatay ömgen/ömgün "the base of the throat and the bone between the the neck and shoulder;" Khakas (a modern Siberian Turkic language) öŋmen "collar bone." For those accustomed to using Arabic dictionaries, Clauson's system (unique to him) is perfectly intelligible.

  3. Polyspaston said,

    May 29, 2022 @ 11:28 am

    Mlk appears occasionally in Egyptian texts to refer to foreign rulers/chieftains (e.g., the biographical inscription fragments of Khnumhotep III at Dahshur – see Allen, in BASOR 352 (2008)).

    Egyptian is of course not Semitic but does share some features. There are a number of what are either biliteral roots in Egyptian, or triliterals with a weak third consonant (e.g., Dd or Dd(j), though so far as I know, nobody has proposed rx(j). Quadriliterals, quinqueliterals, and hexaliterals are known. Quadriliterals are most commonly formed with the causative prefix s-, but some do exist beyond this, consisting of two radicals repeated (e.g., ptpt) or not (e.g., wstn). The hexaliterals are rare and restricted to Old Egyptian for the most part, and consider of a triliteral root doubled (e.g., nDdnDd) and probably indicating a more intense meaning than the triliteral form. As far as I know, most of the 5-radical verbs do the same thing and follow more or less the same pattern, but only partially reduplicate the root (e.g., nDdnDd > Middle Egyptian nDdDd).

  4. Scott P. said,

    May 29, 2022 @ 11:41 am


    Presumably the mlk there would be a Semitic borrowing, anyways?

  5. Ross Presser said,

    May 29, 2022 @ 11:43 am

    What about Malachi (Hebrew: מַלְאָכִי‎‎), normally translated as "messenger" or "angel?

  6. Polyspaston said,

    May 29, 2022 @ 11:55 am


    Yes – if anything the interest is in how rarely it appears, I think. Rereading Allen's article, the term (m3kj when rendered in Egyptian) is actually a hapax, appearing only in this text.

  7. Coby L said,

    May 29, 2022 @ 12:18 pm

    Ross Presser: here is a quote from Wikipedia, "Angels in Judaism", under Etymology. (I usually have trouble inserting links, so I wont even try.)

    Hebrew mal’akh (מַלְאָךְ‎) is the standard word for "messenger", both human and divine, in the Tanakh (Hebrew Bible), though it is rarely used for human messengers in Modern Hebrew[2] as the latter is usually denoted by the term shaliyakh (שליח‎). The noun derives from the verbal consonantal root l-’-k (ל-א-ך‎), meaning specifically "to send with a message" and with time was substituted with more applicable sh-l-h.[3] In Biblical Hebrew this root is attested only in this noun and in the noun "Melakhah" (מְלָאכָה‎), meaning "work", "occupation" or "craftsmanship".

    The morphological structure of the word mal’akh suggests that it is the maqtal form of the root denoting the tool or the means of performing it.[4] The term Mal'akh therefore simply means the one who is sent, often translated as "messenger" when applied to humans; for instance, Mal’akh is the root of the name of the prophet Malachi, whose name means "my messenger". In modern Hebrew, mal’akh is the general word for "angel"; it is also related to the words for "angel" in Arabic (malak ملاك), Aramaic and Ethiopic.

  8. Jamie C said,

    May 29, 2022 @ 4:23 pm

    I'd take issue with the wiktionary translation of Melchior (מלכיאור), and instead of "king of light", I'd say "my king is light" (מלכי = my king). Also, besides the quadriliteral roots, there are a number of biliteral roots. There are also some that were originally biliteral, and acquired a third root letter, sometimes in multiple ways (see 1.1 in the wiki page for Semitic root). Regarding proliferation of quadriliterals, I wonder how much of that is due to borrowings/internationalisms.

  9. Martin Schwartz said,

    May 29, 2022 @ 11:14 pm

    Ms. Long would have done better to look at Wikipedia than Wiktionary; the former at least makes clear that the Heb. has molekh (which would go back to *mulk) and not mōlekh (which would go back to a participle*mālik); the facts get confused in the ensuing discussion above. I recommend readers search Moloch + Lipiński for the 2015 article on sacrifice of children in the Old Testament (and against a god Moloch, about which E, Lipiński has much more to say in his authoritative book on the gods of the Phoenicians). Melchior is "…a rather rare given male name from Old Persian". Whaaa? Missing negative phrase? None of the canonical names of the legendary 3 Magi is from Old Persian; Balthasar is clearly Aramæo-Syriac,and Caspar with its variants is a long story, but not OP. By the way, Mandaic (sic recte) should be put with Aramaic and Syriac, all being variants of the same Semitic sub-language. In Arabic there are two words, malak 'king' and mālik 'possessor', cognate but different. Offhand (can't check now) I think that as an evil angel, Quranic mālik is from the Judaic "Moloch" word/name/ The Hebrew (> Arabic) word for 'angel' in general comes from Heb. √l-'-k (' here = āleph, glottal stop) .The maloka and mlakkha cited for Akkadian are impossible forms for that language.

  10. Ben said,

    May 30, 2022 @ 12:04 am

    @Jamie C

    It has been frequently conjectured that ALL proto-semitic roots were biliteral. Here is a dissertation on the topic:
    And here is a slightly less comprehensive, but imho more convincing, article connecting biliteral roots and "primitive" semantic domains:

    Of course, Proto-Semitic ought to be traceable to Proto-Afro-Asiatic, though the time depth is so great that it is likely unreconstructable.

    Proto-Tibeto-Burman and Proto-Indo-European roots are also often biliteral. I wonder if this is all just a case of information decay over time, such that more complex elements of the proto-language are simply unrecoverable.

  11. Ben said,

    May 30, 2022 @ 12:13 am

    As for why triliteral roots predominated, one suspects this was maintained because the morphological patterns assumed three roots. For instance, the past tense third person singular of MLK in Hebrew is maalak – the template requires three consonants for the vowels to syllabify. Biliteral roots were often extended artificially to accommodate these patterns – and true quadrilateral roots are not present until much later, after Hebrew had been in contact with Greek and other non Afro-Asiatic languages.

  12. Victor Mair said,

    May 30, 2022 @ 12:40 am


    Thanks for coming to grips with the morphological problem I posed.

  13. Victor Mair said,

    May 30, 2022 @ 7:26 am

    In preparation for writing this post, I wrote to my Semiticist colleagues as follows:

    I realize that Lane's massive (more than three thousand pages in 8
    volumes) remained incomplete at his death, and even his great-nephew, Stephen Lane-Poole, was unable to finish it.

    Three questions:

    1. How many words does it contain?

    2. Are they all arranged according to the triliteral roots?

    3. Does anyone know how many triliteral roots it contains?

    Devin Stewart replied as quoted in the "Closing questions" section of the o.p. In a separate message, he added:

    The easy one to answer: the entries are arranged by tri-literal roots.

    The other questions could probably be answered by Manfred Ullmann, but I don't imagine that you will catch him on email.

    To which I replied:

    Thank you for this too, Devin. That was the key question I wanted to have answered about Lane's Lexicon.

    As for the other two, is it safe for me to say:

    1. thousands of words in Lane's Lexicon?

    2. hundreds of triliteral roots in Lane's Lexicon?

    Finally, if Manfred Ullmann (b. 1931) does have an e-mail address, I might give it a try. Whether he answers is up to him.

    Devin answered:

    That was something of an inside Orientalist joke. What you may not know is that Manfred Ullmann still operates with the academic technology of the Zettel, the paper slip, arranged in shoe boxes, and that he regularly publishes books that are hand-written copies. I am not suggesting that he will be reluctant to answer your email, just that I don't think he believes in the entire phenomenon of email to begin with. He would probably know the answers to your questions off-hand because he compiled a meticulous supplement of Lane's dictionary that covered only the volume on the letter lām, which involved careful study of what Lane did and did not do.

    Words: Certainly many thousands of words. There are 300 words for lion, so one can extrapolate.

    Roots: The total possible number of triliteral roots is 27 cubed = 19,683, but many are not used for actual words.
    The used ones certainly are in the thousands.


    Manfred Ullmann sounds like quite a character! Hats off to Manfred!

  14. Scott P. said,

    May 30, 2022 @ 9:43 am

    There are 300 words for lion, so one can extrapolate.

    So we can conclude that lions were to ancient Semites what snow is to the Inuit?

  15. Peter Grubtal said,

    May 30, 2022 @ 11:38 am

    Scott P.
    I think your remark was jocular, and you are probably aware of the urban mythology aspect:

    On the conjectured biliteral roots of some ur-sprache, I wonder. Surely, it wouldn't allow enough permutations to form a fully functional language?

  16. martin schwartz said,

    May 30, 2022 @ 5:22 pm

    Lane: His Manners and Customs of the Modern Egyptians
    remains a delight and may be very cheaply obtained.
    Lion: David Larsen, The Names of the Lion, is a transation
    of a Classical Arabic poem enumerating many of the Arabic words for 'lion'; a PDF is online.
    Martin Schwartz

  17. Victor Mair said,

    May 31, 2022 @ 5:16 am

    From an inquisitive, anonymous colleague:


    Your _Moloch_ article is another tour de force!

    Years ago I think I found the words _sheikh_ and _sharif_ also, and perhaps also _sultan_, in the Sumerian Dictionary. Was amazed how far back these words went.


    VHM: It was easy for me to find the Semitic roots of "sheikh", "sharif / sherif", and "sultan" in the relevant appendix of the American Heritage dictionary, but — for the moment — I have not attempted to trace them back to Sumerian. Perhaps some Language Log readers who are conversant in Sumerian can do so.

    In perusing the Semitic roots appendix of AHD, however, I found two things that caught my attention:

    1. the root ṣpp ("to press down, cover, overlay"), which has a variant ṣwp, also has the quadriliteral form ṣpṣp

    2. "sherbet" has an interesting Arabic derivation (will write a separate post about that)

    Incidentally, English "sheriff" (from Middle English shirreve, from Old English sċīrġerēfa, corresponding to shire +‎ reeve) is unrelated to Arabic "sharif". (source)

  18. Hiroshi Kumamoto said,

    May 31, 2022 @ 6:20 am

    @Professor Schwartz,
    Since in the only occurrence in the NT (Matth. 2.1ff.) μάγοι ἀπὸ άνατολῶν are unnamed and their number unspecified, I'd like to know when they acquired the traditional names and the number of three. The article by Albrecht Dietrich, "Die Weisen aus dem Morgenlande", Zeitschrift für die Neutestamentliche Wissenschaft und die Kunde der Älteren Kirche 3-1, 1902, 1-14 (referred to as "in einem berühten Aufsatz" by J. Dechesne-Guillemin, "Die drei Weisen aus dem Morgenlande und die Anbetung der Zeit" Antaios, Stuttgart, VII, 1965, 234sqq., reprinted in Opera Minora III, Tehran 1978), suggests that the visit of the Parthian (Armenian) king Tiridates in 66CE who paid respect to Nero and returned by the route that was not originally planned (described in detail by Dio Cassius, book 63, ch. 1-7) was such a big event that it gave an inspiration to the author of Matth. some decades later.

  19. Philip Anderson said,

    May 31, 2022 @ 6:41 am

    @Hiroshi Kumamoto
    There is some information here:
    The standard number in Western Europe was three, but differed in other traditions. Three fitted the three gifts.

RSS feed for comments on this post