adpositions.Rmd

---
title: "Adpositions"
author: "Polina Nasledskova, Tatiana Philippova"
date: 'Last update: June 2022'
bibliography: "data/orig_bib/adpositions.bib"
link-citations: true
csl: apa.csl
output:
  html_document:
    number_sections: true
    anchor_sections: true
    pandoc_args: --shift-heading-level-by=-1
---

```{r, include = FALSE}
knitr::opts_chunk$set(echo = FALSE, message = FALSE, warning = FALSE, comment = '')
library(lingglosses)
```

```{r}
library(RefManageR)
BibOptions(check.entries = FALSE, style = 'text', bib.style = 'authoryear')
article_citation <- BibEntry(bibtype = 'Incollection', 
 key='nasledskovaphilippova2022',
 title='Adpositions',
 author='Nasledskova, Polina and Philippova, Tatiana',
 year='2022',
 editor= 'Daniel, Michael and Filatov, Konstantin and Moroz, George and Mukhin, Timofey and Naccarato, Chiara and Verhees, Samira',
 publisher='Linguistic Convergence Laboratory, NRU HSE',
 address='Moscow',
 booktitle= 'Typological Atlas of the Languages of Daghestan (TALD)',
 url='http://lingconlab.ru/dagatlas')
```

## {.tabset .tabset-fade .tabset-pills -} 

### Plain text {-}
```{r, results = 'asis'}
print(article_citation, .opts = list(style = 'text'))
```

### BibTeX {-}

```{r}
print(article_citation, .opts = list(style = 'Bibtex'))
```


## Introduction

This chapter overviews the basic morphological and syntactic properties of East Caucasian adpositions as compared to those of adpositions found in the neighboring languages (Armenian, Tat, Georgian, Azerbaijani, Kumyk and Nogai).
Generally, we relied on grammatical descriptions in deciding what lexical items fall into the category of adpositions, occasionally making very minor adjustments in case we found additional lexemes labeled as adpositions in the dictionaries or in case a certain lexeme did not fit our working definition of an adposition that we adopt from [@hagege2010]: 

“Adpositions may be defined as grammatical tools which mark the relationship between two parts of a sentence: one is the element which an adposition governs. It is traditionally called its complement and is mostly represented by a noun or noun-like word or phrase[...] The other part is an entity which either functions as the predicate of this sentence, or is a non-predicative noun” [@hagege2010: 1].

In terms of their position relative to the dependent noun phrase, East Caucasian languages are uniformly postpositional, similarly to their Turkic neighbors and Georgian. In contrast, (Judeo)-Tat is predominantly a prepositional language, although it features a certain number of postpositions, some of which are borrowings from Azerbaijani [@authier2012: 101]. Armenian has both postpositions and prepositions, the former being more numerous and also more frequent in the East Armenian National Corpus (http://www.eanc.net/)[^1]. In addition, according to [@dum-tragut2009], it features a small set of ambipositions—adpositional elements that can be placed either before or after the dependent noun. The case form of the nominal dependent required by an ambiposition in Armenian may remain stable or change depending on whether the lexeme is used prepositionally or postpositionally. The first scenario is illustrated in (@one) for the ambiposition *šnorhiv* ‘thanks to’ that always requires genitive (syncretic with dative in the nominal, but not in the pronominal declension); the latter is represented by the ambiposition *handep* ‘for’ in (@two) which requires genitive in postposition and nominative in preposition.

(@one) Armenian [@dum-tragut2009: 379, 493]

    a)
```{r, results='asis'}
gloss_example(transliteration = "**Im** **ěnker-oǰ** **šnorhiv** darj-i ē gal-is ew ěndun-um k’ristoneakan havatk’-ě",
              glosses = "my friend-GEN thanks conversion-DAT he_is come-PTCP.PRS.3SG CONJ adopt-PTCP.PRS Christian belief.NOM-the",
              free_translation = "Thanks to my friend, he converts and adopts the Christian belief.")
```

    b)
```{r, results='asis'}
gloss_example(transliteration = "Xač’atryan-i gorc-er-i-n canot’ac’-el ē **šnorhiv** **ir** **usuc’č’-i** Nat’an Milšteyn-i um mot usan-el ē 10 tari",
              glosses = "Xachatryan-DAT work-PL-DAT-the get_acquainted-PTCP.PF he_is thanks his teacher-GEN Nathan Milstein-GEN REL.DAT at learn-PTCP.PF he_is ten year.NOM",
              free_translation = "He became acquainted with the works of Khachatryan thanks to his teacher, Nathan Milstein, with whom he has studied for 10 years.")
```

(@two) Armenian [@dum-tragut2009: 301]

    a)
```{r}
gloss_example(transliteration = "**k’o** **handep** aysteɬ mec hargank’ ka",
              glosses = "you.**GEN** for here big esteem.NOM exist.PRS.3SG",
              free_translation = "There is big esteem for you here.")
```

    b)
```{r}
gloss_example(transliteration = "**Handep** **hayrenik’-ě** nra ser-n anč’ap’eli ē",
              glosses = "for fatherland.**NOM**-the his love.NOM-the immeasurable it_is",
              free_translation = "His love for his fatherland is immeasurable.")
```

## Morphological properties

### Inflection for directionality 

Although cross-linguistically adpositions tend to have a fixed form, East Caucasian postpositions may bear morphemes that could be considered inflectional: they may attach directionality and agreement markers that we consider in turn.

Directionality markers signal the direction or absence of motion of the trajector (figure) with respect to the landmark (ground). They attach to lexemes of different parts of speech that have a spatial meaning: nominals in locative forms, local place names, body part locations, locative converbs, and spatial adverbs and postpositions. The number of values in the directionality category varies from language to language; we find essive (absence of motion), lative (motion to the landmark), elative (motion from the landmark), translative (motion through a landmark), directive (motion towards a landmark), and terminative (motion up to a landmark) forms. [Table 1](#t1) illustrates the inventories of directional forms of postpositions with the meaning ‘under’ in a sample of languages from different branches of East Caucasian; one can see that the repertoire ranges from 2 to 5 forms in this case.

#### **Table 1.** Directional forms of postpositions synonymous to ‘under’ across the branches of East Caucasian {#t1 -}

| Language | Branch | Directional forms | Source |
|:-----------|:----------|:----------|:-----------|
| Hinuq  | Tsezic   | *geɬ* (`r add_gloss("ESS")`), *geɬes* (`r add_gloss("ABL1")`), *geɬzo* (`r add_gloss("ABL2")`), *geɬer* (`r add_gloss("LAT")`), *geɬedo* (`r add_gloss("DIR")`)   | @forker2013: 379 |
| Mehweb | Dargic   | *ʔu* (`r add_gloss("LAT")`), *ʔub* (`r add_gloss("ESS")`), *ʔubadal* (`r add_gloss("EL")`)   | @magometov1982: 126 |
| Bagvalal | Andic | *hiƛ’i* (`r add_gloss("ESS")`), *hiƛ’isː* (`r add_gloss("EL")`), *hiƛ’isːini* (`r add_gloss("TRLAT")`) | @sosenskaja2001pp: 170 |
| Lezgian | Lezgic | *k’anik* (`r add_gloss("ESS")`), *k’anikaj* (`r add_gloss("EL")`) | @haspelmath1993: 219220 |
| Chechen | Nakh | *k’el*  (`r add_gloss("ESS")`/`r add_gloss("LAT")`), *k’eləħ* (`r add_gloss("ESS")`), *k’elərə* (`r add_gloss("EL")`), *k’elxuʊlə* (`r add_gloss("TRLAT")`),  *k’elħaː* (`r add_gloss("DIR")`) | @aliroev1999: 7980 |

The examples in (@three) below illustrate the use of different forms in sentences.

(@three) Hinuq (Tsezic) [@forker2013: 384]

    a)
```{r}
gloss_example(transliteration = "de č’agu goɬ q’ede, **de-de** **purho**=no nox-no",
              glosses = "{I} alive be IRR {I}.OBL-ALOC next=and come-UW.PST",
              free_translation = "If I were alive and you came next to me…")
```

    b)
```{r}
gloss_example(transliteration = "meži **di-de** **purho-s** b-uɬi-yom",
              glosses = "you.PL {I}.OBL-ALOC next-ABL1 HPL-go_out-PROH",
              free_translation = "Do not go away from me.")
```

In the examples above, the form of the postposition differs depending on the type of motion: towards the landmark, (@three a), or from it, as in (@three b).

Outside East Caucasian, something similar to inflection for directionality is found in Modern Eastern Armenian. For instance, the postposition *tak* ‘under’ has an ablative-marked (*takic’*) and an instrumental-marked form (*takov*), which are used in (but not restricted to) elative, (@four a), and translative contexts, respectively. However, the semantic contribution of these case suffixes when attached to adpositions is not as clear-cut as in East Caucasian: for instance, instrumental may also be used if motion does not have a clear direction or its direction is not important, cf. (@four b) [@dum-tragut2009: 304–305].

(@four) Armenian (Indo-European) [@dum-tragut2009: 306]

    a)
```{r}
gloss_example(transliteration = "erexa-n **seʁan-i** **tak-ic’** motec’-av morě",
              glosses = "child.NOM-the table-DAT under-ABL approach-AOR.3SG mother.DAT.the",
              free_translation = "The child approached the mother from under the table.",
              comment = "(the starting point of the action is under the table, in the direction of the mother.)")
```

    b)
```{r}
gloss_example(transliteration = "**getn-i** **tak-ov** mi mec get eː hos-um",
              glosses = "ground-DAT under-INS INDEF big river.NOM it_is flow-PTCP.PRS",
              free_translation = "Under the ground a big river flows.",
              comment = "(there is no starting point and no direction given.)")
```
	
### Agreement morphology  

Turning now to agreement morphemes on postpositions, we note that it is attested in languages of all branches of East Caucasian, except for Nakh and Khinalug. However, it is never the case that all postpositions in a language display agreement: in most cases, less than a half of them do so. Agreement slots may be prefixal, suffixal or (rarely) infixal, as shown in (@five), (@six) and (@seven), respectively. 

(@five) Godoberi (Andic) [@kibriketal1996: 88]
```{r}
gloss_example(transliteration = "waša hamaχi-č’u **w-oχut’u** wu-na",
              glosses = "boy donkey(N)-CONT M-after M-go.PST",
              free_translation = "The boy went after the donkey.")
```

(@six) Avar [@rudnev2020: 833]
```{r}
gloss_example(transliteration = "školal-da ask’o-**w** jasaɬ was w-uχana",
              glosses = "school(N).OBL-LOC near-M girl(F).ERG boy(M).ABS M-beat.PST",
              free_translation = "The girl beat the boy up near the school.")
```

(@seven)   Archi (Lezgic) (Marina Chumakina, p.c.)
```{r, results='asis'}
gloss_example(transliteration = "zari gǝzet o‹**b**›kɬni arχul-ma-k e‹b›q’en",
              glosses = "{I}.ERG newspaper.SG.ABS(III) read<III.SG>.PF the_middle-IN-LAT up_to<III.SG>",
              free_translation = "I read the newspaper up to the middle.")
```

In the examples above the postpositions bear a gender/number affix; one can see that its gender/number value corresponds to that of the absolutive argument. It is generally the case that agreeing postpositions in East Caucasian agree with the absolutive argument. However, agreement with their own complement is also marginally attested, at least for one Mehweb Dargwa postposition (@eight):

(@eight) Mehweb Dargwa (Michael Daniel, p.c.)
```{r}
gloss_example(transliteration = "heš šara=gʷa **ʁir-me-la** **d-ajcana-b**",
              glosses = "this lake(N)=PTC rock(N)-PL-GEN NPL-amidst-N",
              free_translation = "This lake is amidst rocks.")
```

A somewhat similar phenomenon is possessive markers on postpositions that are found in Turkic languages and colloquial Armenian.
In Turkic, some postpositions have a possessive marker indicating the person of its dependent:

(@nine) Azerbaijani [@shiralievsevortyan1971: 301]
```{r}
gloss_example(transliteration = "**qovluǧ-un** **ara-sı-n-dan** bir vərəqə çıx-ar-ıb",
              glosses = "folder-GEN interval-3.POSS-EP-ABL one sheet go_out-CAUS-ANT",
              free_translation = "having put out one sheet from the folder…")
```

In (@nine), the possessive marker on the postposition *arasından* corresponds to the person of its dependent (‘folder’). Note, however, that experts typically do not consider such lexemes as adpositions, attributing them to nouns (e.g., @shiralievsevortyan1971: 165); our dataset, therefore, does not include them either. Adpositions that agree with the absolutive (or nominative, for that matter) are not attested in Turkic languages.
In colloquial Armenian, the dependents of 1st and 2nd person can be expressed by possessive enclitics when used with some spatial postpositions, cf. (@ten a) and (@ten b).

(@ten) Armenian [@dum-tragut2009: 300]

    a)
```{r}
gloss_example(transliteration = "**inj** **het** mišt hayeren eː xos-um",
              glosses = "{I}.DAT with always Armenian he_is speak-PTCP.PRS",
              free_translation = "He always speaks Armenian with me.",
              comment = "(standard Armenian)")
```

    b)
```{r}
gloss_example(transliteration = "**het-s** mišt hayeren eː xos-um",
              glosses = "with-my always Armenian he_is speak-PTCP.PRS",
              free_translation = "He always speaks Armenian with me.",
              comment = "(colloquial Armenian)")
```

However, it is problematic to postulate agreement in these examples, as the free pronoun in the dative or genitive case and the enclitic cannot co-occur. There is also no agreement with the absolutive (nominative) argument. 
	
For a more detailed discussion of agreeing adpositions, see [Agreement of adpositions with the absolutive/nominative](http://lingconlab.ru/dagatlas/031_agreement_of_adpositions.html).

## Syntactic properties

### Adnominal vs. independent use
A feature that is typical of East Caucasian postpositions is that the vast majority of them can be used both adnominally and independently, i.e. with a complement or without it, which is unusual from a cross-linguistic perspective.

(@eleven) Ingush (Nakh) [@nichols2011: 405]

    a)
```{r}
gloss_example(transliteration = "**t'exkaa** **chuhwanahw** t'um joall",
              glosses = "bone.DAT inside marrow II.be_contained.PRS",
              free_translation = "There is marrow inside a bone/Inside of a bone is marrow.")
```

    b)
```{r}
gloss_example(transliteration = "mashen **chuhwanahw** jy",
              glosses = "car inside II.be.PRS",
              free_translation = "The car is inside.")
```

In (@eleven a), the Ingush postposition *chuhwanahw* ‘inside’ is used with a dependent noun (in the dative case), while in (@eleven b) it is used without any dependent and the sentence is still felicitous.
The claim that such postpositions are more typical of East Caucasian languages than the strictly adnominal ones is supported by the following facts:

* There are East Caucasian languages that lack strictly adnominal postpositions, but there are no East Caucasian languages where all postpositions can be used only adnominally;
* Strictly adnominal postpositions often turn out to be borrowed lexemes.

Independent uses of adpositions are also mentioned in the grammars of genealogically unrelated neighboring languages, see, for instance, (@twelve) for the adnominal and independent uses of the Armenian postposition *durs* ‘out, outside’.

(@twelve) Armenian

    a)
```{r, results='asis'}
gloss_example(transliteration = "p’ap’ul pap-n as-um eː or girk’-e erbek’ **tn-ic’** **durs** č’-i han-el",
              glosses = "P’ap’ul grandpa.NOM-the say-PTCP.PRS he_is CONJ book.NOM-the never house-ABL out NEG-he_is take-PTCP.PF",
              free_translation = "P’ap’ul’s grandpa says that he has never taken this book out of the house.")
```
[@dum-tragut2009: 531]

    b)
```{r, results='asis'}
gloss_example(transliteration = "ov č’-i irakana-c’n-i petut’y-an cragir-e na **durs** kmn-a ayd gorčent’ac’-ic’",
              glosses = "who NEG-he_is fulfill-CAUS-COND.FUT.3SG state-DAT program.NOM-the he.NOM outside stay-COND.FUT.3SG that process-ABL",
              free_translation = "Who does not fulfill the state’s program, will remain outside of that process.")
```
[@dum-tragut2009: 318]

However, in these languages they do not seem to be as pervasive as in East Caucasian. More details on the proportion of adpositions that allow independent (adverbial) use can be found in a dedicated chapter and map (in preparation).

### Case assignment
The set of cases that are assigned by postpositions even within one East Caucasian language is pretty large and may include both syntactic (abstract) cases and spatial cases[^2]. This may be illustrated by the case assignment properties of semantically similar postpositions from languages of different branches of East Caucasian.

(@thirteen) Chechen (Nakh) [@jakovlev1940: 143]
```{r}
gloss_example(transliteration = "**gɔʊr-a-na** / **gɔʊr** **t’e** ʡyrre so xiːrə",
              glosses = "horse-OBL-DAT {/} horse.ABS on in_the_morning I sit",
              free_translation = "I sat on a horse in the morning.")
```

(@fourteen) Bagvalal (Andic) [@sosenskaja2001pp: 406]
```{r}
gloss_example(transliteration = "hinc’-ib-a-ɬa č’ihi",
              glosses = "stone-PL-OBL.PL-SUP on",
              free_translation = "on stones")
```

(@fifteen) Tsakhur (Lezgic) [@sosenskaja1999post: 127]
```{r, results='asis'}
gloss_example(transliteration = "sa jiʁ-ɨ-lʲ ǯamaʔaIt-ɨ-k’le **χāI-ni** **aq’ʷalʲ** c’e=t=ta wertalʲot ɢaǯe",
              glosses = "one day-OBL-SUP people-OBL-AFF sky-AOBL on first=IV=ADV.IV helicopter.IV IV.see.PF",
              free_translation = "Once the people saw a helicopter in the sky for the first time.")
```

There is no direct relationship between the postposition’s meaning (spatial or not) and the type of case it assigns (spatial or not): there are abstract postpositions that assign spatial cases and spatial postpositions requiring an abstract case, as can be seen from the examples below.

(@sixteen) Hinuq [@forker2013: 390]
```{r}
gloss_example(transliteration = "xabar-ƛ’o req’un",
              glosses = "story-SUP accordingly",
              free_translation = "in accordance with the story")
```

(@seventeen) Lezgian [@haspelmath1993: 221]
```{r}
gloss_example(transliteration = "čarx-ari-n arada",
              glosses = "rock-PL-GEN between",
              free_translation = "between the rocks")
```

Some postpositions assign several cases which may or may not bring about interpretational differences (@eighteen a, b). 

(@eighteen)	Sanzhi Dargwa [@forker2020dargwa: 156]

    a)
```{r}
gloss_example(transliteration = "nuša šːi-la tːura ag-ur-da",
              glosses = "we village-GEN outside go.PFV-PRET-I",
              free_translation = "We went out of the village.")
```

    b)
```{r}
gloss_example(transliteration = "Rasul-la / Rasul-le-rka tːura cara=ra sa-č-ib-da",
              glosses = "Rasul-GEN {/} Rasul-LOC-EL outside other=ADD hither-lead.PFV-PRET-I",
              free_translation = "Apart from Rasul I also brought another one.")
```

In example (@eighteen), the postposition *tːura* ‘outside’ in its spatial sense is only compatible with a genitive dependent. However, in its abstract sense ‘apart from, except for’, the postposition has different properties: it can assign both genitive and locelative case to its dependent; the two options are not perceivably different in terms of meaning.

Note that several adposition-like elements (typically with the meaning similar to ‘except’, ‘like’, ‘than’) in a number of languages in our sample require a nominal dependent, but the form of the dependent is determined by that of its contextual correlate (@nineteen).

(@nineteen)	Tsez [@polinskydraft2015: 177]

    a)
```{r, results='asis'}
gloss_example(transliteration = "eli mariyat-**qo-r** ƛexuzaƛ’or c’alduqan-za-qo-r xabaryay-nč’i",
              glosses = "we Mariyat-**POSS-LAT** except student-OBL.PL-POSS-LAT talk-PST.W.NEG",
              free_translation = "We did not talk to the students except Mariyat.")
```

    b)
```{r}
gloss_example(transliteration = "eli ɣudes ħalt’-o-ɬ-xor b-ik’i-x ħat’an-**ƛ’o** ƛexuzaƛ’or",
              glosses = "we daily work-OBL-CONT-LAT I.PL-go-PRS Sunday-**SUP.ESS** except",
              free_translation = "We go to work every day except Sunday.")
```

In example (@nineteen), the lexeme *ƛexuzaƛ’or* ‘except’ is used with a noun in the possessive-lative form in (a) and in the superessive form in (b). In (@nineteen a), the form of the noun Mariyat is the same as of its correlate *c’alduqanzaqor* ‘students’. In (@nineteen b), the case of noun *ħat’anƛ’o* ‘Sunday’ is superessive, which is used to mark temporal adjuncts in any context (with or without ‘except’). We treat lexemes like these as not assigning any case.

Also, in Hunzib several (related) postpositions admit both comitative and dative (coinciding with inessive) dependents; the former combination brings about the meaning of motion along a landmark and the latter—through a landmark, as illustrated in (@twenty).

(@twenty) Hunzib [@vandenberg1995: 166]

    a)
```{r}
gloss_example(transliteration = "xõx-ǧur rišo/gišo/tišo",
              glosses = "tree-COM upwards/downwards/horizontal_direction",
              free_translation = "(upwards/downwards) along the tree")
```

    b)
```{r}
gloss_example(transliteration = "xõx-i-i rišo/gišo/tišo",
              glosses = "tree-OBL-DAT upwards/downwards/horizontal_direction",
              free_translation = "(upwards/downwards) through (inside) the tree")
```

In addition, different case forms might be required for nominal vs. pronominal dependents.

(@twentyone) Ingush [@nichols2011: 402]

    a)
```{r}
gloss_example(transliteration = "ħa **c’en** **t’eħa** berkat=ʔa xurgdaːc",
              glosses = "you.SG.GEN house.GEN after abundance=and be.FUT.IV.NEG",
              free_translation = "There will be no abundance in your house.")
```

    b)
```{r}
gloss_example(transliteration = "so čy-iːqːaːča legjolaž jar yz **suːona** **t’eħa**",
              glosses = "{I} in-jump.PTCP.OBL die.FUT.II.SIM II.be.PST she {I}.DAT after",
              free_translation = "She was eager (‘dying’) to go anywhere I went (‘jumped, rushed’).")
```

As illustrated in (@twentyone), the same postposition *t’eħa* ‘after’ requires the genitive on its dependent nouns and the dative on its dependent pronouns. A similar, though more systematic, pattern is observed in Mehweb: while postpositions can govern various forms of nouns (dative, genitive, comitative, absolutive), all pronominal dependents of postpositions take the dative case form [@magometov1982: 127]. 
Outside East Caucasian, this distinction is observed in Armenian: an adposition may require nominative case on nominal dependents and dative case on pronominal ones (for instance, the prepositions *depi* ‘towards’ and *minč'ew* ‘until’ are like this). In addition, a very significant fraction of adpositions require dative-marked nominals and genitive pronominals. However, Modern Eastern Armenian simply does not distinguish between dative and genitive case in nominal forms—this distinction only persists in the pronominal paradigm [@dum-tragut2009: 125], so for the purposes of this study we consider these adpositions to be genitive case assigners. Unambiguously dative adpositional dependents are restricted to 1st and 2nd person pronouns in non-colloquial registers and as such may be treated as exceptions. 

The final point that we would like to make is that many East Caucasian languages have a dominant postpositional case. By dominant case we mean the case that is assigned to postpositional dependents significantly more often than other cases governed by postpositions in a language. For example, Botlikh (<Andic) postpositions govern contessive (8 postpositions) and dative (2 postpositions). We conclude that the dominant postpositional case in Botlikh is contessive.
A dominant case may be instantiated by an abstract case (such as genitive or dative) or by a spatial case (e.g., contessive). According to the data we have collected, we can assume some case to be the dominant one in most East Caucasian languages, with some exceptions. Several neighboring languages outside the family also have dominant adpositional cases: genitive in Georgian and Armenian, and unmarked (nominative) in Azerbaijani and (Judeo-)Tat. Dominant adpositional case is discussed in greater detail in a [dedicated chapter](http://lingconlab.ru/dagatlas/030_dominant_adpositional_case.html). 

[^1]: A simple search for lexemes tagged as prepositions and postpositions yields 605,776 and 2,862,460 results, respectively.
[^2]: It is arguable whether spatial forms of nouns in East Caucasian can be called case forms, but for brevity we will use the term spatial cases to refer to them.


## List of glosses {-}

```{r}
make_gloss_list(readxl::read_xlsx("data/glosses.xlsx"))
```

## References {-}