Lund Language Diversity Forum Blog

by Gerd Carling

Pronouns are fun. They are important, too. Think for a second about what language would look like without pronouns. Every time you use it, he, she, someone, everyone, you would have been forced to repeat the nouns or names of people you refer to: “Where is Gerd’s handbag? Gerd’s handbag has disappeared! Can … (naming everyone name in the house) tell Gerd where Gerd’s handbag is. … (naming everyone in the house) must have taken Gerd’s handbag…” (You understand the problem now, I guess).

All languages have pronouns. As is evident from the hypothetical example above, they replace nouns in therefore fill an important function in speech. Pronouns build systems and pronominal systems worldwide can be considerably diverse and highly complex. We may think of them as reflexes of patterns of our culture, mirroring important functions of our social lives and the ways in which we organize our societies. However, to what extent do they really reflect our culture, social lives, and our communication habits? Is it possible that we just inherit our pronominal systems over generations and cannot do anything to change them, how much we try? It is an interesting question.

The most important characteristics of pronominal systems worldwide is the distinction of person. The second most important distinction is number. This means that most systems have forms for I, you, he/she/it, we, you (pl.), and they. These six basic forms can be combined into many different variants, including

• I+I (two persons saying something at the same time)
• I+you (I and you included)
• I+s/he (I and a third person included)
• you+you (two second persons included)
• you+s/he (you and a third person included)
• s/he+s/he (two third persons)

Even with these, combinations, some systems are very minimalistic, whereas others build up a rich system with many distinctions.

Pronoun system of Sanuma (Yanomami, Brazil/Venezuela):

One way to build a richer system is to increase the numbers (singular and plural), for instance with a dual (including two people, using any of the combinations above), a paral (two people forming a natural pair, such as spouse or twins), or a paucal (a few people).

Another frequent distinction is gender. Gender is normally marked on the third person (it is very unusual to mark gender on the first person, but not on the third). Gender may include a distinction between masculine/feminine, human/non-human, or animate/inanimate. Around 20% of the world’s languages make this distinction in their pronominal system, including languages of the Indo-European family.

Pronoun system of Tocharian A and B (Indo-European, extinct, Central Asia):

Many of the world’s languages distinguish inclusiveness in their system. This means that they use different forms for ‘we’ (dual or plural) if the hearer is included or not.

Pronouns system of Motuna (Buin, Papua New Guinea):

Some languages, again, mark the social status or character of the person in their system. Here, the most well known language is probably Japanese, which makes different distinctions depending on social structure. These forms have come to play an important role in virtual forms of the language, such as in computer games and anime.

Japanese forms for ‘I’ in the phrase “I am a …” in virtual language:

Pronominal systems are very stable over time. They belong to the parts of language that are used most frequently and are therefore highly conservative and very reluctant to borrowing. However, socio-cultural changes and changes in attitude may have an impact on pronominal systems. The complex Japanese system of honorific pronouns has decreased with changing social circumstances. Very few Japanese speakers (outside the virtual world) use the princess or samurai forms of the pronoun ‘I’. Swedish, on the other hand, has increased its pronominal system with a gender-neutral pronoun hen (from Finnish gender-neutral hän), introducing a very complex distinction for third person: han (he), hon (she), hen (gender-neutral, human), den (it, common), det (it, neuter). Hen is relatively commonly used (since 2015 it is listed in the Dictionary of the Swedish Academy), but the future will show whether it will remain in the spoken language.

References

Dixon, Robert M. W. (2010), Basic linguistic theory. Vol. 2, Grammatical topics (Oxford: Oxford University Press).

Johansson, Victoria, Carling, Gerd, and Holmer, Arthur (2013), Språket, människan och världen : människans språk 1-2 (Lund: Studentlitteratur).

Krause, Wolfgang and Thomas, Werner (1960), Tocharisches Elementarbuch. B. 1, Grammatik (Heidelberg).

by Arthur Holmer

Ask a speaker of English what the most salient fact about English is, and it is very unlikely that the answer will be “do-support”. Similarly, we would be surprised if we heard a Gael bragging about initial mutation, or a Finn about consonant gradation, or indeed a German about verb-second word order. These phenomena, insofar as speakers are in any way consciously aware of them, are not considered to be anything special, except perhaps that they are possible sources of difficulty for anyone learning the language.

Not so for Basque ergativity. Ask any speaker of Basque for an important fact about Basque, and they will (in addition to the obvious fact that Basque is a language isolate, i.e. that it has no known relatives) in all probability mention ergativity, that is to say the fact that the subject of a transitive verb like eat gets a special type of case-marking (the -k suffix in the following example), whereas subjects of intransitive verbs like come and go do not. The subject of a verb like come or go behaves like an object, and gets no case-ending.

Sagua    etorri     zen.
mouse    come       AUX
‘The mouse arrived.’

Katua-k   sagua    jan      zuen.
cat-ERG   mouse    eat      AUX
‘The cat ate the mouse.’

Ergativity is often described in terms of three separate roles, usually defined by the abbreviations A, S and O:

A         (=”agent”) subject of transitive verbs like hit or see

S          subject of intranstive verbs like come or go

O         objects of transitive verbs like hit or see

In a language like English, it seems so obvious that A and S behave identically, that we don’t even think of these are separate roles. For us, they are simply the “subject”. For a speaker of an ergative language like Basque, A is the odd one out, the one which gets ergative case, while S and O behave identically.

Why speakers of Basque are consciously aware of ergativity is presumably because they notice how Spanish or French are different from Basque, and almost certainly also because language is such an integral part of Basque identity.

Unfortunately, however, the claims of Basque ergativity are a bit exaggerated. Basque is not nearly as ergative as the Basques themselves would like to believe. Basque (at least the standard language, based on western dialects) is more aptly described as an “active” language, which although it does lend a flavour of dynamic youthfulness, is nowhere near as cool as being an ergative language.

Being an active language means that the subjects of some intransitive verbs (such as dance and sing) end up in ergative, and the subjects of other intransitive verbs (such as come and go) do not.

Haurra-k      abestu     zuen.                 Haurra      joan      zen.
child-ERG     sing       AUX                   child       go        AUX
‘The child sang.’                             ‘The child left.’

Nevertheless, in a simplified classification (shared by Basque speakers and most linguists alike), Basque is an ergative language, with no known relatives.

Ask a Georgian (of Tbilisi, not of Atlanta!) if they know anything about Basque, and surprisingly enough, they probably will, they will probably also tell you that it is not an isolate, since it is in fact related to Georgian! And one of the pieces of evidence you may be given is that both languages share the feature of ergativity.

Now, Georgian is probably an even poorer candidate than Basque for pure ergativity. Not only is Georgian active, like Basque, but it also has a mixed system (called “split ergativity”), whereby it behaves like Basque in the past tense, but like English in the present tense. Even this is in fact a gross oversimplification, as the Georgian system is even more complex, and actually also much more exciting than simple ergativity. But it does raise the question of whether pure ergativity does exist, and if so, where?

What is then this elusive pure ergativity? It would imply that S and O always behave identically, in every respect, in the same sense that S and A always behave identically in English, or Swedish, or Spanish.

Does it indeed matter if pure ergativity exists? Perhaps not. But it is sometimes pointed out that language and thought may be interconnected. If this is the case, it becomes an interesting (if somewhat speculative) question whether or not a purely ergative system might not in some way be reflected in the world view of the speakers. If a transitive action always treats the patient/object as the least marked and default participant of the action, in other words, if the language consistently views an action from the point of view of the patient/object, rather than of the agent, could this somehow be mirrored in how the speakers view their world? This is an intriguing question, but it cannot reasonably be tested on either Basque or Georgian. Instead, access to a maximally ergative language would be required.

In ergative language after ergative language, from Burushaski to Chukchi to Greenlandic, we inevitably find some feature which groups A and S together. In most languages, reflexives (like myself) must refer to A or S. Some kinds of adverbs (like deliberately) must refer to A or S. In almost all ergative languages, an omitted subject in a clause must refer to A or S of the preceding clause, just like in English. The following Basque example is adapted from Ortiz de Urbina 1989.

Ama-k        semea     eskola-n    utzi,     eta     Ø      klasera    joan     zen.
mother-ERG   son       school-in   leave     and            class-to   go       AUX
‘Mother left her son in school and (she / he) went to class.’

However, in Dyirbal, a moribund language spoken by a handful of people south of Cairns in the state of Queensland, in Australia, even this feature groups S and O together.

Nguma      yabu-nggu     buran,    Ø      banaganyu.
father     mother-ERG    see              return
‘Mother saw father and (he (!) / she) returned.’ (Dixon 1994:155)

This phenomenon is called syntactic ergativity, meaning that even sentence structure treats S and O in the same way. This can be recognized easily in examples where two main clauses are juxtaposed, as well as in other more complex constructions.

Unfortunately, even Dyirbal is not a candidate for pure ergativity either, since pronouns still behave like they do in English (S and A have one case form, while O has another), cf. examples from Dixon (1972).

Ngaja       baningu.                     Nginda       baningu.
1SG (S)     arrive                       2SG (S)      arrive
‘I came here.’                           ‘You came here.’

Ngaja       ngin-una       buran.
1SG (A)     2SG-ACC (O)    see
‘I saw you.’

There seem to be a few languages in Queensland (among others Yidiñ, Kalkatungu, Bandjalang and Yalarnnga) which share the syntactic ergativity found in Dyirbal. Unfortunately, most of these also behave like Dyirbal in that pronouns group S and A together, as in English. In Kalkatungu, full pronouns do group S and O together against A, but there is a separate system of bound (short, unemphasized) pronouns that still behave like in Dyirbal.

It is only in one single of these languages, Yalarnnga, that purpose constructions show that the language is syntactically ergative (like Dyirbal), while the pronoun system, unlike Dyirbal, is also fully ergative (there are no bound pronouns that go the other way).

The ergative case-marking of the pronouns is clear from the following example, where the transitive verb hear has the subject ngathu, while the intransitive verb go has the subject ngiya.

Kuntu     nhawa      nga-thu      mangka-mu...  
not       2SG        1SG-ERG      hear-PAST
‘I didn't hear you...’ (Breen & Blake 2007:27)

Kuyirri     nhina-mu,    ngiya     ngani-mu.
boy         be-PAST      1SG       go-PAST
‘When I was a boy, I went.’ (Breen & Blake 2007:67)

To illustrate syntactic ergativity is a little bit more tricky. One way of testing it is the behaviour of purpose clauses (which express the purpose for which another action is performed). In a purpose clause, if the omitted element is to be understood as A, the verb must be specially marked with a marker –li– (“antipassive”) which serves to tell us that the omitted element is neither S nor O.

Ngani-mi     ngiya      [ Ø    manhi-wu      miya-li-ntjata ].
go-FUT       1SG               food-DAT      get-ANTIPAS-PURP
‘I’ll go and [Ø get food].’ (Breen & Blake 2007:59)

There are two contexts where the antipassive marker –li– does not occur. One is when the verb in the purpose clause is intransitive (like sit in the following example), so the omitted element must be S.

Ngiya     laa-kanu     ngana,        [ Ø     wayi-ngali-mpa     nhina-ntjata ].
1SG       now-again    go.NONFUT             that-PL-ALL        sit-PURP
‘I’m going now too, to be with those others.’ (Breen & Blake 2007:64)

If the verb is transitive, like wash, and the antipassive marker –li– does not occur, the omitted element is necessarily understood as referring to the object (O) of the action wash (in other words, the purpose clause can readily be translated as a passive).

Nga-thu   tjaa   ngapa-nha    ngani-ntjata   marnu-yantja-mpa   [ Ø   karri-ntjata ].
1SG-ERG   this   tell-PAST    go-PURP        mother-POSS-ALL          wash-PURP
‘I told him to go to his mother and get washed.’ (Breen & Blake 2007:60)

In Yalarnnga, purpose clauses single out the situation where A is omitted as the special case which must be marked on the verb. If the omitted referent is S or O, this special marking does not occur. As far as can be seen, Yalarnnga syntax treats S and O in one way, and A in another way, and there appears to be no construction which treats S and A alike.

This might not look all that striking, but this is what pure unadulterated ergativity looks like. There is not a lot of it around. In fact, Yalarnnga may be the only language in the world which consistently and exceptionlessly groups S with O to the exclusion of A, in every construction in the language. In other words, the exact mirror image of a language like English, Russian or Latin.

So if we want to investigate the properties of possibly the single currently attested purely ergative language in the world, the ultimate apex of ergativity, we should be prepared to travel to Dajarra, about 1000 km inland, more or less half-way between Cairns and Alice Springs.

Or we should have been. Sadly, the last native speaker of Yalarnnga died in 1978.

Postscript, caveat and a word of comfort

While Yalarnnga appears to be the only purely ergative language in Australia, at least of which we have enough data, (some degree of) syntactic ergativity has been reported at least in two other languages, Päri, spoken by over 20.000 people in South Sudan, and Nadëb, spoken by a few hundred people in the upper reaches of the Amazon, in Brazil. Whether or not these are purely and consistently ergative remains to be seen.

References

Breen, Gavan & Blake, Barry. 2007. The grammar of Yalarnnga. A language of western Queensland. Canberra: Pacific Linguistics.

Dixon, R. M. W. 1972. The Dyirbal language of North Queensland. Cambridge: Cambridge University Press.

Dixon, R. M. W. 1994. Ergativity. Cambridge: Cambridge University Press.

Ortiz de Urbina, Jon. 1989. Parameters in the grammar of Basque. A GB approach to Basque syntax. Dordrecht: Foris.

The art of wining and dining is intimately connected to the language of eating and drinking, which comes with a wide array of delicious lexicon and exotic yet tantalizing etymologies. The words we devour on a daily basis can sometimes have the strangest origins and provenances, so just by opening your fridge or cupboards you are greeted by products which names have travelled far in both space and time. Since we today, on the last day of April, celebrate the Swedish feast day of Valborg’s Eve or Walpurgis Night, it becomes natural to focus on the liquid substances as I would like to offer you a glass of wine-related wisdom.

Most of you might wonder which language the elfish word in the title comes from and some of you have hopefully guessed that it means ‘wine’. The language is the South Caucasian or Kartvelian language Georgian and it reads ‘ɣvino‘. The oldest evidence of wine-making in the Near East has been found in Georgia from around 6000 BCE (McGovern et al. 2017), and the exact origin of the word ‘wine’ is unclear. It is clear however that the English word ‘wine’ and its equivalents in most languages across Europe are either descended or borrowed from Latin vīnum. This is not surprising as the Romans were keen winers and their word for wine thus spread with viticulture and trade. The question is therefore what is the origin of vīnum?

The conventional explanation is that it ultimately comes from the Proto-Indo-European root *ueh₁-i- ‘to turn, bend’ (Beekes 2010), but Beekes also points out that since viticulture emerged in the southern Caucasus it might originally be a loanword from this region and thus is possibly of non-Indo-European origin. This is interesting as the Georgian word ɣvino is considered by many Kartvelologist such as Fähnrich to be a native Georgian word, as it can be reconstructed in Proto-Kartvelian and is connected to the Proto-Kartvelian root *ɣun- ‘to bend, wind’  (Fähnrich 2007). Beekes claims however that the Georgian word is borrowed from Armenian gini (Beekes 2010), as ɣvino is traditionally considered to be borrowed from Indo-European.

We will never know with certainty which of these etymologies is correct, but some etymologies are clearly more likely than others. There is also a tendency for etymologies of food terms to recapitulate the provenance of the concept it refers to, which means that the origin and historical spread of a food item often is reflected in its etymology. A typical example of this is the word ‘pilsner’, which refers to a type of beer originally from Plzeň in Bohemia in the Czech Republic. The German name of the city is Pilsen and Pilsner is an adjective meaning ‘from Plzeň’. English borrowed the word from German as the prestige language of Bohemia was German at the time it entered English. If the concept ‘pilsner’ would be borrowed today, it might have been called the Czech counterpart plzeňské instead!

Wine could therefore potentially be another example of this tendency, carrying the word *ɣvin- from its origins in prehistoric Georgia via the wine-loving cultures of the Mediterranean to the rest of Europe. I must however point out that since the time-depth is several thousand years it is impossible to know for sure, and the question has been discussed in lengths without any conclusive arguments. I will therefore conclude by asking once more: In ღვინო veritas?

If you want to know more about culture vocabulary and etymologies like the example above please visit the website of our project DiACL: Diachronic Atlas of Comparative Linguistics.

References

Beekes, Robert S. P. (ed.) (2010). Etymological Dictionary of Greek. Indo-European Etymological Dictionaries Online. Brill Online. Leiden: Brill.

Fähnrich, Heinz (2007). Kartwelisches etymologisches Wörterbuch. Leiden: Brill.

McGovern, P., Jalabadze, M., Batiuk, S., Callahan, M. P., Smith, K. E., Hall, G. R., Kvavadze, E., Maghradze, D., Rusishvili, N., Bouby, L., Failla, O., Cola, G., Mariani, L., Boaretto, E., Bacilieri, R., This, P., Wales, N., & Lordkipanidze, D. (2017). Early Neolithic wine of Georgia in the South Caucasus. Proceedings of the National Academy of Sciences of the United States of America, 114(48), E10309–E10318.

Everyone likes to talk about the origins of words. For linguists, this is a real boon since it gives us the opportunity to endlessly rant about work. In many ways, talking about the etymologies of words (perhaps specifically in your own language) is similar to taking an ancestry test. It can tell you something about yourself and something about all the places and societies our ancestors might have lived in. Or, like me, you might find out that your ancestors never seem to have left the province where you grew up. Likewise, people like hearing about how words are borrowed across languages and time as a result of language contact. Generally, the story goes like this:

Speakers of one language – for example (American) English – have words for all important concepts in their culture, for example American football. These speakers then encounter speakers of another language, whose culture – for example the Japanese culture – does not include this concept. Sooner or later, a need for communicating about this concept arises, but there is no word for this in Japanese. At this point, it is either possible to invent a new Japanese word to refer to American football, or to simply borrow the word from English and adapt it to Japanese pronunciation. Thus, you end up with amefuto (from Ame(rican) foot(ball)). This process is, and has probably also always been, a very common occurrence around the world. A more intricate example is the word chameleon which comes from Greek khamailéon. This Greek word is constructed by the joining khamaí (‘on the earth, on the ground’) and léōn (‘lion’). This, in turn, comes from a literal translation of Akkadian (an extinct language spoken in ancient Mesopotamia) nēšu ša qaqqari ‘chameleon, reptile’ that basically means ‘lion of the ground’ or ‘predator that crawls upon the ground’.

But what seems to never fail to spark people’s interest are the kind-of-but-not-really-established facts in language which brings us to the wonderful and somewhat speculative world of language substrates. A language substrate is a language that seems to have influenced another language that was somehow more dominant at the time of contact. In the case of amefuto, we know which two languages were involved, English and Japanese, but when we look at languages several thousand years ago, the situation is usually less transparent and only subtle traces of ancient contact between peoples can be detected.

Most languages spoken in Europe belong to the same language family, the Indo-European language family. It is thought that speakers of a common Proto-Indo-European language migrated from the steppes north of the Black Sea into Europe during the Late Neolithic to Early Bronze Age. When the groups split up geographically, different language family branches emerged, including Albanian, Armenian, Balto-Slavic (such as Polish and Russian), Celtic (such as Irish and Welsh), Germanic (such as English and German), Greek, Indo-Iranian (such as Farsi and Hindi) and Italic (such as Italian and Spanish), as well as the now dead branches, Anatolian och Tocharian.

However, at this point in time, Europe was not uninhabited. These Pre-Indo-European peoples living here probably spoke many different languages, of which the only surviving descendant today is Basque. These languages potentially belonged to several different language families and already had names for all the things around them. Since the speakers of the common Proto-Indo-European language and their immediate descendants moved westward from the steppes, we can assume that they lacked words for many things in this new environment and therefore borrowed from neighboring languages.

In the Celtic languages, there are a number of words that seem to have come from non-Indo-European language speakers that inhabited the British Isles and Western Europe, including a couple of words that are conspicuously similar to Basque words. For example, the Old Irish word adarc ‘horn’ closely resembles Basque adar ‘horn’, while words with the same meaning in other Indo-European languages are vastly different (English horn, Latin cornu, Ancient Greek kéras). This suggests that an ancestor or relative of Basque was in close contact with some of the first speakers of Celtic languages that reached the Atlantic coast. This Basque-like language could then be considered a substrate embedded in some Celtic languages.

Some linguists also hypothesize that entire sections of the Germanic vocabulary (primarily related to seafaring, warfare, animals, etc.) are borrowed from an unknown language or several unknown languages, since these words do not seem to fit the corresponding words in other Indo-European languages. Some of these words include sea, ship, strand, ebb, sword, shield, helmet, bow, carp, eel, calf and lamb, and the differences between Germanic languages and other Indo-European branches become apparent if we compare, for example, English lamb and German Lamm to Latin agnus and Russian yagnonok. Similarly, the word for the animal ‘seal’ could have been borrowed from Proto-Finnic, the common ancestor of Finnish and Estonian, *šülkeš ‘seal’, or it was borrowed into both the Germanic and Finnic languages from an unknown third source. Other linguists dismiss many of the suggested substrate words as they could be derived directly from common Indo-European vocabulary. For example, the word strand could have come from Proto-Indo-European *ster-, meaning ‘wide, flat’, and helmet could be derived from Proto-Indo-European *ḱel- ‘to hide, conceal’.

Regardless of the exact number of confirmed substrate words, this mix between attested words from some languages and potential traces of other languages give us a unique glimpse into prehistoric societies, but also leaves enough room for our imagination to try to speculatively fill in some of the knowledge gaps.

References

Kroonen, G. (2013). Etymological dictionary of proto-Germanic. Brill.

Matasović, R. (2012). The substratum in Insular Celtic. Journal of Language Relationship, 8(1), 153-160.

Bouckaert, R., Lemey, P., Dunn, M., Greenhill, S. J., Alekseyenko, A. V., Drummond, A. J., Gray, R. D., Suchard, M. A. & Atkinson, Q. D. (2012). Mapping the origins and expansion of the Indo-European language family. Science, 337(6097), 957-960.

A common morphological process is to add an affix to a lexical root to change its word class. This morphological process is known as derivation. A few English examples to illustrate this process are how the noun energy can be turned into the verb energize, and how the verb to play can be turned into the noun player using another affix. This process is used extensively within the Greenlandic language (and other language in the Eskaleut language family) to create some really impressive words. An example of this is aliikkuserisuillammaassuaanerartassagaluarpaat, meaning something along the lines of ‘However, they will say that he is a great entertainer, but…’ Impressive indeed, but how do they work?

Greenlandic words typically begin with the lexical root, (alikku– ‘entertainment’ in the example above), followed by a set of optional derivation affixes (all of them marked in bold in the example above), and ending in an inflectional affix (a story for another time). A unique part of Greenlandic grammar is the ability to change the word class several times within a single word!

Taking a closer look at the word above, we can see how this process works. In the following, the different affixes are separate by a hyphen, and the superscript is used to show which part of speech the affix is used to derive: aliikku^N-sersu^V-i^V-llamma^N-ssua^N-a^V-nerar^V-ta^V-ssa^V-galuar^V-paat^V. The word begins a noun, then turns into a verb, then turns into a noun again and then finally turning it into a verb. These affixes can have relatively specific meanings, with –sersu^V-meaning ‘to provide’ and -llamma^N– meaning ‘one good at’. It is the last derivational affix that determines whether the word will function as a noun or as a verb in the sentence, not the word class of the lexical root. This phenomenon, also called ‘Ping-Ping recategorization’, is one of the reasons why West Greenlandic can build such long words!

The attentive reader will have noticed that not all affixes appear to change the word from noun to verb or verb to noun. Some of them simply keep the same class, as -ssua^N– and –ssa^V–. Rather than changing the word class, they act as modifiers (in a very board sense), adding more information to the word. For instance, -ssua^N– shows that the noun it modifies is big, and -ssa^V– shows that the event will take place in the future.

However, it is important to keep in mind that the language does not have to use such long words. An English sentence like ‘Peter read the newspaper’ will likewise consist of a set of short words in Greenlandic: Piitap aviisi atuarpaa. We should always be aware of and try to avoid exoticizing languages that happen to have different structure than what we find in familiar European languages!

Greenlandic has approximately 500 of these types of affixes, constituting the source of the language’s impressive ability to construct long and complex words. For readers interested in learning more, see Fortescue (2017) for a general overview of the language family, and Dorais (2010) for a book-length introduction to the language family, with a special focus on the Inuit group.

Dorais, L. J. (2010). Language of the Inuit: syntax, semantics, and society in the Arctic (Vol. 58). McGill-Queen’s Press-MQUP.

Fortescue, M. (2017). The Eskimo-Aleut Language Family. In A. Aikhenvald & R. Dixon (Eds.), The Cambridge Handbook of Linguistic Typology (Cambridge Handbooks in Language and Linguistics, pp. 683-706).