The centum-satem division is an isogloss of the Indo-European language family, related to the different evolution of the three dorsal consonant rows of the mainstream reconstruction of Proto-Indo-European:
| *kʷ, | *gʷ, | *gʷʰ | (labiovelars) |
| *k, | *g, | *gʰ | ("plain velars") |
| *ḱ, | *ǵ, | *ǵʰ | ("palatovelars") |
The terms Centum Group and Satem Group come from the words for the number "one hundred" in a traditional representative language of each group: Latin centum and Avestan satəm. The initial consonant in these two examples comes from the Indo-European "palatovelar" consonant, *ḱ, which became in the first case a simple velar, and in the second a sibilant.
The terms "palatovelar" and "plain velar" are in quotes because they are traditional terms but do not reflect current thinking, which holds that the "palatovelars" were actually plain velars, e.g. [k], while the "plain velars" were pronounced farther back in the mouth, perhaps as uvular consonants (e.g. [q]).
The satem languages (which have the sibilant where the centum equivalents have the velar) include Indo-Iranian, Armenian, Baltic, Slavic, Albanian, and perhaps also a number of barely documented extinct languages, such as Thracian-Dacian.[1] This group changed PIE palatovelars into sibilants, retaining PIE plain velars and merging PIE labiovelars into them to form an expanded velar group. It is sometimes suggested that the plain velars and the labiovelars were not merged in Proto-Albanian,[2] but this is not a mainstream viewpoint. Balto-Slavic is largely satem but evidences centum development in some words, suggesting that "Satemization" was incomplete or operated according to different principles than in the other Satem languages.
The centum group includes all remaining dialects, i.e. Italic, Celtic, Germanic, Hellenic, Tocharian, Anatolian and possibly a number of lesser-known extinct groups (such as Ancient Macedonian and Venetic). This group merged PIE palatovelars and plain velars, yielding plain velars (but see below about Anatolian).
The satem languages share some innovations in common (particularly, the ruki sound law), while the centum languages have no common innovations, and in fact include the two groups that split off the earliest, i.e. Anatolian and Tocharian. Furthermore, the satem languages occur in a contiguous region approximately in the middle of the PIE area, while the centum languages occur in a discontiguous area partly surrounding the Satem languages. This strongly suggests that "satemization" was a single areal sound change, which occurred less completely in Balto-Slavic (at the edge of the area) than elsewhere, while "centumization" was actually a set of unrelated changes occurring independently in multiple language groups. This is easy to understand given the current conception of the PIE values of the three dorsal series, where Centumization involves nothing more than the elimination of the velar-uvular distinction, which is typologically fairly rare and in any case carried very little functional load in PIE. Satemization, however, was a more substantial change, involving the fronting of PIE velars and the unrounding of PIE labiovelars.
In addition, recent evidence from Luwian indicates that all three series were maintained separately in Proto-Anatolian,[3] and the Centumization observed in Hittite only occurred after the breakup of Common Anatolian.[4]
Tocharian is an additional puzzle in that it largely reflects a situation where all three PIE dorsal series as well as all voicing/aspiration distinctions (originally constituting nine separate consonants) have merged into a single phoneme /k/. This has led some writers to suggest that Tocharian does not fit the Centum-Satem model.[5] However, some PIE labiovelars are in fact represented by a labiovelar-like element or by a non-original sequence /ku/. Along with other evidence, this suggests that labiovelars were distinct in Proto-Tocharian and only later merged with velars (as happened independently in Old Irish and to some extent in some other languages), making Tocharian a clearly Centum language.[6]
The isogloss only applies to the parent language with the full inventory of dorsals. Later sound changes within a specific branch of Indo-European that are analogous to one of the centum or satem changes, such as the palatalization of Latin k to s in some Romance languages or the merger of *kʷ with *k in the Goidelic languages, are excluded.
Contents |
August Schleicher, an early Indo-Europeanist, in Part I, "Phonology", of his major work, the 1871 "Compendium of Comparative Grammar of the Indogermanic Language", published a table of original momentane Laute, or "stops", that has only a single velar row, *k, *g, *gʰ, under the name of Gutturalen.[7][8] He does identify four palatals (*ḱ, *ǵ, *ḱʰ, *ǵʰ) but hypothesizes that they came from the gutturals along with the nasal ń and the spirant ç.[9]
Karl Brugmann in his 1886 equivalent work, "Outline of Comparative Grammar of the Indogermanic Language," promotes the palatals to the original language, recognizing two rows of Explosivae, or "stops", the palatal (*ḱ, *ǵ, *ḱʰ, *ǵʰ) and the velar (*k, *g, *kʰ, *gʰ),[10] each of which was simplified to three articulations even in the same work.[11] In that same work Brugmann notices among die velaren Verschlusslaute, "the velar stops", a major contrast between reflexes of the same words in different daughter languages: in some the velar is marked with a u-Sprache, "u-articulation," which he terms a Labialisierung, "labialization," in accordance with the prevailing theory that the labiovelars were velars labialized by combination with a u at some later time and not among the original consonants. He divides languages therefore into die Sprachgruppe mit Labialisierung[12] and die Sprachgruppe ohne Labialisierung, "the language group with (or without) labialization," which are perforce identical to the Centum and Satem groups. He opines that[13]
The doubt introduced in this passage suggests he already suspected the "afterclap" u was not that but was part of an original sound.
In 1890 Peter von Bradke published "Concerning Method and Conclusions of Aryan (Indogermanic) Studies" in which he saw the same division (Trennung) as did Brugmann but he defined it in a different way. He said that the original Aryans knew two kinds of gutturaler Laute, or "guttural sounds," the gutturale oder velare, und die palatale Reihe, "guttural or velar and palatal rows," each of which were aspirated and unaspirated. The velars were to be viewed as gutturals in an engerer Sinn, "narrow sense." They were a reiner K-Laut, "pure K-sound." Palatals were häufig mit nachfolgender Labialisierung, "frequently with subsequent Labialization." This latter distinction led him to divide the palatale Reihe into a Gruppe als Spirant and a reiner K-Laut, typified by the words satem and centum respectively.[14] Later in the book[15] he speaks of an original centum-Gruppe from which on the north of the Black and Caspian Seas the satem-Stämmen dissimilated among the Nomadenvölker, or Steppenvölker, located there by further palatalization of the palatal gutturals.
By the 1897 edition of Grundriss, Brugmann (and Delbrück) had adopted Von Bradke's view. He says[16]
Concerning the labialized velars Brugmann had changed his mind, and there was no more mention of labialized and non-labialized language groups. The labio-velars now appeared under that name as one row of the 5-row Verschlusslaute (Explosivae) containing die labialen V., die dentalen V., die palatalen V., die reinvelaren V. and die labiovelaren V. It was Brugmann who pointed out that labiovelars had merged into the velars in the Satem Group,[17] accounting for the coincidence of the discarded non-labialized group with the Satem Group.
The presence of three dorsal rows in the proto-language is one hypothesis of many.[18] In another hypothesis by Antoine Meillet the original rows were the labiovelars and palatovelars, with the pure velars being allophones of the palatovelars in some cases, such as depalatalization before a resonant.[19] Other possibilities are borrowing between early daughter languages during the process of Satemization, or perhaps the concept of original velars is an artifact based on just plain false etymologies in modern times. Oswald Szemerényi proposed that the "preconsonantal palatals probably owe their origin, at least in part, to a lost palatal vowel;" that is, a velar was palatalized by a following vowel subsequently lost.[20] The palatal row therefore post-dated the original velar and labiovelar but Szemerényi does not give times. He includes the palatals in a table of five rows stops "shortly before the break-up" with a question mark after them. Other scholars who assume two dorsal rows in PIE include Kuryłowicz (1935), Meillet (1937), Lehmann (1952), and Woodhouse (1998).
The Satem languages show characteristic affricate and fricative consonants articulated in the front of the mouth in inherited Indo-European lexical items in which in other languages termed the Centum Languages pure velars and labiovelars, sounds articulated at the back of the mouth, are found. The Satem shift is conveniently illustrated with the word for '100', Proto-Indo-European *(d)ḱm̥tóm, which became Avestan satəm (hence the name of the group), Persian sad, Sanskrit śatam, Latvian simts, Lithuanian šimtas, Old Church Slavonic sъto. Another example is the Slavic prefix sъ(n)- ("with"), which appears in Latin, a centum language, as co(n)-; conjoin is cognate with Russian soyuz("union").
The sources of the satem sounds and the methods by which they became what they are have been debated heavily by Indo-European linguists for many decades. The originator of the concept, Peter von Bradke, believed in a Proto-Indo-European two-row system of four gutturals each row, the pure velar row: *k, *kʰ, *g, *gʰ, and the palatovelar row: *ḱ, *ḱʰ, *ǵ, *ǵʰ. For example, *ḱ became Sanskrit ś [ɕ], Latvian, Avestan, Russian and Armenian s, Lithuanian š [ʃ], and Albanian th [θ] (but k before a resonant). Karl Brugmann added the labio-velar row: *kʷ, *kʷʰ, *gʷ, *gʷʰ, with the proviso that in the Satem languages it merged into the velar row, losing their accompanying lip-rounding. This merger left the Satem group without labio-velars. Regardless of whether satem words were created from those rows with those articulations in that way, they are definable as satem words.
Satem-like features have arisen multiple times during history (e.g. French cent pron. [sã]). As a result, it is sometimes difficult to firmly establish which languages were part of the original Satem diffusion and which were affected by secondary assibilation in a later time period. For instance, it is known that the assibilation found in French and Swedish were later developments as linguists have extensive documentation of Latin and Old Swedish. However, in the case of Dacian and Thracian, there is not enough information on the history of these languages to conclusively settle the issue of when their Satem-like features originated. Extensive lexical borrowing, such as Armenian from Iranian, may also add to the difficulty. The status of Armenian as a Satem language as opposed to a Centum language with secondary assibilation rests on the evidence of a very few words.
The Centum languages show characteristic pure velars and labiovelars articulated at the back of the mouth in inherited Indo-European lexical items in which in other languages termed the Satem Languages affricate and fricative consonants articulated in the front of the mouth are found. The name Centum comes from the Latin word centum (pronounced [kentum]) < PIE *ḱm̥tóm, '100', English hund(red)- (with /h/ from earlier *k, see Grimm's law), Greek (he)katon, Welsh cant, Tocharian B kante. Labiovelars as single phonemes (for example, /kʷ/), as opposed to biphonemes (for example, /kw/) are attested in Greek (the Linear B q- series), Italic (Latin qu), Germanic (Gothic hwair ƕ and qairþra q) and Celtic (Ogham ceirt Q). In the Centum languages, the palatovelar consonants merged into the plain velars (*k, *g, *gʰ). The merger left the Centum Group without palatovelars.
The Centum languages preserve Proto-Indo-European labiovelars (*kʷ, *gʷ, *gʷʰ) or their historical reflexes as distinct from plain velars; for example, PIE *k : *kʷ > Latin c /k/ : qu /kʷ/, Greek κ /k/ : π /p/ (or τ /t/ before front vowels), Gothic /h/ : /hʷ/, etc. Remnants of labial elements from labiovelars in Balto-Slavic include Lithuanian ungurys "eel" < *angʷi- , Lithuanian dygus "pointy" < *dʰeigʷ-. Fewer examples of incomplete Satemization are also known from Indo-Iranian, such as Sanskrit guru "heavy" < *gʷer-, kulam "herd" < *kʷel-; kuru "make" < *kʷer- may be compared, but they arise only post-Rigvedic in attested texts.
When von Bradke first published his views (1890) defining the Centum and Satem Languages, both major theories of language origination, the Tree model and the Wave model already existed. Bradke viewed his classification as "the oldest perceivable division" in Indo-European.[15] Each of these had further "divisions;" that is, Bradke was proposing a tree division, which he elucidated as "a division between eastern and western cultural provinces (Kulturkreises)."[21] The hypothesis came toward the end of Johannes Schmidt's career, innovator of the Wave model. He did not address it and it remained in place as the mainstream hypothesis even though cross-Kulturkreis similarities were noted. For example, referring to the "two sections", Peter Giles in 1901 noted "striking similarities" in languages across them, such as "Italic and Lettic."[22]
The proposed split was undermined by the discoveries of Hittite and Tocharian, which were Centum languages located within the hypothetical Satem range, Tocharian isolated on the Silk Route in the far east, divided from its closest cognates in Europe by thousands of miles of rugged terrain and hostile peoples.[1][23] This proposed first division based on a single isogloss was further weakened by continued research into additional Indo-European isoglosses, many of which seemed of equal or greater importance in the development of daughter languages. Philip Baldi explains:
"...an early dialect split of the type indicated by the centum-satem contrast should be expected to be reflected in other high-order dialect distinctions as well, a pattern which is not evident from an analysis of shared features among eastern and western languages."[23]
Colin Renfrew notes that the satem-centum distinction "is not in itself accorded much significance today"[1] as it is considered "too simplistic".[24] Tree models of the descent of PIE into the various daughter languages still exist but in the absence of historical data for the early stages of those languages the genetic models necessarily rely on comparative data; that is, the isogloss map. The Centum–Satem isogloss is just one of many that criss-cross the Indo-European landscape.[25] Similarly the rejection of Kossinna's Law matching archaeological cultures to language groups in strict correspondence opened a gap between specific tree models and any archaeological support that might be found in prehistory. Archaeology is still useful, as are tree models, but only in a limited way in contexts of preponderance of evidence; that is, the existence of specific cultures in the hypothetical origin area of satemization proves nothing whatever, not even that the sound change originated there. If an origin is to be postulated for that region, however, the archaeology adds that some sort of cultural unity resulting from a common way of life prevailed there.
Application of the wave model is equally difficult. Renfrew's statement of it supposes a Satem center from which a wave of satemization spread, but "the original, earlier Centum remains untouched in the periphery."[26] Centum may at least in many cases (but not Anatolian) have been earlier but it was not original; moreover, Centumization removed the palatovelars from the language, leaving none to satemize. It is necessary to suppose that in any given region Centumization competed with Satemization, which the mixed inventory of Balto-Slavic may support, but the Wave model in this statement of it appears equally as simplistic as the Tree model.
A compromise possibility is that PIE was neither Centum nor Satem, but satemization began in a central area and diffused outward from there. This view falls within the Wave model but begins from a different origin. Where satemization did not reach centumization took place either in different languages, or in a different lexical inventory of the same language, or not at all, allowing for the complexity of outcomes noted by the linguistic community. Winfred Lehmann presents this view based on the work of Antoine Meillet. Meillet had hypothesized a two-row tectal system, labiovelars and velars, but his velars were composed of two allophonic rows, pure velars used before back vowels and palatovelars before front. Lehmann says "Proper treatment of the subject involves consideration of the early Indo-European languages as members of a dialect continuum. In a central area, spirantization of the palatal allophones took place ...."[27]
|
||||||||||||||