"Horse Master" in IE and in Sinitic

« previous post | next post »

This is one in a long series of posts about words for "horse" in various languages, the latest being "Some Mongolian words for 'horse'" (11/7/19) — see also the posts listed under Readings below.  I consider "horse" to be one of the most important diagnostic terms for studying long distance movements of peoples and languages for numerous reasons:

  1. In and of itself, the horse represents the ability to move rapidly across the land.
  2. The spread of horse domestication and associated technology such as the chariot is traceable, affording the opportunity to match datable archeological finds with linguistic data.
  3. The symbolic, religious, military, political, and cultural significance of the horse is salient in widespread human societies outside the normal ecological reach of the animal itself.  In other words, the horse is treasured in areas far beyond its natural habitat (the Eurasian steppeland), such that it is a symbol of royal, aristocratic power and prerogative.  Indeed, for many societies, it is a sacred animal imbued with divine power.
  4. In studying the words for "horse" in various languages, we have been fortunate on Language Log to benefit from the expertise of historical linguists who have been providing cutting edge analysis of data drawn from numerous languages belonging to different groups and families.

And so forth.

As a result, we are poised to achieve some far-reaching, consequential conclusions concerning the development, dispersion, and denotations of human interaction with the horse.  Today, however, I wish to focus on the curious concatenation of terms for "master of the horse" in IE and in Sinitic.

A marshal is a military officer of the highest rank in some countries.

Here's the basic etymology:

From Middle English marchal, mareschal, marchall, from Anglo-Norman marescal, marschal and Old French marescal, mareschal (“farrier; military commander”), from Medieval Latin mariscalcus (“groom, army commander, court dignitary”), either from Frankish *marhskalk, or from Old High German marah-scalc (“horse-servant”), from Proto-Germanic *marhaz + *skalkaz (whence Old Saxon maraskalk, marahscalc). Compare English mare + shalk.


The relevant IE root for the first morpheme is:



    1. marshal from Old French mareschal, from Frankish *marha-skalkaz, horse-servant (*skalkaz, servant).
    2. mare1 from Old English mere, miere, mare, from Germanic feminine *marhjōn-.

[Pokorny marko- 700.]

In addition to "marshal", "mare" is also derived from this root.

A fuller word history of "marshal" may be found in the American Heritage Dictionary:

The Germanic ancestor of Modern English marshal is a compound made up of *marhaz, "horse" (related to the source of our word mare), and *skalkaz, "servant," meaning as a whole literally "horse servant," hence "groom." The Frankish descendant of this Germanic word, *marahskalk, came to designate a high royal official and also a high military commander—not surprising given the importance of cavalry in medieval warfare. Along with many other Frankish words, *marahskalk was borrowed into Old French as mareschal in the early Middle Ages, when much of northern France was ruled by Frankish dynasties. Later, when the Normans established a French-speaking official class in England in the 11th century, the Old French term mareschal came with them. In the first known uses of the word in documents written in England, marshal was used with the meaning "farrier." (It was also recorded as a surname, and in the spelling Marshall, it still survives as such.) The word marshal eventually began to be used in a wider variety of meanings in Middle English, as it had been in Old French, and the term was applied in Middle English to high-ranking officers of the royal court and the courts of law.

As for the second morpheme of "marshal", its etymology is as follows:

From Middle English schalk, scalk, from Old English scealc (“servant; man, soldier, sailor”), from Proto-Germanic *skalkaz (“servant, knight”), from Proto-Indo-European *(s)kelH- (“to cleave, separate, part, divide”). Cognate with German Schalk (“joker”), Old Norse skálkr (“servant, rogue”) ( > Danish and Swedish skalk), Gothic (skalks, “servant”).


Most people know the term sīmǎ 司馬 as a disyllabic surname that is still in use today (see here for a list of persons with this surname from antiquity to the present day).  But before it was a surname, sīmǎ 司馬 was the ancient designation of an official in charge of horses, which is fundamentally what the two morphosyllables that constitute it mean, Minister of War (< in charge of cavalry) already in pre-Han times (202 BC-220 AD).

Sīmǎ 司馬 is an archaic term for "Grand Marshal" (N.B.), an important official post during the Han Dynasty, but also early on could signify "major", a low-level official post within a commandery (the person in this position handled military affairs within the commandery).  The title obviously derived from the power and prestige of the animal.

Here's the complete entry (using Wade-Giles Romanization) from Charles O. Hucker's masterful A Dictionary of Official Titles in Imperial China (Stanford:  Stanford University Press, 1985):

Page 452ab

ssū-mă 司馬
  Lit., to be in charge of horses, i.e., of cavalry; a title de
riving from high antiquity and used through most of imperial history; prefixes are especially to be noted with care, e.g., ta ssu-ma, shao ssu-ma.

(1) CHOU: common abbreviated reference to the Minister of War (ta ssu-ma), head of the Ministry of War (hsia-kuan) and paramount military dignitary in the royal government.

(2) CHOU: Commander, common generic reference to all, or an abbreviated reference to one, of the military officers serving under the Minister of War with such titles as Vice Minister of War (shao ssu-ma), Cavalry Commander of the Army (chün ssu-ma), Commander of Chariots (yü ssu-ma), and Cavalry Commander on Campaign (hsing ssu-ma).

(3) HAN-N-S DIV: common abbreviated reference to the Defender-in-chief (t'ai-wei, ta ssu-ma), one of the eminent central government dignitaries collectively called the Three Dukes (san kung).

(4) HAN-N-S DIV: Commander, title of a 2nd-level military officer found in many agencies, e.g., on the staff of the Chamberlain for the Imperial Insignia (chung-wei, chih chin-wu), who commanded the Northern Army (pei-chün) at the dynastic capital. HB: major.

(5) N-S DIV— SUNG: erratically used as a title for 2nd- or 3rd-IeveI executive officials, i.e., Vice … or Assistant …,in territorial units of administration such as Regions (chou), Area Commands (tu-tu fu), Princely Establishments (wang-fu), Commanderies (chün), Prefectures (chou, fu); normally rank 4b or lower; commonly alternating with the title chih-chung, q.v. RR: administrateur supérieur. SP: administrateur supérieur, sous-directeur du bureau, surintendant-adjoint.

(6) SUI-T'ANG; Adjutant, a 2nd- or 3rd-level executive officer found in most military Guards (wei) stationed at the dynastic capital. RR: administrateur supérieur.

(7) CHIN— YÜAN: Adjutant, rank 6b in Chin, 4 of rank 4a in Yüan, on the staff of each Princely Establishment (wang-fu), specifically in charge of police security. P69.

(8) MING- CH'ING: deriving from the usage described in (5) above, an unofficial reference to a Vice Prefect (t'ung-chih) in a Prefecture (fu) and, in Ch’ing, also to a Vice Magistrate (also t'ung-chih) in a Department (chou).

(9) MING- CH'ING: deriving from the usage described in (1) above, an unofficial reference to executive officials of the Ministry of War (ping-pu), with the prefix ta indicating a Minister of War (ping-pu shang-shu), with the prefix shao indicating a Vice Minister of War (ping-pu shih-lang).

Let's go back to the basics for the construction of sīmǎ 司馬.  It consists of two morphemes meaning respectively "take charge of, control, manage; officer" and, of course, "horse".

The Old Sinitic reconstructions of these two morphosyllables are:


(BaxterSagart): /*s-lə/ (Zhengzhang): /*slɯ/


(BaxterSagart): /*mˤraʔ/ (Zhengzhang): /*mraːʔ/

There were other high ranking official titles of early antiquity that were formed on the same basis as "sīmǎ 司馬" ("marshal"), i.e., "sīX 司X" ("in charge of X") and later became surnames, e.g., "sītú 司徒" ("in charge of disciples or followers; Minister of Education") and "sīkōng 司空" ("in charge of digging", such as the digging of canals; "Minister of Works").

(BaxterSagart): /*[d]ˤa/ (Zhengzhang): /*l'aː/

(BaxterSagart): /*kʰˤoŋ/ (Zhengzhang): /*kʰoːŋ/

The materials presented here are insufficient to demonstrate a direct connection between "marshal" and "sīmǎ 司馬", but the parallels are intriguing.

More to come about the early spread of the domesticated horse from its homeland in the area around the Pontic-Caspian Steppe across Eurasia, including eastward to Northeast Asia, and East Asia, including the southern part of what is now China.



[Thanks to Yixue Yang]


  1. pamela said,

    November 9, 2019 @ 9:40 am

    this is very interesting and of course the first thing one thinks of is Homer's Ἄρχιππος "master of horses," the epithet of a couple of characters in Homer and at least one reference in Sophocles, and it is also in the Greek Old Testament and the New Testament. the semantic architecture is the same as 司馬, and the timing is very roughly coterminous. but is a shift from akw- based IE references to mar- based references related to something? changes in warfare or transportation? in Hammer & Anvil I briefly commented that mar- type lexeis were ubiquitous across Eurasia from late Iron Age on (in English, for instance, "equis" etc only survive as references to classical or Romance languages), but why the shift? have people offered clues on this?

  2. Chris Button said,

    November 9, 2019 @ 9:43 am

    In his Bingbian analysis, Ken-ichi Takashima provides a good discussion of how 馬 could mean "horse" or "marshal" depending on context.

  3. Robert Drews said,

    November 9, 2019 @ 10:09 am

    And on the Greek side, we also have hippodamos, a formulaic adjective that Homer uses for Hektor and the Trojans. It's usually translated, "tamer of horses," but is better translated "master of horses." In the Iliad a chariot driver is hippodamos.

  4. Thomas Hutcheson said,

    November 9, 2019 @ 11:44 am

    I understood "martial" to derive from the Roman war god "Mars." is that in turn derived from the "horse" root?

  5. pamela said,

    November 9, 2019 @ 7:46 pm

    ah! earlier I thought remembered hector as "master of horses" but when I went to check the text the epithet applied to several others, not to him. tamer of horses, I see.

    yes, Manchu morin comes from Mongolian (as do the horses themselves).

    Mars name seems to come from some Mavors, some combination of words for "battle" and "turning" –this idea of "turning" as related to one who controls fate is also a nice puzzle (轉輪王,cakravartin, -vors in Mavors).

  6. Michael Watts said,

    November 9, 2019 @ 8:08 pm

    Martial does derive from Mars. Mars does not derive from the Indo-European "horse" root. But "martial" is also totally unmentioned in the post — why bring it up?

  7. Victor Mair said,

    November 9, 2019 @ 8:22 pm

    Thomas Hutcheson asked about "martial" out of healthy curiosity. pamela answered him in a helpful and fruitful way.

  8. David Marjanović said,

    November 10, 2019 @ 5:33 am

    Horses were also native to the temperate forests of Eurasia, or at least Europe.

    There is no evidence that *márkos can be reconstructed all the way down to Proto-Indo-European. Its descendants are exclusively found in Germanic and Celtic. On top of that, there's a suspicious *a in it – but that issue goes away if we assume (based on no other evidence) that the Celtic word was borrowed from Germanic before Germanic shifted its consonants. If so, the *a could have been a PIE *o; but that doesn't bring us closer to an identified PIE root as far as I know. – If it's not inherited from PIE, it should be a loan, so one wonders if it could have come from the east. But the absence of traces in Iranian argues against that.

    But "martial" is also totally unmentioned in the post — why bring it up?

    Because it's pronounced the same as marshal in English and therefore confused with it a lot. I can't tell you how often I've seen marshal(l) law!

    (Well, Google has seen marshal law 322,000 times and marshall law 1,520,000 times, but both include the names of comics and other names that are probably deliberate puns lost on most of their readers.)

  9. David Marjanović said,

    November 10, 2019 @ 5:35 am

    if we assume (based on no other evidence) that the Celtic word was borrowed from Germanic before Germanic shifted its consonants

    Or, conversely, if we assume (based on no other evidence, but plenty of precedents) that the Germanic word was borrowed from Celtic after Germanic shifted its consonants. In that case, we could be looking at a PIE *mr̩kós… whatever that might mean.

  10. David Marjanović said,

    November 10, 2019 @ 6:35 am

    I come here too irregularly!

    Although my point that there probably never was a *márkos in PIE stands so far, I should have read the comments from here on in the thread on the Korean word first.

  11. Victor Mair said,

    November 10, 2019 @ 8:16 am

    Please do not pontificate before reading the relevant posts and comments.

  12. Gary said,

    November 10, 2019 @ 8:33 am

    Doesn't Mongolian have like 20 words for horse depending on what the horse will be used for?

  13. pamela said,

    November 10, 2019 @ 9:09 am

    I have always been interested in the modulations of ekwa- type and -mar type words for "horse," as well as germanic languages having their own root word in hras-type words (which in fact modern German doesn't use, preferring pferde from a Latin term paraveredus). -mar words seem to me to be eurasian in scope (in my book I associated it with a horse "vector" that was continental), but at the western end the proliferation of words for horse seems to have corresponded to growing uses for horses, even before lineage breeding, after which more specialized words developed.

  14. Chris Button said,

    November 10, 2019 @ 9:19 am

    I think there are two competing discussions here.

    The first is whether a form like *marko- can be reconstructed at the PIE level to parallel *h₁éḱwo-, or whether it needs to be kept at the Proto-Celtic and Proto-Germanic levels. There is a parallel in Tibeto-Burman there where Written Tibetan rta "horse" bears no resemblance to Written Burmese *mraŋ "horse". Then in the 1950s, F.W. Thomas' discovered the Tibetan inscriptional form rmaŋ "horse" (from earlier *mraŋ as Coblin observed). Perhaps other evidence in PIE for *márkos has disappeared (as Pulleyblank suggested), or perhaps it was never there in the first place. Needless to say, the Celtic and Germanic forms are nonetheless very real.

    The second discussion is whether the Celtic and Germanic forms have any connection with Old Chinese 馬 *mráɣʔ and the other similar looking words in Mongolian, Japanese, Korean, etc. (and perhaps also the Burmese and Tibetan forms above, but this also has its own problems since other Tibeto-Burman forms suggest the root may simply have been *raŋ, as Benedict and Pulleyblank mentioned). Personally, on the basis of phonological and historical evidence, I find it highly unlikely that the similarity between Old Chinese 馬 *mráɣʔ and Proto-Celtic *markos at least could have occurred by chance.

    As for Old Chinese 馬 *mráɣʔ, I know I'm beating a dead horse, but most people working in Old Chinese have thrown the baby out with the bath water in (correctly) removing an Old Chinese *-g coda to (incorrectly) leave a phonologically open syllable instead of something like Pulleyblank's -ɣ or -ɰ. People have mistakenly criticized Li Fang-Kuei for reconstructing Old Chinese with no open syllables, but this represents a fundamental misunderstanding of the difference between surface phonetics and underlying phonology. It's that age old question of whether a diphthong at the end of a syllable should be treated as closed by a glide or open as a vocalic diphthong. Unfortunately the distinction between phonetics and phonology does not always make it into the historical discussion. And reconstructions of Old Chinese are very much the worse for it.

  15. Nelson Goering said,

    November 10, 2019 @ 11:28 am

    Since 'horse' itself has been brought up, its most likely etymology is as a derivative related to 'hurry'. The basic verb is found in Latin currō 'run', going back to *√kers- 'run, move quickly', of Western IE distribution. (LIV only recognizes this root in Latin, but Kroonen lists plausible cognates in Germanic, Celtic, and Greek.) A noun like *kŕ̥sos (the position of the accent is secure by Verner's law) has the look of a substantivized adjective (though I don't know of any reflexes of a possible **kr̥sós 'fast, rushing'), so maybe 'fast thing, running thing'. This would regularly give Proto-Germanic *hursaz, as well as Proto-Celtic *karros 'wagon, vehicle'.

    (Kroonen also mentions an older suggestion that the word could be a borrowing from the Alanic precursor to Ossetic wyrs/urs 'stallion' < *uršan- 'male'. But the Germanic initial *h- is a problem for this idea.)

RSS feed for comments on this post