Read Part I http://vetiver.weblog.ro/2012/01/19/mihai-vinereanu-the-origin-of-the-romanian-language-i/#axzz1jki4yVPD and Part II http://vetiver.weblog.ro/2012/01/19/mihai-vinereanu-the-origin-of-the-romanian-language-iii/#axzz1jki4yVPD
The Proto-Indo- European short and long vowel *i had a similar evolution to Proto-Indo-European *e, namely it underwent iotacism affecting the consonant in front of it. The noun in “flax” is considered to be of Latin origin from linum ”flax”, but the term is found in most Indo-European languages and most probably in Thraco-Dacian as well; cf. Greek línon “flax”, Albanian (Tosk liri, Gheg lini), Old Irish lin ”id”, Welsh llin “id”, Breton lien, Gothic lein, OHG lin, Lithuanian linai (pl.), Latvian lini (pl.), Old Prussian linno, all from PIE *līno “flax” (IEW, 691). On the other hand, the verb a lipi “to glue, to stick ” is considered to be of Slavic origin, although there is no appropriate Slavic etymon, but there are close cognate in Baltic language; cf. Lithuanian limpu, lipti “to stick, to glue”, Latvian lipu, lipt “id” from PIE *leip- “to grease, to glue” (IEW, 670). In this case, as one may see the vowel i (< PIE *ei) did not undergo iotacism.
The consonants: The Proto-Indo-European consonantal system suffered a few major changes. Two of them took place apparently long time ago in Thraco-Illyrian, namely, the loss of aspiration of the aspirated stops which collapsed with their non-aspirated counterparts. As I mentioned already, this phonological change is shared with Celtic and Balto-Slavic languages. Another major change is that the Proto-Indo-European labio-velars *gw and *kw turned into labials (b, p) when followed by a back vowel (a, o, u) (see supra). The phenomenon took place most probably in the second half of the last millennium BC. It is attested in Thraco-Illyrian glosses and it will be discussed below.
The Sibilants: The Proto-Indo-European sibilant *s turned into a š (ş), most probably in Thraco-Dacian (sometimes in the 1st millennium BC) before a Proto-Indo-European front vowel (i, e) which underwent iotacism earlier. The same change is consistently found in Latin lexical elements. Its voiced counterpart did not exist in Proto-Indo-European. Romanian şase “six” was derived from Latin *sess < sex “six” by Tiktin (ZRPh. 12, 456) which was accepted by all other linguists, although the presumptive Latin etymon would yield in Romanian *şes or *şas. In some other Indo-European languages the form for “six” sounds similar to Romanian; cf. Lithuanian šeše, Latvian šesi or even Sanskrit shash, all from PIE *seks, *sueks, *kseks (IEW, 1044). Any of this forms would give *šes or šas in Romanian The final vowel -e was added by association with şapte “seven”.
Romanian şopârlă “lizard” was associated with Albanian shapë, sheperillë “lizard” and, therefore, at first, it was considered of Albanian origin (Cihac, Meyer), but for Cioranescu, it is of imitative origin. Instead, Reichenkron (1966) considers it of Thraco-Dacian origin, from PIE *sk’eu- “to gush (out), to spring out”, since for Reichenkron, PIE *sk’ > š in Thraco-Dacian; a hypothesis accepted by Romanian linguists, but he is wrong about it. In this case, Romanian š is the result of a following iota. Thus, Romanian şopârlă originates from PIE *serp- “to crawl” (IEW, 912). Also, Romanian şarpe (var. şerpe) “snake” is considered to be of Latin origin, where ş has similar phonological evolution as şopârlă and it derives from the same Proto-Indo-European root. Needless to say that in Albanian shapë, shepirellë, š (sh) is the result of the same phonological environment.
Instead, when followed by a back vowel or a consonant, PIE *s remains unchanged.
Thus, Romanian samă (var. seamă) “1. reckoning; 2. number, amount; 3. care, interest; 4. like, equal; 5. report. comes from PIE *som(o) “one, together, same” (IEW, 903), with cognates in many Indo-European languages, is of Thraco-Dacian origin.
When followed by a consonant, it has remained unchanged as in the noun sterp “sterile, unfruitful”, considered of Thraco-Dacian origin, since it does not have a correspondent in Latin, but has a close cognate in Albanian shterpë “sterile, unfruitful” and even in some southern Italian dialects stirpa and Venetian dialect sterpa “id”. There is no change of PIE *s in the same phonological environment (in front of a back vowel or a consonant) in the lexical items considered to be of Latin origin as in sare “salt” from PIE *sal-, *sald-, *sal-i, *sal-u ‘salt, sea salt” (IEW, 878). The root is found in many Indo-European languages including Thraco-Dacian and Illyrian place-names; cf. Thraco-Dacian Sald-apa, meaning “salty water” or Salmo-rude, a lake adjacent to the Black Sea ( today ’s Lake Razelm) or Sal-entinai (in Dacia ( Walde, 2, 466), Illyrian Saldae (in Pannonia), or Sal-s-ovia (in Thracia). Also, Romanian a sta “to stay, to stand, to live” from Latin stare < PIE *sta-, stə- “to stay, to stand” (IEW, 1004).
The bi-labial stops: In most cases, the Proto-Indo-European voiced and voiceless bi-labial stops did not change in Thraco-Dacian and Romanian. Thus, Romanian pânză “fabric” was given various “etymologies”, by different linguists. Most of them, opted for a Latin origin, either from Latin pandere “to stretch, to expand, to spread” (Puşcariu, 1373; Tiktin; REW, 6190) or Latin *pandea “fabric” (Cioranescu, 6400), which is not attested nor in Latin, neither in any of the Romance languages. Needless to say, neither of these hypotheses can be accepted. On the other hand, Pascu (1, 191) and I.I. Russu (Elemente, 101) considered it to be of Thraco-Dacian origin. Pascu associates it with Greek péne “fabric”. In fact, Romanian pânză derives from PIE *pand- “fabric (Gewebe) (IEW, 788). The Thraco-Dacian form must have been *pandia > *panza, with cognates in many other Indo-European languages, besides Greek; cf. Latin pannus “fabric, rag”, Gothic fana “fabric”, OHG fano “id”. Indeed, Romanian noun pânză has a cognate in Latin pannus, but it cannot be the etymon of the Romanian noun. Proto-Indo-European non-aspirated voiced bi-labial *b was not so frequent. It was preserved as such in Thraco-Dacian and Romanian, as in buză “lip” from PIE *bu-“lip, kiss”. Instead the voiced aspirated *bh was much more frequent. Romanian brânză “(aged) cheese” is considered Thraco-Dacian since late 19th century by Hasdeu (Cuvente, 1, 190), but no linguist could identify the Proto-Indo-European root it originates from. It derives from PIE *bhrendh- “to swell, to ferment, to ripe”, with cognates in Albanian and Lithuanian; cf. Albanian brenza-t (pl.) “interius, viscera”, brendësat “rennet’, Lithuanian brestu, brendau “to swell, to ripe”. From Romanian it was borrowed into all neighboring languages.
The dental stops: The Proto-Indo-European voiceless dental stop turned into apical ts (ţ) when followed by a front vowel (see supra). Otherwise, it was preserved in Thraco-Dacian and Romanian. The noun ţarc “enclosure” was originally considered to be a loanword from Albanian cark “id” (Treimer, 38, 391, ZRPH.; Pascu, 2, 222): Philippide, 2, 738; Rosetti, 2, 123). On the other hand, Reichenkron (165) considers it to be Thraco-Dacian from PIE *serk- “ enclosure, to enclose” (IEW, 912), but in this case PIE *s would give š, not ts (ţ), but I.I. Russu (An. Muz. de E.T., 1958, 146) correctly connect it to the PIE *tuer- “to enclose”, *terko “enclosure” with cognates in a number of Indo-European languages; cf. Lithuanian tveriù, tvérti “to enclose”, tvártas “fence, enclosure”, Latvian tvāre “fence”, Old Prusian toaris “barn, granary”, OCS zatvoriti “to lock, to enclose”. From Romanian was borrowed in some neighboring languages; cf. Neo-Greek tsarkos and Ukrainian carok, carka, cerkati “to milk”. When followed by a back vowel or a consonant remained unchanged, as in Romanian tare “1. strong, hard, tough; 2. very” from PIE (s)ter-, *(s)tero- “tough, rigid, to be rigid” (IEW, 1022). Although Walde-Pokorny reconstructed a Proto-Indo-European root with vowel *e, all forms of the Indo-European languages are with a, as in Romanian; cf. Sanskrit taras “rapidity, strength, energy”, Hittite tarhuiti “tough, strong”, Germanic *stara “tough, strong, powerful”, except for Greek stereós “tough, strong”. In any case, if we will consider the Proto-Indo-European form given by Walde-Pokorny as correct, it means that the PIE *e turned into *a before iotacism.
The non-aspirated Proto-Indo-European voiced stop *d has have a similar evolution as its voiceless counterpart. Originally, most linguists considered Romanian mânz “colt” to be Illyrian (not Thraco-Dacian) loanword, namely from a Illyrian *mandus or *manzus. However, in the last 50 years some linguists think it to be of Thraco-Dacian origin (I.I. Russu, Elemente…, 180; Poghirc, ILR, 2, 332; Brâncuş, VALR, 97). The root is attested in Celtic place-names; cf. Gaulish Epo-manduo-dunum and Brittanic Mandu-essedum (cf. Walde) and in the Messapic (Illyrian) god-name Jupiter Menzana “Jupiter of the horses” to whom young horses were sacrificed. Besides these ancient languages and Albanian, the root is also present in some modern languages or dialects; cf. Sardinian mandzu “calf”, Italian (dial.) manzo “calf”. De Mauro-Mancini (1176) considers Italian manzu to be of pre-Roman origin. In Romanian, the root has a few other derivatives; cf. mânzat “one year old calf”, mânzare “milking sheep”, as well in Albanian mënd “to suck, to feed”, mëndëshë “wet-nurse”, mëz “colt”. The same phonological change are found in a series of Thraco-Dacian personal names such as Zia or in god-names such as Saba-zios or Gebelei-zis, where the second component is a cognate of Latin deus “god”. In some ancient works Saba-zios is also spelled as Saba-dios (see zeu “god”).
The velar consonants: The Proto-Indo-European velar stop *k followed by a back vowel remained unchanged. Romanian caună “mine, salt mine” (reg.) was considered to come from Latin *cavina < cavus “hollow” (Puşcariu, 324; DAR). The Latin form has no attestation and there are no similar forms in any of the Romance languages. On the other hand, in Romanian there is a multitude of forms deriving from the same root as caună; cf. cavă “depression”, căuc (căuş) “laddle, dipper, scoop”, caval “(little) ditch”, gaură (dial. gavră) “hole, opening, gap, cavity, crack”, găunos “hollow”. All these forms were given, over the years, various etymologies by different authors. There are too many to mention them here, but all these words derive form the same PIE *k’eu-, *k’ou-, *ku- (Bomhard & Kerns), itself from a Proto-Nostratic *k’au-, k’əu- (Bomhard& Kerns, 281). Romanian lexical forms derive from an Thraco-Dacian root *kau-, *kou-; cf. Albanian gavër, (gavr)ë “hole, crack, opening”, Breton keo “cave”, kougon “hole”, Middle Irish cua “hollow”, cuas “hole”, Greek kōos “hole”.
Like all other stops, Proto-Indo-European *k, followed *e or *i was altered, turning into affricate č. Thus, Romanian river-name Cerna (found in different regions of Romania) originates in PIE *kers-, *krsno- “black” (IEW, 583). The Dacian place-name spelled Tsierna in Greek documents and Dierna in Latin documents was a Dacian city situated at the mouth of river Cerna, on the northern bank of Danube River. From the same root derives the verb a cerni “to color in black” and cerneală “ink”.
The Proto-Indo-European voiced non-aspirated velar stop *g had have a similar evolution. The verb a zâmbi “to smile” seems to come from PIE *g’embh- “to show teeth, to bite” (IEW, 369).
There is some evidence from Romanian for the existence of voiceless aspirated velar *kh in Proto-Indo-European, although, it was not as frequent as its voiced counterpart which turned into a laryngeal in Romanian. Neo-grammarians correctly considered that Proto-Indo-European had the voiceless aspirated velar stop *kh, using a small number of data from Indo-Iranian, Armenian and Greek. In most Indo-European languages, one cannot grasp the difference between the voiceless aspirated velar stop *kh and its non-aspirated counterpart *k, since the aspirated one lost its aspiration at an early stage of most Indo-European dialects, just before disintegration of Indo-European (cf. Bomhard& Kerns, 1994). Beside these languages, the laryngeal h is also present in Slavic (x) which could derive only from a Proto-Indo-European *kh. In Romanian, the laryngeal h derives mostly from either Indo-European voiceless aspirated velar stop *kh or the laryngeal *h. I have to mention also that in a few cases, the laryngeal h in Romanian may derive sometimes from Indo-European *gh which, first, lost its aspiration, than it turn into an /h/, as in horn „chimney” (see horn). A number of linguists disagree with Neo-grammarians’ hypothesis regarding the presence of voiceless aspirated in Indo-European. Nevertheless, as Bomhard maintains, voiceless laryngeal *h, present in a number of Indo-European languages should have an explanation.
On the other hand, according to Neo-grammarians, the laryngeals did not exist in Proto-Indo-European. However, today, the Laryngeal Theory is accepted by most linguists, although their opinions are divided regarding the number and the nature of these laryngeals. Furthermore, the elimination of Indo-European *kh creates major typological problems as well. Also, the fact that some Indo-European languages, including Romanian inherited it as a laryngeal, indicates the presence of voiceless aspirated velar stop *kh in Proto-Indo-European. Furthermore, A. Martinet (1970:115) shows that, from a typological point of view, the data from a great number of languages clearly indicate that the voiced aspirated can be added to the non-aspirated voice/voiceless pair, only if in the language, it is present the voiceless aspirated as well. In other words, we cannot have in Indo-European only the pair *k/*g and *gh, without *kh. R. Jakobson (1971:528) takes the same position, while Szemerényi shows that since the Indo-European had voiced aspirated, it should have voiceless aspirated as well. For him , there is just one laryngeal, namely the voiceless glottal fricative *h. He reconstructed the phonetic system of Proto-Indo-European to which he adds *h and eliminating some sounds from the Neo-grammarians’ one (as in Brugmann, 1904:52). A. Bomhard (1994) also indicates that the Laryngeal Theory cannot explain the origin of the laryngeals in most Indo-European languages.
Furthermore, Bomhard (1994:62) shows that in some cases a laryngeal explanation is not possible. In Romanian, the verb a hohoti „to guffaw”, considered to be of Slavic origin, although it does not seem to be, as well as the verb a pufăi „to puff, to blow, to pant” from an older a puhăi “id”, and the adjective puhav „swollen, puffy” which derives from the same root as a puhăi from a PIE *p(h)ukh- “to puff, to blow, to exhale”. if we have to take into consideration only the cognates of a hohoti from PIE *kha-kha “interjection expressing a laughter” (IEW, 634); cf. Sanskrit kákhati “to laugh”, Latin cachino “id”, Armenian xaxank “laughter”, OHG kachazzen “id” and OCS chochotati “id”. Another example where the Nostratic, as well as Indo-European *kh yielded h in Romanian is the noun hoţ „1. thief; 2. crook, charlatan” and it cannot be considered in any way of imitative nature. Its etymology remained controversial to this day. It is quite obvious that Romanian hoţ derives from PN *k(h)aly-/ *k(h)əly- „to rob, to steal, to hide” through the PIE *k(h)elp(h)-/*k(h)olp(h)- „to rob, to steal, to hide” (Bomhard & Kerns, 266); cf. Greek κλέπτω „I steal”, Latin clepō „I steal”, Gothic hlifan „to steal”, hliftus „thief”. In Walde-Pokorny (601) the velar is not aspirated, but palatal *k’lep- „to hide, to steal”, thus, one cannot deduce the origin of Romanian hoţ from this root, unless one looks at the Nostratic root. On the other hand, in Dravidian we have: Tamil kal (katp, katt) „to rob, to steal, to cheat”, kalavani, kalavali, kalvan „thief”, Malayalam kakkuka „to steal”. Forms with the dental /t/, we have Gothic hliftus “thief” and in Tamil. Therefore, we may suppose the existence of an older Thraco-Dacian *holtiu, with a later elision of the lateral /l/. Furthermore, Reichenkron (1966:132) thinks that Romanian hoţ derive from the PIE (s)keud(h)/*(s)keut- „to cover, to hide”; cf. Sankrit kuhaka „crook, charlatan”. The German linguist shows that the meaning changed from „to cover, to hide” to the one of „to cheat” is to be found in Greek and Latin as well. In the case of Romanian language, we may suppose that the two meanings overlapped leading to the overlapping of the two forms or we should think of having in this case just one Indo-European root, instead of two.
The Proto-Indo-European voiced aspirated velar lost its aspiration (see supra). The verbs a găsi “to find”, a gândi “to think”, a ghici “to guess, to predict, to divine” derive from PIE *ghend-, *ghed- “to apprehend, to understand” (IEW, 437). In all these examples, Thraco-Dacian *g followed by *e was not palatalized, because e > ə, before iotacism. Cognates of these three verbs are in Latin, Albanian, Germanic and Slavic languages; cf. Latin apre-haendo, compre-haendo “to grasp, to understand”, Gothic bi-gitan “to find”, Old English be-getan “to be given, to receive”, Albanian gjei “to find”, OCS gadati “to suppose, to guess, to think”.
The Proto-Indo-European labio-velars: As I have shown above Proto-Indo-European labio-velars turned into bi-labial (p, b) in front of a back vowel (a, o, u), but turned into a simple velar in other cases. The verb a lepăda “to drop, to relinquish, to abort” is considered to come from Latin lapidare “to cover with stones, to kill by stoning”, but this etymology should be rejected. Instead, I.I. Russu thinks it is of Thraco-Dacian etymology, but he gives not other details. This verb derives from PIE *lekw- “to leave behind, to drop, to relinquish” (IEW, 669), with cognates in several Indo-European languages; cf. Greek leípō “to leave behind, to abandon”, Latin linquo “to leave behind, to abandon”, Armenian likanem “id”, Lithuanian liekù “1. to leave behind, to abandon; 2. to remain, to stay”. The conjunction şi “and” comes from PIE *kwe “and” (enclitic) > *ke. The velar turned into an affricate in front of a front vowel: *-ke > şi “and”; cf. Sanskrit ca “and”, Avestan, Old Persian ča “id” (enclitic), Latin -que “id” (enclitic), OCS če “and”, Lithuanian -ke “and (enclitic).
The Proto-Indo-European voiced labio-velar *gw turned into b. The noun bou “ox” is considered to be of Latin origin, but it is attested in Illyrian glosses (see supra). Romanian bou “ox” is considered of Latin origin by traditional linguistics, but is it attested in Thraco-Dacian and Illyrian glosses; Dacian bou-dathla “a plant” (in Dioscorides) as well as the Illyrian place-name Bou-dorgis translated as “oxen-tower” by Chantraine (1147), but, I think, a better translation would be “oxen-fair”. The second component should be connected to Illyrian place-name Tergeste (today’s Triest) as well as Romanian târg “fair, market” and city-name Târgovişte. Romanian bou < PIE *gwou-s “cow, ox” (IEW, 482), where the Proto-Indo-European voiced labio-velar *gw turned into b, a widespread transformation found in Thraco-Illyrian, Osco-Umbrian, Continental Celtic as well as in some Greek dialects (see supra); cf. Latin bos, bovis “ox” (a loan-word), Greek bous “id”, Umbrian bum, Welsh buwch, Breton buch, as well as Sankrit gauh “cow”, Avestan gauš “id’, Armenian kov “id”, Latvian guovs, OCS govedo, Tocharian A ko “id”, Albanian ka “cow”. One may see clearly that Albanian form is not a descended of Illyrian. As I mention above, Albanian’s ancestors are the Epirotic dialects (see supra). On the other hand, Latin bos was borrowed from Osco-Umbrian or a Continental Celtic or Illyrian dialect, since in Latin, PIE *gw and *kw never turned into b, respectively p (cf. Latin aqua). In a closed stressed syllable, PIE *o turned into a in Thraco-Dacian and preserved as such in Romanian. Beyond Indo-European languages, the root is found in other Nostratic families such as in Afro-asiatic (Orel, 905) and Dravidian and Sumerian (Bomhard & Kerns, 346) as well. When it was followed by a front vowel, it was palatalized. The noun jar “embers” < PIE *guher-, gwhermo- (IEW, 498); cf. Albanian zjar, zjarm “id”.
The nasals: Both Proto-Indo-European nasals (m, n) were preserved in Thraco-Dacian and Romanian. The noun mire “bridegroom” was given various etymologies over the years, but Poghirc (ILR, 2, 345) and Brâncuş (VALR, 142) consider it to be of Thraco-Dacian origin. Romanian mire derives from PIE *merio- “young man” (IEW, 738); cf. Sanskrit marya “young man, lover, fiancé”, Old Prussian martin “bride”, Lithuanian marti “bride”. Romanian mireasă “bride” is a derivative of the masculine form and the verb a (se) mărita “to marry (about women)” is considered to be of Latin origin, and originates from the same Proto-Indo-European root. The root is found in some other Nostratic families such as Afro-asiatic and Dravidian languages (Bomhard& Kerns, 522).
The verb a necheza (Aroumanian necl’eazare) “to neigh” was considered by traditional linguistics to be a derivative of a râncheza “id” which was given different etymologies either Greek or Latin. In fact, a râncheza is a phonetic variant of a necheza, not the other way around as it was believed. The verb a necheza has cognates with the same meaning in Germanic languages or similar meanings in other Indo-European languages. It derives from PIE *kneug- “imitative formation” (IEW, 608); cf. Greek knizán “to snarl”, Lithuanian kniaukti “to meow”, Old English knaegan “to neigh”, MHG nēgen “id”.
All non-round vowels (a, e, i) turned into î (spelled also â) , which is a mid-central vowel, in front of both nasals (m, n), in all lexical items of both Thraco-Dacian or Latin origin; cf. a spânzura “to hang” < PIE *(s)pend- to draw, to stretch” of Thraco-Dacian origin or împărat “emperor” < Latin imperator “emperor”.
The liquids: The Proto-Indo-European liquids (r, l) did not undergo major changes. The Proto-Indo-European vibrant lateral *r did not change in Thraco-Dacian and Romanian. The noun beregată “throat, esophagus” from Latin *verrucata < verruca “protuberance on the skin” (Puscariu, Dacor., 9, 440; Cioranescu, 776) cannot be taken seriously. This noun comes from PIE *bherug-, *bhrug-, *bhorg- “1. throat, trachea; 2. pharynx” (IEW, 145), cf. Greek pharynx, Armenian beran “mouth”, Lithuanian burna “mouth”. Unlike *r, the liquid *l underwent some changes. It turned into r (rhotacism), in intervocalic position, in most cases. However, there are some exceptions which are difficult to explain. On the other hand, it palatalized in front of an iotacized front vowel and eventually disappeared. Both these conditions apply to lexical items of Thraco-Dacian or Latin origin as well. The noun iepure “hare, rabbit” is considered to derive from Latin leporem < lepus “id”; cf. Italian lebre, Italian Calabrian dialect liepuru, Albanian lepur, Old Sicilian leporine, Ancient Greek (dialect of Massilia, today’s Marseille) leberís “id” (most probably a Celtic loan-word). Romanian iepure (Macedo-Roumanian l’epur(e)) comes much closer to Calabrian and Albanian forms. Italian Calabrian dialect inherits some Oscan phonological features. Instead, the noun femeie “woman”(Aroumanian femeal’e, fumeale “family”) is certainly of Latin origin. It derives from Latin familia “family”, underwent same phonological transformation. In intervocalic position, it underwent rhotacism both in words of both Latin and Thraco-Dacian origin. The noun sare “salt” is considered to be of Latin origin, from Latin sal, sal-is, but we saw that the root is attested in Thraco-Dacian place-names, all from PIE *sal-, *sald-, *sali-, *sal-u “salt, sea water” (IEW, 878) (see supra). On the other hand, the adjective fericit “happy” is of Latin origin, namely form Latin felix “happy”. On the other hand, the verb a feri “to protect, to cover, to avoid” was given several different “etymologies” not worth to be discussed. Reichenkron (120) derives it from PIE *pel-, *ple- “to cover, to hide, to protect” (IEW, 803) where p > f, due to the iotacized e and intervocalic l turned into r, a general rule in Romanian.
Laryngeals: The presence of Proto-Indo-European laryngeal(s) in Romanian was never discussed before. The laryngeal h is quite frequent in Romanian, but it has different origin, a fact that led to the confusion about its origin. It is well known that Late Latin lost all its laryngeals, therefore Romanian could not inherit the laryngeal from Latin. However, it is found in some Slavic loanwords such as odihnă “rest” and duh “spirit” , Turkish loanwords like hal “bad condition or situation”, halva “halva(h)” or Greek origin hartă “map”, hârtie “paper”, hamsie “small fish”. There are also several Gothic words in Romanian which have the laryngeal h: haită “1. pack of wolfs or dogs; 2. bitch”, haldău “cowboy” (reg. Transylvania), haiduc “outlaw”. Although in most cases, laryngeal h is found of Thraco-Dacian origin. Grigore Brâncuş (VALR, 1983) was the first Romanian linguist who understood that there are Romanian words of Thraco-Dacian origin that have the laryngeal h and therefore this sound was inherited from this language. On the other hand, Neo-Grammarians and most Indo-Europeanists after them believe that Proto-Indo-European have no laryngeals.
However, Bomhard (1994) considers that Proto-Nostratic had four larygeals: */?/, */h/, */ħ/ and */ſ/. Before the breaking up of Proto-Nostratic language, these laryngeals started to come closer to each other and to overlap. The new sounds were pronounced with an open glotta, thus, they turned into so called „a colored-A” laryngeals. At this moment, the Anatolian languages separated from the other Indo-European languages. Bomhard drew his conclusion based on the fact that Anatolian languages have the laryngeal, while most of the other Indo-European languages lost it. A better hypothesis would be that most of non-Anatolian Indo-European dialects begun to loose the laryngeals, but not all these dialects did, since the original Indo-European laryngeal still ca be found in a few daughter languages, including Romanian as I have shown above. He shows that in the late post-Anatolian Indo-European (the period when this language begun to disintegrate), all laryngeals turned into *h which, later on, disappeared from most Indo-European dialects. Based on evidence from Armenian language, Bomhard thinks that the only laryngeal of this period was the voiceless fricative laryngeal *h which seems to be correct. However, beside Armenian, it can be found in Romanian and Albanian as well.
The laryngeal /h/ is quite frequent in Romanian. Thus, the laryngeals in the verb a hămesi „to starve, to be hungry”, as well as in Albanian hamës „insatiable, glutton” represent the Proto-Indo-European laryngeal */h/. Forms with laryngeal having similar meanings there are in the Afro-asiatic languages. Thus, one may reconstruct a Proto-Nostratic *ham-/*həm- „to eat, to be hungry, to be insatiable”; cf. Semitic *nVham „to be insatiable”; Arabic nhm, West Chadic *hVm „to eat, to chew”, East Chadic *ham (cf. Orel, 1995:1157). Furthemore, we have Romanian verb hăcui „to cut into, to cut into pieces” which derives from the Proto-Nostratic *ħak-/*ħək- „to cut into” (Bomhard& Kerns, 401) > Proto-Indo-European *ħhek’-w(e)siH „ax”; cf. Greek άκξινη „ax”, Latin ascia < *acsia „ax for carpenters and masons”, Gothic akizi, Old English æx, æsc „ax”, as well as Proto-Afroasiatic *ħak-/*ħək- „to cut into”; cf. Semitic *χak’- > Arabic hakk „crevice in the ground”, Hebrew kākkak „to cut in, to engrave”. Another example is the verb a hurui „to destroy, to demolish”, used mostly in Transylvania, which derives from Proto-Nostratic *ħul-/*ħol- (Bomhard& Kerns, 412) having cognates in Indo-European and Dravidian languages; cf. Hittite hu-ul-la-a-i „to smite, to destroy”, Greek όλυμι „to destroy, to make an end of”, Latin ab-oleō „to destroy”, Tamil ula „to become diminished, to be wasted, to die, to terminate”, ulakkai „end, ruin, death”, Malayalam ulayuka „to be impoverished, ruined” and Sumerian hul „to destroy”. In the case of this last example, from all derivatives in the daughter languages only Sumerian, Hittite and Romanian kept the Proto-Nostratic laryngeal */ħ/ as h.
To sum up, the data from Romanian shows that Proto-Indo-European had the aspirated voiceless velar stop *kh, as well as the laryngeal */h/, both inherited from the Proto-Nostratic. In Romanian, these two Indo-European sounds collapsed into the voiceless laryngeal /h/. In most Indo-European languages the laryngeals have disappeared and the voiceless aspirated velar */kh/ turned into its non-aspirated counterpart. In other words, it is difficult to reconstruct PIE *kh, taking into consideration most of Indo-European languages. Instead, the data from the Romanian language do not leave room for such ambiguities, regarding the presence in Proto-Indo-European, of both: the voiceless aspirated velar */kh/ and at least of one laryngeal */h/. Therefore, one may say that evidence from Romanian shows that post-Anatolian Indo-European had at least one laryngeal which disappeared from most, but not all, Indo-European dialects. On the other hand, it is also obvious that the aspirated voiceless velar stop */kh/ was also present in the Indo-European, which turned into a laryngeal, perhaps in early Thraco-Dacian and inherited by Romanian as such. Although, until now, the scarcity of data led to controversial positions in Indo-European studies, these two phonological aspects found in Romanian represent a turning point in understanding of these two sounds of Indo-European.
