Innateness of grammar

On Language Log, I came across Universal Grammar haters. For some, the debate rages on about nurture versus nature and I, like this blogger, also think the debate is inane.

My personal slant draws from the field of artificial intelligence where it's an already-firm conclusion that "grammars" aren't just abstract concepts pertaining to linguistics only. In a broader sense "grammar" is structure; it determines the sequential order of computation and clarifies the constituents of a functioning dynamic. So grammar also has relevance in other geeky subjects like computer programming, mathematics, systems theory and digital circuitry. Even our very DNA must have the innate capacity to understand a grammar because, without it, it would surely be hit-or-miss whether a string of gene sequences were executed in a timely order. Grammar is order. Neither you nor I would exist otherwise, let alone our complex brains.

Grammar is, on the most fundamental level, part of any coherent system and necessary to maintain that coherency. So, forgetting about whatever political vendettas one personally has against Chomsky et alia, most emphatically there must be some sort of basic "universal grammar" from which specialized grammars for particular languages are drawn. However this universal grammar must not be misunderstood to be only linguistic in nature but part of the fundamental processes of thought and computation. I believe that one innate capacity of the human brain, bland as it may sound, would be the ability to distinguish an object from an action performed on it.

A brain born without any inherent grammar must surely be a lump of thoughtless meat, fundamentally incapable of computation itself. Thought requires grammar. It's a foregone conclusion that an idea is nothing more than a structure of links to other ideas in an infinite sea we call knowledge. The neuronal structure of knowledge requires an innate grammar to parse it before one can comprehend it. To deny that there is a universal grammar in this sense is to say that the very act of thinking can be learned out of thin air. It can't, no more than you can teach a block of ice to think.

Hattic grammar and Proto-Aegean

I'm currently data-mining an excellent article about a very obscure subject, that of Hattic grammar. The article is written by Petra Goedegebuure who gave it a rather verbose title: Central Anatolian languages and language communities in the colony period: A Luwian-Hattian symbiosis and the independent Hittites (2008). It's refreshing that the author has a mature grasp of the subtleties regarding cultural identity and language. Sometimes language shifts while the culture stays largely the same; sometimes culture may alter radically with no large changes to language. A question she explores is: Can certain peculiarities of the Hattic language hint at the specifics of complex, unrecorded shifts in language and culture/cultural identity between the Hattians and the Indo-European speaking population in early Anatolia?

She gives a wealth of thorough examples showing Hattic grammar in action and my eyes have been opened. More frivolously, I believe I can now partially conjugate a Hattic verb with a modest degree of confidence: fa-nifas 'I sit', u-nifas 'you sit', an-nifas 'he/she sits', ai-nifas 'we sit' and nifas '(they) sit'. There are a camp of linguists who believe that Hattic belongs with the Abkhaz-Adyghe languages^[1] that are currently restricted to the northern regions of the Caucasus mountains and I think this most likely.


A Proto-Cyprian connection?

While a few kooks carry on dreaming that Etruscan is actually related to Hattic^[2], the two languages couldn't be any more alien to each other. Hattic is a prefixing language and exhibits an underlying VSO morphology (ie. verb-subject-object, as in Semitic and Egyptian languages) while Etruscan strictly uses suffixes. If there were any prefixes in Etruscan, we can expect them to be very rare, as is in fact typical of any SOV language (compare with other SOV languages like Inuktitut, Japanese and Turkish, for example). We can be certain then that the Cyprian languages, like Etruscan and Eteo-Cypriot, represented an entirely separate language group to Hattic.

Yet, there's still the potential that some traces of Hattian influence lurk in Etruscan through lexical and structural borrowings. Comparing the locations of Hattic (central Anatolia) and of Proto-Cyprian (western Anatolia & Cyprus) alone warrant the thought. And if not with Proto-Cyprian, could there have been an interaction with the older Proto-Aegean stage in the 3rd millennium BCE from which Minoan too would derive? This is why I've been feverishly recording Hattic vocabulary into my computer. Cross-correlation is delicious.

Eyeballing Minoanists and the value of A-SA-SA-RA-ME

Minoanist Leonard Palmer once wrote, “[...] for in questions of genetic relationship the linguist rightly attaches small importance to common elements of vocabulary. Decisive are resemblances of morphological procedures, for these are less readily borrowed.”^[1]

Eyeballing is for the bored and desperate

Rational people deal in facts and probabilities, not in mere possibilities and assumptions. If one's translations depend fundamentally on subjective similarities of words between languages, then one's contribution to the field of paleolinguistics is as useful as toxic sludge.

The eyeballing method, if it can be dignified as a 'method' at all, relies primarily on individual perceptions about phonetic similarity and difference. Since subjectivity is in the eye of the beholder, different individuals will reach different conclusions about the same data according to this strategy, making it disordered and unhelpful. It's best to become familiar with this shameless tactic so as to shun it whenever it surfaces in someone's work.

Internal analysis first

A widely applicable statistic known as Zipf's Law adds to what we should already be able to deduce by experienced linguistic intuition. In any given language, the most frequent words tend to be the shortest. Conversely, less frequent words tend to be longer. Afterall, how often do words like osteochondrodysplasia or spatiotemporal pop up in everyday conversation? This relates to the efficiency of information exchange.

The sheer length of Minoan A-SA-SA-RA-ME and its derivatives also hints that it most likely belongs to a core word class, such as a noun or verb. We can see that a five-syllable term is above the average length of attested Minoan vocabulary which implies that, with all things being equal, its typical frequency in random text or conversation should have been comparatively infrequent. Yet to the contrary, its unexpected ubiquity among libation table inscriptions demonstrates in itself a strong link between its genuine semantic value and its evident religious context. If one agrees that the analyzable suffix -NA belongs specifically to nominal morphology, then since our term is found with this very suffix (KN Za 10: YA-SA-SA-RA-MA-NA), it must likewise be nominal. Hence we can make an informed guess that it's likeliest a name or common noun relevant to Minoan religion.

These rational considerations alone then compel us towards the optimal conclusion that the term conveys the name or title of a deity to which these libations must have been dedicated. Palmer's original comparison to a title for a neighbouring Anatolian goddess (or goddesses) merely supplements the strong deductive foundation of this avenue of investigation. So this identification with the divine can't be so easily downplayed as the kind of immature eyeballing method that sensible linguists work hard to avoid. This view has something richer going for it.

More on the Rhaetic inscription

As per the promise of a previous post, I'll now share my own analyses of the above inscription which I transcribe in more familiar Roman letters below (nb. σ represents the M-like letter, san).

[...]
upiku : tauke
kleimunteis
avaσuerasi : ihi

Following Schumacher, Rex Wallace quotes a translation that makes me cringe a little: "Upiku dedicated (this object) to Kleimunte on behalf of Arvashuera". I suppose I find myself cringing because I know all too well how easy it is to simply assume that hard-to-analyse words are names when they're not. And here, there are far too many quirky names involved that hang in thin air and without any clear historical connection. To be fair though, the translation probably approaches the essential meaning intended by the Rhaetic dedicant.

It's presumed that upiku is a name yet others consider it a noun referring to a religious offering or perhaps a participle in -u meaning 'dedicated' in other contexts. Wallace then quickly classifies kleimunteis as some unknown genitive noun even though -is is normally a marker of the directive case. After correcting Wallace's previously mentioned transcription error, he divides avaσuerasi into avaσue-ra-si interpreting it as a "plural pertinentive". He offers little comment on the significance of the trailing word ihi.


And now for something different

I take the first legible word, upiku, to be both a participle 'offered' and a noun meaning '(that which is) offered; an offering'. I reject Wallace's facile suggestion via Schumacher that this is a personal name. The next word, tauke, is transparently a perfective preterite verb containing the stem tau-. This stem reminds me of a presentive form tva 'shows, demonstrates' (< *taw-a) in the Etruscan inscription TLE 399, written on the back of a mirror with the depiction of Herakles suckling the breast of his mother Uni (the Etruscan equivalent of Greek Hera). So the first two words can immediately be read as '[someone] has presented (tauke) an offering (upiku)'.

In grappling with the next problematic item, kleimunteis, Wallace may have overlooked a possible preposed locative demonstrative klei followed by a noun marked with two case suffixes as per the rules of Suffixaufnahme. This way, it might be read klei mun=te-is 'to [that] in this plot' (cf. Etruscan cle 'at this', muni 'plot', -θi 'in', and -is 'to, towards').

The next line starts with avaσuerasi which is certainly an animate plural in -er (< -ar) with case marker -asi 'for, on behalf of'. This implies an unmarked noun *avaσu which strongly looks to me like a transitive particle in -u of a stative verb in -as- built on a basic verbal root *au. In fact, I've been predicting this verb's existence for a while based on Etruscan avil 'year', seemingly derived from this same verb with common noun formative -il. If *au meant 'to pass on' or 'to depart', avil would literally mean 'that which passes', hence a period of time or 'year', while avaσuerasi would be a reference to 'the departed ones', ie. the ancestral dead. After this, ihi is a cinch to crack when set beside Etruscan ei 'here, there', a locative particle with general dexis. Unfortunately this word is all too often mistranslated as a negative particle leading to much confusion.

When I put all this together, the following translation emerges from the fragmented inscription: "[Someone] has presented an offering to that in the plot on behalf of the departed here."

Translating the Liber linteus religious formula

After Boxing Day, I came across the Wikipedia entry for Liber Linteus. Casey Goranson had in zeal attempted to translate this artifact's repeated religious formula with the use of my dictionary applet last month. Flattered though I am, his translation needed to be revised. "For this soul to endure/remain in the town of night and amidst the people" would be something quite different in Etruscan than what we find attested, perhaps something like *Ca śacni eneri śpureθi cilθl meθlumeθic.

The confusion is expected however since I don't believe I officially gave a clear translation after grappling with its interesting declensional variations back in 2007 in Liber Linteus and religious formulae (read Part 1, Part 2 and Part 3). Published Etruscanists like the Bonfantes have only ever given piecemeal and vague translations.^[1] So it's high time I spilled the beans on this by 2011!


Introduction to case marking variants in the formula

As I previously reckoned, variations in case must have revolved around a same basic jist. In the following the prevailing case markings are respectively locative -e (signifying 'at, with') plus postposition -ri 'for (the benefit/purpose of)', directive -is (signifying 'to, towards') plus postposition -tra 'through' and finally genitive (marking possessor and recipient) without an accompanying postposition:

Śacni-cle-ri cilθl, śpure-ri, meθlume-ri-c enaś.
"For the spirit of night, for (the) city and for (its) people everlasting."

Śacni-cś-treś cilθś, spureś-treś enaś.
"To the night-spirit (and) to (the) city everlasting."

Śacni-cla cilθl, śpural, meθlumeś-c enaś.
"For the spirit of night, for (the) city, and to (its) people everlasting."


Breaking down the grammar

A test of any translation is the theorist's ability to break the whole into its parts as lucidly and thoroughly as possible.

We can see that śacni 'soul, spirit' (literally 'sacred one') is declined only indirectly through the use of postposed deictics with their own case markings (-cle, -cś, -cla) which are to be treated as weakened demonstratives meaning 'the' rather than the more emphatic 'this'. In the variant with directive case, the underlying noun phrase without case marking would be *śacni-ca cilθ, equivalent to an English compound 'the night-spirit', as opposed to the other variants building on *śacni-ca cilθl 'the spirit of (the) night'. These creative variants don't change the meaning but have an impact on grammar. When expressing a compound 'night-spirit' instead of 'spirit of the night', the two consecutive nouns were both declined with the same directive case deliberately, via Suffixaufnahme, since the two nouns in succession were referring to a single idea. This is why first and last names in Etruscan also occasionally display Suffixaufnahme because, again, the two names point to a single person.

The postposition -tra is a transparent, early borrowing from Umbrian tra comparable to other inherited Italic words for 'through', 'over' and 'across' (cf. Latin trans). In this case, this particle in partnership with the directive marking can only be alluding to the transfer of gifts to the recipient and thereby has a terminative nuance indicating that which the action affects. In this case, the action is giving and the listed recipients provide the point where the action ends, so to speak.

The final word enaś seems to be an infinitive verb related to the present-future perfective eniaca of the Pyrgi Tablets, implying a basic root *en-. In both contexts, and through the help of the accompanying Punic translation of those tablets, the word must involve endurance (ie. 'may it outlast the stars'). The termination in -aś seems to mark a state. So a value of 'everlasting' is contextually and grammatically reasonable here. It follows the noun phrase it modifies just like an Etruscan adjective should.


Breaking down the larger religious meaning

Another test of a translation is the greater meaning and context it can provide.

One might debate whether this "night-spirit" is to be understood in this context to refer to a plurality of night spirits in the collective sense (ie. Roman manes) or whether it's best to treat the singular form as it is, a singular deceased person who had recently passed on. Is the reference generic or specific? Either way, the choice of the word śacni establishes a subject of death and resurrection since 'human soul' or 'human spirit' is its strict semantic usage in other documents from what I've seen.

Building on this theme of an immortal soul, and since the "city" here is never overtly specified in the mantra, it probably refers to the City of the Underworld. The "people" then are its non-living citizens held captive by the grip of death for eternity. In fact, LL 11 seems to confirm this interpretation by a pretty clear reference to just such a city of the dead: spur-ta eisna hinθu 'the holy city below'^[2].

The document is already agreed to pertain to the adherence of important rites on specific calendar dates, and so my above translation fits well with the facts. There's a focus in this text on funerary rite, it appears.


NOTES
^[1] According to Bonfante/Bonfante, The Etruscan language: An introduction (2002), 2nd ed., p.93 (see link), sacni means 'priest', a contextual impossibility persisting no doubt by an outdated connection to Latin sacerdos 'priest'. Priests themselves are certainly not to whom offerings are being devoted in these passages. As per Jannot, Religion in Etruria (2005), p.128 (see link), sacnicstra is "a collective term designating men devoted to a god". This mistake together with his other mistake confusing caθesan as a single name and variant of goddess name Caθa (see p.158) shows that Jannot, despite publishing on Etruscology since the 1970s remains persistently naive about basic Etruscan demonstratives. Meanwhile in Pittau, La lingua etrusca (1997), pp.106, 212 & 217 (see link), the author repeats several times that sacni means 'rite'. All of these translations are random and have only blockaded more conscientious attempts at coherently translating full sentences in the Liber Linteus.
^[2] Van der Meer, Liber linteus zagrabiensis (2007), p.149 (see link) lists out verses 11.ix-x as: slapiχun slapinaś. favin. ufli. spurta. eisna. hinθu cla. θesns.

By whom, for whom

According to Bakkum, The Latin dialect of the Ager Faliscus: 150 years of scholarship, vol 2 (2009), p.305 (see link):

"The Etruscan forms in -si and -(a)le can be used to designate both by whom and for whom the object was made (Steinbauer 1999:174-6)."

In case the absurdity isn't noticeable, "by whom" and "for whom" are opposite roles in these inscriptions. How did this nonsensical statement make its way to modernday print? What happened to the concept of structured, coherent grammar? In his solitary example of ET Fa 3.4, a vessel from Vignanello, the solitary name Vultasi doesn't help much to decide the matter. I can't help but appreciate too how, by using an unclear inscription, the author can quickly pass a shaky statement over the reader without being too obvious.


Luckily for us, TLE 651 does decide the matter. As pictured above, this is a statue of a standing nobleman nicknamed The Haranguer. The only names present in the entire inscription at the base of his toga are in the first line and they're declined in this same case: Auleśi Meteliś Ve. Vesial clenśi. Who would think that the image of this mystery man can be anything other than Aule Meteli himself? The inscription must read "For Aule Meteli, for the son of Vel and Vesi." If we interpret this case ending as by whom the statue is made, the inscription fails to explain who this man is while giving us useless information about the statue's creator. This data is hardly as important to an average, literate Etruscan as the man for whom all this metalworking effort was devoted!

Then too, we also find tinśi tiurim avilś χiś repeated several times in the Liber Linteus and since tin, tiur and avil are certain to mean 'day', 'month' and 'year' in these passages, forcing an agentive sense on an inanimate noun like tinśi is absurd. It must therefore mean 'for the day' in the context of the rituals to be performed on specified dates. An Etruscan agentive case has been concocted from nothing.

Bakkum goes on to write something that I'd gauge to be naive wording for a linguist:

"[...] in most other cases, the use of the -si or -(a)le form is due to a verb, usually mulu-."

To observe that this case ending occurs often with the verb mulu- is justified, but to say that the use of this case marker is "due to" a verb, combined with his odd reference to Steinbauer above, seems to disqualify his expertise on structured grammar itself, let alone on Etruscan grammar. Need it be said, the use of this case ending or any case ending isn't incumbent on the verb itself per se but demanded by the overall semantics of what one is expressing. If one doesn't comprehend why -si/-le is used with some verbs like mulu- while not for other verbs, one fails to comprehend the very meaning of these individual grammatical units. The verbs haven't "caused" these endings to occur any more than the Oracle of Delphi.

The correct answer is simple: -si/-le must be consistently translated as 'for', as in 'for the purpose of' or 'on behalf of', never ever 'by' (whether in a locative or agentive sense). This value is evidenced not just in several Etruscan inscriptions but in Lemnian ones too.

Just hang loose, blood

Do you think you know your own language? Sure? Are you also sure about the fine lines between language, dialect and accent? Have you ever pondered on how subcultures shape language? Am I the only geek here? Is this mike on?? Well, whatever. Just watch this comedy skit below, a scene from the 1979 movie Airplane!.


Sometimes comedy keeps it real. Language is an amazingly flexible form of communication and, yes, it's always funny to watch unsuspecting characters transcend behavioural norms and poke fun at society. In the following two clips, Barbara reminisces in a later interview while both the creators and the two jive-talkers (Al White and Norman Alexander Gibbs) explain the collaboration involved in making this classic scene happen.

A little note on Etruscan adjectives and case agreement

A helpful and interesting post by Michael Weiss is to be found at OHCGL Addenda and Corrigenda on the Tyrrheno-Italic name Numasio-. Now for the bad news. The following is more of a tiny quibbling on Etruscan grammar as I see it rather than a major fault but I must question the Italian translation given to the Etruscan inscription found at the bottom of his post which is printed thus:

mi mlac mlakas larθus elaivana araθia numasianas
'Io (sono) il buon/bel (vaso) oleario di Arath Numasiana per il buon Larthu.'
[English: 'I (am) the good (vessel) of oil of Arath Numasiana for the good Larthu.']

I hate to be a stickler for grammar (no wait... I love it!) but for anyone actively studying Etruscan inscriptions, seeing someone miss the common phrase mlac mlakas is glaring. This phrase is repeated in a few other Etruscan inscriptions (eg. ET Cr 2.33, 2.36) and has been compared already to nearly identical formulae in Faliscan (duenom duenas) and in Greek (καλος καλο) as published by Agostiniani (SE 49:1981). This nugget of fact easily shows that overlooking this formula while parsing a sequence mlakas Larθus out of this is nonsense. Such a sequence doesn't even mean 'for the good Larth(u)' but rather *'of the blessed of Larth'.

I object doubly against this translation because I know that Etruscan adjectives (putting aside numerals and such) never precede the noun, nor are they marked with case at all. In Etruscan, there simply is no case agreement between noun and trailing adjective like in neighbouring, unrelated Latin. 'For the good Larth' would be more competently translated into Etruscan as either *Larθus mlac (genitive of giving) or *Larθe-ri mlac (locative with postposition -ri 'for').

Reexamining the inscription above, it's more likely to be translated as follows:

Mi, mlac mlakas, Larθus elaivana Araθia Numasianas.
"I, blessed of the blessed, (am) Larth's oil-vessel with Arath Numasiana."

Notice that now we see that Arath Numasiana is the one in possession of the vessel (in the commitative case ending -a) whereas Larth, whose last name is unspecified, is merely the donor (in the genitive, denoting origin).


UPDATE
(03 September 2010) After just posting this, a clearer translation had come to me by simply moving the comma over, all still in keeping with my grammatical analysis above:

Mi, mlac mlakas Larθus, elaivana Araθia Numasianas.
"I, blessed of the blessed from Larth, (am an) oil-vessel with Arath Numasiana."

Expanding the Etrusco-Lemnian primer

I've expanded my previously announced pdf to include not only my basic sketch of the Etrusco-Lemnian nominal declension but now also the pronouns and demonstratives as well. Sadly, no one has a clue as to what the plural pronouns were like and, as far as anyone knows, they aren't attested, so I'm restricted to the singular pronominal paradigm. Let's hope a happy-go-lucky farmer trips over a new Etruscan artifact with plural pronouns on it some day.

Anyways, lest I ramble on again, I've renamed my pdf Etrusco-Lemnian Declension and it's to be found in the Lingua Files section as always.

Etruscan syntactic inversion

Etruscan word order can be confusing. Despite being clear overall that this language is of the accusative-type with a default SOV word order, I've noticed many times through the course of my studies that the verb sometimes likes to stray to the middle or even to the beginning of the sentence or clause just to confuse me. I haven't read anyone getting into detail about Etruscan word order and it sometimes feels like I'm the only one seriously studying this amazing language. I find no decent answers anywhere, to my frustration, so I guess we have to get our hands dirty and do this ourselves if we want to get it pieced together at all.

The only answer for this I can come up with is that this is some kind of "inversion" like that which occurs in languages like German and Dutch^[1]. If so, there must be a "trigger" that causes the verb to push to the beginning before the subject and object as it does in two relative clauses of the Laris Pulena inscription (TLE 131). I've parsed the relevant sentence with some helpful modern punctuation as follows:

Χi-m, culsl leprnal,
pśl varχ-ti cerine pul alumnaθ, pul hermu huzrna-tre;
pśl tenine eprθnev-c meθlum-t pul hermu.

The two instances of pśl, I believe, are referring back to the genitive phrase culsl leparnal. In each of these relative clauses, we find the verb at the beginning, ahead of any unmarked nomino-accusative nouns. In the first clause, the preterite verb cerine precedes what may be two nouns, pul and alumnaθ. (I've abandoned the idea that pul is a relative pronoun declined in the type-II genitive, which should probably have been written *pl if it were so, because it causes too many structural and semantic difficulties in this passage.)

In the Cippus Perusinus, the phrase ipa ama hen 'that which is forth' might also suggest this same sort of "verb-forwarding". Afterall, if we understand hen to be an adverb, we might expect *ipa hen ama in a more well-behaved SOV language^[2]. Again, we see verb-forwarding in the presence of the relative pronoun.

However, if the relative pronoun triggers word-order inversion, it doesn't appear to be a consistent rule since in TLE 27 we read:

[...] in-pein mler usi ateri.

Here, the verb ateri follows the object mler in a clause introduced by in-pein 'where, at which'.


NOTES
^[1] Odlin, Language transfer: Cross-linguistic influence in language learning (1989), p.94 (see link): "Studies of Dutch and German offer particularly intriguing examples of where word-order transfer can lead to different acquisition patterns. Both languages employ SOV in subordinate clauses and SVO in main clauses, although other main-clause word orders are possible under special circumstances."
^[2] Bomhard/Kerns, The Nostratic macrofamily: A study in distant linguistic relationship (1994), p.161 (see link): "Thus, in a consistent SOV language, an attributive adjective or a genitive precedes its 'head' noun, an adverb precedes its adjective or verb, a noun precedes its case ending or postposition, [...]"

Relative pronouns in Etruscan

A small commentbox coalition developed recently against my Etruscan translation concerning the Cippus Perusinus such that ipa in ipa ama hen agrees in case with its antecedent, tezan 'cippus'. I remain unmoved. Rather than honest criticism, the claims made were boldly exploiting unreferenced half-truths and exaggerating the importance of minutiae while, as always, cloaked in complete anonymity. A bullshitter exposes himself when he approaches the grammar of one language by sole appeal to another unrelated one. Etruscan isn't Latin; apples and oranges. I appreciate the comical absurdity of the attempt though.

Back to linguistics, it's beneficial to explain why their "issues", or what I like to call, "turds", are inane so that no sensible reader could be confused by my judgement calls. The first turd was their imaginary "language universal" such that relative pronouns are always declined according to their role in the subordinate clause and never by their role in the antecedent. The second turd was the general ignorance they had regarding Etruscan's relative pronoun ipa (and about the language as a whole, for that matter).

Concerning these imaginary universal case-agreement rules

Speaking globally, the choice of a relative pronoun's case is *not* necessarily bound to the subordinate clause alone, despite the persistent shouting from this normally silent group of persons. For example, on the grammar of Old English, Gotti/Dossena/Dury, English Historical Linguistics 2006: Syntax and morphology, v.1 (2008), p.11 explains that a relative pronoun may share with its antecedent "features for number and gender, and, optionally, for case". An example is:

Ic wat witodlice ðæt ge secað ðone hælend ðone ðe on rode ahangen wæs.
"I know truly that you seek the Lord (ACC.), who (ACC.) was hung on the cross."

If a relative pronoun was only about its relative clause, we should expect "who" to be declined in the nominative since it's the patientive subject of the participle formation, "was hung". Classical Greek also throws a curve now and then,^[1] and same too for Arabic where the dual relative pronoun agrees with its antecedent in gender, number and *case*.^[2]

Regarding Ancient Ugaritic, Roger Woodard publishes in The Ancient Languages of Syria-Palestine and Arabia (2008), p.31: "The relative pronoun agrees in gender and number with its antecedent; whether the case of the relative pronoun itself is decided by the case of the antecedent or by the function of the relative pronoun in its clause cannot be determined [...]" Even Woodard is unaware of this alleged "language universal" that I'm being harassed with.

There's no need to elaborate further. It's official: my commentbox hecklers are bored latchkey kids.


Etruscan relative pronouns (for real, this time)

If we're going to talk about Etruscan relative pronouns, let's talk about Etruscan relative pronouns. Not Latin ones, Greek ones or Esperanto ones. In Etruscan, there's no question that the relative pronoun is declined for case as are all known pronouns and demonstratives. In the Cippus Perusinus, ipa is certainly in the nominative case (matching corresponding nominatives ita 'that' and ica 'this') but the question is whether this pronoun's declined according to its role in the relative clause or its antecedent, tezan, which I give the value of 'cippus':

Sleleθ caru tezan fuśleri tesnś teiś Raśneś ipa ama hen.

I admit that this is tricky to assert based only on this (despite the fact that my translation is still grammatically valid and contextually sound) but if one is so certain that Etruscan relative pronouns somehow must behave like Latin ones, then I defy such narrow-minded armchair linguists to explain the following on Laris Pulena's sarcophagus (TLE 131):

Χim culsl leprnal pśl varχti cerine pul alumnaθ pul hermu huzrna-tre

Things aren't so simple. First off, we may wager that pśl is an unstressed type of pronoun because it's spelled without vowels just as postclitic demonstratives are (eg. cl 'of this', tś 'to that'). Second, this pronoun appears to be doubly marked which is normal for Etruscan and observed many times elsewhere. In fact, in a language like Etruscan proven to operate under Suffixaufnahme, in what way can we meaningfully avoid interpreting this sentence as a genitive case agreement between three consecutive elements that I've boldfaced in the above phrase (ie. culs-l leprna-l pś-l)? And if we can see this, then we can see that the role of the antecedent in Etruscan relative pronouns might actually be important.

For now, I've never ever seen a decent and complete translation of this artifact so we're in uncharted territory. However, here are some hints I can most securely offer: χi-m 'and next' and culsl 'of the gate'. Have fun pondering on that. As always, everyone is free to have opinions but may yours be only productive ones.


NOTES
^[1] Croy, A primer of Biblical Greek (1999), p.164 (see link) gives the following example:

ἀκούομεν τῶν λόγων ὤν ἡ θυγάτηρ σου λέγει
"We hear the words (GEN.PL.) which (GEN.PL.) your daughter speaks."

^[2] Ryding, A reference grammar of Modern Standard Arabic (2005), p.323 (see link):

li-l-zawj-ayni llað-ayni ya-ntazˤir-āni ħadaθ-an saʕīd-an
"for the couple (OBL.DL.) who (OBL.DL.) are awaiting a happy event"