Indo-European Languages

Literature references and annotations by Dick Grune, dick@dickgrune.com.
Last update: Fri Sep 13 11:14:09 2024.

These references and annotations were originally intended for personal use and are presented here only in the hope that they may be useful to others. There is no claim to completeness or even correctness. Each annotation represents my understanding of the text at the moment I wrote the annotation.
No guarantees given; comments and content criticism welcome.

Rouard, X., Did Indo-European Languages Stem from a Trans-Eurasian Language? An Interdisciplinary Approach: Conclusions of Findings, 2022, pp. 6.
The paper summarizes the linguistic, genetic, historical, archaeological, agricultural and religion considerations from Rouard (2021, see below). The conclusion is that there was a Trans-Eurasian language in Central Asia and that the Indo-European and Dravidian languages and Burushaski derive from it. The paper closes with a list of 50 words that are common to these languages and which could ultimately stem from Trans-Eurasian. Examples are cleu `hear', kahla `speak' and edo `eat'.

Bomhard, A.R., The Origins of Proto-Indo-European -- The Caucasian Substrate Hypothesis, 2019, pp. 88.
Whereas Colarusso sees PIE as an North-West Caucasian language that was influenced heavily by an Uralic language, Bomhard espouses the idea (without giving arguments) that PIE is an Uralic language heavily influenced by PNWC. After a short introduction to PIE and PNWC, the bulk of the paper consists of 198 correspondences.
[DG: Many correspondences link PNWC to a single branch of PIE only, and several are semantically loose: PIE dʰew-r-yo-s = ‘of great value,' (Grm. `teuer', Du. `duur') corresponding to PNWC də́ wə: = `big, great'.]
     It is possible that the influence that PIE exerted on the Caucasian language caused the split between North-East and North-West Caucasian languages, and in fact created PNWC.
[DG: This would be an argument for PIE being originally a non-Caucasian (→ Uralic) language.]

Peyrot, M., The Deviant Typological Profile of the Tocharian Branch of Indo-European May Be Due to Uralic Substrate Influence, 2019, pp. 50.
Some of the non-PIE properties of Tocharian (e.g. consonant system, object marking on the verb) are explained as influence of an early form of Samoyedic. Other factors (Ket, Yukagir) are explored but judged less likely. The contact would have taken place during the Afanasievo period, around 2500 BZ. Maps of the movements of the Tocharians and Samoyeds are supplied.

Fournet, A., Reasons Why Present-Day Indo-Europeanists Are Crooks, 2019, pp. 13.
Vitriolic discussion of the state of the art of Indo-European research. The point seems to be that PIE should not be a spruced up Sanskrit, but boldly consist of 1. Hurro-Urartian, Nakh; 2. Anatolian: Hittite, Luwian; 3. HPA: Hybridized Post-Anatolian. "Hybridized" here means "recognizing the Hurro-Urartian and Anatolian influence".

Kortlandt, F., How I Discovered Proto-Indo-European Glottalic Stops, 2019, pp. 8.
The study of tones and vowel length in the Slavic languages and a paper by Werner Winter led the author to conclude that another influence was at work besides the PIE laryngeals: glottalic consonants. Material from Armenian, Balto-Slavic, Sindhi, Panjabi, Greek, Latin, Germanic, sound symbolism studies by Swadesh, and many other data confirmed the conclusion.
     A complementary study revealed that the aspirated voiced stops in e.g. Indic and some Armenian dialects is an innovation and is not a continuation of the PIE “voiced aspirates” of the traditional theory.
     In the introduction the author asks "Why did nobody ever notice this before?", but does not answer the question. Somewhere else (where??) he notices that glottalics are the stepchildren of Western linguistics and in `Proto-Indo-European Glottalic Consonants' (2017) he mentions an American linguist doing research on Armenian who could not hear the glottalics.

Kortlandt, F., Proto-Indo-European Glottalic Consonants, 2017, pp. 8.
Response to Martin Kümmel, who in 2007 wrote a book "Konsonantenwandel" `Consonant Metamorphosis', in which the author's glottalic theory is rejected. The author points out many errors in Kümmel's evaluation.
     In this paper (and perhaps in earlier papers) the author reconstructs *t^:, *t^ɂ, *t for traditional PIE *t, *d, *dh. which is much more reasonable than the fortis, glottalic lenis, and aspirated lenis from his 1985 paper.

Kortlandt, F., More Evidence for Proto-Indo-European Glottalic Stops, 2016, pp. 4.
A number of examples is very tersely described that show the strong relationship between PIE d and ɂ. For example, many PIE roots are given that are reconstructed both with d and with H1=ɂ from Vedic. Another example is the observation that PIE d and H1dh have the same reflex in proto-Anatolian.
[DG: It is remarkable that all examples involve PIE d, and never g or gw.]
Furthermore the author suggests that originally PIE had just a fortis and a lenis stop, and that it caught the glottalic variant from a North-Caucasian language, e.g. possibly proto-Circassian.

Pyysalo, J., The Solution to the proto-Indo-European Laryngeal Problem, 2016, pp. 21.
Summary of a streamlined version of the author's thesis (2013).
     In the new Glottalized Fricative Theory the proto-Indo-European laryngeal is a guttural fricative h, in two variants: voiceless and voiced; it was lost in all branches except Anatolian. In addition PIE had a vowel α, which occurred only next the the h, gain in two variants, stressed (which turned into a in all languages except Indo-Iranian, where it became i), and unstressed (which disappeared in all languages).
     This system allows all PIE sound rules to be modelled, and many to be clarified. Very many examples are given.

Pyysalo, J., System PIE: The Primary Phoneme Inventory and Sound Law System for Proto-Indo-European, 2013, pp. 504.
Very formal thesis, written in a stern prose. It reads like a cross between a political manifest and the Algol68 Report, with a whiff of Wittgenstein's Tractatus. Example of the mode of expression: "X does not satisfy the requirements of scientific realism" for "X is impossible"; indeed the long phrase makes clear why X is impossible. The thesis strongly emphasizes principles and and does so repetitively.
     The underlying theme is that the laryngeal *h^, reconstructed from Old Anatolian, has not been incorporated correctly and to full effect in the PIE theory.
     Chapter 1 describes the comparative method as applied to PIE in considerable depth, with many examples. E.g.
* Most IE languages have a word for `middle' deriving from PIE medhĭ-, but Polish mie,dzy and Late Avestian mamðya- have an n-like sound before the d. Because these are two witnesses from different branches of PIE, this leads to updating the PIE form to memdhĭo-. This is then considered as resulting from reduplication of the PIE root mdh.
* The fact that Latin sum cannot be derived from PIE *ésmi is explained by showing (through Hittite and Hieroglyphic Luwian) that the PIE paradigm of `to be' was suppletive, with two roots, es- and sa-.
     Chapters 2, 3, and 4 consider all aspects of PIE etymology. For each aspect the most important theories (usually at least three or four) are examined critically, for example: "The comparative Indo-European root theory has been temporarily sidetracked by the laryngeal theory, where empirical theory has been replaced by Møller’s ProtoIndo-Semitic root hypothesis." On the contrary, many PIE roots are mono-literal, made more specific by "extensions"; examples are given. In each case the correct way to introduce the h^ is indicated.
     Chapter 5 integrates these improvements into a new PIE, named "System PIE", summarized as follows:
"For the solution of the laryngeal problem, it is necessary and sufficient to combine PIE *h^ (= Hittite h^) and the cover symbol Neogr. *ə, reinterpreted as vowel PIE *a, in diphonemic PIE *h^a and PIE *ah^." Heere the a supplies the coloring effect.
     The labio-velars and palato-velars are analysed as velar+u and velar+i, and the aspirates are analysed as C+h. Consequently, Neogrammarian *k^h corresponds to System PIE *kiah \/ *kiha \/ *kahi \/ *khai.
[DG: Can separate meanings be attached to words starting with these four segments?]
This reduces the phoneme set of PIE to its bare bones.
     A few noteworthy statements:
* Two defintions: "sound change" == the mechanical, totally regular change of one sound into another, under conditions which we may not (fully) know. "Sound law" == scientific formulation of that change as a process, which may be incomplete.
* The proto-language [DG: + the sound laws] are equivalent to the linguistic data. [DG: I.e., reconstruction is viewed as data compressions.]
* When we reconstruct a word in a proto-language it is a real word in a real language and the sound change is a real historical process (pg 67, bottom).
     The author puts much stock in computational linguistics, and is formalizing all of comparative PIE in a computer project "PIE Lexicon".

Luraghi, S., The Origin of the Proto-Indo-European Gender System: Typological Considerations, 2011, pp. 30.
Dizzying account of all the properties that languages use to classify nouns, in which it turns out that many of the traditionally binary criteria actually show fine-structure. For example, abstractness in nouns comes at least in two flavours: the German das Sitzen (masculine) is an abstract act, die Sitzung (feminine) an abstract event. Such different sub-divisions of a criterion are often associated with different other classification criteria, for example gender, as demonstrated by the above example.
     One class-forming property of nouns is `topic-worthiness'; animate nouns are topic-worthy, inanimate nouns less so. Abstract nouns, many of them characterized by -a (from PIE *-h₂), were also topic-worthy, and became animate by association. This formed a split in the animate noun class. When two animate topic-worthy noun classes became available, it was natural to associate one with masculine and the other with feminine. The resulting 3-class system was more useful than the previous 2-class system because it made referent tracking of topic-worthy entities in discourse easier (= using him and her rather than repeating personal name, family relation, or function every time).
[DG: It may also have allowed reintroducing the more effort-requiring personal name, family relation, and function as honorific forms.]

Kortlandt, F., Indo-Uralic, 2010, pp. 47.
This paper consists of 6 articles, each of a few pages. The papers are densely written, usually without any examples, and use the word `evidently' often.
    
     Eight Indo-Uralic verbs? (1989)
     Indo-European and Uralic have (at least) 8 verbs in common, and, what is more, they are very basic, concrete verbs: `to give, sell' (*miye-), `to wash' (*muske-), `to bring, fetch' (*tagu-), `to drive, hunt' (*gaki-), `to cast, dig', `to do, make' (*deka-), `to lead, draw' (*weda-), and `to take, carry' (*wige-) (Proto-Indo-Uralic reconstruction are give between parentheses where possible). It is unlikely that 8 such basic verbs could be borrowed (and there would be no place to do so), so they must stem from a common source: Indo-Uralic.
    
     The Indo-Uralic verb (2001)
     The Indo-Uralic homeland was situated south of the Ural around 7000-6000 BZ. Pre-Proto-Indo-European then came in contact with North-Caucasian languages, and was transformed in the process, while Proto-Uralic kept the original structure. The goal is to reconstruct the Proto-Indo-Uralic verbal system, but that is difficult because the Proto-Uralic verbal is known only fragmentarily. Using sound laws derived from the seven known Indo-Uralic verbs (`to cast, dig' was dropped), his own reconstruction of the Proto-Indo-European verbal system, and a lot of subtle reasoning, the author manages to draw some conclusions about the Indo-Uralic verbal system. The main conclusions are that Indo-Uralic has an objective conjugation which was preserved in Uralic and turned into the thematic conjugation in Proto-Indo-European; and that the well-known verb endings -m, -s, -t, -me, -te, -nt were original to Indo-Uralic and were supplemented by -q1, -q2, -e, -o, -r, -t-, -dh- to create variants or for disambiguation.
    
     Nivkh as a Uralo-Siberian language (2002)
     The author selects 12 promising Uralic particles (e.g. -nt for participle forming suffix) and searches for their presence in Nivkh. The Indo-Uralic 1st and 2nd person pronouns are easily identified in Nivkh, and the -nt is used to form finite verb forms.
     The conclusion is that there are strong indications for a close relationship between Nivkh and Indo-Uralic.
    
     Indo-Uralic consonant gradation (2003)
     The Uralic languages exhibit `consonant gradation', e.g. Finnish käsi - käden - kättä - kädet -- `hand'. Patterns are reconstructed for Indo-Uralic and then used to explain properties of PIE stems, concerning the relation between tone and consonants, and others.
    
     Indo-Uralic and Altaic (2004)
     Using material from Greenberg's "Indo-European and Its Closest Relatives", the author finds 21 common morphemes to connect Indo-Uralic to Altaic.
    
     Indo-Uralic and Altaic revisited (2008)
     Much of the criticism to deep relationships (Indo-Uralic + Nivkh, Altaic + Korean and Japanese) is directed towards lexical comparisons and indeed these are sometimes shaky. To consolidate the field the author restricts himself to morphemes only, and reiterates the existing material, now including Korean and Japanese.
[DG: Would Vovin be convinced now?]

Vijunas, A., The Proto-Indo-European Sibilant *s, 2010, pp. 16.
Argues that the PIE *s was a broad postalveolar sound, as in Spanish, Danish, or Dutch, rather that the sharp dentialveolar s of English, French, or German.
     One argument is typological: languages that have more than one sibilant usually have a dentialveolar s next to their other sibilant(s), whereas in languages that have only one sibilant this sibilant usually is postalveolar. And PIE has only one sibilant.
     Another argument is that the Hittite only sibilant is usually written with the cuneiforms for š, although Old Babylonian had both s and š.
     The author also argued that the PIE s was still postalveolar in all the proto-daughter-languages, and that having more than one sibilant was a "late" development in many branches. This hypothesis is supported with extensive argumentation and examples.

Luraghi, S., Indo-European Nominal Classification: From Abstract to Feminine, 2009, pp. 17.
In proto-Indo-European the derivational morpheme *-h₂ created abstracta; in the proto daughter languages it had turned into the inflectional morpheme -a for neuter plural (already in Anatolian) and later also into a characterizing inflectional morpheme -a for feminine nouns. The development from abstract to collective to marking neuter plurals is fairly unproblematic but from there to marking feminine nouns requires an explanation. Several have been given in the literature, but none is satisfactory. The author proposes that the two developments were independent, both having the starting point "abstract", rather than the second branching off the first at collective".
     The original PIE gender system had two genders, animate, and inanimate. In the course of time many words from the inanimate class had found their way into the animate class. Since abstracta are neither animate nor inanimate objects, they were conceived as a new class between inanimate and animate. When later a feminine class developed (initially correlating with the standard animate endings on adjectives) this intermediate class, with its -a inflection, was associated with feminine gender. Each of the steps above is supported by evidence of similar developments in a wide array of languages.

Harry A. Hoffner Jr., H. Craig Melchert, A Grammar of the Hittite Language, Vol. 2 Tutorial, Eisenbrauns, Winona Lake, 2008, pp. 75.
Exercise book to Vol. 1, Reference Grammar, to which it is an indispensable aid. It presents fourteen lessons, in a fixed format. Each lesson starts with advice on which parts of the reference grammar to study for the exercise, and how. This is followed by fifteen sentences to translate, most of them made up by the authors, but a considerable number taken from actual inscriptions. The pertinent vocabulary, in three parts: Hittite, Sumerograms, and Akkadograms, closes the lesson. No full translations of the sentences are provided, but most sentences come with translation hints. A combined vocabulary, again in three parts, terminates the booklet.
     I don't know why they call it a `tutorial'. A tutorial is something else in my dictionary.

Michael Meier-Brügger, Indogermanische Sprachwissenschaft, (in German: Indo-European Linguistics), Walter de Gruyter, Berlin, 2002, pp. 96 + 316.
Aptly titled: it is about the linguistics of Indo-European, more than about the language itself -- how we know the things we know about it. Probably at least 30% of the book consists literature references, explicitly on the first 96 pages, and then densely packed on each further page. Still we find paradigms for nouns and verbs, and readability does not suffer too much.
     The book is a veritable treasure trove of information about Indo-European Linguistics. It has a broad coverage, but does not shrink from an occasional deep exposition. It also shows a more personal touch than is usual in such books, in that the author regularly explains with much enthousiasm why this or that feature or phenomenon is intersting. About two IE controversies: 1. The author agrees that the traditional p/b/bh etc. system of IE consonants is improbable and unlikely to be the real system, but finds none of the proposed alternatives convincing enough to replace it. For the moment the p/b/bh etc. system is kept as a notational convenience. 2. The author agrees that Proto-IE had an animate/inanimate distinction; next Proto-Anatolian split off; and then the non-Anatolian languages developed a masculine/feminine distinction in the animate class.

Matthias Fritz, Zur Syntax des Urindogermanischen, (in German: On the Syntax of Proto-Indo-European), in Indogermanische Sprachwissenschaft, ed. by Michael Meier-Brügger, Walter de Gruyter, Berlin, 2002, pp. 241-280.
Description of same, in a very complicated German that would make Tacitus jealous.
     Starts with a very theoretical introduction into the nature of syntax, severely lacking in examples. However irritating such an approach may be, it makes explicit facts that would otherwise be left implicitly obvious; for example: plurality in a verb ending expresses plural actors rather than plural actions. Another non-obvious fact that is pointed out is that 'volition' may express either the will of the speaker or the will of the subject.
     The theory is then applied to PIE; for example, the attitude of the speaker towards the validity of the sentence is analysed into 'old fact', 'new fact', 'possibility', and 'future fact'. These are then mapped onto the Injunctive, Indicative, Optative, and Conjunctive, resp.
     PIE had appositive (extending) and restrictive subordinate clauses, the first introduced by Hio, related to the demonstrative, the second by kwi/kwo, which is related to the interrogatives. (Note that this corresponds to 'which' and 'that' in English, but that the relation is reversed.) Subordinate clauses were characterized by a stressed finite verb (as, remarkably, in Dutch, but not in English).
     The finite verb expresses five dimensions: person (1st, 2nd, 3rd); number (singular, dual, plural); modus (indicative, injunctive, imperative, conjunctive, and optative); tense-aspect (present, aorist, perfect); and voice (active, middle, stative). In addition there is a dimension 'mode of action', (iterative, causative, etc.), expressed less systematically by new stems created by suffixes to the root of the verb, or by suppletion. Tense-aspect and mode of action were already entangled in PIE; paradigms like ferō, tulī, lātum, are late remainders of this entanglement.
     In principle a distinction should be made between 'voice' and 'diathesis'. 'Voice' describes the same situation from a different point of view: 'the cat eats the mouse' (active) vs. 'the mouse is eaten by the cat' (passive). 'Diathesis' describes slightly different situations: 'he makes a meal' (active) vs. 'he makes himself a meal' (middle).
     PIE had eight cases: nominative, vocative, accusative, instrumental, dative, ablative, genitive, and locative. They can in principle be divided in 'supplementary' ones, i.e. those required by the sentence structure (nominative, accusative), and 'indicative' ones, cases that supply new information (the rest).
     The treatment of the PIE cases aims at determining the basic meaning of each, often with surprising results. The accusative was originally a supplement to the finite verb, and its precise meaning was determined by the verb: object, direction, extent ('Wir haben den ganzen Tag (acc.) gewartet' = we waited all day'). The instrumental indicated presence to the action, by a person (company), a thing (tool), or a location (the path along which). The dative denoted involvement; any positive or negative connotation was not part of the PIE dative. The ablative was simple: it denoted point of origin of a motion. The genitive was originally a partitive: cum grano salis = with a grain, a bit of salt. The locative simply denoted the place of action.
     In intransitive sentences the locative pertains to the subject, in transitive sentences it pertains to the object; this can be seen as an ergative trait. The masculine/feminine/neuter system is a development of an earlier common/neuter system. The lack of an ending for the neuter points again to an earlier ergative system.

Winfred P. Lehmann, Pre-Indo-European, Journal of Indo-European Studies Monograph Series, Volume 41, Study of Man, Washington, 2002, pp. 287.
The basic tenet of this book is that Pre-Indo-European was an "active" language, dividing nouns in animate and inanimate. This distinction determines the meaning of the verb gouverning the noun: "John Xs the stone", X=drop, "The stone Xs", X=fall. Such languages are charactized by the ample use of particles to narrow down meaning; several such particles are identified. Given the attempted time depth of about 8000 years, many of the details remain vague, but the author takes pains to indicate the extent of the vagueness.
Several conclusions are drawn from the presence and absence of cultural terms in the newer languages, many of them negative: the home land cannot have been "the northern perifiery of Southwest Asia" or Central Turkey. And the religion of the pre-Indo-Eutopeans was "simple".
The author assumes that the reader has read all pertinent literature.

Colarusso, J., More Pontic -- Further Etymologies between Indo-European and Northwest Caucasian, 2001, pp. 20.
Very extensive etymology of PIE ə3ə1-ek'-w-o-s = `hippos', `equus', showing at the same time that PIE was in contact with PNWC (loan of the PIE form e`kw to the NWC language Udi) and by deriving it from a (typographically challenging) form of the PNWC verb for `to run'. In the same vein the word for `goat' is derived from the verb for `to drive'. Much is done by postulating laryngeals in the fortified (= internally reconstructed) PIE forms, which then correspond to laryngeals in PNWC.

Joseph Harold Greenberg, Indo-European and Its Closest Relatives -- The Eurasiatic Language Family -- Vol. 1. Grammar, Stanford University Press, Stanford, 2000, pp. 326.
Collection of data that support the Eurasian superstock, plus argued account of how this superstock is delineated. Eurasian is comprised of (from west to east) Indo-European, Etruscan, Uralic-Yukaghir, Altaic, Korean-Japanese-Ainu, Gilyak, Chukotian, and Eskimo-Aleut (Etruscan is included after some hesitation, due to lack of sufficient material; but the 'mi' = I, 'mini' = me is taken as a strong indication).
     The supporting material comes in two groups, phonetic reconstructions and common morphemes. Chapter 2 contains the phonetic reconstructions. Eurasian is reconstructed to have alternations between p and m, t and n, etc., and examples in all language families are supplied. Also, vowel harmony is reconstructed for Eurasian, again supported by examples. Three more phonetic reconstructions are given.
     Chapter 3 presents 72 common morphemes. The most prominent ones are of course 'mi' = I, 'ti' = you, 'k-' = who, but others are impressive too: '-k' = dual, '-t' = plural, etc.
     A 37 page appendix (pp. 241-277)contains an analysis of Ainu vowel alternations, which seems somehow out of place.
[DG: I suppose the author performed this study in the course of the preparation of this book, and it was this study that convinced him to group Ainu with Korean and Japanese, rather than with Austronesian (Bengtson) or Austroasiatic (Vovin).]

Anna Giacalone Ramat, Paolo Ramat, The Indo-European Languages, Routledge, London, 1998, pp. 526.
Sixteen chapters by different authors.
     1. IE Culture: points out that reliably reconstructing morphology and a lexicon is doable, but that reconstructing the accompanying semantics involves a lot of guess work. The components of the IE culture and ideology are identified as: the sacred, the military, and the economic, with many examples.
     2. Proto-IE, by Watkins: the best summary of PIE I know, with extensive paradigms and a lot of explanation. Remarkably the author refrains from reconstructing the personal pronouns.
     3. The IE Family: considers relationships to Uralic, Caucasian, Semitic. Is not adverse to the idea of Eurasian.
     4. Sanskrit: based on a PIE without laryngeals. Bartholomae's law: dht → ddh. No explanation of the origin of voiceless aspirates.
     5. Iranian: the data on the Iranian languages are muddled, which is reflected in the text. It is hard to give a coherent account of a language in which the accusative of xratush (= power, cf. demo-cracy) varies between xratuum and xraþβ0m.
     6. Tocharian: given the dearth of material on Tocharian, the text is very welcome, but it is disappointingly short, ending abruptly after 15 pages with an apology about 'A brief summary', whereas the other entries are 25 to 30 on the average. The text is also confusing; the sentence on p. 155, lines 11-12 is incomprehensible; also what are "alternating nouns", and what is "group inflection"? Summary: Almost all consonants exist in normal and palatalized form. The case endings are given in running prose only. It looks as if any resemblance to those of PIE is imaginary, except perhaps for the nom.acc. plural, with a nom. in -i and an acc. in -m. The numerals are clearly IE, and so are the personal pronouns, more or less: sing. ny-, t-/ty-; dual/plural we-, ye-. One and a half page of existing verb forms are given, with minimal explanation. Of the 17 roots involved only one, Toch A. klyos-, B klyaus- = "to hear", is recognizably IE; more IE information would have been welcome. The verb endings, active and medio-passive, are recognizably IE, Toch. A more so than Toch. B: -m, -t, -s (!) vs. -u, -t, -.m.
     7. The Anatolian Languages: There were three Anatolian languages in the 2nd millennium BZ: Hittite, Cuneiform Luwian and Palaic. Only Luwian survived the Sea Peoples, and gave rise to Hieroglyphic Luwian, Lycian and Milyan in the first millennium BZ. The position of the more remotely related Lydian and Carian is less clear. Luwian preserves differences between the three PIE velars, but Hittite has only two, in such a way that it is difficult to reconstruct a Proto-Anatolian, suggesting that Hittite and Luwian already split in PIE times.
     The paper features a number of useful tables in 6 or 7 Anatolian languages and sometimes Proto- Anatolian: the nominal endings, pronouns, and verb endings, active and medio-passive.
     Two pages on the position of the Anatolian within the IE languages, discussing several theories and opinions but not leading to firm conclusions, close the chapter.
     8. Armenian: Although Armenians are known from texts from the sixth century BZ, the language appears in writing only in the fifth century AZ. By then its relationship to IE is hard to recognise: ul = kid (Gr. pōlos, Goth. fula), hayr = father, and the notorious erku = two (from dwo). But we also find ateam = hate (cf. Lat. odium (it's almost too good; is that a loan?)). The problem is that there are rules, but they are unreasonable, f.e. -s → -kh, and dw- → erk-; and they don't apply uniformly: the first rule applies to endings only, not to normal words. So often 'special development' has to be invoked.
     The PIE stops have changed considerably in Armenian, but not unrecognizably so. The author claims the fit is better when we start from the glottalic version, but the difference is not impressive.
     Armenian prefixes and inserts vowel in several places to aid pronunciation, and the author shows a very clever scheme by Venneman, which explains precisely when and where such vowels appear, and also when metathesis (a regional feature) occurs.
     Several pages are spent on the intricate way Armenian words are derived from PIE. The pronouns are recognizably IE, but the exact etymology is often obscure. The research on Armenian etymology is probably severely hindered by the lack of intermediate languages, as are often available for the other IE languages.
     There is no gender difference, not even in the pronouns. Verb endings, present indicative: -m, -s, -y, -mkh, -ykh, -n (with -s → -kh). The aorist 3rd sing. has a prefixed e-: e-git = he found, from gtan= to find.
     In present--day Armenian, the apical and velar l-s form an opposition pair, and so do the tongue and throat r.
     9. Greek: The chapter is firmly based on PIE. It starts with a well-organized description of PIE phonetics, including the accent (pitch), from which first Proto-Greek and then Greek phonetics + accent is derived in detail. A summary of what happened in the dialects follows, including some modern dialects. Lack of examples and the use of highly technical terms make the text difficult to follows, though.
     The reader is assumed to know (classical) Greek well; words are hardly ever translated, and paradigms and endings are amply discussed but not shown; and the lack of examples requires one (me) to consult grammars and dictionaries. Sometimes the text is so abstract that the meaning is hard to discern, f.e. the last paragraph of the chapter.
     The author shows his irritation at the lack of a satisfactory explanation of the -ō ending for 1s, in the phrase "non-thematic -mi, thematic word-end -ω of whatever origin".
     10. Latin: The longest chapter in the book, at 61 pages. The author notes that the apparent relationship between the Latin and Celtic languages does allow the reconstruction of a Proto-Latin-Celtic node: parts of Latin are related to parts of Celtic through rules not shared by other parts. Such rules would both apply and not apply to Proto-Latin-Celtic. The author considers them parallel developments and/or coincidences. Likewise the ae → ē and au → ō need not be attributed to Umbrian influence but are so wide-spread that they may be an independent development in Latin.
     The chapter pays much attention to the Latin dialects, the languages in the country-side around Rome, where we still find such words as dingua, the original of lingua (hence Eng. tongue). Also the Latin in this chapter is not always the Latin we have learned in school: antiquom, equos (nom. sing!), and many unusual forms are given for the demonstrative pronouns, f.e. istae for istī (dat. fem. sing.), and sapsa for ea ipsa. The main forms of Latin used are from Plautus (~200 BZ) and from the beginning of Imperial times (~50 AZ). This gives a nice wide view of the language.
     All long vowels are indicated carefully, so we find lēx and sīc, which goes to show how badly we pronounce Latin. On the other hand, there are no laryngeals. Unlike the chapter on Greek, this one abounds with examples, which is a good thing. Translations are given sparingly; readers are supposed to know their Latin.
     Examples of PIE dh in Latin: dhe- → f- in feci; -dhe- → -b- in werdh- → verb- (cf. Eng. word) and in stHdh- → stab- (cf. Eng. stable).
     Paragraphs stretching over three pages complicate reading, and have induced me to insert additional paragraph breaks. The author suggests that words like ruptus lost their infix -n- rather than that rumpō gained it.
     11. Italic languages: The chapter starts with an unclear and somewhat emotional section on the relationship between the Italic languages and the "national dimension" of Ancient Italy. If I understand it correctly, there was no Proto-Italic from which Umbrian, South Picene, Oscan, etc., developed, but rather Italy started off with zillions of local dialects, which slowly converged to larger and larger clusters, forming the observed languages with their observed variations.
     No translations are given, which may be acceptable for Greek or Latin, but which is absurd for languages like Umbrian or South Picene.
     The author uses X<Y sometimes for 'X derives from Y' and sometimes for 'X leads to Y'; or I don't understand it.
     12. Celtic Languages: Starts with a well readable and level-headed description of the history of the older languages, often pointing out that we don't know all the answers. Appears to use the --sensible-- convention: Old X: we have names and text; Primitive X: we have names but no text; Proto-X: we have reconstruction only. OWCB = Old Welsh/Cornish/Breton, as opposed to Old Irish.
     Detailed pronunciations of Old Irish words are given, including a 'nasalized bilabial fricative', written here μ, without indication how they were obtained. But the notorious 'tau Gallicum' is not mentioned explicitly, and described as 'a new dental phoneme'.
     The primary mutation is lenition. It arose from the difference between single and double consonants: the double ones became single, with spirantization in some branches, and the single ones became lenited at the same time. The morphology of the noun is described extensively, with many derivations from the PIE, through Common Celtic; this gives a reasonable explanation of the modern declinations with their lenitions and nasalisations. These derivations also show how far the modern languages have drifted away from their originals, f.e. Welsh h^yn = 'older' < senjos. As a result of this reduction the plural of many nouns came to coincide with the singular: both donjos = man and donjī = men (donjos < gdonjos < khthonios) became dyn in Welsh, so W. enlisted a second plural, donjones, which yielded the modern plural dynion.
     Similar derivations are given for the Old-Irish verb endings. The stems are more problematic: the present stem has eight formations, AI-AIII, BI-BV, each explained by a suffix from PIE, but the subjunctive, future, and perfective stems require a lot of assumptions to be explained from PIE.
     The Old-Irish verb has a different form at the beginning of a sentence than elsewhere: beirid = he carries vs. -beir, both from PIE bhereti; the difference is explained by an unknown particle in second position, which later disappeared.
     Relative clauses were formed by putting -yo after the verb, and the relation could be subjective or objective: Old-Irish berte < beronti-yo = 'who carry' and = 'whom they carry'.
     13. Germanic: The weak declension of the adjectives is explained as copied from the n-declension of the nouns, which had a 'particularizing, determining function' in PIE: Platōn = the Plato. It was used wherever the adjective referred to something determined (gute Sachen, die guten Sachen), and later made part of the syntax.
     Likewise the prefix ga- > ge- is explained as a PIE particle expressing completion, and related to the Latin com-.
     The reduction of the PIE verb system with 3 moods, 3 tenses, and 3 voices to a system with 2 moods and 2 tenses caused the PIE particles, and later the verb prefixes, to gain importance.
     The predominant word order of Common Germanic was SOV, but the noun inflection still allowed much freedom. Of the four word orders of German and Dutch only two are explained: main sentence order (SVO..), as an innovation; and subsentence order (-S..OV), as a remnant of the old word order. Inverse order (..VSO) and question order (VSO) are not discussed.
     14. Slavic: On their way from PIE the Slavic languages have been befallen by three vowel shifts and three consonant shifts, the more recent ones quite regular. This gives great opportunities for derivations, and zillions of etymologies of Late Common Slavic and Old Church Slavic words are derived directly from PIE via Proto-Slavic: PIE k^erd- > PS sird-ika- > LCS sirdic'e (= heart, cf. Lat. cord-). Some show how far Slavic has deviated from IE: dn̩gh-uh2- (= tongue, Lat. lingua) > PS inzū-ka- > LCS jezyku.
     Detail: PIE -s > PS -x (guttural - not ks): PS damux (Lat. domus) > LCS domu.
     OCS nouns had a dual and the adjectives had a definite form in addition to their normal form. The definite was formed in PS by appending -iax, -iā, -iam (Lat. is, ea, id).
     The paper uses lots of diacritics, both on vowels and on consonants, without defining them, apparently relying on an unspecified transcription.
     The verb has five classes, each subdivided into one to three subclasses, and is described in considerable detail.
     15. Baltic: Place names in Central Europe suggest that about 1500 BZ the Balts as a people lived in an area roughly the size of the Black Sea around the position of present-day Moscow.
     The author emphasizes the conservative IE character of the Baltic languages by frequent comparisons to Hittite etc. The time line of the developments of the Baltic languages can be determined by considering if Latin or German loans are involved. The word order in the modern languages is SVO but in rural areas and folk tales it is still SOV.
     The many diacritics are not explained; the pitch accent is described, but too briefly. But morphology is covered amply, as expected. (Wikipedia: the hooks under vowels used to indicate nasalization (as in Polish), but today signal length, as does the macron; the long i is written y, to avoid an ugly hook under the i or a macron over it. The hooks under the consonants (or over the g) indicate palatalization. The dot over the e turns it into a closed (high) e, as opposed to the normal e, which is open. The accents and tilde indicate pitch, but pitch is no longer an issue in urban Lithuania, and they are not normally written. The hacek over the sibilants has the standard meaning.)
     16. Albanian: From the fact that writers used similar spelling conventions already in the first extant books in Albanian from around the end of the 16-th century, the author concludes that writing tradition started much earlier.
     The southern dialect, Tosk, differs from the northern in rhotacism of intervocalic n, which in the north causes nasalization of the preceding vowel. The author takes great pains to derive Albanian words from PIE, a sometimes difficult task, f.e. Alb. "dore" (=hand) < PIE g^her- (cf. Gr. cheir). The three PIE tectals get different representations in Albanian: k^ > th; k > k/q; kw > s, which may then undergo further developments.
     Many morphological processes involve umlaut (not written with an actual umlaut): plak = old, plur. pleq, where the old plural ending -i shows both in the umlauting of a to e and in the palatalization of the k to q. Similar effects are found in the verb, mainly based on ablaut.
     Albanian still has three genders, but neuter differs from masc. only in that its accusative is equal to its nominative. So a two-gender description is quite possible. There is a rather unsuccessful attempt to explain the prefixed and postfixed article: the 'prefixed' article actually follows the noun it agrees with; and although there is an example 'the book of the pupil', there isn't 'a book of the pupil', 'the book of a pupil' or 'a book of a pupil', so the effects are difficult to observe. Details of the verb system are described briefly in two pages. The system itself and the conjugations are not explained and are apparently assumed to be known, which makes this section less than useful.
     In summary an uneven book, as can be expected for 16 different authors; the sections on PIE and Latin stand out as superb, and the ones on Albanian and Tocharian are the least satisfactory, but even these are not disappointing. In total a very worthwhile book.

Manaster Ramer, A., Armenian -k` < PIE *-(e)s, 1996, pp. 38.
If one sets out to examine how an s could change into a k` (an aspirated k), it is prudent to make sure that that indeed was what happened. The author spends the first 21 pages doing that, by examining all alternative explanations offered in the literature and rejecting each of them. An important point is that the sound change also worked in the (non-morphological) words for 3 and 4: PIE treyes → Arm. erek`, PIE k'etwōres → Arm. č`ork`.
     There are two major problems with this sound law: 1. it is weird; 2. it applies only to the PIE -(e)s of the nominative plural, the PIE -s of the nominative singular having become zero in Modern Armenian. For the first problem the author follows Kortlandt `On the relative chronology of Armenian sound changes' (1980): -s → -x → -k`. The rest of the paper describes several published solutions to the first problem (all pretty complicated) that are not unacceptable to the author.
     The paper closes with an 8-page literature reference list.

Winfred P. Lehmann, Theoretical Bases of Indo-European Linguistics, Routledge, London, 1993, pp. 324.
For each aspect of our knowledge of Proto-Indo-European, historical backgrounds, present opinions and Lehmann's own opinion are given. His own main point is that Proto-Indo-European (or even Pre-Indo-European) was an active language (distinguishing between active and stative verbs and nouns), and that properties of such languages shed valuable light on the Indo-European dialects.

Colarusso, J., Phyletic Links between Proto-Indo-European and ProtoNorthwest Caucasian, in Mother Tongue 21, 1994, 1992, pp. 8-20.
During its history Proto-Indo-European was in close contact with the North-West Caucasian languages for a prolonged period of time, in a region called `Pontus Euxinus'. This influenced PIE considerably, enough for the author to consider PIE as one of the two Pontic language phyla, the other one being PNWC. Time depths are: 9,000 - 7,000 BZ for Pontic, 4,000 - 2,000 BZ for PNWC, 6,000 - 5,000 BZ for internally reconstructed PIE, and 4,000 - 3,000 BZ for comparatively reconstructed PIE. With such time depths lexical evidence is scant, but indications are found in the phonological inventories, which happen to contain much information, both in PIE and in PNWC. Extending PIE's strange phonological inventory in a natural way led to `internally reconstructed' PIE, which bridges the gap to Pontic. The author then gives eight steps which show how the laryngeals transform in the PIE daughter languages. A list of 70 common PIE/PNWC correspondences is produced. Some items are straightforward: PIE -eno, participle, PNWC -nə, dependent verb marker; others are more inventive: PIE -(t)er, kinship suffix, PNWC X-th-ər = `X-be-participle' = `the one who is X'. Another suffix is PIE -ā, feminine/abstract marker, PNWC -xa = `woman', -γa = abstract suffix. Also, the PIE sigmatic aorist is connected to PNWC -z-, accomplished-past particle.

Oswald J.L. Szemerény, Introduction to Indo-European Linguistics, (Einführung in die vergleichende Sprachwissenschaft), Oxford University Press, 1990, pp. 352.
A book in the best German scholar tradition, i.e. 30% of the text consist of footnotes, mainly supplying further details and literature references. And the term 'Introduction' in the title should be taken very, very literally: the author states facts, with one or two arguments, but for the rest immediately refers the reader to the literature, often citing papers that comment on other papers, etc. Also, although the main text is usually very conservative (f.e. hardly any laryngeals), the footnotes often refer to Semitic, Chinese, and even Nostratic.
     Remarkably, S. does not believe in the difference between tonal and non-tonal languages (page 73), claiming that stress and pitch are always related and that the rising and falling accents of Greek and Lithuanian can be explained as stress+pitch on the first half of the (long) syllable (=mora) or on the second. This may be (probably is) true for the IE languages, but 1. does not account for level-pitch differences as found in many languages; 2. does not allow pitch in short syllables; 3. does not account for languages without noticeable stress differences, f.e. Chinese. In short, S. seems to be provincial here.
     Much information about the PIE accent is obtained from Lithuanian: PIE kōrdós, Lith. širdìs.
     On laryngeals: after giving an extensive and objective account of the three-laryngeal theory and slightly ridiculing systems with more than four laryngeals, the author gives his own conclusion: a, e, and o are honest vowels, just as i and u; and there was only one laryngeal, the simple h. So there was no dō (to give), there was no deh3, there was just doh. Makes PIE sound a lot less grim.
     Since no IE language features more than two gutturals, S. accepts at most two gutturals for PIE: velars, with palatalized variants, and labio-velars; and even the latter developed from normal velars by absorbing a following w or u. So in Pre-PIE there was only one velar.
     Morphology. It appears that some PIE roots come in series of CVC_1, CVC_2, CVC_3, etc., all with (vaguely) related meanings. This suggests that in Pre-PIE these roots consisted of two parts, a basic root CV- and a root determinative -C; several root determinatives have been isolated, but no semantics has been suggested for any of them. (Remarkably a similar situation obtains in Hebrew. DG)
     The noun: The ending -bh- of dat., abl., and instr., derives from a postposition related to Eng. by (p. 165) (and the alternative -mo from meta/met?) (and by extension do the other weak case endings also derive from postpositions?).
     The paradigms of the nouns as shown are somewhat confusing because forms equal to previous ones are left out: already in PIE the genitive and ablative singular are equal; and so are the dative, ablative , and instrumental dual; and the dative an ablative plural. So the ablative is never shown since it is always equal to a previous form.
     Adjectives: basically nouns with adaptable gender. PIE had several suffixes for forming comparatives, superlatives, etc.: -yes- (Lat. senior from sen-yos) and -tero- for comparatives; and -isto-, -mo- (Lat. summus from sup-mos), and -sama- (Lat. maximus from mag-simos) for superlatives. But it is unclear what the difference was originally.
     Also, there is a surprising variety of words for value concepts: Lat. bonus-melior-optimus, Eng. good-better-best. And the Indic, Greek, Roman, and Germanic all have different terms for 'good'. Again there is no explanation.
     Pronouns. The treatment of the 1st and 2nd pronouns is less convincing than the other sections. The reason seems to be 1. that no dual is reconstructed, and 2. that S. does not accept suppletion for the 1st sing. and plural, so eghom and wei have to be brought in line with me and nos, resp. The problems also show in the increased use of 'clearly' and 'of course', as f.e. in "The dat. -ebhi has, of course, nothing to do with the instr. -bhi." (p. 218), without further explanation.
     But some things make sense, f.e. why in Old Latin the acc. of egō is mēd instead of mem. In the 2nd the acc. started as tem → tē, which reduplicated to tēte → tēd. Now mēd already existed as the abl., and by analogy to tēd, mem changed to mēd.
     Numerals. PIE seems to have had two words for 'one', one which meant 'alone' and requires three versions to explain the forms in the daughter languages, oinos, oiwos, and oikos; and one, sem, which meant 'single', as in Lat. semel = once.
     Interestingly PIE penkwe = 5 is explained as penk-kwe = 'fist-and', as in counting "1, 2, 3, 4, and fist". A very complicated explanation is given for the ending -ty in 'thirty', 'forty', etc. PIE 1000 was probably gheslo-, leading to Gr. khilo. Germ. tusant- is probably from tuso-kmtom = 'a strong (swollen) hundred'.
     The verb. S. supplies another explanation for the Lat. 2pl pass. ending -minī. Several plural endings show an additional -n: Gr. -men, Hitt. -weni, -teni, and in Ved. -thana occurs next to -tha. This suffix particle is reconstructed as -noi, and a 2pl middle ending -dhwe-noi is assumed next to the traditional -dhwe. With Lat. dw → b this gave -b(e)nei → -mnei → -minī. More plausible than the plural of a passive participle.
     Unfortunately S. does not give an explanation of the 1s ending -ō. (Personal pet theory: PIE had two forms for the 1s pronoun (in suppletion): m and h3; the first led to eghem, Ved. aham, and the -mi ending; the second to eghoh, Lat. egō, to the -ō ending and perhaps to the Hitt. 1s middle -ha ending.)
     Since the Hitt. -hi conjugation does not occur in Luwian, S. concludes that it is an innovation of Hitt. only; but the paradigms seem to allow the idea that Luw. had -hilike endings, but not as a specific conjugation.
     S. gives a good explanation of the fact that the future imperative forms are the same for 2s, 3s, and 2p, and almost the same for 3p: they derive from the normal imperatives + tōd = 'from there', by assimilation of the two ts in the forms.
     In the active, as a rule the accent was on the root in the singular and on the ending in the plural, but there are also verbs in which the accent is always on the root; these are called acrostatic. They are thought to have had original long root vowels; an example is tāks- = 'to work wood'.
     Short definition of the difference between active and middle: the active describes the action as is, objectively; the middle as how the subject experiences it, subjectively; alternative names could be 'objective' and 'subjective' voice.
     The treatment of passive and stative voice is short and inconclusive. No word about the perfect as a stative, although the hesitantly presented endings for the latter are almost equal to those of the first.
     Indicative: the athematic verbs get nothing, the thematic ones get -e- (the th. vowel). Subjunctive: the ath. verbs get a th. vowel, the th. ones get an extra th. vowel. Optative: the ath. verbs get ye/i, the th. ones get -oye, i.e. ye/i with a th. o.
     There is discussion about whether there was an 'injunctive' mood in PIE or not. S.: there was a tendency to use a 'residual/neutral/indefinite' verb form if an earlier verb form already specified mood and tense; the injunctives in Vedic are remnants of these.
     Verb stem formation. PIE had several methods for verb stem formation, of which three stand out: the infix -n-, and the suffixes -sk- and -yo.
     Three theories are given for the origin of the infix -n-, all in the small print. 1. A root suffix -no- or -nu-, related to a word for 'now', and subsequent metathesis (but why only with this suffix? DG). 2. Doubling of the final consonant for emphasis, with subsequent -CC- → -nC- (but kl̩w (to hear) + emphasis → kl̩ww → kn̩w → kl̩nu (as in Ved. sr̩nu-, with double w? DG). 3. if we believe in root determinatives there is third explanation: the -no- from 1. got in between the basic root and the root determinative.
     The suffix -sk- indicated repetition/habit; it survives f.e. in Eng. wash from PIE wod-sk-, 'repeated water action'. The suffix -yo- and its variants implied making or doing things; it still shows in English fill from full. Other suffixes are -t- and -s-, but it is unclear in how far they can be distinguished from root determinatives.
     The endings of the perfect were 1s -a, 2s -tha, 3s -e, 1p -me, 2p -?, 3p -or, and had the tendency to erode. To keep the perfect, it had to be strengthened. Methods included: 1. reduplication (Indic, Gr., some Lat.); 2. long grade throughout (Gmc.; according to S. the Goth. difference brak - brēkum does not go back to PIE); 3. addition of more tangible suffixes: -k- in Gr. (related to Hitt. past -ha?), -vi- in Lat.(from a part. wos?), -f- in Osc.-Umbr. (from -dh-?), -nš- in Umbr. (from -n-ky-?), and -t/d/tt- in Osc., Cont. Celtic, and Gmc. (and in Hung., Jap., and several other Eurasian languages; DG) (from a part. -to-? or from -dh-?). Only very tentative explanations of these suffixes are available.
     There is no reason to assume the existence of aspect in PIE: it exists only in Gr. and the Slavic languages, but the realizations have little in common.
     The participle. The active present participle suffix is well known from many languages: -nt. It may derive from the root em- = 'to take', and the actor suffix -t-. This would then be appended to the noun form of the verb: bhér = 'the carrying' + em- = 'undertake' + -t- = doer → bhér-em-t- → bhéront- = 'who undertakes the carrying'.
     The perfect participle suffix is much less widely spread: -wos-, which occurs in Vedic and in the Gr. eidoos ← weidwos = 'knowing'. Like the present participle it may derive from a verb root: wes- = 'to stay'.
     The reconstructible passive participle suffix is -meno-, well known from Gr. (pheromenos = 'carried'), but also present (but not productive) in Lat.: alumnus ← alo-menos = 'pupil (i.e. fed)'. It may derive from the root men- = 'to remain'.
     Further passive participle endings are -to- (Lat. laudatus) and -no- (Eng. take-n). It is not clear if these are original PIE or later developments.
     The infinitives in the daughter languages are varied in nature; probably PIE had no explicit infinitive, just verbal nouns. They all derive from forms root + nominalizing suffix + case ending, usually the dative, -i: suffix -tu-, f.e. PIE doh-tu-i → Ved. dātavē = Old Pr. dātwei = 'to give', but Lat. datū with abl. ending; suffix -es-, PIE gwīw-es-a-i → Ved. ǰīvasē = Lat. vivere; suffix -no-, PIE bher-o-no-m (acc.) → Goth. bairan = 'the carrying'.
     In total the following endings are reconstructed for PIE. 1. Non-stative endings 1s -m, 2s -s, 3s -t, 1p -me(s), 2p -te(s), 3p -nt, possibly followed by -o- for medio-passive, possibly followed by -i for 'here and now',except that 1p and 2p get -s. 2. The quite different static endings 1s -ha, 2s -tha, 3s -e, 1p -me, 2p -?, 3p -r.
     Basically one would expect the non-stative endings to be postfixed pronouns, as they are f.e. in the Uralic and Semitic languages. For 1s -m, 1p -me(s), 2p -te(s) and even the tentative 1d -we(s) this is easy. But for 3s and 3p this is problematic, and for 2s it is hopeless: there is no believable way to turn -tu into -s. Various (pseudo-)solutions are reported. S. seems to be somewhat partial to the 'ergative explanation': in an (early) ergative state of PIE the verb forms were actually nouns + possessor suffix. The noun was formed from the verb root by the action suffix -t-. So we get: bher-t-m = 'carry-action-mine' = 'my carrying exists', bher-t-t, bher-t, bher-t-me, which contracted into bher-m, bher-s, bher-t, bher-me, with -tt → -s, which is kind of acceptable. (Against: then why didn't 2p bher-t-te contract into bher-se?) (Against: in ergative languages the possessor usually corresponds to the object rather than the subject: 'the carrying I get/have'. DG.) (It seems more reasonable to suppose bher-t-m = 'carry-actor-am' = 'I am carrier'. DG)
     The endings of the perfect are also used in Vedic for the imperf. passive (which is not surprising considering it represents a stative: "exist" means roughly the same as "was created"; this is finally confirmed on page 338, the last page of the text).
     No credible explanation of the perfect endings is given. (Personal pet theory: they are forms of a very old and otherwise lost verb 'to be'. The traditional regular verb hes- cannot have been the standard verb for 'to be' in PIE; no language has a regular verb 'to be'. DG)
     Next to the personal endings, there are the subj. and opt. suffixes to be explained. The opt. suffix -o-ye may derive from the verb root ei- = 'to go': PIE po-o-ye-em → pōyēm = 'drink-go-I' → I go to drink → I want to drink → may I drink. No reasonable origin of the extra thematic vowel for the subj. is suggested.
     In the last paragraph the author warns against 'reduction ad infinitum'.

Kortlandt, F., The Spread of the Indo-Europeans, 1989, pp. 7.
Mainly confirms Mallory's assignment of specific Indo-European waves of migration to specific archeological sites, and corrects many of such assignments by Gimbutas. The last paragraph suggests that there is a (North-)Caucasian substratum in Indo-European.

Kortlandt, F., Proto-Indo-European Tones?, 1986, pp. 8.
If proto-Indo-European had tones they are not related to those of Vedic, Classical Greek, etc., which are later developments.

Luhr, R., Eine weitere Moglichkeit der Genese von anlautendem Germanischem *p-, (in German: Another Possibility for the Origin of the Initial p- in Germanic), 1986, pp. 254-277.
All Indo-European initials p- should have become b- in Germanic, but still initial p- occurs in Germanic. Eight possibilities have been suggested in the literature; they are briefly explained in the paper, without examples:
1. (W. Meid) Germanic initial p- derives from PIE b-.
2. (H. Kuhn) Words with initial p- survived in a non-Germanic PIE "Nordwestblock" language, from which they were borrowed by Germanic languages.
3. (G. Neumann) Such words were borrowed from continental Celtic languages.
4. (Neumann & Meid) They derive from continental Celtic words with b- which then turned into p-, as in 1. above.
Four more suggestions are described even more briefly.
     The new possibility suggested by the author is the PIE (s)p-. More than 20 roots of the form sp-/p- are given, each supported by many examples, mainly from Scandinavian dialects.

Kortlandt, F., Proto-Indo-European Glottalic Stops: The Comparative Evidence, 1985, pp. 19.
On the basis of lots of sometimes very complicated evidence from the Baltic languages (tones), Slavic (vowel length), Armenian (in-depth analysis of the consonant systems of 7 representative Armenian dialects), Vedic and Avestian (vowel length and pre-glottalized stops), Sindhi and Panjabi (glottalics in Sindhi and low tone in Panjabi), Greek (assimilation in numerals), Latin (vowel length and reconstructed pre-glottalized stops), and Germanic (pre-aspiration in Icelandic, Faroese, and Norwegian), the author reinterprets the traditional PIE T, D, and Dh as fortis, glottalic lenis, and aspirated lenis, resp.
[DG: So all three are marked, which is typologically odd.]
     The present reinterpretation provides explanations for the following phenomena:
1. There is no z in PIE because voicedness was not a feature of PIE.
2. PIE b is rare because it was actually p^ɂ, and glottalized p-s are rare cross-linguistically. More in particularly p^ɂ was the first to lose its glottalization and merged with p.
3. There are no PIE roots with DVD because D is a glottalic and cross-linguistically two glottalics in one word are rare.
4. [DG: in another Kortlandt paper: t is used extensively in PIE verb conjugations but d is not because d is actually t^ɂ and glottalized consonants are avoided in conjugation morphemes in languages.

J. Colarusso, Typological Parallels between Proto-Indo-European and the Northwest Caucasian Languages, in Bono Homini Donum, ed. by Y.L. Arbeitman, 1981, pp. 475-558.
Summary of Sect. 10-16. The behavior of laryngeals in the North-Caucasian languages is analyzed and the results are used to get a new view on the PIE laryngeals. More in particular, the phonemes are decomposed in their features--low, lax, back, aspirated, etc.--, and the effects of these on vowels and assimilation are analyzed and compared to effects observed in the IE languages.
     On the basis of this the author arrives at the following most likely phonemes for the PIE laryngeals: h1 = χ, h2 = ħ, h3 = ħw, and h4 = h, the first three of them with their voiced counterparts.

W.R. Schmalstieg, Ergativity in Indo-European, in Bono Homini Donum, ed. by Y.L. Arbeitman, 1981, pp. 243-258.
The author uses the term 'diathesis', which apparently means the axis active, passive, medium. Some languages have no diathesis: a sentence like 'carrying man-X horse-Y' can then be interpreted/translated in several ways: 1. actively, as 'the horse carries the man', with -Y as -nom. and -X as -acc.; 2. passively, as 'the man is carried by the horse', with -X as -subj., and -Y as -agent; 3. Ergatively, with -Y as -erg. and -X as -abs.; 4. Genitively, as 'horse's carrying for the man', with -X as -gen. and -Y as -benefic . Note that the sentence does not contain a verb: 'carrying' is at most a verbal noun.
     The author starts with the fourth interpretation for an early form of PIE: bhr-to patr-os wir-om = 'the carrying of the father for the man', where -to is a determinative, -os is the subjective genitive ending, and -om the benefactive one. Later this was reinterpreted as active: the -t became a verb ending, the -os became a nominative ending, and the -om became an accusative ending. This explains why nominative and genitive both end in -s.

Paolo Ramat, Linguistic reconstruction and Indo-European syntax, Current Issues in Linguistic Theory, Vol. 19, John Benjamins, Amsterdam, 1980, pp. 236.
There is much disagreement on the subject, but two things seem to be more or less clear. 1. We do not have the theory and the techniques to reconstruct syntax with the same confidence as we reconstruct lexicons and morphology. 2. The syntax of Proto- (Pre-?) Indo-European is SOV.

Paul Kent Andersen, On the reconstruction of the syntax of comparison in Proto-Indo-European, in Linguistic reconstruction and Indo-European syntax, Current Issues in Linguistic Theory, Vol. 19, ed. by Paolo Ramat, John Benjamins, Amsterdam, 1980, pp. 225-236.
A very systematic approach which confirms the form adjective--yes + standard-ablative. Since this form is doubly marked -- the -yes and the ablative -- there was room for development and later dialects could come up with differing forms.

Catherine A. Callaghan, An `Indo-European' type paradigm in proto Eastern Miwok, in American Indian and Indoeuropean studies, Trends in Linguistics, #16, Mouton, The Hague, 1980, pp. 31-41.
The pronominal series for nouns and some verbs in proto Eastern Miwok (-ma, -s, -, -mas, -tok, -phu) shows remarkable similarity to the Indo-European sequence, but can be shown to be a relatively new development. Other series show no similarities.

Martha B. Kendall, The unethical dative, in American Indian and Indoeuropean studies, Trends in Linguistics, #16, ed. by Kathryn Klar et al., Mouton, The Hague, 1980, pp. 383-394.
An amusing collection of expressions that use the `unethical' dative, of the type: `Man hat mir mein Fahrrad geklaut', in many Indo-European languages.

Hasmig Seropian, Indo-European, Classical Armenian, and Modern Armenian, in American Indian and Indoeuropean studies, Trends in Linguistics, #16, ed. by Kathryn Klar et al., Mouton, The Hague, 1980, pp. 469-476.
Shows that Modern Armenian is not derived from Classical Armenian, but is rather an independent development of Proto-Armenian, by showing several features which are present both in Proto-Indo-European and in Modern Armenian, but which are absent in Classical Armenian. These features include word order, suffix composition and the tendency to omit the subject of a sentence.

Hans Krahe, Grundzüge der vergleichenden Syntax der indogermanischen Sprachen, (in German: Fundamentals of the Comparative Syntax of the Indo-European Languages), Innsbrucker Beiträge zur Sprachwissenschaft, Institut für Vergleichende Sprachwissenschaften der Universität Innsbruck, Innsbruck, 1972, pp. 136.
Series of lectures on same, published by his pupils. Very detailed but eclectic information; in spite of its title Albanian, Armenian and Tocharian are not even mentioned. It is remarkable to see the author, a big name in Proto-Indo-European, show on several occasions that he thought of Proto-Indo-European as more primitive than present languages, for example on p. 85, where he turns to child language to adduce properties of Proto-Indo-European.

W.B. Lockwood, Überblick über die indogermanischen Sprachen, (translated from English: `A Panorama of Indo-European Languages'), Gunter Narr Verlag, Tübingen, c1972, 1979, pp. 319.
Fairly superficial treatment of the 13 (!) branches of Indo-European, describing for each branch its history, the history of its research, some morphological details and short sample texts. The structuring is extremely conservative: Osco-Umbrian and Venetic are considered separate branches and the Indo-Hittite hypothesis is not mentioned. No literature references. Not a serious book.

Winfred P. Lehmann, Proto-Indo-European Phonology, University of Texas Press, Austin, Tx., 1952, pp. 129.
After a history of and introduction to the laryngeal theory, proto-Germanic data are squeezed for effects of the laryngeals; some additional information is obtained from Greek and Indo-Iranian. All this results in the reconstruction of seven(!) stages of the phonology of pre- and proto-Indo-European:
the pre-stess stage of pre-Indo-European,
the stage of pre-Indo-European with phonemic stress,
the period of non-distinctive stress,
the stage of pre-Indo-European with distinctive pitch,
the period of non-distinctive pitch,
the stage with partial loss of laryngeals and with long vowels, and
the proto-Indo-European phonemic system.