HOFFMANN, KARL (26 February 1915–21 May 1996; Figure 1), German Indo-Europeanist and Indo-Iranist. Karl Hoffmann was born at Hof am Regen (Upper Palatinate, Bavaria), the son of a railway official. He spent his childhood and youth in Munich; there he began his Indo-European and Indo-Iranian studies in 1934. At the beginning of the war in 1939 he was called up for military service; but, while on leave in 1941, he was able to obtain a doctorate with his (unpublished) dissertation Die altindoarischen Wörter mit -ṇḍ- besonders im Ṛgveda. It was not until the end of the war in 1945 that he was able to resume his academic work. In 1951 he habilitated (qualified as a university lecturer) in Munich; the title of his habilitation thesis, which is an authoritative work as a synchronic functional investigation, was Der Injunktiv im Veda. In 1952 he was appointed to teach comparative Indo-European linguistics in Saarbrücken, and in 1955 he was offered the chair of comparative linguistics in Erlangen. Declining an appointment at the University of Bonn in 1962, he remained at the University of Erlangen until his retirement in 1983.

Hoffmann was mainly interested in Indo-Iranian studies, which he did not conceive of as a mere combination of Indology and Iranian studies, but as a distinct subject comprising historical philology and comparative linguistics. His studies are essentially devoted to Vedic, transmitted in India, and to Avestan and Old Persian on the Iranian side. Collected in the three volumes Aufsätze zur Indoiranistik (1975, 1976, 1992), they impressively show how effective the close union of linguistics and philology can be, as practiced by Hoffmann.

The two Old Iranian languages and Vedic are closely related, as is shown by the numerous specific correspondences, which go far beyond those between other Indo-European languages, thus leading to the assumption of a common pre-historical, intermediate language, namely Indo-Iranian. It is true that documentation for the two linguistic branches, Old Indic and Old Iranian, differs notably. Vedic, as the religious language of ancient India, is amply attested; but the evidence for the two Old Iranian languages is very limited. Old Persian is only documented by the inscriptions of the Achaemenid kings. Avestan is the language found in the Avesta, the not very extensive corpus of the preserved sacred texts of the Zoroastrians. The language is preserved in two forms—Old Avestan, which is limited to the Zarathustrian Gāthās, the liturgical Yasna Haptaŋhāiti, and a few prayers, and the Younger Avestan corpus consisting largely of ritual and legal texts.

Due to the different states of the evidence, Hoffmann’s works deal mainly with the complex (in chronology and contents) and reliably transmitted Vedic texts. His procedure is, in general, philological-linguistic in nature. In numerous separate studies he concerned himself with comprehending in a philologically exact manner linguistic phenomena in the relevant literature which had not been interpreted, or not yet to a satisfactory extent. This procedure often required preliminary text-critical investigations. The material first was carefully edited and plausibly interpreted as to its contents; then it was analyzed linguistically. According to need, the appropriate fields were applied—i.e., phonology, morphology, syntax, semantics, and etymology. In this way new or improved conclusions could be reached, and these have long since made their way into the handbooks and etymological dictionaries.

Hoffmann proceeds in a similar way in his investigation of many problems of language or content in Old Iranian. Certainly the situation for Avestan or Old Persian is significantly more difficult. Not only is there far less linguistic evidence transmitted, compared to that for Vedic, but the written tradition of Avestan is highly problematic and often quite unreliable. Old Persian is preserved in inscriptions which are contemporary documents of the language of the Achaemenid kings, but the notation of the Old Persian cuneiform script is not consistent enough to provide an assured reading throughout. For both Avestan and Old Persian texts, a single-language analysis would leave much in the dark. Therefore Hoffmann often brought Vedic, as the most closely related language, into his arguments. This practice made it possible for him to release Old Iranian problems from their isolation, that is, to put content-related or linguistic questions in a new light or to rearrange them by approaching them from Vedic. In this way he was also able to apply conclusions he had reached on the Vedic side to explain Old Persian or Avestan problems.

Hoffmann’s investigations in Old Iranian, as in Vedic, often started from a single problem, which is suggested in the article’s title. The problem may be a single word or a phrase or some other linguistic phenomenon. The methodology of his text-critical, philological, and linguistic inquiry is impressive in the way, not only the immediate problem, but a whole complex of questions is reasoned out. The word or the linguistic phenomenon he was dealing with was placed in a wider context, and the solution to the problem could then be approached from various directions; each step in the argument was explained and made understandable.

Hoffmann’s method is well illustrated by even a brief article such as one on the Avestan hapax jauua (“Die av. Verbalformen jauua Yt. 5,63 . . . ,” 1969; Aufsätze, pp. 258-61). His investigation shows that the linguistically and philologically isolated term jauua (supposedly “hurry!”) can be emended to the common verbal form jasa (“come!”), which is also supported by the context. The uu appearing in the manuscripts instead of s is explained as a mistake in the base manuscript; the Avestan letter s was misread as uu, as occasionally happened. Hoffmann repeatedly manages to reconstruct the original form after having proved manuscript readings as faulty; the textual context always plays a major part. A more extensive example is the article “Avest. upa.mraōdə̄sca” (1966; Aufsätze, pp. 195-206) about another hapax form, which likewise must contain a spelling mistake, as may be proved on the basis of linguistic arguments. The correction to upa.mraoka- and other corrections in context with it restore a piece of ritual text.

In another case (“The Avesta Fragment FrD.3,” 1968; Aufsätze, pp. 221-27), Hoffmann succeeds in interpreting an entire fragment by resorting to Vedic. He cites the Vedic verb “to abandon, leave behind” and the present form hāsa- “to race, run for a prize, gain” and connects the Avestan verb“to abandon” and its reduplicated present zaz-. Another article deals with the infrequently occurring, sometimes augmented YAv. optative verb form, by means of which a unique OPers. optative form can be explained (“Präterital Optativ im Altiranischen,” 1975; Aufsätze, pp. 605-19). His remarks about the preservation of a much-disputed phrase from Indo-Iranian ritual (“Avestisch haoma yō gauua,” 1967; Aufsätze, pp. 475-82) are convincing; two plausible syntactic explanations are offered. A particularly instructive example of Hoffmann’s methodical treatment of manuscript readings is the article “Zur avestischen Textkritik: Der Akk. Pl. Mask. der a-Stämme” (1970; Aufsätze, pp. 274-87). Here reference should be particularly made to the extensive note (pp. 275 f.) on the possible stages of the Avesta tradition.

Hoffmann wrote fewer studies on Old Persian, due to the deficient evidence for it in comparison with Avestan. However, a good example is the article “Altpersisch afuvāyā"(1955; Aufsätze, pp. 52-57). The hapax is interpreted by Hoffmann sensibly, that is, adequately for the context, as the abl. sing. of a two-syllable, feminine noun afuvā- “panic, mortal agony” (= apuvā-,with anaptyxis for resolution of the rare sound group -pv-; cf. OI. apvāˊ). For an Old Persian construction, the “naming phrase,” to which he devoted two articles (1956, 1961; Aufsätze, pp. 120-29; 403-10), Hoffmann cites Vedic correspondences; thus he suggests an Indo-Iranian age for the this stylistic feature.

A general survey entitled “Altiranisch” in the Handbuch der Orientalistik (HO I/4/1, Leiden, 1958, pp. 1-19) contains many important observations and judgements in limited space and in keeping with the state of research at the time. The same is true for the entries “Avestan Language. i. The Avestan script” and “ii. The phonology of Avestan” in Encyclopaedia Iranica (1988; EIr. III/1, pp. 47-56). Both essays were included in the Aufsätze zur Indoiranistik (pp. 58-76, 864-79). Hoffmann’s book on Avestan appeared after his death: Avestische Laut- und Flexionslehre (1996, with B. Forssmann). This is a very carefully prepared, historical-comparative grammar of Avestan, in which ancient Indo-Iranian on the one hand and the two most closely related languages, Old Persian and Vedic, on the other play essential roles. The work was mainly based on Avestan word forms which are fairly secure in their interpretation. It is distinguished by a clearly organized and systematic presentation, with many cross-references. Existing difficulties and research gaps are not concealed in the assessment of linguistic facts. For future research, this grammar will serve as an indispenable aid, and at the same time it proposes valuable ideas.

From the 1960s on, Hoffmann particularly devoted his attention to the history of the Avesta tradition, above all to the Avestan script: an alphabetic right-to-left script containing many letters; according to Hoffmann and other scholars, it originated in the Sasanian period. For example, we find in a congress paper, “The Avestan Script” (1967; Aufsätze, p. 710), the statement that the first version of the texts in the Avestan script could be called the “Sasanian archetype.” The creation of the script on the basis of the Pahlavi script and the method used by its inventor are discussed in “Zum Zeicheninventar der Avesta-Schrift” (1971; Aufsätze, pp. 316-26); the question of when and how the Avesta was transmitted is studied in “Das Avesta in der Persis” (1979; Aufsätze, pp. 736-40). A comprehensive account of the entire subject can be found in the monograph Der Sasanidische Archetypus. Untersuchungen zu Schreibung und Lautgestalt des Avestischen (1989, with J. Narten).

The problem in studying Avestan lies in the written tradition, which is very defective. The entire Avesta tradition is founded on the narrow basis of Avesta manuscripts belonging to the second millennium C.E.; the presumed oldest one belongs to the late 13th century, the most recent ones to the 18th and 19th centuries. The extant Avesta manuscripts probably go back to only a few basic ones that were not essentially different from one another (ca. 1000 C.E.); the existing text is practically always constant in length and word order. However, there exist a confusing number of variants, which, according to Hoffmann, are not only reading or writing mistakes, but partly also variants of pronunciation, due to the inconsistent “vulgate pronunciation” of the authors. These writing variants often make it difficult or impossible to interpret and understand a text passage, when they do not enable one to recognize the underlying word or grammatical form.

All Avesta manuscripts are written in a special script, the so-called Avesta script. As was already known previously, this script is not the result of a long evolution but is to be considered as the creation of a script inventor and possible collaborators. The model for the Avesta script was for the most part the Pahlavi book script in its early stage of development reached in the Sasanian period. Hence the creation of the Avesta script must have taken place in the Sasanian period. Hoffmann’s arguments lead to the conclusion that this script, which possesses an extensive inventory of vowel and consonant signs, was a phonetic alphabet, which was especially created to record the contemporary oral tradition of the Avesta with phonetic precision. Hoffmann further concludes that the inventor of this special script for Avestan also set down the Avestan texts, i.e., the “Sasanian archetype,” which therefore is to be considered as the phonetic representation of the Avesta recitation in the Sasanian period, that is, the Avesta in “Sasanian pronunciation.” This means that there is no direct testimony for the language of the author of the text. However, the Sasanian archetype represents an important intermediate stage; because, with its specially created, extensive alphabetic script, it produces an impression of the sound produced by the Avesta texts in the traditional recitations during the Sasanian period. Answering the question as to which of the confusingly numerous, late manuscript variants might correspond to the supposed text of the Sasanian archetype, that is, how the Sasanian archetype sounded, is a step on the way to the original composer’s speech form, which is posited through comparative linguistics.

Because of the particular part played by the Avestan script in our knowledge of the “Sasanian pronunciation” of the Avesta texts, Hoffmann tried to produce an exact description of all of the letters and their sound values. He considered it as a principal requirement that Avestan, which was recorded in a highly differentiated alphabetic script, should be transliterated letter by letter. A clear example is the writing of the Old Iranian vowel-consonants [ḭ] and [ṷ]. In word-initial position, the manuscripts have distinct signs: y, v. In medial position these are not found; the manuscripts consistently have: ii, uu. (Formerly these digraphs also were transcribed with y, v; the transliterations ii, uu have become widely accepted.) This consistent differentiation in the manuscripts probably corresponds to a reality of the “Sasanian pronunciation” of the Avesta. Since the same combinations of letters ii, uu serve to produce the Old Iranian sound sequences [iḭ], [uṷ] (formerly transcribed with iy, uv), Hoffmann assumes that, at the time of the first recording in the Avesta script, the “Sasanian pronunciation” developed each inter-vocalic or post-consonantal [ḭ] or [ṷ], if preserved, into [iḭ], [uṷ].

Hoffmann also extensively discussed the Old Persian cuneiform script in his essay “Zur altpersischen Schrift” (1975; Aufsätze, pp. 620-45). Here, too, his explanations met with wide approval; but there also exist partially or completely differing views. Hoffmann starts with the premise that, in view of the stylistic principle obviously underlying the signs of the Old Persian cuneiform script, the latter cannot have developed from the Babylonian-Assyrian cuneiform script and its variants but represents a conscious new creation, probably invented especially for Old Persian under Darius I. The question is how precisely this cuneiform script reflected the Old Persian language. In the attempt to better judge this matter, comparative linguistics (Avestan, Vedic) and linguistic history (Old Iranian, and Indo-Iranian on the one hand, Middle and New Persian and other Middle and New Iranian languages on the other) play an essential part; so does the contemporary foreign tradition (above all, Elamite). No matter how we assess the Old Persian cuneiform script, consisting of 36 phonetic symbols, one point cannot be ignored: the sign system is built inconsistently. Hoffmann interprets the phonetic symbols as syllabic symbols, with the exception of three vowel symbols, a, i, u, which function as consonant-vowel signs. (For comparison Hoffmann refers to the Cypriote syllabic script with a similarly constructed set of characters.) This leads to the following distribution: For each of the 22 consonants represented in Old Persian there is one consonant-a sign (Ka); for two of these Ka signs there are also Ki signs (ji, vi); for 5 Ka signs there are also Ku signs (ku, gu, tu, nu, ru); for two Ka signs there are both Ki and Ku signs (di, du; mi, mu). If the system had been carried out consistently, one would expect Ka, Ki, and Ku signs to exist for each consonant (except for those for which the sequence Ki or Ku does not occur in Old Persian), as happens with da, di, du and ma, mi, mu. However, this would have required the creation of more than twenty additional cuneiform symbols, and a number of such Ki and Ku signs would have been only rarely used.

Against the hypothesis, which is sometimes advanced, that the Old Persian cuneiform script was a purely alphabetic script, perhaps comparable to the Aramaic alphabetic script, Hoffmann offers two arguments. One is the existence of the Ki and Ku signs, which do not have a phonetic basis. If we assumed that these were symbols for the palatalization or labialization of the consonants (K), such a phonetic reason would be improbable precisely for the nasal m, which in fact has the characters mi and mu.On the other hand, the absence of any signs for palatalization or labialization of dentals would be hard to understand: we have da, di, du, but only ta, tu; na, nu, and yet only θa. The second argument is the existence of the vowel signs i and u, which have no equivalent in the consonant letters yōd and wāw of the Aramaic alphabet of the middle of the first millennium B.C.E. Hoffmann assumes that a system of Ka, Ki, Ku symbols was originally intended for Old Persian, but the creation of the necessary inventory of cuneiform signs was for some reason abandoned, and a less explicit system of vowel signs began to be used. In his opinion, some written forms in the Behistun [see BISOTUN] inscription, which probably represents the first Old Persian inscription, may express the original intention of the creators of the script, namely that Ki and Ku signs alone, without the addition of the signs i and u, were to mark the short syllabic vowel.

In spite of the evident lack of structure in the inventory of signs, Hoffmann’s account leads to recognition of certain rules in the use of Old Persian cuneiform signs on the basis of internal linguistic and historical linguistic criteria. These rules are used quite consistently; yet the writing is for the most part ambiguous because of the inconsistent system of notation. This is shown, for instance, in his rules for describing vowel notation (based on the Darius and Xerxes inscriptions): i after a Ka signfor which there exists a Ki indicates [Kai]; i after a Ka for which there is no Ki indicates both [Kai] and [Ki, Kī]. The same pattern is true for u aftera Ka sign: if there exists a corresponding Ku sign, the combination expresses [Kau]; otherwise it indicates both [Kau] and [Ku, Kū]. In addition, i after a Ki sign indicates [Ki, Kī]; correspondingly u after a Ku indicates [Ku, Kū]. A Ka sign can indicate the consonant alone, when it is written before another Ka or at the end of a word.

Altogether, Hoffmann tried to explain peculiarities or supposed deficiencies of the Old Persian script phonetically, rather than orthographically. Particularly striking, for example, is the fact that nasals are not indicated before plosives and fricatives, even though they are verified as existing in Old Persian by historical linguistic origin or later development and also by the contemporary foreign tradition. The phenomenon of not indicating a nasal before a stop consonant is in fact also found in the Elamite language, but here as an exception rather than a rule. Otherwise, nasals in these positions are indicated in Elamite, Babylonian, and also in Aramaic. Hoffmann concludes that in Old Persian the nasals, when preceding plosives and fricatives, are somehow articulated otherwise than as m or n—for example, as uvular nasals. They may have been interpreted as a variation of the preceding syllabic vowel, so that expression with the syllabic sign ma or na did not apply. Another conspicuous fact in Old Persian inscriptions is that, before u,an h required on comparative linguistic grounds is not written. Here, too, Hoffmann sees no graphic phenomenon but prefers a phonetic explanation, namely the disappearance of h before u; the feature can also be explained as the result of Old Persian h being a faintly articulated sound. However, Turfan Middle Persian shows that the sequence hu was preserved. According to Hoffmann the absence of h before u may have occurred in the dialect to which the Achaemenid court language belonged. In other respects, too, there are occasional signs of differences in dialect within Old Persian, and the linguistic evolution from Darius to Middle Persian may not have proceeded in a direct line. Various phenomena which are usually considered as strange spelling conventions or “orthographica” are to be traced back, according to Hoffmann, to specific Old Persian articulatory features. In spite of the structural deficiency of the Old Persian cuneiform script, they can all be explained from a linguistic point of view. Hoff-mann therefore concludes that the Old Persian script is, within the framework of its possibilities, a faithful representation of the sound system of the Old Persian language.

Nine of Hoffmann’s students received doctoral degrees in various fields of Indo-European studies; and some of them have habilitated: J. Narten (Ph.D. in Vedic, habilitation in Avestan), B. Forssman (Ph.D. in Greek), R. Lühr (Ph.D. and habilitation in Germanic), G. Klingenschmitt (Ph.D. in Pahlavi, habilitation in Armenian), M. Witzel (Ph.D. in Vedic), E. Tichy (Ph.D. in Greek, habilitation in Vedic), H. Eichner (Ph.D. in Hittite, habilitation in proto-Indo-European), N. Oettinger ( Ph.D. in Hethitic, habilitation in Avestan), T. Gotō (Ph.D. in Vedic).



Works of Karl Hoffmann. Der Injunktiv im Veda. Eine synchronische Funktionsuntersuchung, Heidelberg, 1967.

Aufsätze sur Indoiranistik, vols. I and II, ed. by J. Narten, Wiesbaden, 1975-76; vol. III, ed. by S. Glauch et al., Wiesbaden, 1992.

With Johanna Nar-ten: Der sasanidische Archetypus. Untersuchungen zu Schreibung und Lautgestalt des Avestischen, Wiesbaden, 1989.

With Bernhard Forssman: Avestische Laut- und Flexionslehre, Innsbruck, 1996.


Obituary notices. Klaus Strunk, Jahrbuch der Bayerischen Akademie der Wissenschaften, 1996, pp. 238-43.

Michael Witzel, IIJ 40, 1997, pp. 245-53.

Bernhard Forssman, Kratylos 42, 1997, pp. 214-18.

Heiner Eichner, Almanach der Österreichischen Akademie der Wissenschaften, 1999, pp. 483-92.

(Johanna Narten)

Originally Published: December 15, 2004

Last Updated: March 22, 2012

This article is available in print.
Vol. XII, Fasc. 4, pp. 420-423